1
0
Fork 0
remarkable-linux/drivers/gpu/drm/i915/i915_drv.h

4340 lines
130 KiB
C
Raw Normal View History

/* i915_drv.h -- Private header for the I915 driver -*- linux-c -*-
*/
/*
*
* Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
* All Rights Reserved.
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the
* "Software"), to deal in the Software without restriction, including
* without limitation the rights to use, copy, modify, merge, publish,
* distribute, sub license, and/or sell copies of the Software, and to
* permit persons to whom the Software is furnished to do so, subject to
* the following conditions:
*
* The above copyright notice and this permission notice (including the
* next paragraph) shall be included in all copies or substantial portions
* of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
* OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
* IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
* ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
* TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
*
*/
#ifndef _I915_DRV_H_
#define _I915_DRV_H_
#include <uapi/drm/i915_drm.h>
#include <uapi/drm/drm_fourcc.h>
#include <linux/io-mapping.h>
#include <linux/i2c.h>
#include <linux/i2c-algo-bit.h>
#include <linux/backlight.h>
#include <linux/hash.h>
#include <linux/intel-iommu.h>
#include <linux/kref.h>
drm/i915: irq-drive the dp aux communication At least on the platforms that have a dp aux irq and also have it enabled - vlvhsw should have one, too. But I don't have a machine to test this on. Judging from docs there's no dp aux interrupt for gm45. Also, I only have an ivb cpu edp machine, so the dp aux A code for snb/ilk is untested. For dpcd probing when nothing is connected it slashes about 5ms of cpu time (cpu time is now negligible), which agrees with 3 * 5 400 usec timeouts. A previous version of this patch increases the time required to go through the dp_detect cycle (which includes reading the edid) from around 33 ms to around 40 ms. Experiments indicated that this is purely due to the irq latency - the hw doesn't allow us to queue up dp aux transactions and hence irq latency directly affects throughput. gmbus is much better, there we have a 8 byte buffer, and we get the irq once another 4 bytes can be queued up. But by using the pm_qos interface to request the lowest possible cpu wake-up latency this slowdown completely disappeared. Since all our output detection logic is single-threaded with the mode_config mutex right now anyway, I've decide not ot play fancy and to just reuse the gmbus wait queue. But this would definitely prep the way to run dp detection on different ports in parallel v2: Add a timeout for dp aux transfers when using interrupts - the hw _does_ prevent this with the hw-based 400 usec timeout, but if the irq somehow doesn't arrive we're screwed. Lesson learned while developing this ;-) v3: While at it also convert the busy-loop to wait_for_atomic, so that we don't run the risk of an infinite loop any more. v4: Ensure we have the smallest possible irq latency by using the pm_qos interface. v5: Add a comment to the code to explain why we frob pm_qos. Suggested by Chris Wilson. v6: Disable dp irq for vlv, that's easier than trying to get at docs and hw. v7: Squash in a fix for Haswell that Paulo Zanoni tracked down - the dp aux registers aren't at a fixed offset any more, but can be on the PCH while the DP port is on the cpu die. Reviewed-by: Imre Deak <imre.deak@intel.com> (v6) Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2012-12-01 05:53:48 -07:00
#include <linux/pm_qos.h>
drm/i915: Move GEM activity tracking into a common struct reservation_object In preparation to support many distinct timelines, we need to expand the activity tracking on the GEM object to handle more than just a request per engine. We already use the struct reservation_object on the dma-buf to handle many fence contexts, so integrating that into the GEM object itself is the preferred solution. (For example, we can now share the same reservation_object between every consumer/producer using this buffer and skip the manual import/export via dma-buf.) v2: Reimplement busy-ioctl (by walking the reservation object), postpone the ABI change for another day. Similarly use the reservation object to find the last_write request (if active and from i915) for choosing display CS flips. Caveats: * busy-ioctl: busy-ioctl only reports on the native fences, it will not warn of stalls (in set-domain-ioctl, pread/pwrite etc) if the object is being rendered to by external fences. It also will not report the same busy state as wait-ioctl (or polling on the dma-buf) in the same circumstances. On the plus side, it does retain reporting of which *i915* engines are engaged with this object. * non-blocking atomic modesets take a step backwards as the wait for render completion blocks the ioctl. This is fixed in a subsequent patch to use a fence instead for awaiting on the rendering, see "drm/i915: Restore nonblocking awaits for modesetting" * dynamic array manipulation for shared-fences in reservation is slower than the previous lockless static assignment (e.g. gem_exec_lut_handle runtime on ivb goes from 42s to 66s), mainly due to atomic operations (maintaining the fence refcounts). * loss of object-level retirement callbacks, emulated by VMA retirement tracking. * minor loss of object-level last activity information from debugfs, could be replaced with per-vma information if desired Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/20161028125858.23563-21-chris@chris-wilson.co.uk
2016-10-28 06:58:44 -06:00
#include <linux/reservation.h>
#include <linux/shmem_fs.h>
#include <drm/drmP.h>
#include <drm/intel-gtt.h>
#include <drm/drm_legacy.h> /* for struct drm_dma_handle */
#include <drm/drm_gem.h>
#include <drm/drm_auth.h>
#include <drm/drm_cache.h>
#include "i915_params.h"
#include "i915_reg.h"
#include "i915_utils.h"
#include "intel_uncore.h"
#include "intel_bios.h"
#include "intel_dpll_mgr.h"
#include "intel_uc.h"
#include "intel_lrc.h"
#include "intel_ringbuffer.h"
#include "i915_gem.h"
#include "i915_gem_context.h"
#include "i915_gem_fence_reg.h"
#include "i915_gem_object.h"
#include "i915_gem_gtt.h"
#include "i915_gem_render_state.h"
#include "i915_gem_request.h"
#include "i915_gem_timeline.h"
#include "i915_vma.h"
drm/i915: gvt: Introduce the basic architecture of GVT-g This patch introduces the very basic framework of GVT-g device model, includes basic prototypes, definitions, initialization. v12: - Call intel_gvt_init() in driver early initialization stage. (Chris) v8: - Remove the GVT idr and mutex in intel_gvt_host. (Joonas) v7: - Refine the URL link in Kconfig. (Joonas) - Refine the introduction of GVT-g host support in Kconfig. (Joonas) - Remove the macro GVT_ALIGN(), use round_down() instead. (Joonas) - Make "struct intel_gvt" a data member in struct drm_i915_private.(Joonas) - Remove {alloc, free}_gvt_device() - Rename intel_gvt_{create, destroy}_gvt_device() - Expost intel_gvt_init_host() - Remove the dummy "struct intel_gvt" declaration in intel_gvt.h (Joonas) v6: - Refine introduction in Kconfig. (Chris) - The exposed API functions will take struct intel_gvt * instead of void *. (Chris/Tvrtko) - Remove most memebers of strct intel_gvt_device_info. Will add them in the device model patches.(Chris) - Remove gvt_info() and gvt_err() in debug.h. (Chris) - Move GVT kernel parameter into i915_params. (Chris) - Remove include/drm/i915_gvt.h, as GVT-g will be built within i915. - Remove the redundant struct i915_gvt *, as the functions in i915 will directly take struct intel_gvt *. - Add more comments for reviewer. v5: Take Tvrtko's comments: - Fix the misspelled words in Kconfig - Let functions take drm_i915_private * instead of struct drm_device * - Remove redundant prints/local varible initialization v3: Take Joonas' comments: - Change file name i915_gvt.* to intel_gvt.* - Move GVT kernel parameter into intel_gvt.c - Remove redundant debug macros - Change error handling style - Add introductions for some stub functions - Introduce drm/i915_gvt.h. Take Kevin's comments: - Move GVT-g host/guest check into intel_vgt_balloon in i915_gem_gtt.c v2: - Introduce i915_gvt.c. It's necessary to introduce the stubs between i915 driver and GVT-g host, as GVT-g components is configurable in kernel config. When disabled, the stubs here do nothing. Take Joonas' comments: - Replace boolean return value with int. - Replace customized info/warn/debug macros with DRM macros. - Document all non-static functions like i915. - Remove empty and unused functions. - Replace magic number with marcos. - Set GVT-g in kernel config to "n" by default. Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> Cc: Kevin Tian <kevin.tian@intel.com> Signed-off-by: Zhi Wang <zhi.a.wang@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1466078825-6662-5-git-send-email-zhi.a.wang@intel.com Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
2016-06-16 06:07:00 -06:00
#include "intel_gvt.h"
/* General customization:
*/
#define DRIVER_NAME "i915"
#define DRIVER_DESC "Intel Graphics"
#define DRIVER_DATE "20170818"
#define DRIVER_TIMESTAMP 1503088845
/* Use I915_STATE_WARN(x) and I915_STATE_WARN_ON() (rather than WARN() and
* WARN_ON()) for hw state sanity checks to check for unexpected conditions
* which may not necessarily be a user visible problem. This will either
* WARN() or DRM_ERROR() depending on the verbose_checks moduleparam, to
* enable distros and users to tailor their preferred amount of i915 abrt
* spam.
*/
#define I915_STATE_WARN(condition, format...) ({ \
int __ret_warn_on = !!(condition); \
if (unlikely(__ret_warn_on)) \
if (!WARN(i915.verbose_state_checks, format)) \
DRM_ERROR(format); \
unlikely(__ret_warn_on); \
})
#define I915_STATE_WARN_ON(x) \
I915_STATE_WARN((x), "%s", "WARN_ON(" __stringify(x) ")")
bool __i915_inject_load_failure(const char *func, int line);
#define i915_inject_load_failure() \
__i915_inject_load_failure(__func__, __LINE__)
typedef struct {
uint32_t val;
} uint_fixed_16_16_t;
#define FP_16_16_MAX ({ \
uint_fixed_16_16_t fp; \
fp.val = UINT_MAX; \
fp; \
})
static inline bool is_fixed16_zero(uint_fixed_16_16_t val)
{
if (val.val == 0)
return true;
return false;
}
static inline uint_fixed_16_16_t u32_to_fixed16(uint32_t val)
{
uint_fixed_16_16_t fp;
WARN_ON(val >> 16);
fp.val = val << 16;
return fp;
}
static inline uint32_t fixed16_to_u32_round_up(uint_fixed_16_16_t fp)
{
return DIV_ROUND_UP(fp.val, 1 << 16);
}
static inline uint32_t fixed16_to_u32(uint_fixed_16_16_t fp)
{
return fp.val >> 16;
}
static inline uint_fixed_16_16_t min_fixed16(uint_fixed_16_16_t min1,
uint_fixed_16_16_t min2)
{
uint_fixed_16_16_t min;
min.val = min(min1.val, min2.val);
return min;
}
static inline uint_fixed_16_16_t max_fixed16(uint_fixed_16_16_t max1,
uint_fixed_16_16_t max2)
{
uint_fixed_16_16_t max;
max.val = max(max1.val, max2.val);
return max;
}
static inline uint_fixed_16_16_t clamp_u64_to_fixed16(uint64_t val)
{
uint_fixed_16_16_t fp;
WARN_ON(val >> 32);
fp.val = clamp_t(uint32_t, val, 0, ~0);
return fp;
}
static inline uint32_t div_round_up_fixed16(uint_fixed_16_16_t val,
uint_fixed_16_16_t d)
{
return DIV_ROUND_UP(val.val, d.val);
}
static inline uint32_t mul_round_up_u32_fixed16(uint32_t val,
uint_fixed_16_16_t mul)
{
uint64_t intermediate_val;
intermediate_val = (uint64_t) val * mul.val;
intermediate_val = DIV_ROUND_UP_ULL(intermediate_val, 1 << 16);
WARN_ON(intermediate_val >> 32);
return clamp_t(uint32_t, intermediate_val, 0, ~0);
}
static inline uint_fixed_16_16_t mul_fixed16(uint_fixed_16_16_t val,
uint_fixed_16_16_t mul)
{
uint64_t intermediate_val;
intermediate_val = (uint64_t) val.val * mul.val;
intermediate_val = intermediate_val >> 16;
return clamp_u64_to_fixed16(intermediate_val);
}
static inline uint_fixed_16_16_t div_fixed16(uint32_t val, uint32_t d)
{
uint64_t interm_val;
interm_val = (uint64_t)val << 16;
interm_val = DIV_ROUND_UP_ULL(interm_val, d);
return clamp_u64_to_fixed16(interm_val);
}
static inline uint32_t div_round_up_u32_fixed16(uint32_t val,
uint_fixed_16_16_t d)
{
uint64_t interm_val;
interm_val = (uint64_t)val << 16;
interm_val = DIV_ROUND_UP_ULL(interm_val, d.val);
WARN_ON(interm_val >> 32);
return clamp_t(uint32_t, interm_val, 0, ~0);
}
static inline uint_fixed_16_16_t mul_u32_fixed16(uint32_t val,
uint_fixed_16_16_t mul)
{
uint64_t intermediate_val;
intermediate_val = (uint64_t) val * mul.val;
return clamp_u64_to_fixed16(intermediate_val);
}
static inline uint_fixed_16_16_t add_fixed16(uint_fixed_16_16_t add1,
uint_fixed_16_16_t add2)
{
uint64_t interm_sum;
interm_sum = (uint64_t) add1.val + add2.val;
return clamp_u64_to_fixed16(interm_sum);
}
static inline uint_fixed_16_16_t add_fixed16_u32(uint_fixed_16_16_t add1,
uint32_t add2)
{
uint64_t interm_sum;
uint_fixed_16_16_t interm_add2 = u32_to_fixed16(add2);
interm_sum = (uint64_t) add1.val + interm_add2.val;
return clamp_u64_to_fixed16(interm_sum);
}
static inline const char *yesno(bool v)
{
return v ? "yes" : "no";
}
static inline const char *onoff(bool v)
{
return v ? "on" : "off";
}
static inline const char *enableddisabled(bool v)
{
return v ? "enabled" : "disabled";
}
enum pipe {
INVALID_PIPE = -1,
PIPE_A = 0,
PIPE_B,
PIPE_C,
_PIPE_EDP,
I915_MAX_PIPES = _PIPE_EDP
};
#define pipe_name(p) ((p) + 'A')
drm/i915: add TRANSCODER_EDP Before Haswell we used to have the CPU pipes and the PCH transcoders. We had the same amount of pipes and transcoders, and there was a 1:1 mapping between them. After Haswell what we used to call CPU pipe was split into CPU pipe and CPU transcoder. So now we have 3 CPU pipes (A, B and C), 4 CPU transcoders (A, B, C and EDP) and 1 PCH transcoder (only used for VGA). For all the outputs except for EDP we have an 1:1 mapping on the CPU pipes and CPU transcoders, so if you're using CPU pipe A you have to use CPU transcoder A. When have an eDP output you have to use transcoder EDP and you can attach this CPU transcoder to any of the 3 CPU pipes. When using VGA you need to select a pair of matching CPU pipes/transcoders (A/A, B/B, C/C) and you also need to enable/use the PCH transcoder. For now we're just creating the cpu_transcoder definitions and setting cpu_transcoder to TRANSCODER_EDP on DDI eDP code, but none of the registers was ported to use transcoder instead of pipe. The goal is to keep the code backwards-compatible since on all cases except when using eDP we must have pipe == cpu_transcoder. V2: Comment the haswell_crtc_off chunk, suggested by Damien Lespiau and Daniel Vetter. We currently need the haswell_crtc_off chunk because TRANSCODER_EDP can be used by any CRTC, so when you stop using it you have to stop saying you're using it, otherwise you may have at some point 2 CRTCs claiming they're using TRANSCODER_EDP (a disabled CRTC and an enabled one), then the HW state readout code will get completely confused. In other words: Imagine the following case: xrandr --output eDP1 --auto --crtc 0 xrandr --output eDP1 --off xrandr --output eDP1 --auto --crtc 2 After the last command you could get a "pipe A assertion failure (expected off, current on)" because CRTC 0 still claims it's using TRANSCODER_EDP, so the HW state readout function will read it (through PIPECONF) and expect it to be off, when it's actually on because it's being used by CRTC 2. So when we make "intel_crtc->cpu_transcoder = intel_crtc->pipe" we make sure we're pointing to our own original CRTC which is certainly not used by any other CRTC. Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com> Reviewed-by: Damien Lespiau <damien.lespiau@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2012-10-24 11:59:34 -06:00
enum transcoder {
TRANSCODER_A = 0,
TRANSCODER_B,
TRANSCODER_C,
TRANSCODER_EDP,
TRANSCODER_DSI_A,
TRANSCODER_DSI_C,
I915_MAX_TRANSCODERS
drm/i915: add TRANSCODER_EDP Before Haswell we used to have the CPU pipes and the PCH transcoders. We had the same amount of pipes and transcoders, and there was a 1:1 mapping between them. After Haswell what we used to call CPU pipe was split into CPU pipe and CPU transcoder. So now we have 3 CPU pipes (A, B and C), 4 CPU transcoders (A, B, C and EDP) and 1 PCH transcoder (only used for VGA). For all the outputs except for EDP we have an 1:1 mapping on the CPU pipes and CPU transcoders, so if you're using CPU pipe A you have to use CPU transcoder A. When have an eDP output you have to use transcoder EDP and you can attach this CPU transcoder to any of the 3 CPU pipes. When using VGA you need to select a pair of matching CPU pipes/transcoders (A/A, B/B, C/C) and you also need to enable/use the PCH transcoder. For now we're just creating the cpu_transcoder definitions and setting cpu_transcoder to TRANSCODER_EDP on DDI eDP code, but none of the registers was ported to use transcoder instead of pipe. The goal is to keep the code backwards-compatible since on all cases except when using eDP we must have pipe == cpu_transcoder. V2: Comment the haswell_crtc_off chunk, suggested by Damien Lespiau and Daniel Vetter. We currently need the haswell_crtc_off chunk because TRANSCODER_EDP can be used by any CRTC, so when you stop using it you have to stop saying you're using it, otherwise you may have at some point 2 CRTCs claiming they're using TRANSCODER_EDP (a disabled CRTC and an enabled one), then the HW state readout code will get completely confused. In other words: Imagine the following case: xrandr --output eDP1 --auto --crtc 0 xrandr --output eDP1 --off xrandr --output eDP1 --auto --crtc 2 After the last command you could get a "pipe A assertion failure (expected off, current on)" because CRTC 0 still claims it's using TRANSCODER_EDP, so the HW state readout function will read it (through PIPECONF) and expect it to be off, when it's actually on because it's being used by CRTC 2. So when we make "intel_crtc->cpu_transcoder = intel_crtc->pipe" we make sure we're pointing to our own original CRTC which is certainly not used by any other CRTC. Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com> Reviewed-by: Damien Lespiau <damien.lespiau@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2012-10-24 11:59:34 -06:00
};
static inline const char *transcoder_name(enum transcoder transcoder)
{
switch (transcoder) {
case TRANSCODER_A:
return "A";
case TRANSCODER_B:
return "B";
case TRANSCODER_C:
return "C";
case TRANSCODER_EDP:
return "EDP";
case TRANSCODER_DSI_A:
return "DSI A";
case TRANSCODER_DSI_C:
return "DSI C";
default:
return "<invalid>";
}
}
drm/i915: add TRANSCODER_EDP Before Haswell we used to have the CPU pipes and the PCH transcoders. We had the same amount of pipes and transcoders, and there was a 1:1 mapping between them. After Haswell what we used to call CPU pipe was split into CPU pipe and CPU transcoder. So now we have 3 CPU pipes (A, B and C), 4 CPU transcoders (A, B, C and EDP) and 1 PCH transcoder (only used for VGA). For all the outputs except for EDP we have an 1:1 mapping on the CPU pipes and CPU transcoders, so if you're using CPU pipe A you have to use CPU transcoder A. When have an eDP output you have to use transcoder EDP and you can attach this CPU transcoder to any of the 3 CPU pipes. When using VGA you need to select a pair of matching CPU pipes/transcoders (A/A, B/B, C/C) and you also need to enable/use the PCH transcoder. For now we're just creating the cpu_transcoder definitions and setting cpu_transcoder to TRANSCODER_EDP on DDI eDP code, but none of the registers was ported to use transcoder instead of pipe. The goal is to keep the code backwards-compatible since on all cases except when using eDP we must have pipe == cpu_transcoder. V2: Comment the haswell_crtc_off chunk, suggested by Damien Lespiau and Daniel Vetter. We currently need the haswell_crtc_off chunk because TRANSCODER_EDP can be used by any CRTC, so when you stop using it you have to stop saying you're using it, otherwise you may have at some point 2 CRTCs claiming they're using TRANSCODER_EDP (a disabled CRTC and an enabled one), then the HW state readout code will get completely confused. In other words: Imagine the following case: xrandr --output eDP1 --auto --crtc 0 xrandr --output eDP1 --off xrandr --output eDP1 --auto --crtc 2 After the last command you could get a "pipe A assertion failure (expected off, current on)" because CRTC 0 still claims it's using TRANSCODER_EDP, so the HW state readout function will read it (through PIPECONF) and expect it to be off, when it's actually on because it's being used by CRTC 2. So when we make "intel_crtc->cpu_transcoder = intel_crtc->pipe" we make sure we're pointing to our own original CRTC which is certainly not used by any other CRTC. Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com> Reviewed-by: Damien Lespiau <damien.lespiau@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2012-10-24 11:59:34 -06:00
static inline bool transcoder_is_dsi(enum transcoder transcoder)
{
return transcoder == TRANSCODER_DSI_A || transcoder == TRANSCODER_DSI_C;
}
/*
* Global legacy plane identifier. Valid only for primary/sprite
* planes on pre-g4x, and only for primary planes on g4x+.
*/
enum plane {
PLANE_A,
PLANE_B,
PLANE_C,
};
#define plane_name(p) ((p) + 'A')
#define sprite_name(p, s) ((p) * INTEL_INFO(dev_priv)->num_sprites[(p)] + (s) + 'A')
/*
* Per-pipe plane identifier.
* I915_MAX_PLANES in the enum below is the maximum (across all platforms)
* number of planes per CRTC. Not all platforms really have this many planes,
* which means some arrays of size I915_MAX_PLANES may have unused entries
* between the topmost sprite plane and the cursor plane.
*
* This is expected to be passed to various register macros
* (eg. PLANE_CTL(), PS_PLANE_SEL(), etc.) so adjust with care.
*/
enum plane_id {
PLANE_PRIMARY,
PLANE_SPRITE0,
PLANE_SPRITE1,
PLANE_SPRITE2,
PLANE_CURSOR,
I915_MAX_PLANES,
};
#define for_each_plane_id_on_crtc(__crtc, __p) \
for ((__p) = PLANE_PRIMARY; (__p) < I915_MAX_PLANES; (__p)++) \
for_each_if ((__crtc)->plane_ids_mask & BIT(__p))
enum port {
PORT_NONE = -1,
PORT_A = 0,
PORT_B,
PORT_C,
PORT_D,
PORT_E,
I915_MAX_PORTS
};
#define port_name(p) ((p) + 'A')
#define I915_NUM_PHYS_VLV 2
enum dpio_channel {
DPIO_CH0,
DPIO_CH1
};
enum dpio_phy {
DPIO_PHY0,
DPIO_PHY1,
DPIO_PHY2,
};
enum intel_display_power_domain {
POWER_DOMAIN_PIPE_A,
POWER_DOMAIN_PIPE_B,
POWER_DOMAIN_PIPE_C,
POWER_DOMAIN_PIPE_A_PANEL_FITTER,
POWER_DOMAIN_PIPE_B_PANEL_FITTER,
POWER_DOMAIN_PIPE_C_PANEL_FITTER,
POWER_DOMAIN_TRANSCODER_A,
POWER_DOMAIN_TRANSCODER_B,
POWER_DOMAIN_TRANSCODER_C,
POWER_DOMAIN_TRANSCODER_EDP,
POWER_DOMAIN_TRANSCODER_DSI_A,
POWER_DOMAIN_TRANSCODER_DSI_C,
POWER_DOMAIN_PORT_DDI_A_LANES,
POWER_DOMAIN_PORT_DDI_B_LANES,
POWER_DOMAIN_PORT_DDI_C_LANES,
POWER_DOMAIN_PORT_DDI_D_LANES,
POWER_DOMAIN_PORT_DDI_E_LANES,
POWER_DOMAIN_PORT_DDI_A_IO,
POWER_DOMAIN_PORT_DDI_B_IO,
POWER_DOMAIN_PORT_DDI_C_IO,
POWER_DOMAIN_PORT_DDI_D_IO,
POWER_DOMAIN_PORT_DDI_E_IO,
POWER_DOMAIN_PORT_DSI,
POWER_DOMAIN_PORT_CRT,
POWER_DOMAIN_PORT_OTHER,
POWER_DOMAIN_VGA,
POWER_DOMAIN_AUDIO,
POWER_DOMAIN_PLLS,
POWER_DOMAIN_AUX_A,
POWER_DOMAIN_AUX_B,
POWER_DOMAIN_AUX_C,
POWER_DOMAIN_AUX_D,
POWER_DOMAIN_GMBUS,
POWER_DOMAIN_MODESET,
drm/i915: use power get/put instead of set for power on after init Currently we make sure that all power domains are enabled during driver init and turn off unneded ones only after the first modeset. Similarly during suspend we enable all power domains, which will remain on through the following resume until the first modeset. This logic is supported by intel_set_power_well() in the power domain framework. It would be nice to simplify the API, so that we only have get/put functions and make it more explicit on the higher level how this "power well on during init" logic works. This will make it also easier if in the future we want to shorten the time the power wells are on. For this add a new device private flag tracking whether we have the power wells on because of init/suspend and use only intel_display_power_get()/put(). As nothing else uses intel_set_power_well() we can remove it. This also fixes commit 6efdf354ddb186c6604d1692075421e8d2c740e9 Author: Imre Deak <imre.deak@intel.com> Date: Wed Oct 16 17:25:52 2013 +0300 drm/i915: enable only the needed power domains during modeset where removing intel_set_power_well() resulted in not releasing the reference on the power well that was taken during init and thus leaving the power well on all the time. Regression reported by Paulo. v2: - move the init_power_on flag to the power_domains struct (Daniel) v3: - add note about this being a regression fix too (Paulo) Signed-off-by: Imre Deak <imre.deak@intel.com> Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2013-10-25 08:36:48 -06:00
POWER_DOMAIN_INIT,
POWER_DOMAIN_NUM,
};
#define POWER_DOMAIN_PIPE(pipe) ((pipe) + POWER_DOMAIN_PIPE_A)
#define POWER_DOMAIN_PIPE_PANEL_FITTER(pipe) \
((pipe) + POWER_DOMAIN_PIPE_A_PANEL_FITTER)
#define POWER_DOMAIN_TRANSCODER(tran) \
((tran) == TRANSCODER_EDP ? POWER_DOMAIN_TRANSCODER_EDP : \
(tran) + POWER_DOMAIN_TRANSCODER_A)
enum hpd_pin {
HPD_NONE = 0,
HPD_TV = HPD_NONE, /* TV is known to be unreliable */
HPD_CRT,
HPD_SDVO_B,
HPD_SDVO_C,
HPD_PORT_A,
HPD_PORT_B,
HPD_PORT_C,
HPD_PORT_D,
HPD_PORT_E,
HPD_NUM_PINS
};
#define for_each_hpd_pin(__pin) \
for ((__pin) = (HPD_NONE + 1); (__pin) < HPD_NUM_PINS; (__pin)++)
#define HPD_STORM_DEFAULT_THRESHOLD 5
struct i915_hotplug {
struct work_struct hotplug_work;
struct {
unsigned long last_jiffies;
int count;
enum {
HPD_ENABLED = 0,
HPD_DISABLED = 1,
HPD_MARK_DISABLED = 2
} state;
} stats[HPD_NUM_PINS];
u32 event_bits;
struct delayed_work reenable_work;
struct intel_digital_port *irq_port[I915_MAX_PORTS];
u32 long_port_mask;
u32 short_port_mask;
struct work_struct dig_port_work;
drm/i915: Enable polling when we don't have hpd Unfortunately, there's two situations where we lose hpd right now: - Runtime suspend - When we've shut off all of the power wells on Valleyview/Cherryview While it would be nice if this didn't cause issues, this has the ability to get us in some awkward states where a user won't be able to get their display to turn on. For instance; if we boot a Valleyview system without any monitors connected, it won't need any of it's power wells and thus shut them off. Since this causes us to lose HPD, this means that unless the user knows how to ssh into their machine and do a manual reprobe for monitors, none of the monitors they connect after booting will actually work. Eventually we should come up with a better fix then having to enable polling for this, since this makes rpm a lot less useful, but for now the infrastructure in i915 just isn't there yet to get hpd in these situations. Changes since v1: - Add comment explaining the addition of the if (!mode_config->poll_running) in intel_hpd_init() - Remove unneeded if (!dev->mode_config.poll_enabled) in i915_hpd_poll_init_work() - Call to drm_helper_hpd_irq_event() after we disable polling - Add cancel_work_sync() call to intel_hpd_cancel_work() Changes since v2: - Apparently dev->mode_config.poll_running doesn't actually reflect whether or not a poll is currently in progress, and is actually used for dynamic module paramter enabling/disabling. So now we instead keep track of our own poll_running variable in dev_priv->hotplug - Clean i915_hpd_poll_init_work() a little bit Changes since v3: - Remove the now-redundant connector loop in intel_hpd_init(), just rely on intel_hpd_poll_enable() for setting connector->polled correctly on each connector - Get rid of poll_running - Don't assign enabled in i915_hpd_poll_init_work before we actually lock dev->mode_config.mutex - Wrap enabled assignment in i915_hpd_poll_init_work() in READ_ONCE() for doc purposes - Do the same for dev_priv->hotplug.poll_enabled with WRITE_ONCE in intel_hpd_poll_enable() - Add some comments about racing not mattering in intel_hpd_poll_enable Changes since v4: - Rename intel_hpd_poll_enable() to intel_hpd_poll_init() - Drop the bool argument from intel_hpd_poll_init() - Remove redundant calls to intel_hpd_poll_init() - Rename poll_enable_work to poll_init_work - Add some kerneldoc for intel_hpd_poll_init() - Cross-reference intel_hpd_poll_init() in intel_hpd_init() - Just copy the loop from intel_hpd_init() in intel_hpd_poll_init() Changes since v5: - Minor kerneldoc nitpicks Cc: stable@vger.kernel.org Cc: Ville Syrjälä <ville.syrjala@linux.intel.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Lyude <cpaul@redhat.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2016-06-21 15:03:44 -06:00
struct work_struct poll_init_work;
bool poll_enabled;
unsigned int hpd_storm_threshold;
/*
* if we get a HPD irq from DP and a HPD irq from non-DP
* the non-DP HPD could block the workqueue on a mode config
* mutex getting, that userspace may have taken. However
* userspace is waiting on the DP workqueue to run which is
* blocked behind the non-DP one.
*/
struct workqueue_struct *dp_wq;
};
#define I915_GEM_GPU_DOMAINS \
(I915_GEM_DOMAIN_RENDER | \
I915_GEM_DOMAIN_SAMPLER | \
I915_GEM_DOMAIN_COMMAND | \
I915_GEM_DOMAIN_INSTRUCTION | \
I915_GEM_DOMAIN_VERTEX)
#define for_each_pipe(__dev_priv, __p) \
for ((__p) = 0; (__p) < INTEL_INFO(__dev_priv)->num_pipes; (__p)++)
#define for_each_pipe_masked(__dev_priv, __p, __mask) \
for ((__p) = 0; (__p) < INTEL_INFO(__dev_priv)->num_pipes; (__p)++) \
for_each_if ((__mask) & (1 << (__p)))
#define for_each_universal_plane(__dev_priv, __pipe, __p) \
for ((__p) = 0; \
(__p) < INTEL_INFO(__dev_priv)->num_sprites[(__pipe)] + 1; \
(__p)++)
#define for_each_sprite(__dev_priv, __p, __s) \
for ((__s) = 0; \
(__s) < INTEL_INFO(__dev_priv)->num_sprites[(__p)]; \
(__s)++)
#define for_each_port_masked(__port, __ports_mask) \
for ((__port) = PORT_A; (__port) < I915_MAX_PORTS; (__port)++) \
for_each_if ((__ports_mask) & (1 << (__port)))
#define for_each_crtc(dev, crtc) \
list_for_each_entry(crtc, &(dev)->mode_config.crtc_list, head)
#define for_each_intel_plane(dev, intel_plane) \
list_for_each_entry(intel_plane, \
&(dev)->mode_config.plane_list, \
base.head)
2016-05-12 08:06:01 -06:00
#define for_each_intel_plane_mask(dev, intel_plane, plane_mask) \
list_for_each_entry(intel_plane, \
&(dev)->mode_config.plane_list, \
2016-05-12 08:06:01 -06:00
base.head) \
for_each_if ((plane_mask) & \
(1 << drm_plane_index(&intel_plane->base)))
#define for_each_intel_plane_on_crtc(dev, intel_crtc, intel_plane) \
list_for_each_entry(intel_plane, \
&(dev)->mode_config.plane_list, \
base.head) \
for_each_if ((intel_plane)->pipe == (intel_crtc)->pipe)
#define for_each_intel_crtc(dev, intel_crtc) \
list_for_each_entry(intel_crtc, \
&(dev)->mode_config.crtc_list, \
base.head)
#define for_each_intel_crtc_mask(dev, intel_crtc, crtc_mask) \
list_for_each_entry(intel_crtc, \
&(dev)->mode_config.crtc_list, \
base.head) \
drm/i915/gen9: Compute DDB allocation at atomic check time (v4) Calculate the DDB blocks needed to satisfy the current atomic transaction at atomic check time. This is a prerequisite to calculating SKL watermarks during the 'check' phase and rejecting any configurations that we can't find valid watermarks for. Due to the nature of DDB allocation, it's possible for the addition of a new CRTC to make the watermark configuration already in use on another, unchanged CRTC become invalid. A change in which CRTC's are active triggers a recompute of the entire DDB, which unfortunately means we need to disallow any other atomic commits from racing with such an update. If the active CRTC's change, we need to grab the lock on all CRTC's and run all CRTC's through their 'check' handler to recompute and re-check their per-CRTC DDB allocations. Note that with this patch we only compute the DDB allocation but we don't actually use the computed values during watermark programming yet. For ease of review/testing/bisecting, we still recompute the DDB at watermark programming time and just WARN() if it doesn't match the precomputed values. A future patch will switch over to using the precomputed values once we're sure they're being properly computed. Another clarifying note: DDB allocation itself shouldn't ever fail with the algorithm we use today (i.e., we have enough DDB blocks on BXT to support the minimum needs of the worst-case scenario of every pipe/plane enabled at full size). However the watermarks calculations based on the DDB may fail and we'll be moving those to the atomic check as well in future patches. v2: - Skip DDB calculations in the rare case where our transaction doesn't actually touch any CRTC's at all. Assuming at least one CRTC state is present in our transaction, then it means we can't race with any transactions that would update dev_priv->active_crtcs (which requires _all_ CRTC locks). v3: - Also calculate DDB during initial hw readout, to prevent using incorrect bios values. (Maarten) v4: - Use new distrust_bios_wm flag instead of skip_initial_wm (which was never actually set). - Set intel_state->active_pipe_changes instead of just realloc_pipes Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Cc: Lyude Paul <cpaul@redhat.com> Cc: Radhakrishna Sripada <radhakrishna.sripada@intel.com> Signed-off-by: Matt Roper <matthew.d.roper@intel.com> Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1463061971-19638-10-git-send-email-matthew.d.roper@intel.com
2016-05-12 08:06:03 -06:00
for_each_if ((crtc_mask) & (1 << drm_crtc_index(&intel_crtc->base)))
#define for_each_intel_encoder(dev, intel_encoder) \
list_for_each_entry(intel_encoder, \
&(dev)->mode_config.encoder_list, \
base.head)
#define for_each_intel_connector_iter(intel_connector, iter) \
while ((intel_connector = to_intel_connector(drm_connector_list_iter_next(iter))))
#define for_each_encoder_on_crtc(dev, __crtc, intel_encoder) \
list_for_each_entry((intel_encoder), &(dev)->mode_config.encoder_list, base.head) \
for_each_if ((intel_encoder)->base.crtc == (__crtc))
#define for_each_connector_on_encoder(dev, __encoder, intel_connector) \
list_for_each_entry((intel_connector), &(dev)->mode_config.connector_list, base.head) \
for_each_if ((intel_connector)->base.encoder == (__encoder))
#define for_each_power_domain(domain, mask) \
for ((domain) = 0; (domain) < POWER_DOMAIN_NUM; (domain)++) \
for_each_if (BIT_ULL(domain) & (mask))
#define for_each_power_well(__dev_priv, __power_well) \
for ((__power_well) = (__dev_priv)->power_domains.power_wells; \
(__power_well) - (__dev_priv)->power_domains.power_wells < \
(__dev_priv)->power_domains.power_well_count; \
(__power_well)++)
#define for_each_power_well_rev(__dev_priv, __power_well) \
for ((__power_well) = (__dev_priv)->power_domains.power_wells + \
(__dev_priv)->power_domains.power_well_count - 1; \
(__power_well) - (__dev_priv)->power_domains.power_wells >= 0; \
(__power_well)--)
#define for_each_power_domain_well(__dev_priv, __power_well, __domain_mask) \
for_each_power_well(__dev_priv, __power_well) \
for_each_if ((__power_well)->domains & (__domain_mask))
#define for_each_power_domain_well_rev(__dev_priv, __power_well, __domain_mask) \
for_each_power_well_rev(__dev_priv, __power_well) \
for_each_if ((__power_well)->domains & (__domain_mask))
#define for_each_intel_plane_in_state(__state, plane, plane_state, __i) \
for ((__i) = 0; \
(__i) < (__state)->base.dev->mode_config.num_total_plane && \
((plane) = to_intel_plane((__state)->base.planes[__i].ptr), \
(plane_state) = to_intel_plane_state((__state)->base.planes[__i].state), 1); \
(__i)++) \
for_each_if (plane_state)
struct drm_i915_private;
drm/i915: Prevent recursive deadlock on releasing a busy userptr During release of the GEM object we hold the struct_mutex. As the object may be holding onto the last reference for the task->mm, calling mmput() may trigger exit_mmap() which close the vma which will call drm_gem_vm_close() and attempt to reacquire the struct_mutex. In order to avoid that recursion, we have to defer the mmput() until after we drop the struct_mutex, i.e. we need to schedule a worker to do the clean up. A further issue spotted by Tvrtko was caused when we took a GTT mmapping of a userptr buffer object. In that case, we would never call mmput as the object would be cyclically referenced by the GTT mmapping and not freed upon process exit - keeping the entire process mm alive after the process task was reaped. The fix employed is to replace the mm_users/mmput() reference handling to mm_count/mmdrop() for the shared i915_mm_struct. INFO: task test_surfaces:1632 blocked for more than 120 seconds.       Tainted: GF          O 3.14.5+ #1 "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. test_surfaces   D 0000000000000000     0  1632   1590 0x00000082  ffff88014914baa8 0000000000000046 0000000000000000 ffff88014914a010  0000000000012c40 0000000000012c40 ffff8800a0058210 ffff88014784b010  ffff88014914a010 ffff880037b1c820 ffff8800a0058210 ffff880037b1c824 Call Trace:  [<ffffffff81582499>] schedule+0x29/0x70  [<ffffffff815825fe>] schedule_preempt_disabled+0xe/0x10  [<ffffffff81583b93>] __mutex_lock_slowpath+0x183/0x220  [<ffffffff81583c53>] mutex_lock+0x23/0x40  [<ffffffffa005c2a3>] drm_gem_vm_close+0x33/0x70 [drm]  [<ffffffff8115a483>] remove_vma+0x33/0x70  [<ffffffff8115a5dc>] exit_mmap+0x11c/0x170  [<ffffffff8104d6eb>] mmput+0x6b/0x100  [<ffffffffa00f44b9>] i915_gem_userptr_release+0x89/0xc0 [i915]  [<ffffffffa00e6706>] i915_gem_free_object+0x126/0x250 [i915]  [<ffffffffa005c06a>] drm_gem_object_free+0x2a/0x40 [drm]  [<ffffffffa005cc32>] drm_gem_object_handle_unreference_unlocked+0xe2/0x120 [drm]  [<ffffffffa005ccd4>] drm_gem_object_release_handle+0x64/0x90 [drm]  [<ffffffff8127ffeb>] idr_for_each+0xab/0x100  [<ffffffffa005cc70>] ? drm_gem_object_handle_unreference_unlocked+0x120/0x120 [drm]  [<ffffffff81583c46>] ? mutex_lock+0x16/0x40  [<ffffffffa005c354>] drm_gem_release+0x24/0x40 [drm]  [<ffffffffa005b82b>] drm_release+0x3fb/0x480 [drm]  [<ffffffff8118d482>] __fput+0xb2/0x260  [<ffffffff8118d6de>] ____fput+0xe/0x10  [<ffffffff8106f27f>] task_work_run+0x8f/0xf0  [<ffffffff81052228>] do_exit+0x1a8/0x480  [<ffffffff81052551>] do_group_exit+0x51/0xc0  [<ffffffff810525d7>] SyS_exit_group+0x17/0x20  [<ffffffff8158e092>] system_call_fastpath+0x16/0x1b v2: Incorporate feedback from Tvrtko and remove the unnessary mm referencing when creating the i915_mm_struct and improve some of the function names and comments. Reported-by: Jacek Danecki <jacek.danecki@intel.com> Test-case: igt/gem_userptr_blits/process-exit* Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Tested-by: "Gong, Zhipeng" <zhipeng.gong@intel.com> Cc: Jacek Danecki <jacek.danecki@intel.com> Cc: "Ursulin, Tvrtko" <tvrtko.ursulin@intel.com> Reviewed-by: "Ursulin, Tvrtko" <tvrtko.ursulin@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: stable@vger.kernel.org # hold off until 3.17 ships for additional testing Signed-off-by: Jani Nikula <jani.nikula@intel.com>
2014-08-07 07:20:40 -06:00
struct i915_mm_struct;
drm/i915: Introduce mapping of user pages into video memory (userptr) ioctl By exporting the ability to map user address and inserting PTEs representing their backing pages into the GTT, we can exploit UMA in order to utilize normal application data as a texture source or even as a render target (depending upon the capabilities of the chipset). This has a number of uses, with zero-copy downloads to the GPU and efficient readback making the intermixed streaming of CPU and GPU operations fairly efficient. This ability has many widespread implications from faster rendering of client-side software rasterisers (chromium), mitigation of stalls due to read back (firefox) and to faster pipelining of texture data (such as pixel buffer objects in GL or data blobs in CL). v2: Compile with CONFIG_MMU_NOTIFIER v3: We can sleep while performing invalidate-range, which we can utilise to drop our page references prior to the kernel manipulating the vma (for either discard or cloning) and so protect normal users. v4: Only run the invalidate notifier if the range intercepts the bo. v5: Prevent userspace from attempting to GTT mmap non-page aligned buffers v6: Recheck after reacquire mutex for lost mmu. v7: Fix implicit padding of ioctl struct by rounding to next 64bit boundary. v8: Fix rebasing error after forwarding porting the back port. v9: Limit the userptr to page aligned entries. We now expect userspace to handle all the offset-in-page adjustments itself. v10: Prevent vma from being copied across fork to avoid issues with cow. v11: Drop vma behaviour changes -- locking is nigh on impossible. Use a worker to load user pages to avoid lock inversions. v12: Use get_task_mm()/mmput() for correct refcounting of mm. v13: Use a worker to release the mmu_notifier to avoid lock inversion v14: Decouple mmu_notifier from struct_mutex using a custom mmu_notifer with its own locking and tree of objects for each mm/mmu_notifier. v15: Prevent overlapping userptr objects, and invalidate all objects within the mmu_notifier range v16: Fix a typo for iterating over multiple objects in the range and rearrange error path to destroy the mmu_notifier locklessly. Also close a race between invalidate_range and the get_pages_worker. v17: Close a race between get_pages_worker/invalidate_range and fresh allocations of the same userptr range - and notice that struct_mutex was presumed to be held when during creation it wasn't. v18: Sigh. Fix the refactor of st_set_pages() to allocate enough memory for the struct sg_table and to clear it before reporting an error. v19: Always error out on read-only userptr requests as we don't have the hardware infrastructure to support them at the moment. v20: Refuse to implement read-only support until we have the required infrastructure - but reserve the bit in flags for future use. v21: use_mm() is not required for get_user_pages(). It is only meant to be used to fix up the kernel thread's current->mm for use with copy_user(). v22: Use sg_alloc_table_from_pages for that chunky feeling v23: Export a function for sanity checking dma-buf rather than encode userptr details elsewhere, and clean up comments based on suggestions by Bradley. Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> Cc: "Gong, Zhipeng" <zhipeng.gong@intel.com> Cc: Akash Goel <akash.goel@intel.com> Cc: "Volkin, Bradley D" <bradley.d.volkin@intel.com> Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> Reviewed-by: Brad Volkin <bradley.d.volkin@intel.com> [danvet: Frob ioctl allocation to pick the next one - will cause a bit of fuss with create2 apparently, but such are the rules.] [danvet2: oops, forgot to git add after manual patch application] [danvet3: Appease sparse.] Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2014-05-16 07:22:37 -06:00
struct i915_mmu_object;
struct drm_i915_file_private {
struct drm_i915_private *dev_priv;
struct drm_file *file;
struct {
spinlock_t lock;
struct list_head request_list;
/* 20ms is a fairly arbitrary limit (greater than the average frame time)
* chosen to prevent the CPU getting more than a frame ahead of the GPU
* (when using lax throttling for the frontbuffer). We also use it to
* offer free GPU waitboosts for severely congested workloads.
*/
#define DRM_I915_THROTTLE_JIFFIES msecs_to_jiffies(20)
} mm;
struct idr context_idr;
struct intel_rps_client {
drm/i915: Avoid keeping waitboost active for signaling threads Once a client has requested a waitboost, we keep that waitboost active until all clients are no longer waiting. This is because we don't distinguish which waiter deserves the boost. However, with the advent of fence signaling, the signaler threads appear as waiters to the RPS interrupt handler. So instead of using a single boolean to track when to keep the waitboost active, use a counter of all outstanding waitboosted requests. At this point, I have removed all vestiges of the rate limiting on clients. Whilst this means that compositors should remain more fluid, it also means that boosts are more prevalent. See commit b29c19b64528 ("drm/i915: Boost RPS frequency for CPU stalls") for a longer discussion on the pros and cons of both approaches. A drawback of this implementation is that it requires constant request submission to keep the waitboost trimmed (as it is now cancelled when the request is completed). This will be fine for a busy system, but near idle the boosts may be kept for longer than desired (effectively tens of vblanks worstcase) and there is a reliance on rc6 instead. v2: Remove defunct rps.client_lock Reported-by: Michał Winiarski <michal.winiarski@intel.com> Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Michał Winiarski <michal.winiarski@intel.com> Reviewed-by: Michał Winiarski <michal.winiarski@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/20170628123548.9236-1-chris@chris-wilson.co.uk
2017-06-28 06:35:48 -06:00
atomic_t boosts;
} rps;
unsigned int bsd_engine;
/* Client can have a maximum of 3 contexts banned before
* it is denied of creating new contexts. As one context
* ban needs 4 consecutive hangs, and more if there is
* progress in between, this is a last resort stop gap measure
* to limit the badly behaving clients access to gpu.
*/
#define I915_MAX_CLIENT_CONTEXT_BANS 3
atomic_t context_bans;
};
/* Used by dp and fdi links */
struct intel_link_m_n {
uint32_t tu;
uint32_t gmch_m;
uint32_t gmch_n;
uint32_t link_m;
uint32_t link_n;
};
void intel_link_compute_m_n(int bpp, int nlanes,
int pixel_clock, int link_clock,
struct intel_link_m_n *m_n,
bool reduce_m_n);
/* Interface history:
*
* 1.1: Original.
* 1.2: Add Power Management
* 1.3: Add vblank support
* 1.4: Fix cmdbuffer path, add heap destroy
* 1.5: Add vblank pipe configuration
* 1.6: - New ioctl for scheduling buffer swaps on vertical blank
* - Support vertical blank on secondary display pipe
*/
#define DRIVER_MAJOR 1
#define DRIVER_MINOR 6
#define DRIVER_PATCHLEVEL 0
struct opregion_header;
struct opregion_acpi;
struct opregion_swsci;
struct opregion_asle;
struct intel_opregion {
struct opregion_header *header;
struct opregion_acpi *acpi;
struct opregion_swsci *swsci;
u32 swsci_gbda_sub_functions;
u32 swsci_sbcb_sub_functions;
struct opregion_asle *asle;
void *rvda;
void *vbt_firmware;
const void *vbt;
u32 vbt_size;
u32 *lid_state;
struct work_struct asle_work;
};
#define OPREGION_SIZE (8*1024)
struct intel_overlay;
struct intel_overlay_error_state;
struct sdvo_device_mapping {
u8 initialized;
u8 dvo_port;
u8 slave_addr;
u8 dvo_wiring;
u8 i2c_pin;
u8 ddc_pin;
};
struct intel_connector;
struct intel_encoder;
struct intel_atomic_state;
struct intel_crtc_state;
struct intel_initial_plane_config;
struct intel_crtc;
drm/i915: move find_pll callback to dev_priv->display Now that the DP madness is cleared out, this is all only per-platform. So move it out from the intel clock limits structure. While at it drop the intel prefix on the static functions, call the vtable entry find_dpll (since it's for the display pll) and rip out the now unnecessary forward declarations. Note that the parameters of ->find_dpll are still unchanged, but they eventually need to be moved over to just take in a pipe configuration. But currently a lot of things are still missing from the pipe configuration (reflock, output-specific dpll limits and preferences, downclocked dotclock). So this will happen in a later step. Note that intel_g4x_limit has a peculiar case where it selects intel_limits_i9xx_sdvo as the limit. This is pretty bogus and also not used since the only output types left are DP and native TV-out which both use special pre-tuned dpll values. v2: Re-add comment for the find_pll callback (requested by Paulo) and elaborate on why the transformation is correct for g4x platforms (to clarify a review question from Paulo). Double up on that by adding a WARN as suggested by Paulo Zanoni on irc. v3: Initialize limits to NULL since gcc is now unhappy. v4: v2/3 will blow up with a NULL dereference in ->find_dpll for dp and TV-out ports, spotted by Paulo on irc. So just give up on this madness for now, and leave this to be fixed in a later patch. v5: Since the ever-so-slight change for g4x might result in some dpll parameter computation failing spuriously where before it didn't for ports with preset dpll settings (DP & TV-out) override this. For paranoia also do it in the ilk+ code. Cc: Paulo Zanoni <przanoni@gmail.com> Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2013-06-03 14:40:22 -06:00
struct intel_limit;
struct dpll;
struct intel_cdclk_state;
struct drm_i915_display_funcs {
void (*get_cdclk)(struct drm_i915_private *dev_priv,
struct intel_cdclk_state *cdclk_state);
void (*set_cdclk)(struct drm_i915_private *dev_priv,
const struct intel_cdclk_state *cdclk_state);
int (*get_fifo_size)(struct drm_i915_private *dev_priv, int plane);
int (*compute_pipe_wm)(struct intel_crtc_state *cstate);
drm/i915: Add two-stage ILK-style watermark programming (v11) In addition to calculating final watermarks, let's also pre-calculate a set of intermediate watermark values at atomic check time. These intermediate watermarks are a combination of the watermarks for the old state and the new state; they should satisfy the requirements of both states which means they can be programmed immediately when we commit the atomic state (without waiting for a vblank). Once the vblank does happen, we can then re-program watermarks to the more optimal final value. v2: Significant rebasing/rewriting. v3: - Move 'need_postvbl_update' flag to CRTC state (Daniel) - Don't forget to check intermediate watermark values for validity (Maarten) - Don't due async watermark optimization; just do it at the end of the atomic transaction, after waiting for vblanks. We do want it to be async eventually, but adding that now will cause more trouble for Maarten's in-progress work. (Maarten) - Don't allocate space in crtc_state for intermediate watermarks on platforms that don't need it (gen9+). - Move WaCxSRDisabledForSpriteScaling:ivb into intel_begin_crtc_commit now that ilk_update_wm is gone. v4: - Add a wm_mutex to cover updates to intel_crtc->active and the need_postvbl_update flag. Since we don't have async yet it isn't terribly important yet, but might as well add it now. - Change interface to program watermarks. Platforms will now expose .initial_watermarks() and .optimize_watermarks() functions to do watermark programming. These should lock wm_mutex, copy the appropriate state values into intel_crtc->active, and then call the internal program watermarks function. v5: - Skip intermediate watermark calculation/check during initial hardware readout since we don't trust the existing HW values (and don't have valid values of our own yet). - Don't try to call .optimize_watermarks() on platforms that don't have atomic watermarks yet. (Maarten) v6: - Rebase v7: - Further rebase v8: - A few minor indentation and line length fixes v9: - Yet another rebase since Maarten's patches reworked a bunch of the code (wm_pre, wm_post, etc.) that this was previously based on. v10: - Move wm_mutex to dev_priv to protect against racing commits against disjoint CRTC sets. (Maarten) - Drop unnecessary clearing of cstate->wm.need_postvbl_update (Maarten) v11: - Now that we've moved to atomic watermark updates, make sure we call the proper function to program watermarks in {ironlake,haswell}_crtc_enable(); the failure to do so on the previous patch iteration led to us not actually programming the watermarks before turning on the CRTC, which was the cause of the underruns that the CI system was seeing. - Fix inverted logic for determining when to optimize watermarks. We were needlessly optimizing when the intermediate/optimal values were the same (harmless), but not actually optimizing when they differed (also harmless, but wasteful from a power/bandwidth perspective). Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Signed-off-by: Matt Roper <matthew.d.roper@intel.com> Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1456276813-5689-1-git-send-email-matthew.d.roper@intel.com
2016-02-23 18:20:13 -07:00
int (*compute_intermediate_wm)(struct drm_device *dev,
struct intel_crtc *intel_crtc,
struct intel_crtc_state *newstate);
void (*initial_watermarks)(struct intel_atomic_state *state,
struct intel_crtc_state *cstate);
void (*atomic_update_watermarks)(struct intel_atomic_state *state,
struct intel_crtc_state *cstate);
void (*optimize_watermarks)(struct intel_atomic_state *state,
struct intel_crtc_state *cstate);
drm/i915/gen9: Compute DDB allocation at atomic check time (v4) Calculate the DDB blocks needed to satisfy the current atomic transaction at atomic check time. This is a prerequisite to calculating SKL watermarks during the 'check' phase and rejecting any configurations that we can't find valid watermarks for. Due to the nature of DDB allocation, it's possible for the addition of a new CRTC to make the watermark configuration already in use on another, unchanged CRTC become invalid. A change in which CRTC's are active triggers a recompute of the entire DDB, which unfortunately means we need to disallow any other atomic commits from racing with such an update. If the active CRTC's change, we need to grab the lock on all CRTC's and run all CRTC's through their 'check' handler to recompute and re-check their per-CRTC DDB allocations. Note that with this patch we only compute the DDB allocation but we don't actually use the computed values during watermark programming yet. For ease of review/testing/bisecting, we still recompute the DDB at watermark programming time and just WARN() if it doesn't match the precomputed values. A future patch will switch over to using the precomputed values once we're sure they're being properly computed. Another clarifying note: DDB allocation itself shouldn't ever fail with the algorithm we use today (i.e., we have enough DDB blocks on BXT to support the minimum needs of the worst-case scenario of every pipe/plane enabled at full size). However the watermarks calculations based on the DDB may fail and we'll be moving those to the atomic check as well in future patches. v2: - Skip DDB calculations in the rare case where our transaction doesn't actually touch any CRTC's at all. Assuming at least one CRTC state is present in our transaction, then it means we can't race with any transactions that would update dev_priv->active_crtcs (which requires _all_ CRTC locks). v3: - Also calculate DDB during initial hw readout, to prevent using incorrect bios values. (Maarten) v4: - Use new distrust_bios_wm flag instead of skip_initial_wm (which was never actually set). - Set intel_state->active_pipe_changes instead of just realloc_pipes Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Cc: Lyude Paul <cpaul@redhat.com> Cc: Radhakrishna Sripada <radhakrishna.sripada@intel.com> Signed-off-by: Matt Roper <matthew.d.roper@intel.com> Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1463061971-19638-10-git-send-email-matthew.d.roper@intel.com
2016-05-12 08:06:03 -06:00
int (*compute_global_watermarks)(struct drm_atomic_state *state);
void (*update_wm)(struct intel_crtc *crtc);
int (*modeset_calc_cdclk)(struct drm_atomic_state *state);
/* Returns the active state of the crtc, and if the crtc is active,
* fills out the pipe-config with the hw state. */
bool (*get_pipe_config)(struct intel_crtc *,
struct intel_crtc_state *);
void (*get_initial_plane_config)(struct intel_crtc *,
struct intel_initial_plane_config *);
int (*crtc_compute_clock)(struct intel_crtc *crtc,
struct intel_crtc_state *crtc_state);
void (*crtc_enable)(struct intel_crtc_state *pipe_config,
struct drm_atomic_state *old_state);
void (*crtc_disable)(struct intel_crtc_state *old_crtc_state,
struct drm_atomic_state *old_state);
void (*update_crtcs)(struct drm_atomic_state *state,
unsigned int *crtc_vblank_mask);
void (*audio_codec_enable)(struct drm_connector *connector,
struct intel_encoder *encoder,
const struct drm_display_mode *adjusted_mode);
void (*audio_codec_disable)(struct intel_encoder *encoder);
void (*fdi_link_train)(struct intel_crtc *crtc,
const struct intel_crtc_state *crtc_state);
void (*init_clock_gating)(struct drm_i915_private *dev_priv);
drm/i915: Small display interrupt handlers tidy I have noticed some of our interrupt handlers use both dev and dev_priv while they could get away with only dev_priv in the huge majority of cases. Tidying that up had a cascading effect on changing functions prototypes, so relatively big churn factor, but I think it is for the better. For example even where changes cascade out of i915_irq.c, for functions prefixed with intel_, genX_ or <plat>_, it makes more sense to take dev_priv directly anyway. This allows us to eliminate local variables and intermixed usage of dev and dev_priv where only one is good enough. End result is shrinkage of both source and the resulting binary. i915.ko: - .text 000b0899 + .text 000b0619 Or if we look at the Gen8 display irq chain: -00000000000006ad t gen8_irq_handler +0000000000000663 t gen8_irq_handler -0000000000000028 T intel_opregion_asle_intr +0000000000000024 T intel_opregion_asle_intr -000000000000008c t ilk_hpd_irq_handler +000000000000007f t ilk_hpd_irq_handler -0000000000000116 T intel_check_page_flip +0000000000000112 T intel_check_page_flip -000000000000011a T intel_prepare_page_flip +0000000000000119 T intel_prepare_page_flip -0000000000000014 T intel_finish_page_flip_plane +0000000000000013 T intel_finish_page_flip_plane -0000000000000053 t hsw_pipe_crc_irq_handler +000000000000004c t hsw_pipe_crc_irq_handler -000000000000022e t cpt_irq_handler +0000000000000213 t cpt_irq_handler So small shrinkage but it is all fast paths so doesn't harm. Situation is similar in other interrupt handlers as well. v2: Tidy intel_queue_rps_boost_for_request as well. (Chris Wilson) Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
2016-05-06 07:48:28 -06:00
void (*hpd_irq_setup)(struct drm_i915_private *dev_priv);
/* clock updates for mode set */
/* cursor updates */
/* render clock increase/decrease */
/* display clock increase/decrease */
/* pll clock increase/decrease */
void (*load_csc_matrix)(struct drm_crtc_state *crtc_state);
void (*load_luts)(struct drm_crtc_state *crtc_state);
};
#define CSR_VERSION(major, minor) ((major) << 16 | (minor))
#define CSR_VERSION_MAJOR(version) ((version) >> 16)
#define CSR_VERSION_MINOR(version) ((version) & 0xffff)
drm/i915/skl: Add support to load SKL CSR firmware. Display Context Save and Restore support is needed for various SKL Display C states like DC5, DC6. This implementation is added based on first version of DMC CSR program that we received from h/w team. Here we are using request_firmware based design. Finally this firmware should end up in linux-firmware tree. For SKL platform its mandatory to ensure that we load this csr program before enabling DC states like DC5/DC6. As CSR program gets reset on various conditions, we should ensure to load it during boot and in future change to be added to load this system resume sequence too. v1: Initial relese as RFC patch v2: Design change as per Daniel, Damien and Shobit's review comments request firmware method followed. v3: Some optimization and functional changes. Pulled register defines into drivers/gpu/drm/i915/i915_reg.h Used kmemdup to allocate and duplicate firmware content. Ensured to free allocated buffer. v4: Modified as per review comments from Satheesh and Daniel Removed temporary buffer. Optimized number of writes by replacing I915_WRITE with I915_WRITE64. v5: Modified as per review comemnts from Damien. - Changed name for functions and firmware. - Introduced HAS_CSR. - Reverted back previous change and used csr_buf with u8 size. - Using cpu_to_be64 for endianness change. Modified as per review comments from Imre. - Modified registers and macro names to be a bit closer to bspec terminology and the existing register naming in the driver. - Early return for non SKL platforms in intel_load_csr_program function. - Added locking around CSR program load function as it may be called concurrently during system/runtime resume. - Releasing the fw before loading the program for consistency - Handled error path during f/w load. v6: Modified as per review comments from Imre. - Corrected out_freecsr sequence. v7: Modified as per review comments from Imre. Fail loading fw if fw->size%8!=0. v8: Rebase to latest. v9: Rebase on top of -nightly (Damien) v10: Enabled support for dmc firmware ver 1.0. According to ver 1.0 in a single binary package all the firmware's that are required for different stepping's of the product will be stored. The package contains the css header, followed by the package header and the actual dmc firmwares. Package header contains the firmware/stepping mapping table and the corresponding firmware offsets to the individual binaries, within the package. Each individual program binary contains the header and the payload sections whose size is specified in the header section. This changes are done to extract the specific firmaware from the package. (Animesh) v11: Modified as per review comemnts from Imre. - Added code comment from bpec for header structure elements. - Added __packed to avoid structure padding. - Added helper functions for stepping and substepping info. - Added code comment for CSR_MAX_FW_SIZE. - Disabled BXT firmware loading, will be enabled with dmc 1.0 support. - Changed skl_stepping_info based on bspec, earlier used from config DB. - Removed duplicate call of cpu_to_be* from intel_csr_load_program function. - Used cpu_to_be32 instead of cpu_to_be64 as firmware binary in dword aligned. - Added sanity check for header length. - Added sanity check for mmio address got from firmware binary. - kmalloc done separately for dmc header and dmc firmware. (Animesh) v12: Modified as per review comemnts from Imre. - Corrected the typo error in skl stepping info structure. - Added out-of-bound access for skl_stepping_info. - Sanity check for mmio address modified. - Sanity check added for stepping and substeppig. - Modified the intel_dmc_info structure, cache only the required header info. (Animesh) v13: clarify firmware load error message. The reason for a firmware loading failure can be obscure if the driver is built-in. Provide an explanation to the user about the likely reason for the failure and how to resolve it. (Imre) v14: Suggested by Jani. - fix s/I915/CONFIG_DRM_I915/ typo - add fw_path to the firmware object instead of using a static ptr (Jani) v15: 1) Changed the firmware name as dmc_gen9.bin, everytime for a new firmware version a symbolic link with same name will help not to build kernel again. 2) Changes done as per review comments from Imre. - Error check removed for intel_csr_ucode_init. - Moved csr-specific data structure to intel_csr.h and optimization done on structure definition. - fw->data used directly for parsing the header info & memory allocation only done separately for payload. (Animesh) v16: - No need for out_regs label in i915_driver_load(), so removed it. - Changed the firmware name as skl_dmc_ver1.bin, followed naming convention <platform>_dmc_<api-version>.bin (Animesh) Issue: VIZ-2569 Signed-off-by: A.Sunil Kamath <sunil.kamath@intel.com> Signed-off-by: Damien Lespiau <damien.lespiau@intel.com> Signed-off-by: Animesh Manna <animesh.manna@intel.com> Signed-off-by: Imre Deak <imre.deak@intel.com> Reviewed-by: Imre Deak <imre.deak@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2015-05-04 06:58:44 -06:00
struct intel_csr {
struct work_struct work;
drm/i915/skl: Add support to load SKL CSR firmware. Display Context Save and Restore support is needed for various SKL Display C states like DC5, DC6. This implementation is added based on first version of DMC CSR program that we received from h/w team. Here we are using request_firmware based design. Finally this firmware should end up in linux-firmware tree. For SKL platform its mandatory to ensure that we load this csr program before enabling DC states like DC5/DC6. As CSR program gets reset on various conditions, we should ensure to load it during boot and in future change to be added to load this system resume sequence too. v1: Initial relese as RFC patch v2: Design change as per Daniel, Damien and Shobit's review comments request firmware method followed. v3: Some optimization and functional changes. Pulled register defines into drivers/gpu/drm/i915/i915_reg.h Used kmemdup to allocate and duplicate firmware content. Ensured to free allocated buffer. v4: Modified as per review comments from Satheesh and Daniel Removed temporary buffer. Optimized number of writes by replacing I915_WRITE with I915_WRITE64. v5: Modified as per review comemnts from Damien. - Changed name for functions and firmware. - Introduced HAS_CSR. - Reverted back previous change and used csr_buf with u8 size. - Using cpu_to_be64 for endianness change. Modified as per review comments from Imre. - Modified registers and macro names to be a bit closer to bspec terminology and the existing register naming in the driver. - Early return for non SKL platforms in intel_load_csr_program function. - Added locking around CSR program load function as it may be called concurrently during system/runtime resume. - Releasing the fw before loading the program for consistency - Handled error path during f/w load. v6: Modified as per review comments from Imre. - Corrected out_freecsr sequence. v7: Modified as per review comments from Imre. Fail loading fw if fw->size%8!=0. v8: Rebase to latest. v9: Rebase on top of -nightly (Damien) v10: Enabled support for dmc firmware ver 1.0. According to ver 1.0 in a single binary package all the firmware's that are required for different stepping's of the product will be stored. The package contains the css header, followed by the package header and the actual dmc firmwares. Package header contains the firmware/stepping mapping table and the corresponding firmware offsets to the individual binaries, within the package. Each individual program binary contains the header and the payload sections whose size is specified in the header section. This changes are done to extract the specific firmaware from the package. (Animesh) v11: Modified as per review comemnts from Imre. - Added code comment from bpec for header structure elements. - Added __packed to avoid structure padding. - Added helper functions for stepping and substepping info. - Added code comment for CSR_MAX_FW_SIZE. - Disabled BXT firmware loading, will be enabled with dmc 1.0 support. - Changed skl_stepping_info based on bspec, earlier used from config DB. - Removed duplicate call of cpu_to_be* from intel_csr_load_program function. - Used cpu_to_be32 instead of cpu_to_be64 as firmware binary in dword aligned. - Added sanity check for header length. - Added sanity check for mmio address got from firmware binary. - kmalloc done separately for dmc header and dmc firmware. (Animesh) v12: Modified as per review comemnts from Imre. - Corrected the typo error in skl stepping info structure. - Added out-of-bound access for skl_stepping_info. - Sanity check for mmio address modified. - Sanity check added for stepping and substeppig. - Modified the intel_dmc_info structure, cache only the required header info. (Animesh) v13: clarify firmware load error message. The reason for a firmware loading failure can be obscure if the driver is built-in. Provide an explanation to the user about the likely reason for the failure and how to resolve it. (Imre) v14: Suggested by Jani. - fix s/I915/CONFIG_DRM_I915/ typo - add fw_path to the firmware object instead of using a static ptr (Jani) v15: 1) Changed the firmware name as dmc_gen9.bin, everytime for a new firmware version a symbolic link with same name will help not to build kernel again. 2) Changes done as per review comments from Imre. - Error check removed for intel_csr_ucode_init. - Moved csr-specific data structure to intel_csr.h and optimization done on structure definition. - fw->data used directly for parsing the header info & memory allocation only done separately for payload. (Animesh) v16: - No need for out_regs label in i915_driver_load(), so removed it. - Changed the firmware name as skl_dmc_ver1.bin, followed naming convention <platform>_dmc_<api-version>.bin (Animesh) Issue: VIZ-2569 Signed-off-by: A.Sunil Kamath <sunil.kamath@intel.com> Signed-off-by: Damien Lespiau <damien.lespiau@intel.com> Signed-off-by: Animesh Manna <animesh.manna@intel.com> Signed-off-by: Imre Deak <imre.deak@intel.com> Reviewed-by: Imre Deak <imre.deak@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2015-05-04 06:58:44 -06:00
const char *fw_path;
uint32_t *dmc_payload;
drm/i915/skl: Add support to load SKL CSR firmware. Display Context Save and Restore support is needed for various SKL Display C states like DC5, DC6. This implementation is added based on first version of DMC CSR program that we received from h/w team. Here we are using request_firmware based design. Finally this firmware should end up in linux-firmware tree. For SKL platform its mandatory to ensure that we load this csr program before enabling DC states like DC5/DC6. As CSR program gets reset on various conditions, we should ensure to load it during boot and in future change to be added to load this system resume sequence too. v1: Initial relese as RFC patch v2: Design change as per Daniel, Damien and Shobit's review comments request firmware method followed. v3: Some optimization and functional changes. Pulled register defines into drivers/gpu/drm/i915/i915_reg.h Used kmemdup to allocate and duplicate firmware content. Ensured to free allocated buffer. v4: Modified as per review comments from Satheesh and Daniel Removed temporary buffer. Optimized number of writes by replacing I915_WRITE with I915_WRITE64. v5: Modified as per review comemnts from Damien. - Changed name for functions and firmware. - Introduced HAS_CSR. - Reverted back previous change and used csr_buf with u8 size. - Using cpu_to_be64 for endianness change. Modified as per review comments from Imre. - Modified registers and macro names to be a bit closer to bspec terminology and the existing register naming in the driver. - Early return for non SKL platforms in intel_load_csr_program function. - Added locking around CSR program load function as it may be called concurrently during system/runtime resume. - Releasing the fw before loading the program for consistency - Handled error path during f/w load. v6: Modified as per review comments from Imre. - Corrected out_freecsr sequence. v7: Modified as per review comments from Imre. Fail loading fw if fw->size%8!=0. v8: Rebase to latest. v9: Rebase on top of -nightly (Damien) v10: Enabled support for dmc firmware ver 1.0. According to ver 1.0 in a single binary package all the firmware's that are required for different stepping's of the product will be stored. The package contains the css header, followed by the package header and the actual dmc firmwares. Package header contains the firmware/stepping mapping table and the corresponding firmware offsets to the individual binaries, within the package. Each individual program binary contains the header and the payload sections whose size is specified in the header section. This changes are done to extract the specific firmaware from the package. (Animesh) v11: Modified as per review comemnts from Imre. - Added code comment from bpec for header structure elements. - Added __packed to avoid structure padding. - Added helper functions for stepping and substepping info. - Added code comment for CSR_MAX_FW_SIZE. - Disabled BXT firmware loading, will be enabled with dmc 1.0 support. - Changed skl_stepping_info based on bspec, earlier used from config DB. - Removed duplicate call of cpu_to_be* from intel_csr_load_program function. - Used cpu_to_be32 instead of cpu_to_be64 as firmware binary in dword aligned. - Added sanity check for header length. - Added sanity check for mmio address got from firmware binary. - kmalloc done separately for dmc header and dmc firmware. (Animesh) v12: Modified as per review comemnts from Imre. - Corrected the typo error in skl stepping info structure. - Added out-of-bound access for skl_stepping_info. - Sanity check for mmio address modified. - Sanity check added for stepping and substeppig. - Modified the intel_dmc_info structure, cache only the required header info. (Animesh) v13: clarify firmware load error message. The reason for a firmware loading failure can be obscure if the driver is built-in. Provide an explanation to the user about the likely reason for the failure and how to resolve it. (Imre) v14: Suggested by Jani. - fix s/I915/CONFIG_DRM_I915/ typo - add fw_path to the firmware object instead of using a static ptr (Jani) v15: 1) Changed the firmware name as dmc_gen9.bin, everytime for a new firmware version a symbolic link with same name will help not to build kernel again. 2) Changes done as per review comments from Imre. - Error check removed for intel_csr_ucode_init. - Moved csr-specific data structure to intel_csr.h and optimization done on structure definition. - fw->data used directly for parsing the header info & memory allocation only done separately for payload. (Animesh) v16: - No need for out_regs label in i915_driver_load(), so removed it. - Changed the firmware name as skl_dmc_ver1.bin, followed naming convention <platform>_dmc_<api-version>.bin (Animesh) Issue: VIZ-2569 Signed-off-by: A.Sunil Kamath <sunil.kamath@intel.com> Signed-off-by: Damien Lespiau <damien.lespiau@intel.com> Signed-off-by: Animesh Manna <animesh.manna@intel.com> Signed-off-by: Imre Deak <imre.deak@intel.com> Reviewed-by: Imre Deak <imre.deak@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2015-05-04 06:58:44 -06:00
uint32_t dmc_fw_size;
uint32_t version;
drm/i915/skl: Add support to load SKL CSR firmware. Display Context Save and Restore support is needed for various SKL Display C states like DC5, DC6. This implementation is added based on first version of DMC CSR program that we received from h/w team. Here we are using request_firmware based design. Finally this firmware should end up in linux-firmware tree. For SKL platform its mandatory to ensure that we load this csr program before enabling DC states like DC5/DC6. As CSR program gets reset on various conditions, we should ensure to load it during boot and in future change to be added to load this system resume sequence too. v1: Initial relese as RFC patch v2: Design change as per Daniel, Damien and Shobit's review comments request firmware method followed. v3: Some optimization and functional changes. Pulled register defines into drivers/gpu/drm/i915/i915_reg.h Used kmemdup to allocate and duplicate firmware content. Ensured to free allocated buffer. v4: Modified as per review comments from Satheesh and Daniel Removed temporary buffer. Optimized number of writes by replacing I915_WRITE with I915_WRITE64. v5: Modified as per review comemnts from Damien. - Changed name for functions and firmware. - Introduced HAS_CSR. - Reverted back previous change and used csr_buf with u8 size. - Using cpu_to_be64 for endianness change. Modified as per review comments from Imre. - Modified registers and macro names to be a bit closer to bspec terminology and the existing register naming in the driver. - Early return for non SKL platforms in intel_load_csr_program function. - Added locking around CSR program load function as it may be called concurrently during system/runtime resume. - Releasing the fw before loading the program for consistency - Handled error path during f/w load. v6: Modified as per review comments from Imre. - Corrected out_freecsr sequence. v7: Modified as per review comments from Imre. Fail loading fw if fw->size%8!=0. v8: Rebase to latest. v9: Rebase on top of -nightly (Damien) v10: Enabled support for dmc firmware ver 1.0. According to ver 1.0 in a single binary package all the firmware's that are required for different stepping's of the product will be stored. The package contains the css header, followed by the package header and the actual dmc firmwares. Package header contains the firmware/stepping mapping table and the corresponding firmware offsets to the individual binaries, within the package. Each individual program binary contains the header and the payload sections whose size is specified in the header section. This changes are done to extract the specific firmaware from the package. (Animesh) v11: Modified as per review comemnts from Imre. - Added code comment from bpec for header structure elements. - Added __packed to avoid structure padding. - Added helper functions for stepping and substepping info. - Added code comment for CSR_MAX_FW_SIZE. - Disabled BXT firmware loading, will be enabled with dmc 1.0 support. - Changed skl_stepping_info based on bspec, earlier used from config DB. - Removed duplicate call of cpu_to_be* from intel_csr_load_program function. - Used cpu_to_be32 instead of cpu_to_be64 as firmware binary in dword aligned. - Added sanity check for header length. - Added sanity check for mmio address got from firmware binary. - kmalloc done separately for dmc header and dmc firmware. (Animesh) v12: Modified as per review comemnts from Imre. - Corrected the typo error in skl stepping info structure. - Added out-of-bound access for skl_stepping_info. - Sanity check for mmio address modified. - Sanity check added for stepping and substeppig. - Modified the intel_dmc_info structure, cache only the required header info. (Animesh) v13: clarify firmware load error message. The reason for a firmware loading failure can be obscure if the driver is built-in. Provide an explanation to the user about the likely reason for the failure and how to resolve it. (Imre) v14: Suggested by Jani. - fix s/I915/CONFIG_DRM_I915/ typo - add fw_path to the firmware object instead of using a static ptr (Jani) v15: 1) Changed the firmware name as dmc_gen9.bin, everytime for a new firmware version a symbolic link with same name will help not to build kernel again. 2) Changes done as per review comments from Imre. - Error check removed for intel_csr_ucode_init. - Moved csr-specific data structure to intel_csr.h and optimization done on structure definition. - fw->data used directly for parsing the header info & memory allocation only done separately for payload. (Animesh) v16: - No need for out_regs label in i915_driver_load(), so removed it. - Changed the firmware name as skl_dmc_ver1.bin, followed naming convention <platform>_dmc_<api-version>.bin (Animesh) Issue: VIZ-2569 Signed-off-by: A.Sunil Kamath <sunil.kamath@intel.com> Signed-off-by: Damien Lespiau <damien.lespiau@intel.com> Signed-off-by: Animesh Manna <animesh.manna@intel.com> Signed-off-by: Imre Deak <imre.deak@intel.com> Reviewed-by: Imre Deak <imre.deak@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2015-05-04 06:58:44 -06:00
uint32_t mmio_count;
drm/i915: Type safe register read/write Make I915_READ and I915_WRITE more type safe by wrapping the register offset in a struct. This should eliminate most of the fumbles we've had with misplaced parens. This only takes care of normal mmio registers. We could extend the idea to other register types and define each with its own struct. That way you wouldn't be able to accidentally pass the wrong thing to a specific register access function. The gpio_reg setup is probably the ugliest thing left. But I figure I'd just leave it for now, and wait for some divine inspiration to strike before making it nice. As for the generated code, it's actually a bit better sometimes. Eg. looking at i915_irq_handler(), we can see the following change: lea 0x70024(%rdx,%rax,1),%r9d mov $0x1,%edx - movslq %r9d,%r9 - mov %r9,%rsi - mov %r9,-0x58(%rbp) - callq *0xd8(%rbx) + mov %r9d,%esi + mov %r9d,-0x48(%rbp) callq *0xd8(%rbx) So previously gcc thought the register offset might be signed and decided to sign extend it, just in case. The rest appears to be mostly just minor shuffling of instructions. v2: i915_mmio_reg_{offset,equal,valid}() helpers added s/_REG/_MMIO/ in the register defines mo more switch statements left to worry about ring_emit stuff got sorted in a prep patch cmd parser, lrc context and w/a batch buildup also in prep patch vgpu stuff cleaned up and moved to a prep patch all other unrelated changes split out v3: Rebased due to BXT DSI/BLC, MOCS, etc. v4: Rebased due to churn, s/i915_mmio_reg_t/i915_reg_t/ Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Link: http://patchwork.freedesktop.org/patch/msgid/1447853606-2751-1-git-send-email-ville.syrjala@linux.intel.com
2015-11-18 06:33:26 -07:00
i915_reg_t mmioaddr[8];
drm/i915/skl: Add support to load SKL CSR firmware. Display Context Save and Restore support is needed for various SKL Display C states like DC5, DC6. This implementation is added based on first version of DMC CSR program that we received from h/w team. Here we are using request_firmware based design. Finally this firmware should end up in linux-firmware tree. For SKL platform its mandatory to ensure that we load this csr program before enabling DC states like DC5/DC6. As CSR program gets reset on various conditions, we should ensure to load it during boot and in future change to be added to load this system resume sequence too. v1: Initial relese as RFC patch v2: Design change as per Daniel, Damien and Shobit's review comments request firmware method followed. v3: Some optimization and functional changes. Pulled register defines into drivers/gpu/drm/i915/i915_reg.h Used kmemdup to allocate and duplicate firmware content. Ensured to free allocated buffer. v4: Modified as per review comments from Satheesh and Daniel Removed temporary buffer. Optimized number of writes by replacing I915_WRITE with I915_WRITE64. v5: Modified as per review comemnts from Damien. - Changed name for functions and firmware. - Introduced HAS_CSR. - Reverted back previous change and used csr_buf with u8 size. - Using cpu_to_be64 for endianness change. Modified as per review comments from Imre. - Modified registers and macro names to be a bit closer to bspec terminology and the existing register naming in the driver. - Early return for non SKL platforms in intel_load_csr_program function. - Added locking around CSR program load function as it may be called concurrently during system/runtime resume. - Releasing the fw before loading the program for consistency - Handled error path during f/w load. v6: Modified as per review comments from Imre. - Corrected out_freecsr sequence. v7: Modified as per review comments from Imre. Fail loading fw if fw->size%8!=0. v8: Rebase to latest. v9: Rebase on top of -nightly (Damien) v10: Enabled support for dmc firmware ver 1.0. According to ver 1.0 in a single binary package all the firmware's that are required for different stepping's of the product will be stored. The package contains the css header, followed by the package header and the actual dmc firmwares. Package header contains the firmware/stepping mapping table and the corresponding firmware offsets to the individual binaries, within the package. Each individual program binary contains the header and the payload sections whose size is specified in the header section. This changes are done to extract the specific firmaware from the package. (Animesh) v11: Modified as per review comemnts from Imre. - Added code comment from bpec for header structure elements. - Added __packed to avoid structure padding. - Added helper functions for stepping and substepping info. - Added code comment for CSR_MAX_FW_SIZE. - Disabled BXT firmware loading, will be enabled with dmc 1.0 support. - Changed skl_stepping_info based on bspec, earlier used from config DB. - Removed duplicate call of cpu_to_be* from intel_csr_load_program function. - Used cpu_to_be32 instead of cpu_to_be64 as firmware binary in dword aligned. - Added sanity check for header length. - Added sanity check for mmio address got from firmware binary. - kmalloc done separately for dmc header and dmc firmware. (Animesh) v12: Modified as per review comemnts from Imre. - Corrected the typo error in skl stepping info structure. - Added out-of-bound access for skl_stepping_info. - Sanity check for mmio address modified. - Sanity check added for stepping and substeppig. - Modified the intel_dmc_info structure, cache only the required header info. (Animesh) v13: clarify firmware load error message. The reason for a firmware loading failure can be obscure if the driver is built-in. Provide an explanation to the user about the likely reason for the failure and how to resolve it. (Imre) v14: Suggested by Jani. - fix s/I915/CONFIG_DRM_I915/ typo - add fw_path to the firmware object instead of using a static ptr (Jani) v15: 1) Changed the firmware name as dmc_gen9.bin, everytime for a new firmware version a symbolic link with same name will help not to build kernel again. 2) Changes done as per review comments from Imre. - Error check removed for intel_csr_ucode_init. - Moved csr-specific data structure to intel_csr.h and optimization done on structure definition. - fw->data used directly for parsing the header info & memory allocation only done separately for payload. (Animesh) v16: - No need for out_regs label in i915_driver_load(), so removed it. - Changed the firmware name as skl_dmc_ver1.bin, followed naming convention <platform>_dmc_<api-version>.bin (Animesh) Issue: VIZ-2569 Signed-off-by: A.Sunil Kamath <sunil.kamath@intel.com> Signed-off-by: Damien Lespiau <damien.lespiau@intel.com> Signed-off-by: Animesh Manna <animesh.manna@intel.com> Signed-off-by: Imre Deak <imre.deak@intel.com> Reviewed-by: Imre Deak <imre.deak@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2015-05-04 06:58:44 -06:00
uint32_t mmiodata[8];
uint32_t dc_state;
uint32_t allowed_dc_mask;
drm/i915/skl: Add support to load SKL CSR firmware. Display Context Save and Restore support is needed for various SKL Display C states like DC5, DC6. This implementation is added based on first version of DMC CSR program that we received from h/w team. Here we are using request_firmware based design. Finally this firmware should end up in linux-firmware tree. For SKL platform its mandatory to ensure that we load this csr program before enabling DC states like DC5/DC6. As CSR program gets reset on various conditions, we should ensure to load it during boot and in future change to be added to load this system resume sequence too. v1: Initial relese as RFC patch v2: Design change as per Daniel, Damien and Shobit's review comments request firmware method followed. v3: Some optimization and functional changes. Pulled register defines into drivers/gpu/drm/i915/i915_reg.h Used kmemdup to allocate and duplicate firmware content. Ensured to free allocated buffer. v4: Modified as per review comments from Satheesh and Daniel Removed temporary buffer. Optimized number of writes by replacing I915_WRITE with I915_WRITE64. v5: Modified as per review comemnts from Damien. - Changed name for functions and firmware. - Introduced HAS_CSR. - Reverted back previous change and used csr_buf with u8 size. - Using cpu_to_be64 for endianness change. Modified as per review comments from Imre. - Modified registers and macro names to be a bit closer to bspec terminology and the existing register naming in the driver. - Early return for non SKL platforms in intel_load_csr_program function. - Added locking around CSR program load function as it may be called concurrently during system/runtime resume. - Releasing the fw before loading the program for consistency - Handled error path during f/w load. v6: Modified as per review comments from Imre. - Corrected out_freecsr sequence. v7: Modified as per review comments from Imre. Fail loading fw if fw->size%8!=0. v8: Rebase to latest. v9: Rebase on top of -nightly (Damien) v10: Enabled support for dmc firmware ver 1.0. According to ver 1.0 in a single binary package all the firmware's that are required for different stepping's of the product will be stored. The package contains the css header, followed by the package header and the actual dmc firmwares. Package header contains the firmware/stepping mapping table and the corresponding firmware offsets to the individual binaries, within the package. Each individual program binary contains the header and the payload sections whose size is specified in the header section. This changes are done to extract the specific firmaware from the package. (Animesh) v11: Modified as per review comemnts from Imre. - Added code comment from bpec for header structure elements. - Added __packed to avoid structure padding. - Added helper functions for stepping and substepping info. - Added code comment for CSR_MAX_FW_SIZE. - Disabled BXT firmware loading, will be enabled with dmc 1.0 support. - Changed skl_stepping_info based on bspec, earlier used from config DB. - Removed duplicate call of cpu_to_be* from intel_csr_load_program function. - Used cpu_to_be32 instead of cpu_to_be64 as firmware binary in dword aligned. - Added sanity check for header length. - Added sanity check for mmio address got from firmware binary. - kmalloc done separately for dmc header and dmc firmware. (Animesh) v12: Modified as per review comemnts from Imre. - Corrected the typo error in skl stepping info structure. - Added out-of-bound access for skl_stepping_info. - Sanity check for mmio address modified. - Sanity check added for stepping and substeppig. - Modified the intel_dmc_info structure, cache only the required header info. (Animesh) v13: clarify firmware load error message. The reason for a firmware loading failure can be obscure if the driver is built-in. Provide an explanation to the user about the likely reason for the failure and how to resolve it. (Imre) v14: Suggested by Jani. - fix s/I915/CONFIG_DRM_I915/ typo - add fw_path to the firmware object instead of using a static ptr (Jani) v15: 1) Changed the firmware name as dmc_gen9.bin, everytime for a new firmware version a symbolic link with same name will help not to build kernel again. 2) Changes done as per review comments from Imre. - Error check removed for intel_csr_ucode_init. - Moved csr-specific data structure to intel_csr.h and optimization done on structure definition. - fw->data used directly for parsing the header info & memory allocation only done separately for payload. (Animesh) v16: - No need for out_regs label in i915_driver_load(), so removed it. - Changed the firmware name as skl_dmc_ver1.bin, followed naming convention <platform>_dmc_<api-version>.bin (Animesh) Issue: VIZ-2569 Signed-off-by: A.Sunil Kamath <sunil.kamath@intel.com> Signed-off-by: Damien Lespiau <damien.lespiau@intel.com> Signed-off-by: Animesh Manna <animesh.manna@intel.com> Signed-off-by: Imre Deak <imre.deak@intel.com> Reviewed-by: Imre Deak <imre.deak@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2015-05-04 06:58:44 -06:00
};
#define DEV_INFO_FOR_EACH_FLAG(func) \
func(is_mobile); \
func(is_lp); \
func(is_alpha_support); \
/* Keep has_* in alphabetical order */ \
func(has_64bit_reloc); \
func(has_aliasing_ppgtt); \
func(has_csr); \
func(has_ddi); \
func(has_dp_mst); \
drm/i915: Modify error handler for per engine hang recovery This is a preparatory patch which modifies error handler to do per engine hang recovery. The actual patch which implements this sequence follows later in the series. The aim is to prepare existing recovery function to adapt to this new function where applicable (which fails at this point because core implementation is lacking) and continue recovery using legacy full gpu reset. A helper function is also added to query the availability of engine reset. A subsequent patch will add the capability to query which type of reset is present (engine -> full -> no-reset) via the get-param ioctl. It has been decided that the error events that are used to notify user of reset will only be sent in case if full chip reset. In case of just single (or multiple) engine resets, userspace won't be notified by these events. Note that this implementation of engine reset is for i915 directly submitting to the ELSP, where the driver manages the hang detection, recovery and resubmission. With GuC submission these tasks are shared between driver and firmware; i915 will still responsible for detecting a hang, and when it does it will have to request GuC to reset that Engine and remind the firmware about the outstanding submissions. This will be added in different patch. v2: rebase, advertise engine reset availability in platform definition, add note about GuC submission. v3: s/*engine_reset*/*reset_engine*/. (Chris) Handle reset as 2 level resets, by first going to engine only and fall backing to full/chip reset as needed, i.e. reset_engine will need the struct_mutex. v4: Pass the engine mask to i915_reset. (Chris) v5: Rebase, update selftests. v6: Rebase, prepare for mutex-less reset engine. v7: Pass reset_engine mask as a function parameter, and iterate over the engine mask for reset_engine. (Chris) v8: Use i915.reset >=2 in has_reset_engine; remove redundant reset logging; add a reset-engine-in-progress flag to prevent concurrent resets, and avoid dual purposing of reset-backoff. (Chris) v9: Support reset of different engines in parallel (Chris) v10: Handle reset-engine flag locking better (Chris) v11: Squash in reporting of per-engine-reset availability. Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com> Signed-off-by: Ian Lister <ian.lister@intel.com> Signed-off-by: Tomas Elf <tomas.elf@intel.com> Signed-off-by: Arun Siluvery <arun.siluvery@linux.intel.com> Signed-off-by: Michel Thierry <michel.thierry@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/20170615201828.23144-4-michel.thierry@intel.com Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Link: http://patchwork.freedesktop.org/patch/msgid/20170620095751.13127-5-chris@chris-wilson.co.uk
2017-06-20 03:57:46 -06:00
func(has_reset_engine); \
func(has_fbc); \
func(has_fpga_dbg); \
func(has_full_ppgtt); \
func(has_full_48bit_ppgtt); \
func(has_gmbus_irq); \
func(has_gmch_display); \
func(has_guc); \
func(has_guc_ct); \
func(has_hotplug); \
func(has_l3_dpf); \
func(has_llc); \
func(has_logical_ring_contexts); \
func(has_overlay); \
func(has_pipe_cxsr); \
func(has_pooled_eu); \
func(has_psr); \
func(has_rc6); \
func(has_rc6p); \
func(has_resource_streamer); \
func(has_runtime_pm); \
func(has_snoop); \
func(unfenced_needs_alignment); \
func(cursor_needs_physical); \
func(hws_needs_physical); \
func(overlay_needs_physical); \
func(supports_tv);
struct sseu_dev_info {
u8 slice_mask;
u8 subslice_mask;
u8 eu_total;
u8 eu_per_subslice;
u8 min_eu_in_pool;
/* For each slice, which subslice(s) has(have) 7 EUs (bitfield)? */
u8 subslice_7eu[3];
u8 has_slice_pg:1;
u8 has_subslice_pg:1;
u8 has_eu_pg:1;
};
static inline unsigned int sseu_subslice_total(const struct sseu_dev_info *sseu)
{
return hweight8(sseu->slice_mask) * hweight8(sseu->subslice_mask);
}
/* Keep in gen based order, and chronological order within a gen */
enum intel_platform {
INTEL_PLATFORM_UNINITIALIZED = 0,
INTEL_I830,
INTEL_I845G,
INTEL_I85X,
INTEL_I865G,
INTEL_I915G,
INTEL_I915GM,
INTEL_I945G,
INTEL_I945GM,
INTEL_G33,
INTEL_PINEVIEW,
INTEL_I965G,
INTEL_I965GM,
INTEL_G45,
INTEL_GM45,
INTEL_IRONLAKE,
INTEL_SANDYBRIDGE,
INTEL_IVYBRIDGE,
INTEL_VALLEYVIEW,
INTEL_HASWELL,
INTEL_BROADWELL,
INTEL_CHERRYVIEW,
INTEL_SKYLAKE,
INTEL_BROXTON,
INTEL_KABYLAKE,
INTEL_GEMINILAKE,
INTEL_COFFEELAKE,
INTEL_CANNONLAKE,
INTEL_MAX_PLATFORMS
};
struct intel_device_info {
u32 display_mmio_offset;
u16 device_id;
u8 num_pipes;
u8 num_sprites[I915_MAX_PIPES];
u8 num_scalers[I915_MAX_PIPES];
u8 gen;
u16 gen_mask;
enum intel_platform platform;
u8 ring_mask; /* Rings supported by the HW */
u8 num_rings;
#define DEFINE_FLAG(name) u8 name:1
DEV_INFO_FOR_EACH_FLAG(DEFINE_FLAG);
#undef DEFINE_FLAG
u16 ddb_size; /* in blocks */
/* Register offsets for the various display pipes and transcoders */
int pipe_offsets[I915_MAX_TRANSCODERS];
int trans_offsets[I915_MAX_TRANSCODERS];
int palette_offsets[I915_MAX_PIPES];
int cursor_offsets[I915_MAX_PIPES];
/* Slice/subslice/EU info */
struct sseu_dev_info sseu;
struct color_luts {
u16 degamma_lut_size;
u16 gamma_lut_size;
} color;
};
struct intel_display_error_state;
struct i915_gpu_state {
struct kref ref;
struct timeval time;
struct timeval boottime;
struct timeval uptime;
2016-10-12 03:05:19 -06:00
struct drm_i915_private *i915;
char error_msg[128];
bool simulated;
bool awake;
bool wakelock;
bool suspended;
int iommu;
u32 reset_count;
u32 suspend_count;
struct intel_device_info device_info;
struct i915_params params;
/* Generic register state */
u32 eir;
u32 pgtbl_er;
u32 ier;
u32 gtier[4], ngtier;
u32 ccid;
u32 derrmr;
u32 forcewake;
u32 error; /* gen6+ */
u32 err_int; /* gen7 */
u32 fault_data0; /* gen8, gen9 */
u32 fault_data1; /* gen8, gen9 */
u32 done_reg;
u32 gac_eco;
u32 gam_ecochk;
u32 gab_ctl;
u32 gfx_mode;
u32 nfence;
u64 fence[I915_MAX_NUM_FENCES];
struct intel_overlay_error_state *overlay;
struct intel_display_error_state *display;
struct drm_i915_error_object *semaphore;
struct drm_i915_error_object *guc_log;
struct drm_i915_error_engine {
int engine_id;
/* Software tracked state */
bool waiting;
int num_waiters;
drm/i915: Decouple hang detection from hangcheck period Hangcheck state accumulation has gained more steps along the years, like head movement and more recently the subunit inactivity check. As the subunit sampling is only done if the previous state check showed inactivity, we have added more stages (and time) to reach a hang verdict. Asymmetric engine states led to different actual weight of 'one hangcheck unit' and it was demonstrated in some hangs that due to difference in stages, simpler engines were accused falsely of a hang as their scoring was much more quicker to accumulate above the hang treshold. To completely decouple the hangcheck guilty score from the hangcheck period, convert hangcheck score to a rough period of inactivity measurement. As these are tracked as jiffies, they are meaningful also across reset boundaries. This makes finding a guilty engine more accurate across multi engine activity scenarios, especially across asymmetric engines. We lose the ability to detect cross batch malicious attempts to hinder the progress. Plan is to move this functionality to be part of context banning which is more natural fit, later in the series. v2: use time_before macros (Chris) reinstate the pardoning of moving engine after hc (Chris) v3: avoid global state for per engine stall detection (Chris) v4: take timeline last retirement into account (Chris) v5: do debug print on pardoning, split out retirement timestamp (Chris) Cc: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
2016-11-18 06:09:04 -07:00
unsigned long hangcheck_timestamp;
bool hangcheck_stalled;
enum intel_engine_hangcheck_action hangcheck_action;
struct i915_address_space *vm;
int num_requests;
u32 reset_count;
/* position of active request inside the ring */
u32 rq_head, rq_post, rq_tail;
/* our own tracking of ring head and tail */
u32 cpu_ring_head;
u32 cpu_ring_tail;
u32 last_seqno;
/* Register state */
u32 start;
u32 tail;
u32 head;
u32 ctl;
u32 mode;
u32 hws;
u32 ipeir;
u32 ipehr;
u32 bbstate;
u32 instpm;
u32 instps;
u32 seqno;
u64 bbaddr;
u64 acthd;
u32 fault_reg;
u64 faddr;
u32 rc_psmi; /* sleep state */
u32 semaphore_mboxes[I915_NUM_ENGINES - 1];
struct intel_instdone instdone;
struct drm_i915_error_context {
char comm[TASK_COMM_LEN];
pid_t pid;
u32 handle;
u32 hw_id;
int ban_score;
int active;
int guilty;
} context;
struct drm_i915_error_object {
u64 gtt_offset;
u64 gtt_size;
int page_count;
int unused;
u32 *pages[0];
} *ringbuffer, *batchbuffer, *wa_batchbuffer, *ctx, *hws_page;
struct drm_i915_error_object **user_bo;
long user_bo_count;
struct drm_i915_error_object *wa_ctx;
struct drm_i915_error_request {
long jiffies;
pid_t pid;
u32 context;
int ban_score;
u32 seqno;
u32 head;
u32 tail;
} *requests, execlist[2];
struct drm_i915_error_waiter {
char comm[TASK_COMM_LEN];
pid_t pid;
u32 seqno;
} *waiters;
struct {
u32 gfx_mode;
union {
u64 pdp[4];
u32 pp_dir_base;
};
} vm_info;
} engine[I915_NUM_ENGINES];
struct drm_i915_error_buffer {
u32 size;
u32 name;
u32 rseqno[I915_NUM_ENGINES], wseqno;
u64 gtt_offset;
u32 read_domains;
u32 write_domain;
s32 fence_reg:I915_MAX_NUM_FENCE_BITS;
u32 tiling:2;
u32 dirty:1;
u32 purgeable:1;
u32 userptr:1;
s32 engine:4;
u32 cache_level:3;
} *active_bo[I915_NUM_ENGINES], *pinned_bo;
u32 active_bo_count[I915_NUM_ENGINES], pinned_bo_count;
struct i915_address_space *active_vm[I915_NUM_ENGINES];
};
enum i915_cache_level {
I915_CACHE_NONE = 0,
I915_CACHE_LLC, /* also used for snoopable memory on non-LLC */
I915_CACHE_L3_LLC, /* gen7+, L3 sits between the domain specifc
caches, eg sampler/render caches, and the
large Last-Level-Cache. LLC is coherent with
the CPU, but L3 is only visible to the GPU. */
I915_CACHE_WT, /* hsw:gt3e WriteThrough for scanouts */
};
#define I915_COLOR_UNEVICTABLE (-1) /* a non-vma sharing the address space */
enum fb_op_origin {
ORIGIN_GTT,
ORIGIN_CPU,
ORIGIN_CS,
ORIGIN_FLIP,
ORIGIN_DIRTYFB,
};
struct intel_fbc {
/* This is always the inner lock when overlapping with struct_mutex and
* it's the outer lock when overlapping with stolen_lock. */
struct mutex lock;
unsigned threshold;
unsigned int possible_framebuffer_bits;
unsigned int busy_bits;
unsigned int visible_pipes_mask;
struct intel_crtc *crtc;
struct drm_mm_node compressed_fb;
struct drm_mm_node *compressed_llb;
bool false_color;
drm/i915: introduce intel_fbc_{enable,disable} The goal is to call FBC enable/disable only once per modeset, while activate/deactivate/update will be called multiple times. The enable() function will be responsible for deciding if a CRTC will have FBC on it and then it will "lock" FBC on this CRTC: it won't be possible to change FBC's CRTC until disable(). With this, all checks and resource acquisition that only need to be done once per modeset can be moved from update() to enable(). And then the update(), activate() and deactivate() code will also get simpler since they won't need to worry about the CRTC being changed. The disable() function will do the reverse operation of enable(). One of its features is that it should only be called while the pipe is already off. This guarantees that FBC is stopped and nothing is using the CFB. With this, the activate() and deactivate() functions just start and temporarily stop FBC. They are the ones touching the hardware enable bit, so HW state reflects dev_priv->crtc.active. The last function remaining is update(). A lot of times I thought about renaming update() to activate() or try_to_activate() since it's called when we want to activate FBC. The thing is that update() may not only decide to activate FBC, but also deactivate or keep it on the same state, so I'll leave this name for now. Moving code to enable() and disable() will also help in case we decide to move FBC to pipe_config or something else later. The current patch only puts the very basic code on enable() and disable(). The next commits will take care of moving more stuff from update() to the new functions. v2: - Rebase. - Improve commit message (Chris). v3: Rebase after changing the patch order. v4: Rebase again after upstream changes. Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/
2015-10-15 07:44:46 -06:00
bool enabled;
bool active;
drm/i915/fbc: disable FBC on FIFO underruns Ever since I started working on FBC I was already aware that FBC can really amplify the FIFO underrun symptoms. On systems where FIFO underruns were harmless error messages, enabling FBC would cause the underruns to give black screens. We recently tried to enable FBC on Haswell and got reports of a system that would hang after some hours of uptime, and the first bad commit was the one that enabled FBC. We also observed that this system had FIFO underrun error messages on its dmesg. Although we don't have any evidence that fixing the underruns would solve the bug and make FBC work properly on this machine, IMHO it's better if we minimize the amount of possible problems by just giving up FBC whenever we detect an underrun. v2: New version, different implementation and commit message. v3: Clarify the fact that we run from an IRQ handler (Chris). v4: Also add the underrun_detected check at can_choose() to avoid misleading dmesg messages (DK). v5: Fix Engrish, use READ_ONCE on the unlocked read (Chris). Cc: Stefan Richter <stefanr@s5r6.in-berlin.de> Cc: Lyude <cpaul@redhat.com> Cc: stevenhoneyman@gmail.com <stevenhoneyman@gmail.com> Cc: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1473773937-19758-1-git-send-email-paulo.r.zanoni@intel.com
2016-09-13 07:38:57 -06:00
bool underrun_detected;
struct work_struct underrun_work;
/*
* Due to the atomic rules we can't access some structures without the
* appropriate locking, so we cache information here in order to avoid
* these problems.
*/
struct intel_fbc_state_cache {
struct i915_vma *vma;
struct {
unsigned int mode_flags;
uint32_t hsw_bdw_pixel_rate;
} crtc;
struct {
unsigned int rotation;
int src_w;
int src_h;
bool visible;
} plane;
struct {
const struct drm_format_info *format;
unsigned int stride;
} fb;
} state_cache;
/*
* This structure contains everything that's relevant to program the
* hardware registers. When we want to figure out if we need to disable
* and re-enable FBC for a new configuration we just check if there's
* something different in the struct. The genx_fbc_activate functions
* are supposed to read from it in order to program the registers.
*/
struct intel_fbc_reg_params {
struct i915_vma *vma;
struct {
enum pipe pipe;
enum plane plane;
unsigned int fence_y_offset;
} crtc;
struct {
const struct drm_format_info *format;
unsigned int stride;
} fb;
int cfb_size;
} params;
struct intel_fbc_work {
drm/i915: use a single intel_fbc_work struct This was already on my TODO list, and was requested both by Chris and Ville, for different reasons. The advantages are avoiding a frequent malloc/free pair, and the locality of having the work structure embedded in dev_priv. The maximum used memory is also smaller since previously we could have multiple allocated intel_fbc_work structs at the same time, and now we'll always have a single one - the one embedded on dev_priv. Of course, we're now using a little more memory on the cases where there's nothing scheduled. The biggest challenge here is to keep everything synchronized the way it was before. Currently, when we try to activate FBC, we allocate a new intel_fbc_work structure. Then later when we conclude we must delay the FBC activation a little more, we allocate a new intel_fbc_work struct, and then adjust dev_priv->fbc.fbc_work to point to the new struct. So when the old work runs - at intel_fbc_work_fn() - it will check that dev_priv->fbc.fbc_work points to something else, so it does nothing. Everything is also protected by fbc.lock. Just cancelling the old delayed work doesn't work because we might just cancel it after the work function already started to run, but while it is still waiting to grab fbc.lock. That's why we use the "dev_priv->fbc.fbc_work == work" check described in the paragraph above. So now that we have a single work struct we have to introduce a new way to synchronize everything. So we're making the work function a normal work instead of a delayed work, and it will be responsible for sleeping the appropriate amount of time itself. This way, after it wakes up it can grab the lock, ask "were we delayed or cancelled?" and then go back to sleep, enable FBC or give up. v2: - Spelling fixes. - Rebase after changing the patch order. - Fix ms/jiffies confusion. Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> (v1) Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/
2015-10-26 12:27:49 -06:00
bool scheduled;
u32 scheduled_vblank;
drm/i915: use a single intel_fbc_work struct This was already on my TODO list, and was requested both by Chris and Ville, for different reasons. The advantages are avoiding a frequent malloc/free pair, and the locality of having the work structure embedded in dev_priv. The maximum used memory is also smaller since previously we could have multiple allocated intel_fbc_work structs at the same time, and now we'll always have a single one - the one embedded on dev_priv. Of course, we're now using a little more memory on the cases where there's nothing scheduled. The biggest challenge here is to keep everything synchronized the way it was before. Currently, when we try to activate FBC, we allocate a new intel_fbc_work structure. Then later when we conclude we must delay the FBC activation a little more, we allocate a new intel_fbc_work struct, and then adjust dev_priv->fbc.fbc_work to point to the new struct. So when the old work runs - at intel_fbc_work_fn() - it will check that dev_priv->fbc.fbc_work points to something else, so it does nothing. Everything is also protected by fbc.lock. Just cancelling the old delayed work doesn't work because we might just cancel it after the work function already started to run, but while it is still waiting to grab fbc.lock. That's why we use the "dev_priv->fbc.fbc_work == work" check described in the paragraph above. So now that we have a single work struct we have to introduce a new way to synchronize everything. So we're making the work function a normal work instead of a delayed work, and it will be responsible for sleeping the appropriate amount of time itself. This way, after it wakes up it can grab the lock, ask "were we delayed or cancelled?" and then go back to sleep, enable FBC or give up. v2: - Spelling fixes. - Rebase after changing the patch order. - Fix ms/jiffies confusion. Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> (v1) Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/
2015-10-26 12:27:49 -06:00
struct work_struct work;
} work;
const char *no_fbc_reason;
};
/*
* HIGH_RR is the highest eDP panel refresh rate read from EDID
* LOW_RR is the lowest eDP panel refresh rate found from EDID
* parsing for same resolution.
*/
enum drrs_refresh_rate_type {
DRRS_HIGH_RR,
DRRS_LOW_RR,
DRRS_MAX_RR, /* RR count */
};
enum drrs_support_type {
DRRS_NOT_SUPPORTED = 0,
STATIC_DRRS_SUPPORT = 1,
SEAMLESS_DRRS_SUPPORT = 2
drm/i915: Add support for DRRS to switch RR This patch computes and stored 2nd M/N/TU for switching to different refresh rate dynamically. PIPECONF_EDP_RR_MODE_SWITCH bit helps toggle between alternate refresh rates programmed in 2nd M/N/TU registers. v2: Daniel's review comments Computing M2/N2 in compute_config and storing it in crtc_config v3: Modified reference to edp_downclock and edp_downclock_avail based on the changes made to move them from dev_private to intel_panel. v4: Modified references to is_drrs_supported based on the changes made to rename it to drrs_support. v5: Jani's review comments Removed superfluous return statements. Changed support for Gen 7 and above. Corrected indentation. Re-structured the code which finds crtc and connector from encoder. Changed some logs to be less verbose. v6: Modifying i915_drrs to include only intel connector as intel_dp can be derived from intel connector when required. v7: As per internal review comments, acquiring mutex just before accessing drrs RR. As per Chris's review comments, added documentation about the use of locking in the function. v8: Incorporated Jani's review comments. Removed reference to edp_downclock. v9: Jani's review comments. Modified comment in set_drrs. Changed index to type edp_drrs_refresh_rate_type. Check if PSR is enabled before setting registers fo DRRS. Signed-off-by: Pradeep Bhat <pradeep.bhat@intel.com> Signed-off-by: Vandana Kannan <vandana.kannan@intel.com> Cc: Jani Nikula <jani.nikula@linux.intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2014-04-05 00:43:28 -06:00
};
struct intel_dp;
struct i915_drrs {
struct mutex mutex;
struct delayed_work work;
struct intel_dp *dp;
unsigned busy_frontbuffer_bits;
enum drrs_refresh_rate_type refresh_rate_type;
enum drrs_support_type type;
};
struct i915_psr {
struct mutex lock;
bool sink_support;
bool source_ok;
struct intel_dp *enabled;
bool active;
struct delayed_work work;
drm/i915: Fix up PSR frontbuffer tracking I've tried to split this up, but all the changes are so tightly related that I didn't find a good way to do this without breaking bisecting. Essentially this completely changes how psr is glued into the overall driver, and there's not much you can do to soften such a paradigm change. - Use frontbuffer tracking bits stuff to separate disable and re-enable. - Don't re-check everything in the psr work. We have now accurate tracking for everything, so no need to check for sprites or tiling really. Allows us to ditch tons of locks. - That in turn allows us to properly cancel the work in the disable function - no more deadlocks. - Add a check for HSW sprites and force a flush. Apparently the hardware doesn't forward the flushing when updating the sprite base address. We can do the same trick everywhere else we have such issues, e.g. on baytrail with ... everything. - Don't re-enable psr with a delay in psr_exit. It really must be turned off forever if we detect a gtt write. At least with the current frontbuffer render tracking. Userspace can do a busy ioctl call or no-op pageflip to re-enable psr. - Drop redundant checks for crtc and crtc->active - now that they're only called from enable this is guaranteed. - Fix up the hsw port check. eDP can also happen on port D, but the issue is exactly that it doesn't work there. So an || check is wrong. - We still schedule the psr work with a delay. The frontbuffer flushing interface mandates that we upload the next full frame, so need to wait a bit. Once we have single-shot frame uploads we can do better here. v2: Don't enable psr initially, rely upon the fb flush of the initial plane setup for that. Gives us more unified code flow and makes the crtc enable sequence less a special case. v3: s/psr_exit/psr_invalidate/ for consistency v4: Fixup whitespace. v5: Correctly bail out of psr_invalidate/flush when dev_priv->psr.enabled is NULL. Spotted by Rodrigo. v6: - Only schedule work when there's work to do. Fixes WARNINGs reported by Rodrigo. - Comments Chris requested to clarify the code. v7: Fix conflict on rebase (Rodrigo) Cc: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> (v6) Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2014-07-11 11:30:16 -06:00
unsigned busy_frontbuffer_bits;
bool psr2_support;
bool aux_frame_sync;
bool link_standby;
bool y_cord_support;
bool colorimetry_support;
bool alpm;
};
enum intel_pch {
drm/i915: add PCH_NONE to enum intel_pch And rely on the fact that it's 0 to assume that machines without a PCH will have PCH_NONE as dev_priv->pch_type. Just today I finally realized that HAS_PCH_IBX is true for machines without a PCH. IMHO this is totally counter-intuitive and I don't think it's a good idea to assume that we're going to check for HAS_PCH_IBX only after we check for HAS_PCH_SPLIT. I believe that in the future we'll have more PCH types and checks like: if (HAS_PCH_IBX(dev) || HAS_PCH_CPT(dev)) will become more and more common. There's a good chance that we may break non-PCH machines by adding these checks in code that runs on all machines. I also believe that the HAS_PCH_SPLIT check will become less common as we add more and more different PCH types. We'll probably start replacing checks like: if (HAS_PCH_SPLIT(dev)) foo(); else bar(); with: if (HAS_PCH_NEW(dev)) baz(); else if (HAS_PCH_OLD(dev) || HAS_PCH_IBX(dev)) foo(); else bar(); and this may break gen 2/3/4. As far as we have investigated, this patch will affect the behavior of intel_hdmi_dpms and intel_dp_link_down on gen 4. In both functions the code inside the HAS_PCH_IBX check is for IBX-specific workarounds, so we should be safe. If we start bisecting gen 2/3/4 bugs to this commit we should consider replacing the HAS_PCH_IBX checks with something else. V2: Improve commit message, list possible side effects and solution. Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2012-07-03 15:48:16 -06:00
PCH_NONE = 0, /* No PCH present */
PCH_IBX, /* Ibexpeak PCH */
PCH_CPT, /* Cougarpoint/Pantherpoint PCH */
PCH_LPT, /* Lynxpoint/Wildcatpoint PCH */
PCH_SPT, /* Sunrisepoint PCH */
PCH_KBP, /* Kaby Lake PCH */
PCH_CNP, /* Cannon Lake PCH */
PCH_NOP,
};
enum intel_sbi_destination {
SBI_ICLK,
SBI_MPHY,
};
#define QUIRK_LVDS_SSC_DISABLE (1<<1)
#define QUIRK_INVERT_BRIGHTNESS (1<<2)
#define QUIRK_BACKLIGHT_PRESENT (1<<3)
#define QUIRK_PIN_SWIZZLED_PAGES (1<<5)
#define QUIRK_INCREASE_T12_DELAY (1<<6)
struct intel_fbdev;
struct intel_fbc_work;
struct intel_gmbus {
struct i2c_adapter adapter;
#define GMBUS_FORCE_BIT_RETRY (1U << 31)
u32 force_bit;
u32 reg0;
drm/i915: Type safe register read/write Make I915_READ and I915_WRITE more type safe by wrapping the register offset in a struct. This should eliminate most of the fumbles we've had with misplaced parens. This only takes care of normal mmio registers. We could extend the idea to other register types and define each with its own struct. That way you wouldn't be able to accidentally pass the wrong thing to a specific register access function. The gpio_reg setup is probably the ugliest thing left. But I figure I'd just leave it for now, and wait for some divine inspiration to strike before making it nice. As for the generated code, it's actually a bit better sometimes. Eg. looking at i915_irq_handler(), we can see the following change: lea 0x70024(%rdx,%rax,1),%r9d mov $0x1,%edx - movslq %r9d,%r9 - mov %r9,%rsi - mov %r9,-0x58(%rbp) - callq *0xd8(%rbx) + mov %r9d,%esi + mov %r9d,-0x48(%rbp) callq *0xd8(%rbx) So previously gcc thought the register offset might be signed and decided to sign extend it, just in case. The rest appears to be mostly just minor shuffling of instructions. v2: i915_mmio_reg_{offset,equal,valid}() helpers added s/_REG/_MMIO/ in the register defines mo more switch statements left to worry about ring_emit stuff got sorted in a prep patch cmd parser, lrc context and w/a batch buildup also in prep patch vgpu stuff cleaned up and moved to a prep patch all other unrelated changes split out v3: Rebased due to BXT DSI/BLC, MOCS, etc. v4: Rebased due to churn, s/i915_mmio_reg_t/i915_reg_t/ Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Link: http://patchwork.freedesktop.org/patch/msgid/1447853606-2751-1-git-send-email-ville.syrjala@linux.intel.com
2015-11-18 06:33:26 -07:00
i915_reg_t gpio_reg;
struct i2c_algo_bit_data bit_algo;
struct drm_i915_private *dev_priv;
};
struct i915_suspend_saved_registers {
u32 saveDSPARB;
u32 saveFBC_CONTROL;
u32 saveCACHE_MODE_0;
u32 saveMI_ARB_STATE;
u32 saveSWF0[16];
u32 saveSWF1[16];
u32 saveSWF3[3];
uint64_t saveFENCE[I915_MAX_NUM_FENCES];
u32 savePCH_PORT_HOTPLUG;
u16 saveGCDGMBUS;
};
struct vlv_s0ix_state {
/* GAM */
u32 wr_watermark;
u32 gfx_prio_ctrl;
u32 arb_mode;
u32 gfx_pend_tlb0;
u32 gfx_pend_tlb1;
u32 lra_limits[GEN7_LRA_LIMITS_REG_NUM];
u32 media_max_req_count;
u32 gfx_max_req_count;
u32 render_hwsp;
u32 ecochk;
u32 bsd_hwsp;
u32 blt_hwsp;
u32 tlb_rd_addr;
/* MBC */
u32 g3dctl;
u32 gsckgctl;
u32 mbctl;
/* GCP */
u32 ucgctl1;
u32 ucgctl3;
u32 rcgctl1;
u32 rcgctl2;
u32 rstctl;
u32 misccpctl;
/* GPM */
u32 gfxpause;
u32 rpdeuhwtc;
u32 rpdeuc;
u32 ecobus;
u32 pwrdwnupctl;
u32 rp_down_timeout;
u32 rp_deucsw;
u32 rcubmabdtmr;
u32 rcedata;
u32 spare2gh;
/* Display 1 CZ domain */
u32 gt_imr;
u32 gt_ier;
u32 pm_imr;
u32 pm_ier;
u32 gt_scratch[GEN7_GT_SCRATCH_REG_NUM];
/* GT SA CZ domain */
u32 tilectl;
u32 gt_fifoctl;
u32 gtlc_wake_ctrl;
u32 gtlc_survive;
u32 pmwgicz;
/* Display 2 CZ domain */
u32 gu_ctl0;
u32 gu_ctl1;
u32 pcbr;
u32 clock_gate_dis2;
};
struct intel_rps_ei {
ktime_t ktime;
u32 render_c0;
u32 media_c0;
};
struct intel_gen6_power_mgmt {
drm/i915: sanitize rps irq disabling When disabling the RPS interrupts there is a tricky dependency between the thread disabling the interrupts, the RPS interrupt handler and the corresponding RPS work. The RPS work can reenable the interrupts, so there is no straightforward order in the disabling thread to (1) make sure that any RPS work is flushed and to (2) disable all RPS interrupts. Currently this is solved by masking the interrupts using two separate mask registers (first level display IMR and PM IMR) and doing the disabling when all first level interrupts are disabled. This works, but the requirement to run with all first level interrupts disabled is unnecessary making the suspend / unload time ordering of RPS disabling wrt. other unitialization steps difficult and error prone. Removing this restriction allows us to disable RPS early during suspend / unload and forget about it for the rest of the sequence. By adding a more explicit method for avoiding the above race, it also becomes easier to prove its correctness. Finally currently we can hit the WARN in snb_update_pm_irq(), when a final RPS work runs with the first level interrupts already disabled. This won't lead to any problem (due to the separate interrupt masks), but with the change in this and the next patch we can get rid of the WARN, while leaving it in place for other scenarios. To address the above points, add a new RPS interrupts_enabled flag and use this during RPS disabling to avoid requeuing the RPS work and reenabling of the RPS interrupts. Since the interrupt disabling happens now in intel_suspend_gt_powersave(), we will disable RPS interrupts explicitly during suspend (and not just through the first level mask), but there is no problem doing so, it's also more consistent and allows us to unify more of the RPS disabling during suspend and unload time in the next patch. v2/v3: - rebase on patch "drm/i915: move rps irq disable one level up" in the patchset Signed-off-by: Imre Deak <imre.deak@intel.com> Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2014-11-19 06:30:04 -07:00
/*
* work, interrupts_enabled and pm_iir are protected by
* dev_priv->irq_lock
*/
struct work_struct work;
drm/i915: sanitize rps irq disabling When disabling the RPS interrupts there is a tricky dependency between the thread disabling the interrupts, the RPS interrupt handler and the corresponding RPS work. The RPS work can reenable the interrupts, so there is no straightforward order in the disabling thread to (1) make sure that any RPS work is flushed and to (2) disable all RPS interrupts. Currently this is solved by masking the interrupts using two separate mask registers (first level display IMR and PM IMR) and doing the disabling when all first level interrupts are disabled. This works, but the requirement to run with all first level interrupts disabled is unnecessary making the suspend / unload time ordering of RPS disabling wrt. other unitialization steps difficult and error prone. Removing this restriction allows us to disable RPS early during suspend / unload and forget about it for the rest of the sequence. By adding a more explicit method for avoiding the above race, it also becomes easier to prove its correctness. Finally currently we can hit the WARN in snb_update_pm_irq(), when a final RPS work runs with the first level interrupts already disabled. This won't lead to any problem (due to the separate interrupt masks), but with the change in this and the next patch we can get rid of the WARN, while leaving it in place for other scenarios. To address the above points, add a new RPS interrupts_enabled flag and use this during RPS disabling to avoid requeuing the RPS work and reenabling of the RPS interrupts. Since the interrupt disabling happens now in intel_suspend_gt_powersave(), we will disable RPS interrupts explicitly during suspend (and not just through the first level mask), but there is no problem doing so, it's also more consistent and allows us to unify more of the RPS disabling during suspend and unload time in the next patch. v2/v3: - rebase on patch "drm/i915: move rps irq disable one level up" in the patchset Signed-off-by: Imre Deak <imre.deak@intel.com> Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2014-11-19 06:30:04 -07:00
bool interrupts_enabled;
u32 pm_iir;
/* PM interrupt bits that should never be masked */
u32 pm_intrmsk_mbz;
/* Frequencies are stored in potentially platform dependent multiples.
* In other words, *_freq needs to be multiplied by X to be interesting.
* Soft limits are those which are used for the dynamic reclocking done
* by the driver (raise frequencies under heavy loads, and lower for
* lighter loads). Hard limits are those imposed by the hardware.
*
* A distinction is made for overclocking, which is never enabled by
* default, and is considered to be above the hard limit if it's
* possible at all.
*/
u8 cur_freq; /* Current frequency (cached, may not == HW) */
u8 min_freq_softlimit; /* Minimum frequency permitted by the driver */
u8 max_freq_softlimit; /* Max frequency permitted by the driver */
u8 max_freq; /* Maximum frequency, RP0 if not overclocking */
u8 min_freq; /* AKA RPn. Minimum frequency */
u8 boost_freq; /* Frequency to request when wait boosting */
u8 idle_freq; /* Frequency to request when we are idle */
u8 efficient_freq; /* AKA RPe. Pre-determined balanced frequency */
u8 rp1_freq; /* "less than" RP0 power/freqency */
u8 rp0_freq; /* Non-overclocked max frequency. */
u16 gpll_ref_freq; /* vlv/chv GPLL reference frequency */
u8 up_threshold; /* Current %busy required to uplock */
u8 down_threshold; /* Current %busy required to downclock */
drm/i915: Tweak RPS thresholds to more aggressively downclock After applying wait-boost we often find ourselves stuck at higher clocks than required. The current threshold value requires the GPU to be continuously and completely idle for 313ms before it is dropped by one bin. Conversely, we require the GPU to be busy for an average of 90% over a 84ms period before we upclock. So the current thresholds almost never downclock the GPU, and respond very slowly to sudden demands for more power. It is easy to observe that we currently lock into the wrong bin and both underperform in benchmarks and consume more power than optimal (just by repeating the task and measuring the different results). An alternative approach, as discussed in the bspec, is to use a continuous threshold for upclocking, and an average value for downclocking. This is good for quickly detecting and reacting to state changes within a frame, however it fails with the common throttling method of waiting upon the outstanding frame - at least it is difficult to choose a threshold that works well at 15,000fps and at 60fps. So continue to use average busy/idle loads to determine frequency change. v2: Use 3 power zones to keep frequencies low in steady-state mostly idle (e.g. scrolling, interactive 2D drawing), and frequencies high for demanding games. In between those end-states, we use a fast-reclocking algorithm to converge more quickly on the desired bin. v3: Bug fixes - make sure we reset adj after switching power zones. v4: Tune - drop the continuous busy thresholds as it prevents us from choosing the right frequency for glxgears style swap benchmarks. Instead the goal is to be able to find the right clocks irrespective of the wait-boost. Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Kenneth Graunke <kenneth@whitecape.org> Cc: Stéphane Marchesin <stephane.marchesin@gmail.com> Cc: Owen Taylor <otaylor@redhat.com> Cc: "Meng, Mengmeng" <mengmeng.meng@intel.com> Cc: "Zhuang, Lena" <lena.zhuang@intel.com> Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2013-09-25 10:34:57 -06:00
int last_adj;
enum { LOW_POWER, BETWEEN, HIGH_POWER } power;
bool enabled;
struct delayed_work autoenable_work;
drm/i915: Avoid keeping waitboost active for signaling threads Once a client has requested a waitboost, we keep that waitboost active until all clients are no longer waiting. This is because we don't distinguish which waiter deserves the boost. However, with the advent of fence signaling, the signaler threads appear as waiters to the RPS interrupt handler. So instead of using a single boolean to track when to keep the waitboost active, use a counter of all outstanding waitboosted requests. At this point, I have removed all vestiges of the rate limiting on clients. Whilst this means that compositors should remain more fluid, it also means that boosts are more prevalent. See commit b29c19b64528 ("drm/i915: Boost RPS frequency for CPU stalls") for a longer discussion on the pros and cons of both approaches. A drawback of this implementation is that it requires constant request submission to keep the waitboost trimmed (as it is now cancelled when the request is completed). This will be fine for a busy system, but near idle the boosts may be kept for longer than desired (effectively tens of vblanks worstcase) and there is a reliance on rc6 instead. v2: Remove defunct rps.client_lock Reported-by: Michał Winiarski <michal.winiarski@intel.com> Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Michał Winiarski <michal.winiarski@intel.com> Reviewed-by: Michał Winiarski <michal.winiarski@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/20170628123548.9236-1-chris@chris-wilson.co.uk
2017-06-28 06:35:48 -06:00
atomic_t num_waiters;
atomic_t boosts;
/* manual wa residency calculations */
struct intel_rps_ei ei;
/*
* Protects RPS/RC6 register access and PCU communication.
* Must be taken after struct_mutex if nested. Note that
* this lock may be held for long periods of time when
* talking to hw - so only take it when talking to hw!
*/
struct mutex hw_lock;
};
/* defined intel_pm.c */
extern spinlock_t mchdev_lock;
struct intel_ilk_power_mgmt {
u8 cur_delay;
u8 min_delay;
u8 max_delay;
u8 fmax;
u8 fstart;
u64 last_count1;
unsigned long last_time1;
unsigned long chipset_power;
u64 last_count2;
u64 last_time2;
unsigned long gfx_power;
u8 corr;
int c_m;
int r_t;
};
struct drm_i915_private;
struct i915_power_well;
struct i915_power_well_ops {
/*
* Synchronize the well's hw state to match the current sw state, for
* example enable/disable it based on the current refcount. Called
* during driver init and resume time, possibly after first calling
* the enable/disable handlers.
*/
void (*sync_hw)(struct drm_i915_private *dev_priv,
struct i915_power_well *power_well);
/*
* Enable the well and resources that depend on it (for example
* interrupts located on the well). Called after the 0->1 refcount
* transition.
*/
void (*enable)(struct drm_i915_private *dev_priv,
struct i915_power_well *power_well);
/*
* Disable the well and resources that depend on it. Called after
* the 1->0 refcount transition.
*/
void (*disable)(struct drm_i915_private *dev_priv,
struct i915_power_well *power_well);
/* Returns the hw enabled state. */
bool (*is_enabled)(struct drm_i915_private *dev_priv,
struct i915_power_well *power_well);
};
/* Power well structure for haswell */
struct i915_power_well {
const char *name;
bool always_on;
/* power well enable/disable usage count */
int count;
/* cached hw enabled state */
bool hw_enabled;
u64 domains;
/* unique identifier for this power well */
enum i915_power_well_id id;
/*
* Arbitraty data associated with this power well. Platform and power
* well specific.
*/
union {
struct {
enum dpio_phy phy;
} bxt;
struct {
/* Mask of pipes whose IRQ logic is backed by the pw */
u8 irq_pipe_mask;
/* The pw is backing the VGA functionality */
bool has_vga:1;
bool has_fuses:1;
} hsw;
};
const struct i915_power_well_ops *ops;
};
struct i915_power_domains {
drm/i915: use power get/put instead of set for power on after init Currently we make sure that all power domains are enabled during driver init and turn off unneded ones only after the first modeset. Similarly during suspend we enable all power domains, which will remain on through the following resume until the first modeset. This logic is supported by intel_set_power_well() in the power domain framework. It would be nice to simplify the API, so that we only have get/put functions and make it more explicit on the higher level how this "power well on during init" logic works. This will make it also easier if in the future we want to shorten the time the power wells are on. For this add a new device private flag tracking whether we have the power wells on because of init/suspend and use only intel_display_power_get()/put(). As nothing else uses intel_set_power_well() we can remove it. This also fixes commit 6efdf354ddb186c6604d1692075421e8d2c740e9 Author: Imre Deak <imre.deak@intel.com> Date: Wed Oct 16 17:25:52 2013 +0300 drm/i915: enable only the needed power domains during modeset where removing intel_set_power_well() resulted in not releasing the reference on the power well that was taken during init and thus leaving the power well on all the time. Regression reported by Paulo. v2: - move the init_power_on flag to the power_domains struct (Daniel) v3: - add note about this being a regression fix too (Paulo) Signed-off-by: Imre Deak <imre.deak@intel.com> Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2013-10-25 08:36:48 -06:00
/*
* Power wells needed for initialization at driver init and suspend
* time are on. They are kept on until after the first modeset.
*/
bool init_power_on;
bool initializing;
int power_well_count;
drm/i915: use power get/put instead of set for power on after init Currently we make sure that all power domains are enabled during driver init and turn off unneded ones only after the first modeset. Similarly during suspend we enable all power domains, which will remain on through the following resume until the first modeset. This logic is supported by intel_set_power_well() in the power domain framework. It would be nice to simplify the API, so that we only have get/put functions and make it more explicit on the higher level how this "power well on during init" logic works. This will make it also easier if in the future we want to shorten the time the power wells are on. For this add a new device private flag tracking whether we have the power wells on because of init/suspend and use only intel_display_power_get()/put(). As nothing else uses intel_set_power_well() we can remove it. This also fixes commit 6efdf354ddb186c6604d1692075421e8d2c740e9 Author: Imre Deak <imre.deak@intel.com> Date: Wed Oct 16 17:25:52 2013 +0300 drm/i915: enable only the needed power domains during modeset where removing intel_set_power_well() resulted in not releasing the reference on the power well that was taken during init and thus leaving the power well on all the time. Regression reported by Paulo. v2: - move the init_power_on flag to the power_domains struct (Daniel) v3: - add note about this being a regression fix too (Paulo) Signed-off-by: Imre Deak <imre.deak@intel.com> Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2013-10-25 08:36:48 -06:00
struct mutex lock;
int domain_use_count[POWER_DOMAIN_NUM];
struct i915_power_well *power_wells;
};
#define MAX_L3_SLICES 2
struct intel_l3_parity {
u32 *remap_info[MAX_L3_SLICES];
struct work_struct error_work;
int which_slice;
};
struct i915_gem_mm {
/** Memory allocator for GTT stolen memory */
struct drm_mm stolen;
/** Protects the usage of the GTT stolen memory allocator. This is
* always the inner lock when overlapping with struct_mutex. */
struct mutex stolen_lock;
/** List of all objects in gtt_space. Used to restore gtt
* mappings on resume */
struct list_head bound_list;
/**
* List of objects which are not bound to the GTT (thus
* are idle and not used by the GPU). These objects may or may
* not actually have any pages attached.
*/
struct list_head unbound_list;
/** List of all objects in gtt_space, currently mmaped by userspace.
* All objects within this list must also be on bound_list.
*/
struct list_head userfault_list;
/**
* List of objects which are pending destruction.
*/
struct llist_head free_list;
struct work_struct free_work;
/** Usable portion of the GTT for GEM */
dma_addr_t stolen_base; /* limited to low memory (32-bit) */
/** PPGTT used for aliasing the PPGTT with the GTT */
struct i915_hw_ppgtt *aliasing_ppgtt;
struct notifier_block oom_notifier;
struct notifier_block vmap_notifier;
struct shrinker shrinker;
/** LRU list of objects with fence regs on them. */
struct list_head fence_list;
/**
* Workqueue to fault in userptr pages, flushed by the execbuf
* when required but otherwise left to userspace to try again
* on EAGAIN.
*/
struct workqueue_struct *userptr_wq;
u64 unordered_timeline;
/* the indicator for dispatch video commands on two BSD rings */
atomic_t bsd_engine_dispatch_index;
/** Bit 6 swizzling required for X tiling */
uint32_t bit_6_swizzle_x;
/** Bit 6 swizzling required for Y tiling */
uint32_t bit_6_swizzle_y;
/* accounting, useful for userland debugging */
spinlock_t object_stat_lock;
u64 object_memory;
u32 object_count;
};
struct drm_i915_error_state_buf {
struct drm_i915_private *i915;
unsigned bytes;
unsigned size;
int err;
u8 *buf;
loff_t start;
loff_t pos;
};
#define I915_RESET_TIMEOUT (10 * HZ) /* 10s */
#define I915_FENCE_TIMEOUT (10 * HZ) /* 10s */
drm/i915: Decouple hang detection from hangcheck period Hangcheck state accumulation has gained more steps along the years, like head movement and more recently the subunit inactivity check. As the subunit sampling is only done if the previous state check showed inactivity, we have added more stages (and time) to reach a hang verdict. Asymmetric engine states led to different actual weight of 'one hangcheck unit' and it was demonstrated in some hangs that due to difference in stages, simpler engines were accused falsely of a hang as their scoring was much more quicker to accumulate above the hang treshold. To completely decouple the hangcheck guilty score from the hangcheck period, convert hangcheck score to a rough period of inactivity measurement. As these are tracked as jiffies, they are meaningful also across reset boundaries. This makes finding a guilty engine more accurate across multi engine activity scenarios, especially across asymmetric engines. We lose the ability to detect cross batch malicious attempts to hinder the progress. Plan is to move this functionality to be part of context banning which is more natural fit, later in the series. v2: use time_before macros (Chris) reinstate the pardoning of moving engine after hc (Chris) v3: avoid global state for per engine stall detection (Chris) v4: take timeline last retirement into account (Chris) v5: do debug print on pardoning, split out retirement timestamp (Chris) Cc: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
2016-11-18 06:09:04 -07:00
#define I915_ENGINE_DEAD_TIMEOUT (4 * HZ) /* Seqno, head and subunits dead */
#define I915_SEQNO_DEAD_TIMEOUT (12 * HZ) /* Seqno dead with active head */
struct i915_gpu_error {
/* For hangcheck timer */
#define DRM_I915_HANGCHECK_PERIOD 1500 /* in ms */
#define DRM_I915_HANGCHECK_JIFFIES msecs_to_jiffies(DRM_I915_HANGCHECK_PERIOD)
struct delayed_work hangcheck_work;
/* For reset and error_state handling. */
spinlock_t lock;
/* Protected by the above dev->gpu_error.lock. */
struct i915_gpu_state *first_error;
drm/i915: More surgically unbreak the modeset vs reset deadlock There's no reason to entirely wedge the gpu, for the minimal deadlock bugfix we only need to unbreak/decouple the atomic commit from the gpu reset. The simplest way to fix that is by replacing the unconditional fence wait a the top of commit_tail by a wait which completes either when the fences are done (normal case, or when a reset doesn't need to touch the display state). Or when the gpu reset needs to force-unblock all pending modeset states. The lesser source of deadlocks is when we try to pin a new framebuffer and run into a stall. There's a bunch of places this can happen, like eviction, changing the caching mode, acquiring a fence on older platforms. And we can't just break the depency loop and keep going, the only way would be to break out and restart. But the problem with that approach is that we must stall for the reset to complete before we grab any locks, and with the atomic infrastructure that's a bit tricky. The only place is the ioctl code, and we don't want to insert code into e.g. the BUSY ioctl. Hence for that problem just create a critical section, and if any code is in there, wedge the GPU. For the steady-state this should never be a problem. Note that in both cases TDR itself keeps working, so from a userspace pov this trickery isn't observable. Users themselvs might spot a short glitch while the rendering is catching up again, but that's still better than pre-TDR where we've thrown away all the rendering, including innocent batches. Also, this fixes the regression TDR introduced of making gpu resets deadlock-prone when we do need to touch the display. One thing I noticed is that gpu_error.flags seems to use both our own wait-queue in gpu_error.wait_queue, and the generic wait_on_bit facilities. Not entirely sure why this inconsistency exists, I just picked one style. A possible future avenue could be to insert the gpu reset in-between ongoing modeset changes, which would avoid the momentary glitch. But that's a lot more work to implement in the atomic commit machinery, and given that we only need this for pre-g4x hw, of questionable utility just for the sake of polishing gpu reset even more on those old boxes. It might be useful for other features though. v2: Rebase onto 4.13 with a s/wait_queue_t/struct wait_queue_entry/. v3: Really emabarrassing fixup, I checked the wrong bit and broke the unbreak/wakeup logic. v4: Also handle deadlocks in pin_to_display. v5: Review from Michel: - Fixup the BUILD_BUG_ON - Don't forget about the overlay Cc: Michel Thierry <michel.thierry@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Mika Kuoppala <mika.kuoppala@intel.com> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com> (v2) Cc: Michel Thierry <michel.thierry@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20170808080828.23650-3-daniel.vetter@ffwll.ch Reviewed-by: Michel Thierry <michel.thierry@intel.com>
2017-08-08 02:08:28 -06:00
atomic_t pending_fb_pin;
unsigned long missed_irq_rings;
drm/i915: clear up wedged transitions We have two important transitions of the wedged state in the current code: - 0 -> 1: This means a hang has been detected, and signals to everyone that they please get of any locks, so that the reset work item can do its job. - 1 -> 0: The reset handler has completed. Now the last transition mixes up two states: "Reset completed and successful" and "Reset failed". To distinguish these two we do some tricks with the reset completion, but I simply could not convince myself that this doesn't race under odd circumstances. Hence split this up, and add a new terminal state indicating that the hw is gone for good. Also add explicit #defines for both states, update comments. v2: Split out the reset handling bugfix for the throttle ioctl. v3: s/tmp/wedged/ sugested by Chris Wilson. Also fixup up a rebase error which prevented this patch from actually compiling. v4: To unify the wedged state with the reset counter, keep the reset-in-progress state just as a flag. The terminally-wedged state is now denoted with a big number. v5: Add a comment to the reset_counter special values explaining that WEDGED & RESET_IN_PROGRESS needs to be true for the code to be correct. v6: Fixup logic errors introduced with the wedged+reset_counter unification. Since WEDGED implies reset-in-progress (in a way we're terminally stuck in the dead-but-reset-not-completed state), we need ensure that we check for this everywhere. The specific bug was in wait_for_error, which would simply have timed out. v7: Extract an inline i915_reset_in_progress helper to make the code more readable. Also annote the reset-in-progress case with an unlikely, to help the compiler optimize the fastpath. Do the same for the terminally wedged case with i915_terminally_wedged. Reviewed-by: Damien Lespiau <damien.lespiau@intel.com> Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2012-11-15 09:17:22 -07:00
/**
* State variable controlling the reset flow and count
drm/i915: clear up wedged transitions We have two important transitions of the wedged state in the current code: - 0 -> 1: This means a hang has been detected, and signals to everyone that they please get of any locks, so that the reset work item can do its job. - 1 -> 0: The reset handler has completed. Now the last transition mixes up two states: "Reset completed and successful" and "Reset failed". To distinguish these two we do some tricks with the reset completion, but I simply could not convince myself that this doesn't race under odd circumstances. Hence split this up, and add a new terminal state indicating that the hw is gone for good. Also add explicit #defines for both states, update comments. v2: Split out the reset handling bugfix for the throttle ioctl. v3: s/tmp/wedged/ sugested by Chris Wilson. Also fixup up a rebase error which prevented this patch from actually compiling. v4: To unify the wedged state with the reset counter, keep the reset-in-progress state just as a flag. The terminally-wedged state is now denoted with a big number. v5: Add a comment to the reset_counter special values explaining that WEDGED & RESET_IN_PROGRESS needs to be true for the code to be correct. v6: Fixup logic errors introduced with the wedged+reset_counter unification. Since WEDGED implies reset-in-progress (in a way we're terminally stuck in the dead-but-reset-not-completed state), we need ensure that we check for this everywhere. The specific bug was in wait_for_error, which would simply have timed out. v7: Extract an inline i915_reset_in_progress helper to make the code more readable. Also annote the reset-in-progress case with an unlikely, to help the compiler optimize the fastpath. Do the same for the terminally wedged case with i915_terminally_wedged. Reviewed-by: Damien Lespiau <damien.lespiau@intel.com> Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2012-11-15 09:17:22 -07:00
*
* This is a counter which gets incremented when reset is triggered,
*
* Before the reset commences, the I915_RESET_BACKOFF bit is set
* meaning that any waiters holding onto the struct_mutex should
* relinquish the lock immediately in order for the reset to start.
*
* If reset is not completed succesfully, the I915_WEDGE bit is
* set meaning that hardware is terminally sour and there is no
* recovery. All waiters on the reset_queue will be woken when
* that happens.
*
* This counter is used by the wait_seqno code to notice that reset
* event happened and it needs to restart the entire ioctl (since most
* likely the seqno it waited for won't ever signal anytime soon).
drm/i915: create a race-free reset detection With the previous patch the state transition handling of the reset code itself is now (hopefully) race free and solid. But that still leaves out everyone else - with the various lock-free wait paths we have there's the possibility that the reset happens between the point where we read the seqno we should wait on and the actual wait. And if __wait_seqno then never sees the RESET_IN_PROGRESS state, we'll happily wait for a seqno which will in all likelyhood never signal. In practice this is not a big problem since the X server gets constantly interrupted, and can then submit more work (hopefully) to unblock everyone else: As soon as a new seqno write lands, all waiters will unblock. But running the i-g-t reset testcase ZZ_hangman can expose this race, especially on slower hw with fewer cpu cores. Now looking forward to ARB_robustness and friends that's not the best possible behaviour, hence this patch adds a reset_counter to be able to detect any reset, even if a given thread never observed the in-progress state. The important part is to correctly order things: - The write side needs to increment the counter after any seqno gets reset. Hence we need to do that at the end of the reset work, and again wake everyone up. We also need to place a barrier in between any possible seqno changes and the counter increment, since any unlock operations only guarantee that nothing leaks out, but not that at later load operation gets moved ahead. - On the read side we need to ensure that no reset can sneak in and invalidate the seqno. In all cases we can use the one-sided barrier that unlock operations guarantee (of the lock protecting the respective seqno/ring pair) to ensure correct ordering. Hence it is sufficient to place the atomic read before the mutex/spin_unlock and no additional barriers are required. The end-result of all this is that we need to wake up everyone twice in a reset operation: - First, before the reset starts, to get any lockholders of the locks, so that the reset can proceed. - Second, after the reset is completed, to allow waiters to properly and reliably detect the reset condition and bail out. I admit that this entire reset_counter thing smells a bit like overkill, but I think it's justified since it makes it really explicit what the bail-out condition is. And we need a reset counter anyway to implement ARB_robustness, and imo with finer-grained locking on the horizont this is the most resilient scheme I could think of. v2: Drop spurious change in the wait_for_error EXIT_COND - we only need to wait until we leave the reset-in-progress wedged state. v3: Don't play tricks with barriers in the throttle ioctl, the spin_unlock is barrier enough. I've also considered using a little helper to grab the current reset_counter, but then decided that hiding the atomic_read isn't a great idea, since having it explicitly show up in the code is a nice remainder to reviews to check the memory barriers. v4: Add a comment to explain why we need to fall through in __wait_seqno in the end variable assignments. v5: Review from Damien: - s/smb/smp/ in a comment - don't increment the reset counter after we've set it to WEDGED. Now we (again) properly wedge the gpu when the reset fails. Reviewed-by: Damien Lespiau <damien.lespiau@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2012-12-06 01:01:42 -07:00
*
* This is important for lock-free wait paths, where no contended lock
* naturally enforces the correct ordering between the bail-out of the
* waiter and the gpu reset work code.
drm/i915: clear up wedged transitions We have two important transitions of the wedged state in the current code: - 0 -> 1: This means a hang has been detected, and signals to everyone that they please get of any locks, so that the reset work item can do its job. - 1 -> 0: The reset handler has completed. Now the last transition mixes up two states: "Reset completed and successful" and "Reset failed". To distinguish these two we do some tricks with the reset completion, but I simply could not convince myself that this doesn't race under odd circumstances. Hence split this up, and add a new terminal state indicating that the hw is gone for good. Also add explicit #defines for both states, update comments. v2: Split out the reset handling bugfix for the throttle ioctl. v3: s/tmp/wedged/ sugested by Chris Wilson. Also fixup up a rebase error which prevented this patch from actually compiling. v4: To unify the wedged state with the reset counter, keep the reset-in-progress state just as a flag. The terminally-wedged state is now denoted with a big number. v5: Add a comment to the reset_counter special values explaining that WEDGED & RESET_IN_PROGRESS needs to be true for the code to be correct. v6: Fixup logic errors introduced with the wedged+reset_counter unification. Since WEDGED implies reset-in-progress (in a way we're terminally stuck in the dead-but-reset-not-completed state), we need ensure that we check for this everywhere. The specific bug was in wait_for_error, which would simply have timed out. v7: Extract an inline i915_reset_in_progress helper to make the code more readable. Also annote the reset-in-progress case with an unlikely, to help the compiler optimize the fastpath. Do the same for the terminally wedged case with i915_terminally_wedged. Reviewed-by: Damien Lespiau <damien.lespiau@intel.com> Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2012-11-15 09:17:22 -07:00
*/
unsigned long reset_count;
drm/i915: clear up wedged transitions We have two important transitions of the wedged state in the current code: - 0 -> 1: This means a hang has been detected, and signals to everyone that they please get of any locks, so that the reset work item can do its job. - 1 -> 0: The reset handler has completed. Now the last transition mixes up two states: "Reset completed and successful" and "Reset failed". To distinguish these two we do some tricks with the reset completion, but I simply could not convince myself that this doesn't race under odd circumstances. Hence split this up, and add a new terminal state indicating that the hw is gone for good. Also add explicit #defines for both states, update comments. v2: Split out the reset handling bugfix for the throttle ioctl. v3: s/tmp/wedged/ sugested by Chris Wilson. Also fixup up a rebase error which prevented this patch from actually compiling. v4: To unify the wedged state with the reset counter, keep the reset-in-progress state just as a flag. The terminally-wedged state is now denoted with a big number. v5: Add a comment to the reset_counter special values explaining that WEDGED & RESET_IN_PROGRESS needs to be true for the code to be correct. v6: Fixup logic errors introduced with the wedged+reset_counter unification. Since WEDGED implies reset-in-progress (in a way we're terminally stuck in the dead-but-reset-not-completed state), we need ensure that we check for this everywhere. The specific bug was in wait_for_error, which would simply have timed out. v7: Extract an inline i915_reset_in_progress helper to make the code more readable. Also annote the reset-in-progress case with an unlikely, to help the compiler optimize the fastpath. Do the same for the terminally wedged case with i915_terminally_wedged. Reviewed-by: Damien Lespiau <damien.lespiau@intel.com> Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2012-11-15 09:17:22 -07:00
/**
* flags: Control various stages of the GPU reset
*
* #I915_RESET_BACKOFF - When we start a reset, we want to stop any
* other users acquiring the struct_mutex. To do this we set the
* #I915_RESET_BACKOFF bit in the error flags when we detect a reset
* and then check for that bit before acquiring the struct_mutex (in
* i915_mutex_lock_interruptible()?). I915_RESET_BACKOFF serves a
* secondary role in preventing two concurrent global reset attempts.
*
* #I915_RESET_HANDOFF - To perform the actual GPU reset, we need the
* struct_mutex. We try to acquire the struct_mutex in the reset worker,
* but it may be held by some long running waiter (that we cannot
* interrupt without causing trouble). Once we are ready to do the GPU
* reset, we set the I915_RESET_HANDOFF bit and wakeup any waiters. If
* they already hold the struct_mutex and want to participate they can
* inspect the bit and do the reset directly, otherwise the worker
* waits for the struct_mutex.
*
drm/i915: Modify error handler for per engine hang recovery This is a preparatory patch which modifies error handler to do per engine hang recovery. The actual patch which implements this sequence follows later in the series. The aim is to prepare existing recovery function to adapt to this new function where applicable (which fails at this point because core implementation is lacking) and continue recovery using legacy full gpu reset. A helper function is also added to query the availability of engine reset. A subsequent patch will add the capability to query which type of reset is present (engine -> full -> no-reset) via the get-param ioctl. It has been decided that the error events that are used to notify user of reset will only be sent in case if full chip reset. In case of just single (or multiple) engine resets, userspace won't be notified by these events. Note that this implementation of engine reset is for i915 directly submitting to the ELSP, where the driver manages the hang detection, recovery and resubmission. With GuC submission these tasks are shared between driver and firmware; i915 will still responsible for detecting a hang, and when it does it will have to request GuC to reset that Engine and remind the firmware about the outstanding submissions. This will be added in different patch. v2: rebase, advertise engine reset availability in platform definition, add note about GuC submission. v3: s/*engine_reset*/*reset_engine*/. (Chris) Handle reset as 2 level resets, by first going to engine only and fall backing to full/chip reset as needed, i.e. reset_engine will need the struct_mutex. v4: Pass the engine mask to i915_reset. (Chris) v5: Rebase, update selftests. v6: Rebase, prepare for mutex-less reset engine. v7: Pass reset_engine mask as a function parameter, and iterate over the engine mask for reset_engine. (Chris) v8: Use i915.reset >=2 in has_reset_engine; remove redundant reset logging; add a reset-engine-in-progress flag to prevent concurrent resets, and avoid dual purposing of reset-backoff. (Chris) v9: Support reset of different engines in parallel (Chris) v10: Handle reset-engine flag locking better (Chris) v11: Squash in reporting of per-engine-reset availability. Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com> Signed-off-by: Ian Lister <ian.lister@intel.com> Signed-off-by: Tomas Elf <tomas.elf@intel.com> Signed-off-by: Arun Siluvery <arun.siluvery@linux.intel.com> Signed-off-by: Michel Thierry <michel.thierry@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/20170615201828.23144-4-michel.thierry@intel.com Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Link: http://patchwork.freedesktop.org/patch/msgid/20170620095751.13127-5-chris@chris-wilson.co.uk
2017-06-20 03:57:46 -06:00
* #I915_RESET_ENGINE[num_engines] - Since the driver doesn't need to
* acquire the struct_mutex to reset an engine, we need an explicit
* flag to prevent two concurrent reset attempts in the same engine.
* As the number of engines continues to grow, allocate the flags from
* the most significant bits.
*
* #I915_WEDGED - If reset fails and we can no longer use the GPU,
* we set the #I915_WEDGED bit. Prior to command submission, e.g.
* i915_gem_request_alloc(), this bit is checked and the sequence
* aborted (with -EIO reported to userspace) if set.
*/
unsigned long flags;
#define I915_RESET_BACKOFF 0
#define I915_RESET_HANDOFF 1
drm/i915: More surgically unbreak the modeset vs reset deadlock There's no reason to entirely wedge the gpu, for the minimal deadlock bugfix we only need to unbreak/decouple the atomic commit from the gpu reset. The simplest way to fix that is by replacing the unconditional fence wait a the top of commit_tail by a wait which completes either when the fences are done (normal case, or when a reset doesn't need to touch the display state). Or when the gpu reset needs to force-unblock all pending modeset states. The lesser source of deadlocks is when we try to pin a new framebuffer and run into a stall. There's a bunch of places this can happen, like eviction, changing the caching mode, acquiring a fence on older platforms. And we can't just break the depency loop and keep going, the only way would be to break out and restart. But the problem with that approach is that we must stall for the reset to complete before we grab any locks, and with the atomic infrastructure that's a bit tricky. The only place is the ioctl code, and we don't want to insert code into e.g. the BUSY ioctl. Hence for that problem just create a critical section, and if any code is in there, wedge the GPU. For the steady-state this should never be a problem. Note that in both cases TDR itself keeps working, so from a userspace pov this trickery isn't observable. Users themselvs might spot a short glitch while the rendering is catching up again, but that's still better than pre-TDR where we've thrown away all the rendering, including innocent batches. Also, this fixes the regression TDR introduced of making gpu resets deadlock-prone when we do need to touch the display. One thing I noticed is that gpu_error.flags seems to use both our own wait-queue in gpu_error.wait_queue, and the generic wait_on_bit facilities. Not entirely sure why this inconsistency exists, I just picked one style. A possible future avenue could be to insert the gpu reset in-between ongoing modeset changes, which would avoid the momentary glitch. But that's a lot more work to implement in the atomic commit machinery, and given that we only need this for pre-g4x hw, of questionable utility just for the sake of polishing gpu reset even more on those old boxes. It might be useful for other features though. v2: Rebase onto 4.13 with a s/wait_queue_t/struct wait_queue_entry/. v3: Really emabarrassing fixup, I checked the wrong bit and broke the unbreak/wakeup logic. v4: Also handle deadlocks in pin_to_display. v5: Review from Michel: - Fixup the BUILD_BUG_ON - Don't forget about the overlay Cc: Michel Thierry <michel.thierry@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Mika Kuoppala <mika.kuoppala@intel.com> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com> (v2) Cc: Michel Thierry <michel.thierry@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20170808080828.23650-3-daniel.vetter@ffwll.ch Reviewed-by: Michel Thierry <michel.thierry@intel.com>
2017-08-08 02:08:28 -06:00
#define I915_RESET_MODESET 2
#define I915_WEDGED (BITS_PER_LONG - 1)
drm/i915: Modify error handler for per engine hang recovery This is a preparatory patch which modifies error handler to do per engine hang recovery. The actual patch which implements this sequence follows later in the series. The aim is to prepare existing recovery function to adapt to this new function where applicable (which fails at this point because core implementation is lacking) and continue recovery using legacy full gpu reset. A helper function is also added to query the availability of engine reset. A subsequent patch will add the capability to query which type of reset is present (engine -> full -> no-reset) via the get-param ioctl. It has been decided that the error events that are used to notify user of reset will only be sent in case if full chip reset. In case of just single (or multiple) engine resets, userspace won't be notified by these events. Note that this implementation of engine reset is for i915 directly submitting to the ELSP, where the driver manages the hang detection, recovery and resubmission. With GuC submission these tasks are shared between driver and firmware; i915 will still responsible for detecting a hang, and when it does it will have to request GuC to reset that Engine and remind the firmware about the outstanding submissions. This will be added in different patch. v2: rebase, advertise engine reset availability in platform definition, add note about GuC submission. v3: s/*engine_reset*/*reset_engine*/. (Chris) Handle reset as 2 level resets, by first going to engine only and fall backing to full/chip reset as needed, i.e. reset_engine will need the struct_mutex. v4: Pass the engine mask to i915_reset. (Chris) v5: Rebase, update selftests. v6: Rebase, prepare for mutex-less reset engine. v7: Pass reset_engine mask as a function parameter, and iterate over the engine mask for reset_engine. (Chris) v8: Use i915.reset >=2 in has_reset_engine; remove redundant reset logging; add a reset-engine-in-progress flag to prevent concurrent resets, and avoid dual purposing of reset-backoff. (Chris) v9: Support reset of different engines in parallel (Chris) v10: Handle reset-engine flag locking better (Chris) v11: Squash in reporting of per-engine-reset availability. Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com> Signed-off-by: Ian Lister <ian.lister@intel.com> Signed-off-by: Tomas Elf <tomas.elf@intel.com> Signed-off-by: Arun Siluvery <arun.siluvery@linux.intel.com> Signed-off-by: Michel Thierry <michel.thierry@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/20170615201828.23144-4-michel.thierry@intel.com Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Link: http://patchwork.freedesktop.org/patch/msgid/20170620095751.13127-5-chris@chris-wilson.co.uk
2017-06-20 03:57:46 -06:00
#define I915_RESET_ENGINE (I915_WEDGED - I915_NUM_ENGINES)
drm/i915: clear up wedged transitions We have two important transitions of the wedged state in the current code: - 0 -> 1: This means a hang has been detected, and signals to everyone that they please get of any locks, so that the reset work item can do its job. - 1 -> 0: The reset handler has completed. Now the last transition mixes up two states: "Reset completed and successful" and "Reset failed". To distinguish these two we do some tricks with the reset completion, but I simply could not convince myself that this doesn't race under odd circumstances. Hence split this up, and add a new terminal state indicating that the hw is gone for good. Also add explicit #defines for both states, update comments. v2: Split out the reset handling bugfix for the throttle ioctl. v3: s/tmp/wedged/ sugested by Chris Wilson. Also fixup up a rebase error which prevented this patch from actually compiling. v4: To unify the wedged state with the reset counter, keep the reset-in-progress state just as a flag. The terminally-wedged state is now denoted with a big number. v5: Add a comment to the reset_counter special values explaining that WEDGED & RESET_IN_PROGRESS needs to be true for the code to be correct. v6: Fixup logic errors introduced with the wedged+reset_counter unification. Since WEDGED implies reset-in-progress (in a way we're terminally stuck in the dead-but-reset-not-completed state), we need ensure that we check for this everywhere. The specific bug was in wait_for_error, which would simply have timed out. v7: Extract an inline i915_reset_in_progress helper to make the code more readable. Also annote the reset-in-progress case with an unlikely, to help the compiler optimize the fastpath. Do the same for the terminally wedged case with i915_terminally_wedged. Reviewed-by: Damien Lespiau <damien.lespiau@intel.com> Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2012-11-15 09:17:22 -07:00
/** Number of times an engine has been reset */
u32 reset_engine_count[I915_NUM_ENGINES];
/**
* Waitqueue to signal when a hang is detected. Used to for waiters
* to release the struct_mutex for the reset to procede.
*/
wait_queue_head_t wait_queue;
drm/i915: clear up wedged transitions We have two important transitions of the wedged state in the current code: - 0 -> 1: This means a hang has been detected, and signals to everyone that they please get of any locks, so that the reset work item can do its job. - 1 -> 0: The reset handler has completed. Now the last transition mixes up two states: "Reset completed and successful" and "Reset failed". To distinguish these two we do some tricks with the reset completion, but I simply could not convince myself that this doesn't race under odd circumstances. Hence split this up, and add a new terminal state indicating that the hw is gone for good. Also add explicit #defines for both states, update comments. v2: Split out the reset handling bugfix for the throttle ioctl. v3: s/tmp/wedged/ sugested by Chris Wilson. Also fixup up a rebase error which prevented this patch from actually compiling. v4: To unify the wedged state with the reset counter, keep the reset-in-progress state just as a flag. The terminally-wedged state is now denoted with a big number. v5: Add a comment to the reset_counter special values explaining that WEDGED & RESET_IN_PROGRESS needs to be true for the code to be correct. v6: Fixup logic errors introduced with the wedged+reset_counter unification. Since WEDGED implies reset-in-progress (in a way we're terminally stuck in the dead-but-reset-not-completed state), we need ensure that we check for this everywhere. The specific bug was in wait_for_error, which would simply have timed out. v7: Extract an inline i915_reset_in_progress helper to make the code more readable. Also annote the reset-in-progress case with an unlikely, to help the compiler optimize the fastpath. Do the same for the terminally wedged case with i915_terminally_wedged. Reviewed-by: Damien Lespiau <damien.lespiau@intel.com> Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2012-11-15 09:17:22 -07:00
/**
* Waitqueue to signal when the reset has completed. Used by clients
* that wait for dev_priv->mm.wedged to settle.
*/
wait_queue_head_t reset_queue;
/* For missed irq/seqno simulation. */
drm/i915: Slaughter the thundering i915_wait_request herd One particularly stressful scenario consists of many independent tasks all competing for GPU time and waiting upon the results (e.g. realtime transcoding of many, many streams). One bottleneck in particular is that each client waits on its own results, but every client is woken up after every batchbuffer - hence the thunder of hooves as then every client must do its heavyweight dance to read a coherent seqno to see if it is the lucky one. Ideally, we only want one client to wake up after the interrupt and check its request for completion. Since the requests must retire in order, we can select the first client on the oldest request to be woken. Once that client has completed his wait, we can then wake up the next client and so on. However, all clients then incur latency as every process in the chain may be delayed for scheduling - this may also then cause some priority inversion. To reduce the latency, when a client is added or removed from the list, we scan the tree for completed seqno and wake up all the completed waiters in parallel. Using igt/benchmarks/gem_latency, we can demonstrate this effect. The benchmark measures the number of GPU cycles between completion of a batch and the client waking up from a call to wait-ioctl. With many concurrent waiters, with each on a different request, we observe that the wakeup latency before the patch scales nearly linearly with the number of waiters (before external factors kick in making the scaling much worse). After applying the patch, we can see that only the single waiter for the request is being woken up, providing a constant wakeup latency for every operation. However, the situation is not quite as rosy for many waiters on the same request, though to the best of my knowledge this is much less likely in practice. Here, we can observe that the concurrent waiters incur extra latency from being woken up by the solitary bottom-half, rather than directly by the interrupt. This appears to be scheduler induced (having discounted adverse effects from having a rbtree walk/erase in the wakeup path), each additional wake_up_process() costs approximately 1us on big core. Another effect of performing the secondary wakeups from the first bottom-half is the incurred delay this imposes on high priority threads - rather than immediately returning to userspace and leaving the interrupt handler to wake the others. To offset the delay incurred with additional waiters on a request, we could use a hybrid scheme that did a quick read in the interrupt handler and dequeued all the completed waiters (incurring the overhead in the interrupt handler, not the best plan either as we then incur GPU submission latency) but we would still have to wake up the bottom-half every time to do the heavyweight slow read. Or we could only kick the waiters on the seqno with the same priority as the current task (i.e. in the realtime waiter scenario, only it is woken up immediately by the interrupt and simply queues the next waiter before returning to userspace, minimising its delay at the expense of the chain, and also reducing contention on its scheduler runqueue). This is effective at avoid long pauses in the interrupt handler and at avoiding the extra latency in realtime/high-priority waiters. v2: Convert from a kworker per engine into a dedicated kthread for the bottom-half. v3: Rename request members and tweak comments. v4: Use a per-engine spinlock in the breadcrumbs bottom-half. v5: Fix race in locklessly checking waiter status and kicking the task on adding a new waiter. v6: Fix deciding when to force the timer to hide missing interrupts. v7: Move the bottom-half from the kthread to the first client process. v8: Reword a few comments v9: Break the busy loop when the interrupt is unmasked or has fired. v10: Comments, unnecessary churn, better debugging from Tvrtko v11: Wake all completed waiters on removing the current bottom-half to reduce the latency of waking up a herd of clients all waiting on the same request. v12: Rearrange missed-interrupt fault injection so that it works with igt/drv_missed_irq_hang v13: Rename intel_breadcrumb and friends to intel_wait in preparation for signal handling. v14: RCU commentary, assert_spin_locked v15: Hide BUG_ON behind the compiler; report on gem_latency findings. v16: Sort seqno-groups by priority so that first-waiter has the highest task priority (and so avoid priority inversion). v17: Add waiters to post-mortem GPU hang state. v18: Return early for a completed wait after acquiring the spinlock. Avoids adding ourselves to the tree if the is already complete, and skips the awkward question of why we don't do completion wakeups for waits earlier than or equal to ourselves. v19: Prepare for init_breadcrumbs to fail. Later patches may want to allocate during init, so be prepared to propagate back the error code. Testcase: igt/gem_concurrent_blit Testcase: igt/benchmarks/gem_latency Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: "Rogozhkin, Dmitry V" <dmitry.v.rogozhkin@intel.com> Cc: "Gong, Zhipeng" <zhipeng.gong@intel.com> Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> Cc: Dave Gordon <david.s.gordon@intel.com> Cc: "Goel, Akash" <akash.goel@intel.com> Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> #v18 Link: http://patchwork.freedesktop.org/patch/msgid/1467390209-3576-6-git-send-email-chris@chris-wilson.co.uk
2016-07-01 10:23:15 -06:00
unsigned long test_irq_rings;
};
i915: ignore lid open event when resuming i915 driver needs to do modeset when 1. system resumes from sleep 2. lid is opened In PM_SUSPEND_MEM state, all the GPEs are cleared when system resumes, thus it is the i915_resume code does the modeset rather than intel_lid_notify(). But in PM_SUSPEND_FREEZE state, this will be broken because system is still responsive to the lid events. 1. When we close the lid in Freeze state, intel_lid_notify() sets modeset_on_lid. 2. When we reopen the lid, intel_lid_notify() will do a modeset, before the system is resumed. here is the error log, [92146.548074] WARNING: at drivers/gpu/drm/i915/intel_display.c:1028 intel_wait_for_pipe_off+0x184/0x190 [i915]() [92146.548076] Hardware name: VGN-Z540N [92146.548078] pipe_off wait timed out [92146.548167] Modules linked in: hid_generic usbhid hid snd_hda_codec_realtek snd_hda_intel snd_hda_codec parport_pc snd_hwdep ppdev snd_pcm_oss i915 snd_mixer_oss snd_pcm arc4 iwldvm snd_seq_dummy mac80211 snd_seq_oss snd_seq_midi fbcon tileblit font bitblit softcursor drm_kms_helper snd_rawmidi snd_seq_midi_event coretemp drm snd_seq kvm btusb bluetooth snd_timer iwlwifi pcmcia tpm_infineon i2c_algo_bit joydev snd_seq_device intel_agp cfg80211 snd intel_gtt yenta_socket pcmcia_rsrc sony_laptop agpgart microcode psmouse tpm_tis serio_raw mxm_wmi soundcore snd_page_alloc tpm acpi_cpufreq lpc_ich pcmcia_core tpm_bios mperf processor lp parport firewire_ohci firewire_core crc_itu_t sdhci_pci sdhci thermal e1000e [92146.548173] Pid: 4304, comm: kworker/0:0 Tainted: G W 3.8.0-rc3-s0i3-v3-test+ #9 [92146.548175] Call Trace: [92146.548189] [<c10378e2>] warn_slowpath_common+0x72/0xa0 [92146.548227] [<f86398b4>] ? intel_wait_for_pipe_off+0x184/0x190 [i915] [92146.548263] [<f86398b4>] ? intel_wait_for_pipe_off+0x184/0x190 [i915] [92146.548270] [<c10379b3>] warn_slowpath_fmt+0x33/0x40 [92146.548307] [<f86398b4>] intel_wait_for_pipe_off+0x184/0x190 [i915] [92146.548344] [<f86399c2>] intel_disable_pipe+0x102/0x190 [i915] [92146.548380] [<f8639ea4>] ? intel_disable_plane+0x64/0x80 [i915] [92146.548417] [<f8639f7c>] i9xx_crtc_disable+0xbc/0x150 [i915] [92146.548456] [<f863ebee>] intel_crtc_update_dpms+0x5e/0x90 [i915] [92146.548493] [<f86437cf>] intel_modeset_setup_hw_state+0x42f/0x8f0 [i915] [92146.548535] [<f8645b0b>] intel_lid_notify+0x9b/0xc0 [i915] [92146.548543] [<c15610d3>] notifier_call_chain+0x43/0x60 [92146.548550] [<c105d1e1>] __blocking_notifier_call_chain+0x41/0x80 [92146.548556] [<c105d23f>] blocking_notifier_call_chain+0x1f/0x30 [92146.548563] [<c131a684>] acpi_lid_send_state+0x78/0xa4 [92146.548569] [<c131aa9e>] acpi_button_notify+0x3b/0xf1 [92146.548577] [<c12df56a>] ? acpi_os_execute+0x17/0x19 [92146.548582] [<c12e591a>] ? acpi_ec_sync_query+0xa5/0xbc [92146.548589] [<c12e2b82>] acpi_device_notify+0x16/0x18 [92146.548595] [<c12f4904>] acpi_ev_notify_dispatch+0x38/0x4f [92146.548600] [<c12df0e8>] acpi_os_execute_deferred+0x20/0x2b [92146.548607] [<c1051208>] process_one_work+0x128/0x3f0 [92146.548613] [<c1564f73>] ? common_interrupt+0x33/0x38 [92146.548618] [<c104f8c0>] ? wake_up_worker+0x30/0x30 [92146.548624] [<c12df0c8>] ? acpi_os_wait_events_complete+0x1e/0x1e [92146.548629] [<c10524f9>] worker_thread+0x119/0x3b0 [92146.548634] [<c10523e0>] ? manage_workers+0x240/0x240 [92146.548640] [<c1056e84>] kthread+0x94/0xa0 [92146.548647] [<c1060000>] ? ftrace_raw_output_sched_stat_runtime+0x70/0xf0 [92146.548652] [<c15649b7>] ret_from_kernel_thread+0x1b/0x28 [92146.548658] [<c1056df0>] ? kthread_create_on_node+0xc0/0xc0 three different modeset flags are introduced in this patch MODESET_ON_LID_OPEN: do modeset on next lid open event MODESET_DONE: modeset already done MODESET_SUSPENDED: suspended, only do modeset when system is resumed In this way, 1. when lid is closed, MODESET_ON_LID_OPEN is set so that we'll do modeset on next lid open event. 2. when lid is opened, MODESET_DONE is set so that duplicate lid open events will be ignored. 3. when system suspends, MODESET_SUSPENDED is set. In this case, we will not do modeset on any lid events. Plus, locking mechanism is also introduced to avoid racing. Signed-off-by: Zhang Rui <rui.zhang@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2013-02-05 00:41:53 -07:00
enum modeset_restore {
MODESET_ON_LID_OPEN,
MODESET_DONE,
MODESET_SUSPENDED,
};
#define DP_AUX_A 0x40
#define DP_AUX_B 0x10
#define DP_AUX_C 0x20
#define DP_AUX_D 0x30
#define DDC_PIN_B 0x05
#define DDC_PIN_C 0x04
#define DDC_PIN_D 0x06
struct ddi_vbt_port_info {
/*
* This is an index in the HDMI/DVI DDI buffer translation table.
* The special value HDMI_LEVEL_SHIFT_UNKNOWN means the VBT didn't
* populate this field.
*/
#define HDMI_LEVEL_SHIFT_UNKNOWN 0xff
uint8_t hdmi_level_shift;
uint8_t supports_dvi:1;
uint8_t supports_hdmi:1;
uint8_t supports_dp:1;
uint8_t supports_edp:1;
uint8_t alternate_aux_channel;
uint8_t alternate_ddc_pin;
uint8_t dp_boost_level;
uint8_t hdmi_boost_level;
};
enum psr_lines_to_wait {
PSR_0_LINES_TO_WAIT = 0,
PSR_1_LINE_TO_WAIT,
PSR_4_LINES_TO_WAIT,
PSR_8_LINES_TO_WAIT
};
struct intel_vbt_data {
struct drm_display_mode *lfp_lvds_vbt_mode; /* if any */
struct drm_display_mode *sdvo_lvds_vbt_mode; /* if any */
/* Feature bits */
unsigned int int_tv_support:1;
unsigned int lvds_dither:1;
unsigned int lvds_vbt:1;
unsigned int int_crt_support:1;
unsigned int lvds_use_ssc:1;
unsigned int display_clock_mode:1;
unsigned int fdi_rx_polarity_inverted:1;
unsigned int panel_type:4;
int lvds_ssc_freq;
unsigned int bios_lvds_val; /* initial [PCH_]LVDS reg val in VBIOS */
enum drrs_support_type drrs_type;
struct {
int rate;
int lanes;
int preemphasis;
int vswing;
bool low_vswing;
bool initialized;
bool support;
int bpp;
struct edp_power_seq pps;
} edp;
struct {
bool full_link;
bool require_aux_wakeup;
int idle_frames;
enum psr_lines_to_wait lines_to_wait;
int tp1_wakeup_time;
int tp2_tp3_wakeup_time;
} psr;
struct {
u16 pwm_freq_hz;
bool present;
bool active_low_pwm;
u8 min_brightness; /* min_brightness/255 of max */
u8 controller; /* brightness controller number */
enum intel_backlight_type type;
} backlight;
/* MIPI DSI */
struct {
u16 panel_id;
struct mipi_config *config;
struct mipi_pps_data *pps;
u8 seq_version;
u32 size;
u8 *data;
const u8 *sequence[MIPI_SEQ_MAX];
} dsi;
int crt_ddc_pin;
int child_dev_num;
union child_device_config *child_dev;
struct ddi_vbt_port_info ddi_port_info[I915_MAX_PORTS];
struct sdvo_device_mapping sdvo_mappings[2];
};
enum intel_ddb_partitioning {
INTEL_DDB_PART_1_2,
INTEL_DDB_PART_5_6, /* IVB+ */
};
struct intel_wm_level {
bool enable;
uint32_t pri_val;
uint32_t spr_val;
uint32_t cur_val;
uint32_t fbc_val;
};
struct ilk_wm_values {
uint32_t wm_pipe[3];
uint32_t wm_lp[3];
uint32_t wm_lp_spr[3];
uint32_t wm_linetime[3];
bool enable_fbc_wm;
enum intel_ddb_partitioning partitioning;
};
struct g4x_pipe_wm {
uint16_t plane[I915_MAX_PLANES];
drm/i915: Two stage watermarks for g4x Implement proper two stage watermark programming for g4x. As with other pre-SKL platforms, the watermark registers aren't double buffered on g4x. Hence we must sequence the watermark update carefully around plane updates. The code is quite heavily modelled on the VLV/CHV code, with some fairly significant differences due to the different hardware architecture: * g4x doesn't use inverted watermark values * CxSR actually affects the watermarks since it controls memory self refresh in addition to the max FIFO mode * A further HPLL SR mode is possible with higher memory wakeup latency * g4x has FBC2 and so it also has FBC watermarks * max FIFO mode for primary plane only (cursor is allowed, sprite is not) * g4x has no manual FIFO repartitioning * some TLB miss related workarounds are needed for the watermarks Actually the hardware is quite similar to ILK+ in many ways. The most visible differences are in the actual watermakr register layout. ILK revamped that part quite heavily whereas g4x is still using the layout inherited from earlier platforms. Note that we didn't previously enable the HPLL SR on g4x. So in order to not introduce too many functional changes in this patch I've not actually enabled it here either, even though the code is now fully ready for it. We'll enable it separately later on. Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/20170421181432.15216-13-ville.syrjala@linux.intel.com Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
2017-04-21 12:14:29 -06:00
uint16_t fbc;
};
drm/i915: Rewrite VLV/CHV watermark code Assuming the PND deadline mechanism works reasonably we should do memory requests as early as possible so that PND has schedule the requests more intelligently. Currently we're still calculating the watermarks as if VLV/CHV are identical to g4x, which isn't the case. The current code also seems to calculate insufficient watermarks and hence we're seeing some underruns, especially on high resolution displays. To fix it just rip out the current code and replace is with something that tries to utilize PND as efficiently as possible. We now calculate the WM watermark to trigger when the FIFO still has 256us worth of data. 256us is the maximum deadline value supoorted by PND, so issuing memory requests earlier would mean we probably couldn't utilize the full FIFO as PND would attempt to return the data at least in at least 256us. We also clamp the watermark to at least 8 cachelines as that's the magic watermark that enabling trickle feed would also impose. I'm assuming it matches some burst size. In theory we could just enable trickle feed and ignore the WM values, except trickle feed doesn't work with max fifo mode anyway, so we'd still need to calculate the SR watermarks. It seems cleaner to just disable trickle feed and calculate all watermarks the same way. Also trickle feed wouldn't account for the 256us max deadline value, thoguh that may be a moot point in non-max fifo mode sicne the FIFOs are fairly small. On VLV max fifo mode can be used with either primary or sprite planes. So the code now also checks all the planes (apart from the cursor) when calculating the SR plane watermark. We don't have to worry about the WM1 watermarks since we're using the PND deadline scheme which means the hardware ignores WM1 values. v2: Use plane->state->fb instead of plane->fb Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2015-03-05 12:19:49 -07:00
struct g4x_sr_wm {
uint16_t plane;
uint16_t cursor;
drm/i915: Two stage watermarks for g4x Implement proper two stage watermark programming for g4x. As with other pre-SKL platforms, the watermark registers aren't double buffered on g4x. Hence we must sequence the watermark update carefully around plane updates. The code is quite heavily modelled on the VLV/CHV code, with some fairly significant differences due to the different hardware architecture: * g4x doesn't use inverted watermark values * CxSR actually affects the watermarks since it controls memory self refresh in addition to the max FIFO mode * A further HPLL SR mode is possible with higher memory wakeup latency * g4x has FBC2 and so it also has FBC watermarks * max FIFO mode for primary plane only (cursor is allowed, sprite is not) * g4x has no manual FIFO repartitioning * some TLB miss related workarounds are needed for the watermarks Actually the hardware is quite similar to ILK+ in many ways. The most visible differences are in the actual watermakr register layout. ILK revamped that part quite heavily whereas g4x is still using the layout inherited from earlier platforms. Note that we didn't previously enable the HPLL SR on g4x. So in order to not introduce too many functional changes in this patch I've not actually enabled it here either, even though the code is now fully ready for it. We'll enable it separately later on. Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/20170421181432.15216-13-ville.syrjala@linux.intel.com Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
2017-04-21 12:14:29 -06:00
uint16_t fbc;
};
struct vlv_wm_ddl_values {
uint8_t plane[I915_MAX_PLANES];
};
drm/i915: Rewrite VLV/CHV watermark code Assuming the PND deadline mechanism works reasonably we should do memory requests as early as possible so that PND has schedule the requests more intelligently. Currently we're still calculating the watermarks as if VLV/CHV are identical to g4x, which isn't the case. The current code also seems to calculate insufficient watermarks and hence we're seeing some underruns, especially on high resolution displays. To fix it just rip out the current code and replace is with something that tries to utilize PND as efficiently as possible. We now calculate the WM watermark to trigger when the FIFO still has 256us worth of data. 256us is the maximum deadline value supoorted by PND, so issuing memory requests earlier would mean we probably couldn't utilize the full FIFO as PND would attempt to return the data at least in at least 256us. We also clamp the watermark to at least 8 cachelines as that's the magic watermark that enabling trickle feed would also impose. I'm assuming it matches some burst size. In theory we could just enable trickle feed and ignore the WM values, except trickle feed doesn't work with max fifo mode anyway, so we'd still need to calculate the SR watermarks. It seems cleaner to just disable trickle feed and calculate all watermarks the same way. Also trickle feed wouldn't account for the 256us max deadline value, thoguh that may be a moot point in non-max fifo mode sicne the FIFOs are fairly small. On VLV max fifo mode can be used with either primary or sprite planes. So the code now also checks all the planes (apart from the cursor) when calculating the SR plane watermark. We don't have to worry about the WM1 watermarks since we're using the PND deadline scheme which means the hardware ignores WM1 values. v2: Use plane->state->fb instead of plane->fb Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2015-03-05 12:19:49 -07:00
struct vlv_wm_values {
struct g4x_pipe_wm pipe[3];
struct g4x_sr_wm sr;
struct vlv_wm_ddl_values ddl[3];
uint8_t level;
bool cxsr;
};
drm/i915: Two stage watermarks for g4x Implement proper two stage watermark programming for g4x. As with other pre-SKL platforms, the watermark registers aren't double buffered on g4x. Hence we must sequence the watermark update carefully around plane updates. The code is quite heavily modelled on the VLV/CHV code, with some fairly significant differences due to the different hardware architecture: * g4x doesn't use inverted watermark values * CxSR actually affects the watermarks since it controls memory self refresh in addition to the max FIFO mode * A further HPLL SR mode is possible with higher memory wakeup latency * g4x has FBC2 and so it also has FBC watermarks * max FIFO mode for primary plane only (cursor is allowed, sprite is not) * g4x has no manual FIFO repartitioning * some TLB miss related workarounds are needed for the watermarks Actually the hardware is quite similar to ILK+ in many ways. The most visible differences are in the actual watermakr register layout. ILK revamped that part quite heavily whereas g4x is still using the layout inherited from earlier platforms. Note that we didn't previously enable the HPLL SR on g4x. So in order to not introduce too many functional changes in this patch I've not actually enabled it here either, even though the code is now fully ready for it. We'll enable it separately later on. Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/20170421181432.15216-13-ville.syrjala@linux.intel.com Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
2017-04-21 12:14:29 -06:00
struct g4x_wm_values {
struct g4x_pipe_wm pipe[2];
struct g4x_sr_wm sr;
struct g4x_sr_wm hpll;
bool cxsr;
bool hpll_en;
bool fbc_en;
};
struct skl_ddb_entry {
uint16_t start, end; /* in number of blocks, 'end' is exclusive */
};
static inline uint16_t skl_ddb_entry_size(const struct skl_ddb_entry *entry)
{
return entry->end - entry->start;
}
static inline bool skl_ddb_entry_equal(const struct skl_ddb_entry *e1,
const struct skl_ddb_entry *e2)
{
if (e1->start == e2->start && e1->end == e2->end)
return true;
return false;
}
struct skl_ddb_allocation {
struct skl_ddb_entry plane[I915_MAX_PIPES][I915_MAX_PLANES]; /* packed/uv */
drm/i915/skl: Simplify wm structures slightly (v2) A bunch of SKL watermark-related structures have the cursor plane as a separate entry from the rest of the planes. Since a previous patch updated I915_MAX_PLANES such that those plane arrays now have a slot for the cursor, update the code to use the new slot in the existing plane arrays and kill off the cursor-specific structures. There shouldn't be any functional change here; this is just shuffling around how the data is stored in some of the data structures. The whole patch is generated with Coccinelle via the following semantic patch: @@ struct skl_pipe_wm_parameters WMP; @@ - WMP.cursor + WMP.plane[PLANE_CURSOR] @@ struct skl_pipe_wm_parameters *WMP; @@ - WMP->cursor + WMP->plane[PLANE_CURSOR] @@ @@ struct skl_pipe_wm_parameters { ... - struct intel_plane_wm_parameters cursor; ... }; @@ struct skl_ddb_allocation DDB; expression E; @@ - DDB.cursor[E] + DDB.plane[E][PLANE_CURSOR] @@ struct skl_ddb_allocation *DDB; expression E; @@ - DDB->cursor[E] + DDB->plane[E][PLANE_CURSOR] @@ @@ struct skl_ddb_allocation { ... - struct skl_ddb_entry cursor[I915_MAX_PIPES]; ... }; @@ struct skl_wm_values WMV; expression E1, E2; @@ ( - WMV.cursor[E1][E2] + WMV.plane[E1][PLANE_CURSOR][E2] | - WMV.cursor_trans[E1] + WMV.plane_trans[E1][PLANE_CURSOR] ) @@ struct skl_wm_values *WMV; expression E1, E2; @@ ( - WMV->cursor[E1][E2] + WMV->plane[E1][PLANE_CURSOR][E2] | - WMV->cursor_trans[E1] + WMV->plane_trans[E1][PLANE_CURSOR] ) @@ @@ struct skl_wm_values { ... - uint32_t cursor[I915_MAX_PIPES][8]; ... - uint32_t cursor_trans[I915_MAX_PIPES]; ... }; @@ struct skl_wm_level WML; @@ ( - WML.cursor_en + WML.plane_en[PLANE_CURSOR] | - WML.cursor_res_b + WML.plane_res_b[PLANE_CURSOR] | - WML.cursor_res_l + WML.plane_res_l[PLANE_CURSOR] ) @@ struct skl_wm_level *WML; @@ ( - WML->cursor_en + WML->plane_en[PLANE_CURSOR] | - WML->cursor_res_b + WML->plane_res_b[PLANE_CURSOR] | - WML->cursor_res_l + WML->plane_res_l[PLANE_CURSOR] ) @@ @@ struct skl_wm_level { ... - bool cursor_en; ... - uint16_t cursor_res_b; - uint8_t cursor_res_l; ... }; v2: Use a PLANE_CURSOR enum entry rather than making the code reference I915_MAX_PLANES or I915_MAX_PLANES+1, which was confusing. (Ander) Signed-off-by: Matt Roper <matthew.d.roper@intel.com> Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2015-09-24 16:53:10 -06:00
struct skl_ddb_entry y_plane[I915_MAX_PIPES][I915_MAX_PLANES];
};
struct skl_wm_values {
unsigned dirty_pipes;
struct skl_ddb_allocation ddb;
};
struct skl_wm_level {
bool plane_en;
uint16_t plane_res_b;
uint8_t plane_res_l;
};
drm/i915: allow package C8+ states on Haswell (disabled) This patch allows PC8+ states on Haswell. These states can only be reached when all the display outputs are disabled, and they allow some more power savings. The fact that the graphics device is allowing PC8+ doesn't mean that the machine will actually enter PC8+: all the other devices also need to allow PC8+. For now this option is disabled by default. You need i915.allow_pc8=1 if you want it. This patch adds a big comment inside i915_drv.h explaining how it works and how it tracks things. Read it. v2: (this is not really v2, many previous versions were already sent, but they had different names) - Use the new functions to enable/disable GTIMR and GEN6_PMIMR - Rename almost all variables and functions to names suggested by Chris - More WARNs on the IRQ handling code - Also disable PC8 when there's GPU work to do (thanks to Ben for the help on this), so apps can run caster - Enable PC8 on a delayed work function that is delayed for 5 seconds. This makes sure we only enable PC8+ if we're really idle - Make sure we're not in PC8+ when suspending v3: - WARN if IRQs are disabled on __wait_seqno - Replace some DRM_ERRORs with WARNs - Fix calls to restore GT and PM interrupts - Use intel_mark_busy instead of intel_ring_advance to disable PC8 v4: - Use the force_wake, Luke! v5: - Remove the "IIR is not zero" WARNs - Move the force_wake chunk to its own patch - Only restore what's missing from RC6, not everything Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2013-08-19 10:18:09 -06:00
/*
* This struct helps tracking the state needed for runtime PM, which puts the
* device in PCI D3 state. Notice that when this happens, nothing on the
* graphics device works, even register access, so we don't get interrupts nor
* anything else.
drm/i915: allow package C8+ states on Haswell (disabled) This patch allows PC8+ states on Haswell. These states can only be reached when all the display outputs are disabled, and they allow some more power savings. The fact that the graphics device is allowing PC8+ doesn't mean that the machine will actually enter PC8+: all the other devices also need to allow PC8+. For now this option is disabled by default. You need i915.allow_pc8=1 if you want it. This patch adds a big comment inside i915_drv.h explaining how it works and how it tracks things. Read it. v2: (this is not really v2, many previous versions were already sent, but they had different names) - Use the new functions to enable/disable GTIMR and GEN6_PMIMR - Rename almost all variables and functions to names suggested by Chris - More WARNs on the IRQ handling code - Also disable PC8 when there's GPU work to do (thanks to Ben for the help on this), so apps can run caster - Enable PC8 on a delayed work function that is delayed for 5 seconds. This makes sure we only enable PC8+ if we're really idle - Make sure we're not in PC8+ when suspending v3: - WARN if IRQs are disabled on __wait_seqno - Replace some DRM_ERRORs with WARNs - Fix calls to restore GT and PM interrupts - Use intel_mark_busy instead of intel_ring_advance to disable PC8 v4: - Use the force_wake, Luke! v5: - Remove the "IIR is not zero" WARNs - Move the force_wake chunk to its own patch - Only restore what's missing from RC6, not everything Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2013-08-19 10:18:09 -06:00
*
* Every piece of our code that needs to actually touch the hardware needs to
* either call intel_runtime_pm_get or call intel_display_power_get with the
* appropriate power domain.
drm/i915: make PC8 be part of runtime PM suspend/resume Currently, when our driver becomes idle for i915.pc8_timeout (default: 5s) we enable PC8, so we save some power, but not everything we can. Then, while PC8 is enabled, if we stay idle for more autosuspend_delay_ms (default: 10s) we'll enter runtime PM and put the graphics device in D3 state, saving even more power. The two features are separate things with increasing levels of power savings, but if we disable PC8 we'll never get into D3. While from the modularity point of view it would be nice to keep these features as separate, we have reasons to merge them: - We are not aware of anybody wanting a "PC8 without D3" environment. - If we keep both features as separate, we'll have to to test both PC8 and PC8+D3 code paths. We're already having a major pain to make QA do automated testing of just one thing, testing both paths will cost even more. - Only Haswell+ supports PC8, so if we want to add runtime PM support to, for example, IVB, we'll have to copy some code from the PC8 feature to runtime PM, so merging both features as a single thing will make it easier for enabling runtime PM on other platforms. This patch only does the very basic steps required to have PC8 and runtime PM merged on a single feature: the next patches will take care of cleaning up everything. v2: - Rebase. v3: - Rebase. - Fully remove the deprecated i915 params since Daniel doesn't consider them as part of the ABI. v4: - Rebase. - Fix typo in the commit message. v5: - Rebase, again. - Add a huge comment explaining the different forcewake usage (Chris, Daniel). - Use open-coded forcewake functions (Daniel). Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com> Reviewed-by: Imre Deak <imre.deak@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2014-03-07 16:08:05 -07:00
*
* Our driver uses the autosuspend delay feature, which means we'll only really
* suspend if we stay with zero refcount for a certain amount of time. The
* default value is currently very conservative (see intel_runtime_pm_enable), but
* it can be changed with the standard runtime PM files from sysfs.
drm/i915: allow package C8+ states on Haswell (disabled) This patch allows PC8+ states on Haswell. These states can only be reached when all the display outputs are disabled, and they allow some more power savings. The fact that the graphics device is allowing PC8+ doesn't mean that the machine will actually enter PC8+: all the other devices also need to allow PC8+. For now this option is disabled by default. You need i915.allow_pc8=1 if you want it. This patch adds a big comment inside i915_drv.h explaining how it works and how it tracks things. Read it. v2: (this is not really v2, many previous versions were already sent, but they had different names) - Use the new functions to enable/disable GTIMR and GEN6_PMIMR - Rename almost all variables and functions to names suggested by Chris - More WARNs on the IRQ handling code - Also disable PC8 when there's GPU work to do (thanks to Ben for the help on this), so apps can run caster - Enable PC8 on a delayed work function that is delayed for 5 seconds. This makes sure we only enable PC8+ if we're really idle - Make sure we're not in PC8+ when suspending v3: - WARN if IRQs are disabled on __wait_seqno - Replace some DRM_ERRORs with WARNs - Fix calls to restore GT and PM interrupts - Use intel_mark_busy instead of intel_ring_advance to disable PC8 v4: - Use the force_wake, Luke! v5: - Remove the "IIR is not zero" WARNs - Move the force_wake chunk to its own patch - Only restore what's missing from RC6, not everything Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2013-08-19 10:18:09 -06:00
*
* The irqs_disabled variable becomes true exactly after we disable the IRQs and
* goes back to false exactly before we reenable the IRQs. We use this variable
* to check if someone is trying to enable/disable IRQs while they're supposed
* to be disabled. This shouldn't happen and we'll print some error messages in
* case it happens.
drm/i915: allow package C8+ states on Haswell (disabled) This patch allows PC8+ states on Haswell. These states can only be reached when all the display outputs are disabled, and they allow some more power savings. The fact that the graphics device is allowing PC8+ doesn't mean that the machine will actually enter PC8+: all the other devices also need to allow PC8+. For now this option is disabled by default. You need i915.allow_pc8=1 if you want it. This patch adds a big comment inside i915_drv.h explaining how it works and how it tracks things. Read it. v2: (this is not really v2, many previous versions were already sent, but they had different names) - Use the new functions to enable/disable GTIMR and GEN6_PMIMR - Rename almost all variables and functions to names suggested by Chris - More WARNs on the IRQ handling code - Also disable PC8 when there's GPU work to do (thanks to Ben for the help on this), so apps can run caster - Enable PC8 on a delayed work function that is delayed for 5 seconds. This makes sure we only enable PC8+ if we're really idle - Make sure we're not in PC8+ when suspending v3: - WARN if IRQs are disabled on __wait_seqno - Replace some DRM_ERRORs with WARNs - Fix calls to restore GT and PM interrupts - Use intel_mark_busy instead of intel_ring_advance to disable PC8 v4: - Use the force_wake, Luke! v5: - Remove the "IIR is not zero" WARNs - Move the force_wake chunk to its own patch - Only restore what's missing from RC6, not everything Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2013-08-19 10:18:09 -06:00
*
* For more, read the Documentation/power/runtime_pm.txt.
drm/i915: allow package C8+ states on Haswell (disabled) This patch allows PC8+ states on Haswell. These states can only be reached when all the display outputs are disabled, and they allow some more power savings. The fact that the graphics device is allowing PC8+ doesn't mean that the machine will actually enter PC8+: all the other devices also need to allow PC8+. For now this option is disabled by default. You need i915.allow_pc8=1 if you want it. This patch adds a big comment inside i915_drv.h explaining how it works and how it tracks things. Read it. v2: (this is not really v2, many previous versions were already sent, but they had different names) - Use the new functions to enable/disable GTIMR and GEN6_PMIMR - Rename almost all variables and functions to names suggested by Chris - More WARNs on the IRQ handling code - Also disable PC8 when there's GPU work to do (thanks to Ben for the help on this), so apps can run caster - Enable PC8 on a delayed work function that is delayed for 5 seconds. This makes sure we only enable PC8+ if we're really idle - Make sure we're not in PC8+ when suspending v3: - WARN if IRQs are disabled on __wait_seqno - Replace some DRM_ERRORs with WARNs - Fix calls to restore GT and PM interrupts - Use intel_mark_busy instead of intel_ring_advance to disable PC8 v4: - Use the force_wake, Luke! v5: - Remove the "IIR is not zero" WARNs - Move the force_wake chunk to its own patch - Only restore what's missing from RC6, not everything Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2013-08-19 10:18:09 -06:00
*/
struct i915_runtime_pm {
drm/i915: add support for checking if we hold an RPM reference Atm, we assert that the device is not suspended until the point when the device is truly put to a suspended state. This is fine, but we can catch more problems if we check that RPM refcount is non-zero. After that one drops to zero we shouldn't access the device any more, even if the actual device suspend may be delayed. Change assert_rpm_wakelock_held() accordingly to check for a non-zero RPM refcount in addition to the current device-not-suspended check. For the new asserts to work we need to annotate every place explicitly in the code where we expect that the device is powered. The places where we only assume this, but may not hold an RPM reference: - driver load We assume the device to be powered until we enable RPM. Make this explicit by taking an RPM reference around the load function. - system and runtime sudpend/resume handlers These handlers are called when the RPM reference becomes 0 and know the exact point after which the device can get powered off. Disable the RPM-reference-held check for their duration. - the IRQ, hangcheck and RPS work handlers These handlers are flushed in the system/runtime suspend handler before the device is powered off, so it's guaranteed that they won't run while the device is powered off even though they don't hold any RPM reference. Disable the RPM-reference-held check for their duration. In all these cases we still check that the device is not suspended. These explicit annotations also have the positive side effect of documenting our assumptions better. This caught additional WARNs from the atomic modeset path, those should be fixed separately. v2: - remove the redundant HAS_RUNTIME_PM check (moved to patch 1) (Ville) v3: - use a new dedicated RPM wakelock refcount to also catch cases where our own RPM get/put functions were not called (Chris) - assert also that the new RPM wakelock refcount is 0 in the RPM suspend handler (Chris) - change the assert error message to be more meaningful (Chris) - prevent false assert errors and check that the RPM wakelock is 0 in the RPM resume handler too - prevent false assert errors in the hangcheck work too - add a device not suspended assert check to the hangcheck work v4: - rename disable/enable_rpm_asserts to disable/enable_rpm_wakeref_asserts and wakelock_count to wakeref_count - disable the wakeref asserts in the IRQ handlers and RPS work too - update/clarify commit message v5: - mark places we plan to change to use proper RPM refcounting with separate DISABLE/ENABLE_RPM_WAKEREF_ASSERTS aliases (Chris) Signed-off-by: Imre Deak <imre.deak@intel.com> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Link: http://patchwork.freedesktop.org/patch/msgid/1450227139-13471-1-git-send-email-imre.deak@intel.com
2015-12-15 17:52:19 -07:00
atomic_t wakeref_count;
bool suspended;
bool irqs_enabled;
drm/i915: allow package C8+ states on Haswell (disabled) This patch allows PC8+ states on Haswell. These states can only be reached when all the display outputs are disabled, and they allow some more power savings. The fact that the graphics device is allowing PC8+ doesn't mean that the machine will actually enter PC8+: all the other devices also need to allow PC8+. For now this option is disabled by default. You need i915.allow_pc8=1 if you want it. This patch adds a big comment inside i915_drv.h explaining how it works and how it tracks things. Read it. v2: (this is not really v2, many previous versions were already sent, but they had different names) - Use the new functions to enable/disable GTIMR and GEN6_PMIMR - Rename almost all variables and functions to names suggested by Chris - More WARNs on the IRQ handling code - Also disable PC8 when there's GPU work to do (thanks to Ben for the help on this), so apps can run caster - Enable PC8 on a delayed work function that is delayed for 5 seconds. This makes sure we only enable PC8+ if we're really idle - Make sure we're not in PC8+ when suspending v3: - WARN if IRQs are disabled on __wait_seqno - Replace some DRM_ERRORs with WARNs - Fix calls to restore GT and PM interrupts - Use intel_mark_busy instead of intel_ring_advance to disable PC8 v4: - Use the force_wake, Luke! v5: - Remove the "IIR is not zero" WARNs - Move the force_wake chunk to its own patch - Only restore what's missing from RC6, not everything Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2013-08-19 10:18:09 -06:00
};
enum intel_pipe_crc_source {
INTEL_PIPE_CRC_SOURCE_NONE,
INTEL_PIPE_CRC_SOURCE_PLANE1,
INTEL_PIPE_CRC_SOURCE_PLANE2,
INTEL_PIPE_CRC_SOURCE_PF,
INTEL_PIPE_CRC_SOURCE_PIPE,
/* TV/DP on pre-gen5/vlv can't use the pipe source. */
INTEL_PIPE_CRC_SOURCE_TV,
INTEL_PIPE_CRC_SOURCE_DP_B,
INTEL_PIPE_CRC_SOURCE_DP_C,
INTEL_PIPE_CRC_SOURCE_DP_D,
INTEL_PIPE_CRC_SOURCE_AUTO,
INTEL_PIPE_CRC_SOURCE_MAX,
};
struct intel_pipe_crc_entry {
uint32_t frame;
uint32_t crc[5];
};
#define INTEL_PIPE_CRC_ENTRIES_NR 128
struct intel_pipe_crc {
spinlock_t lock;
bool opened; /* exclusive access to the result file */
struct intel_pipe_crc_entry *entries;
enum intel_pipe_crc_source source;
int head, tail;
wait_queue_head_t wq;
drm/i915: Use new CRC debugfs API The core provides now an ABI to userspace for generation of frame CRCs, so implement the ->set_crc_source() callback and reuse as much code as possible with the previous ABI implementation. When handling the pageflip interrupt, we skip 1 or 2 frames depending on the HW because they contain wrong values. For the legacy ABI for generating frame CRCs, this was done in userspace but now that we have a generic ABI it's better if it's not exposed by the kernel. v2: - Leave the legacy implementation in place as the ABI implementation in the core is incompatible with it. v3: - Use the "cooked" vblank counter so we have a whole 32 bits. - Make sure we don't mess with the state of the legacy CRC capture ABI implementation. v4: - Keep use of get_vblank_counter as in the legacy code, will be changed in a followup commit. v5: - Skip first frame or two as it's known that they contain wrong data. - A few fixes suggested by Emil Velikov. v6: - Rework programming of the HW registers to preserve previous behavior. v7: - Address whitespace issue. - Added a comment on why in the implementation of the new ABI we skip the 1st or 2nd frames. v9: - Add stub for intel_crtc_set_crc_source. v12: - Rebased. - Remove stub for intel_crtc_set_crc_source and instead set the callback to NULL (Jani Nikula). v15: - Rebased. Signed-off-by: Tomeu Vizoso <tomeu.vizoso@collabora.com> Reviewed-by: Emil Velikov <emil.velikov@collabora.com> Reviewed-by: Robert Foss <robert.foss@collabora.com> irq Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/20170110134305.26326-2-tomeu.vizoso@collabora.com
2017-01-10 06:43:04 -07:00
int skipped;
};
drm/i915: Track frontbuffer invalidation/flushing So these are the guts of the new beast. This tracks when a frontbuffer gets invalidated (due to frontbuffer rendering) and hence should be constantly scaned out, and when it's flushed again and can be compressed/one-shot-upload. Rules for flushing are simple: The frontbuffer needs one more full upload starting from the next vblank. Which means that the flushing can _only_ be called once the frontbuffer update has been latched. But this poses a problem for pageflips: We can't just delay the flushing until the pageflip is latched, since that would pose the risk that we override frontbuffer rendering that has been scheduled in-between the pageflip ioctl and the actual latching. To handle this track asynchronous invalidations (and also pageflip) state per-ring and delay any in-between flushing until the rendering has completed. And also cancel any delayed flushing if we get a new invalidation request (whether delayed or not). Also call intel_mark_fb_busy in both cases in all cases to make sure that we keep the screen at the highest refresh rate both on flips, synchronous plane updates and for frontbuffer rendering. v2: Lots of improvements Suggestions from Chris: - Move invalidate/flush in flush_*_domain and set_to_*_domain. - Drop the flush in busy_ioctl since it's redundant. Was a leftover from an earlier concept to track flips/delayed flushes. - Don't forget about the initial modeset enable/final disable. Suggested by Chris. Track flips accurately, too. Since flips complete independently of rendering we need to track pending flips in a separate mask. Again if an invalidate happens we need to cancel the evenutal flush to avoid races. v3: Provide correct header declarations for flip functions. Currently not needed outside of intel_display.c, but part of the proper interface. v4: Add proper domain management to fbcon so that the fbcon buffer is also tracked correctly. v5: Fixup locking around the fbcon set_to_gtt_domain call. v6: More comments from Chris: - Split out fbcon changes. - Drop superflous checks for potential scanout before calling intel_fb functions - we can micro-optimize this later. - s/intel_fb_/intel_fb_obj_/ to make it clear that this deals in gem object. We already have precedence for fb_obj in the pin_and_fence functions. v7: Clarify the semantics of the flip flush handling by renaming things a bit: - Don't go through a gem object but take the relevant frontbuffer bits directly. These functions center on the plane, the actual object is irrelevant - even a flip to the same object as already active should cause a flush. - Add a new intel_frontbuffer_flip for synchronous plane updates. It currently just calls intel_frontbuffer_flush since the implemenation differs. This way we achieve a clear split between one-shot update events on one side and frontbuffer rendering with potentially a very long delay between the invalidate and flush. Chris and I also had some discussions about mark_busy and whether it is appropriate to call from flush. But mark busy is a state which should be derived from the 3 events (invalidate, flush, flip) we now have by the users, like psr does by tracking relevant information in psr.busy_frontbuffer_bits. DRRS (the only real use of mark_busy for frontbuffer) needs to have similar logic. With that the overall mark_busy in the core could be removed. v8: Only when retiring gpu buffers only flush frontbuffer bits we actually invalidated in a batch. Just for safety since before any additional usage/invalidate we should always retire current rendering. Suggested by Chris Wilson. v9: Actually use intel_frontbuffer_flip in all appropriate places. Spotted by Chris. v10: Address more comments from Chris: - Don't call _flip in set_base when the crtc is inactive, avoids redunancy in the modeset case with the initial enabling of all planes. - Add comments explaining that the initial/final plane enable/disable still has work left to do before it's fully generic. v11: Only invalidate for gtt/cpu access when writing. Spotted by Chris. v12: s/_flush/_flip/ in intel_overlay.c per Chris' comment. Cc: Rodrigo Vivi <rodrigo.vivi@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2014-06-19 08:01:59 -06:00
struct i915_frontbuffer_tracking {
spinlock_t lock;
drm/i915: Track frontbuffer invalidation/flushing So these are the guts of the new beast. This tracks when a frontbuffer gets invalidated (due to frontbuffer rendering) and hence should be constantly scaned out, and when it's flushed again and can be compressed/one-shot-upload. Rules for flushing are simple: The frontbuffer needs one more full upload starting from the next vblank. Which means that the flushing can _only_ be called once the frontbuffer update has been latched. But this poses a problem for pageflips: We can't just delay the flushing until the pageflip is latched, since that would pose the risk that we override frontbuffer rendering that has been scheduled in-between the pageflip ioctl and the actual latching. To handle this track asynchronous invalidations (and also pageflip) state per-ring and delay any in-between flushing until the rendering has completed. And also cancel any delayed flushing if we get a new invalidation request (whether delayed or not). Also call intel_mark_fb_busy in both cases in all cases to make sure that we keep the screen at the highest refresh rate both on flips, synchronous plane updates and for frontbuffer rendering. v2: Lots of improvements Suggestions from Chris: - Move invalidate/flush in flush_*_domain and set_to_*_domain. - Drop the flush in busy_ioctl since it's redundant. Was a leftover from an earlier concept to track flips/delayed flushes. - Don't forget about the initial modeset enable/final disable. Suggested by Chris. Track flips accurately, too. Since flips complete independently of rendering we need to track pending flips in a separate mask. Again if an invalidate happens we need to cancel the evenutal flush to avoid races. v3: Provide correct header declarations for flip functions. Currently not needed outside of intel_display.c, but part of the proper interface. v4: Add proper domain management to fbcon so that the fbcon buffer is also tracked correctly. v5: Fixup locking around the fbcon set_to_gtt_domain call. v6: More comments from Chris: - Split out fbcon changes. - Drop superflous checks for potential scanout before calling intel_fb functions - we can micro-optimize this later. - s/intel_fb_/intel_fb_obj_/ to make it clear that this deals in gem object. We already have precedence for fb_obj in the pin_and_fence functions. v7: Clarify the semantics of the flip flush handling by renaming things a bit: - Don't go through a gem object but take the relevant frontbuffer bits directly. These functions center on the plane, the actual object is irrelevant - even a flip to the same object as already active should cause a flush. - Add a new intel_frontbuffer_flip for synchronous plane updates. It currently just calls intel_frontbuffer_flush since the implemenation differs. This way we achieve a clear split between one-shot update events on one side and frontbuffer rendering with potentially a very long delay between the invalidate and flush. Chris and I also had some discussions about mark_busy and whether it is appropriate to call from flush. But mark busy is a state which should be derived from the 3 events (invalidate, flush, flip) we now have by the users, like psr does by tracking relevant information in psr.busy_frontbuffer_bits. DRRS (the only real use of mark_busy for frontbuffer) needs to have similar logic. With that the overall mark_busy in the core could be removed. v8: Only when retiring gpu buffers only flush frontbuffer bits we actually invalidated in a batch. Just for safety since before any additional usage/invalidate we should always retire current rendering. Suggested by Chris Wilson. v9: Actually use intel_frontbuffer_flip in all appropriate places. Spotted by Chris. v10: Address more comments from Chris: - Don't call _flip in set_base when the crtc is inactive, avoids redunancy in the modeset case with the initial enabling of all planes. - Add comments explaining that the initial/final plane enable/disable still has work left to do before it's fully generic. v11: Only invalidate for gtt/cpu access when writing. Spotted by Chris. v12: s/_flush/_flip/ in intel_overlay.c per Chris' comment. Cc: Rodrigo Vivi <rodrigo.vivi@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2014-06-19 08:01:59 -06:00
/*
* Tracking bits for delayed frontbuffer flushing du to gpu activity or
* scheduled flips.
*/
unsigned busy_bits;
unsigned flip_bits;
};
struct i915_wa_reg {
drm/i915: Type safe register read/write Make I915_READ and I915_WRITE more type safe by wrapping the register offset in a struct. This should eliminate most of the fumbles we've had with misplaced parens. This only takes care of normal mmio registers. We could extend the idea to other register types and define each with its own struct. That way you wouldn't be able to accidentally pass the wrong thing to a specific register access function. The gpio_reg setup is probably the ugliest thing left. But I figure I'd just leave it for now, and wait for some divine inspiration to strike before making it nice. As for the generated code, it's actually a bit better sometimes. Eg. looking at i915_irq_handler(), we can see the following change: lea 0x70024(%rdx,%rax,1),%r9d mov $0x1,%edx - movslq %r9d,%r9 - mov %r9,%rsi - mov %r9,-0x58(%rbp) - callq *0xd8(%rbx) + mov %r9d,%esi + mov %r9d,-0x48(%rbp) callq *0xd8(%rbx) So previously gcc thought the register offset might be signed and decided to sign extend it, just in case. The rest appears to be mostly just minor shuffling of instructions. v2: i915_mmio_reg_{offset,equal,valid}() helpers added s/_REG/_MMIO/ in the register defines mo more switch statements left to worry about ring_emit stuff got sorted in a prep patch cmd parser, lrc context and w/a batch buildup also in prep patch vgpu stuff cleaned up and moved to a prep patch all other unrelated changes split out v3: Rebased due to BXT DSI/BLC, MOCS, etc. v4: Rebased due to churn, s/i915_mmio_reg_t/i915_reg_t/ Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Link: http://patchwork.freedesktop.org/patch/msgid/1447853606-2751-1-git-send-email-ville.syrjala@linux.intel.com
2015-11-18 06:33:26 -07:00
i915_reg_t addr;
u32 value;
/* bitmask representing WA bits */
u32 mask;
};
drm/i915/gen9: Add framework to whitelist specific GPU registers Some of the HW registers are privileged and cannot be written to from non-privileged batch buffers coming from userspace unless they are added to the HW whitelist. This whitelist is maintained by HW and it is different from SW whitelist. Userspace need write access to them to implement preemption related WA. The reason for using this approach is, the register bits that control preemption granularity at the HW level are not context save/restored; so even if we set these bits always in kernel they are going to change once the context is switched out. We can consider making them non-privileged by default but these registers also contain other chicken bits which should not be allowed to be modified. In the later revisions controlling bits are save/restored at context level but in the existing revisions these are exported via other debug registers and should be on the whitelist. This patch adds changes to provide HW with a list of registers to be whitelisted. HW checks this list during execution and provides access accordingly. HW imposes a limit on the number of registers on whitelist and it is per-engine. At this point we are only enabling whitelist for RCS and we don't foresee any requirement for other engines. The registers to be whitelisted are added using generic workaround list mechanism, even these are only enablers for userspace workarounds. But by sharing this mechanism we get some test assets without additional cost (Mika). v2: rebase v3: parameterize RING_FORCE_TO_NONPRIV() as _MMIO() should be limited to i915_reg.h (Ville), drop inline for wa_ring_whitelist_reg (Mika). v4: improvements suggested by Chris Wilson. Clarify that this is HW whitelist and different from the one maintained in driver. This list is engine specific but it gets initialized along with other WA which is RCS specific thing, so make it clear that we are not doing any cross engine setup during initialization. Make HW whitelist count of each engine available in debugfs. Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com> Cc: Mika Kuoppala <mika.kuoppala@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Arun Siluvery <arun.siluvery@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1453412634-29238-2-git-send-email-arun.siluvery@linux.intel.com Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2016-01-21 14:43:47 -07:00
/*
* RING_MAX_NONPRIV_SLOTS is per-engine but at this point we are only
* allowing it for RCS as we don't foresee any requirement of having
* a whitelist for other engines. When it is really required for
* other engines then the limit need to be increased.
*/
#define I915_MAX_WA_REGS (16 + RING_MAX_NONPRIV_SLOTS)
struct i915_workarounds {
struct i915_wa_reg reg[I915_MAX_WA_REGS];
u32 count;
u32 hw_whitelist_count[I915_NUM_ENGINES];
};
drm/i915: Introduce a PV INFO page structure for Intel GVT-g. Introduce a PV INFO structure, to facilitate the Intel GVT-g technology, which is a GPU virtualization solution with mediated pass-through. This page contains the shared information between i915 driver and the host emulator. For now, this structure utilizes an area of 4K bytes on HSW GPU's unused MMIO space. Future hardware will have the reserved window architecturally defined, and layout of the page will be added in future BSpec. The i915 driver load routine detects if it is running in a VM by reading the contents of this PV INFO page. Thereafter a flag, vgpu.active is set, and intel_vgpu_active() is used by checking this flag to conclude if GPU is virtualized with Intel GVT-g. By now, intel_vgpu_active() will return true, only when the driver is running as a guest in the Intel GVT-g enhanced environment on HSW platform. v2: take Chris' comments: - call the i915_check_vgpu() in intel_uncore_init() - sanitize i915_check_vgpu() by adding BUILD_BUG_ON() and debug info take Daniel's comments: - put the definition of PV INFO into a new header - i915_vgt_if.h other changes: - access mmio regs by readq/readw in i915_check_vgpu() v3: take Daniel's comments: - move the i915/vgt interfaces into a new i915_vgpu.c - update makefile - add kerneldoc to functions which are non-static - add a DOC: section describing some of the high-level design - update drm docbook other changes: - rename i915_vgt_if.h to i915_vgpu.h v4: take Tvrtko's comments: - fix a typo in commit message - add debug message when vgt version mismatches - rename low_gmadr/high_gmadr to mappable/non-mappable in PV INFO structure Signed-off-by: Yu Zhang <yu.c.zhang@linux.intel.com> Signed-off-by: Jike Song <jike.song@intel.com> Signed-off-by: Eddie Dong <eddie.dong@intel.com> Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2015-02-10 04:05:47 -07:00
struct i915_virtual_gpu {
bool active;
drm/i915: Enable guest i915 full ppgtt functionality Enable the guest i915 full ppgtt functionality when host can provide this capability. vgt_caps is introduced to guest i915 driver to get the vgpu capabilities from the device model. VGT_CPAS_FULL_PPGTT is one of the capabilities type to let guest i915 dirver know that the guest i915 full ppgtt is supported by device model. Notice that the minor version of pvinfo isn't bumped because of this vgt_caps introduction, due to older guest would be broken by simply increasing the pvinfo version. Although the pvinfo minor version doesn't increase, the compatibility won't be blocked. The compatibility is ensured by checking the value of caps field in pvinfo. Zero means no full ppgtt support and BIT(2) means this feature is provided. Changes since v1: - Use u32 instead of uint32_t (Joonas) - Move VGT_CAPS_FULL_PPGTT introduction to this patch and use #define instead of enum (Joonas) - Rewrite the vgpu full ppgtt capability checking logic. (Joonas) - Some coding style refine. (Joonas) Changes since v2: - Divide the whole patch set into two separate patch series, with one patch in i915 side to check guest i915 full ppgtt capability and enable it when this capability is supported by the device model, and the other one in gvt side which fixs the blocking issue and enables the device model to provide the capability to guest. And this patch focuses on guest i915 side. (Joonas) - Change the title from "introduce vgt_caps to pvinfo" to "Enable guest i915 full ppgtt functionality". (Tina) Change since v3: - Add some comments about pvinfo caps and version. (Joonas) Change since v4: - Tested by Tina Zhang. Change since v5: - Add limitation about supporting 32bit full ppgtt. Change since v6: - Change the fallback to 48bit full ppgtt if i915.ppgtt_enable=2. (Zhenyu) Change in v9: - Remove the fixme comment due to no plan for 32bit full ppgtt support. (Zhenyu) - Reorder the patch-set to fix compiling issue with git-bisect. (Zhenyu) - Add print log when forcing guest 48bit full ppgtt. (Zhenyu) v10: - Update against Joonas's has_full_ppgtt and has_full_48bit_ppgtt disconnect change. (Zhenyu) Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> # in v2 Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Tina Zhang <tina.zhang@intel.com> Signed-off-by: Tina Zhang <tina.zhang@intel.com> Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
2017-08-14 01:20:46 -06:00
u32 caps;
drm/i915: Introduce a PV INFO page structure for Intel GVT-g. Introduce a PV INFO structure, to facilitate the Intel GVT-g technology, which is a GPU virtualization solution with mediated pass-through. This page contains the shared information between i915 driver and the host emulator. For now, this structure utilizes an area of 4K bytes on HSW GPU's unused MMIO space. Future hardware will have the reserved window architecturally defined, and layout of the page will be added in future BSpec. The i915 driver load routine detects if it is running in a VM by reading the contents of this PV INFO page. Thereafter a flag, vgpu.active is set, and intel_vgpu_active() is used by checking this flag to conclude if GPU is virtualized with Intel GVT-g. By now, intel_vgpu_active() will return true, only when the driver is running as a guest in the Intel GVT-g enhanced environment on HSW platform. v2: take Chris' comments: - call the i915_check_vgpu() in intel_uncore_init() - sanitize i915_check_vgpu() by adding BUILD_BUG_ON() and debug info take Daniel's comments: - put the definition of PV INFO into a new header - i915_vgt_if.h other changes: - access mmio regs by readq/readw in i915_check_vgpu() v3: take Daniel's comments: - move the i915/vgt interfaces into a new i915_vgpu.c - update makefile - add kerneldoc to functions which are non-static - add a DOC: section describing some of the high-level design - update drm docbook other changes: - rename i915_vgt_if.h to i915_vgpu.h v4: take Tvrtko's comments: - fix a typo in commit message - add debug message when vgt version mismatches - rename low_gmadr/high_gmadr to mappable/non-mappable in PV INFO structure Signed-off-by: Yu Zhang <yu.c.zhang@linux.intel.com> Signed-off-by: Jike Song <jike.song@intel.com> Signed-off-by: Eddie Dong <eddie.dong@intel.com> Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2015-02-10 04:05:47 -07:00
};
/* used in computing the new watermarks state */
struct intel_wm_config {
unsigned int num_pipes_active;
bool sprites_enabled;
bool sprites_scaled;
};
struct i915_oa_format {
u32 format;
int size;
};
struct i915_oa_reg {
i915_reg_t addr;
u32 value;
};
struct i915_oa_config {
char uuid[UUID_STRING_LEN + 1];
int id;
const struct i915_oa_reg *mux_regs;
u32 mux_regs_len;
const struct i915_oa_reg *b_counter_regs;
u32 b_counter_regs_len;
const struct i915_oa_reg *flex_regs;
u32 flex_regs_len;
struct attribute_group sysfs_metric;
struct attribute *attrs[2];
struct device_attribute sysfs_metric_id;
drm/i915/perf: Implement I915_PERF_ADD/REMOVE_CONFIG interface The motivation behind this new interface is expose at runtime the creation of new OA configs which can be used as part of the i915 perf open interface. This will enable the kernel to learn new configs which may be experimental, or otherwise not part of the core set currently available through the i915 perf interface. v2: Drop DRM_ERROR for userspace errors (Matthew) Add padding to userspace structure (Matthew) s/guid/uuid/ (Matthew) v3: Use u32 instead of int to iterate through registers (Matthew) v4: Lock access to dynamic config list (Lionel) v5: by Matthew: Fix uninitialized error values Fix incorrect unwiding when opening perf stream Use kmalloc_array() to store register Use uuid_is_valid() to valid config uuids Declare ioctls as write only Check padding members are set to 0 by Lionel: Return ENOENT rather than EINVAL when trying to remove non existing config v6: by Chris: Use ref counts for OA configs Store UUID in drm_i915_perf_oa_config rather then using pointer Shuffle fields of drm_i915_perf_oa_config to avoid padding v7: by Chris Rename uapi pointers fields to end with '_ptr' v8: by Andrzej, Marek, Sebastian Update register whitelisting by Lionel Add more register names for documentation Allow configuration programming in non-paranoid mode Add support for value filter for a couple of registers already programmed in other part of the kernel v9: Documentation fix (Lionel) Allow writing WAIT_FOR_RC6_EXIT only on Gen8+ (Andrzej) v10: Perform read access_ok() on register pointers (Lionel) Signed-off-by: Matthew Auld <matthew.auld@intel.com> Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com> Signed-off-by: Andrzej Datczuk <andrzej.datczuk@intel.com> Reviewed-by: Andrzej Datczuk <andrzej.datczuk@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20170803165812.2373-2-lionel.g.landwerlin@intel.com
2017-08-03 11:05:50 -06:00
atomic_t ref_count;
};
drm/i915: Add i915 perf infrastructure Adds base i915 perf infrastructure for Gen performance metrics. This adds a DRM_IOCTL_I915_PERF_OPEN ioctl that takes an array of uint64 properties to configure a stream of metrics and returns a new fd usable with standard VFS system calls including read() to read typed and sized records; ioctl() to enable or disable capture and poll() to wait for data. A stream is opened something like: uint64_t properties[] = { /* Single context sampling */ DRM_I915_PERF_PROP_CTX_HANDLE, ctx_handle, /* Include OA reports in samples */ DRM_I915_PERF_PROP_SAMPLE_OA, true, /* OA unit configuration */ DRM_I915_PERF_PROP_OA_METRICS_SET, metrics_set_id, DRM_I915_PERF_PROP_OA_FORMAT, report_format, DRM_I915_PERF_PROP_OA_EXPONENT, period_exponent, }; struct drm_i915_perf_open_param parm = { .flags = I915_PERF_FLAG_FD_CLOEXEC | I915_PERF_FLAG_FD_NONBLOCK | I915_PERF_FLAG_DISABLED, .properties_ptr = (uint64_t)properties, .num_properties = sizeof(properties) / 16, }; int fd = drmIoctl(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param); Records read all start with a common { type, size } header with DRM_I915_PERF_RECORD_SAMPLE being of most interest. Sample records contain an extensible number of fields and it's the DRM_I915_PERF_PROP_SAMPLE_xyz properties given when opening that determine what's included in every sample. No specific streams are supported yet so any attempt to open a stream will return an error. v2: use i915_gem_context_get() - Chris Wilson v3: update read() interface to avoid passing state struct - Chris Wilson fix some rebase fallout, with i915-perf init/deinit v4: s/DRM_IORW/DRM_IOW/ - Emil Velikov Signed-off-by: Robert Bragg <robert@sixbynine.org> Reviewed-by: Matthew Auld <matthew.auld@intel.com> Reviewed-by: Sourab Gupta <sourab.gupta@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/20161107194957.3385-2-robert@sixbynine.org
2016-11-07 12:49:47 -07:00
struct i915_perf_stream;
/**
* struct i915_perf_stream_ops - the OPs to support a specific stream type
*/
drm/i915: Add i915 perf infrastructure Adds base i915 perf infrastructure for Gen performance metrics. This adds a DRM_IOCTL_I915_PERF_OPEN ioctl that takes an array of uint64 properties to configure a stream of metrics and returns a new fd usable with standard VFS system calls including read() to read typed and sized records; ioctl() to enable or disable capture and poll() to wait for data. A stream is opened something like: uint64_t properties[] = { /* Single context sampling */ DRM_I915_PERF_PROP_CTX_HANDLE, ctx_handle, /* Include OA reports in samples */ DRM_I915_PERF_PROP_SAMPLE_OA, true, /* OA unit configuration */ DRM_I915_PERF_PROP_OA_METRICS_SET, metrics_set_id, DRM_I915_PERF_PROP_OA_FORMAT, report_format, DRM_I915_PERF_PROP_OA_EXPONENT, period_exponent, }; struct drm_i915_perf_open_param parm = { .flags = I915_PERF_FLAG_FD_CLOEXEC | I915_PERF_FLAG_FD_NONBLOCK | I915_PERF_FLAG_DISABLED, .properties_ptr = (uint64_t)properties, .num_properties = sizeof(properties) / 16, }; int fd = drmIoctl(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param); Records read all start with a common { type, size } header with DRM_I915_PERF_RECORD_SAMPLE being of most interest. Sample records contain an extensible number of fields and it's the DRM_I915_PERF_PROP_SAMPLE_xyz properties given when opening that determine what's included in every sample. No specific streams are supported yet so any attempt to open a stream will return an error. v2: use i915_gem_context_get() - Chris Wilson v3: update read() interface to avoid passing state struct - Chris Wilson fix some rebase fallout, with i915-perf init/deinit v4: s/DRM_IORW/DRM_IOW/ - Emil Velikov Signed-off-by: Robert Bragg <robert@sixbynine.org> Reviewed-by: Matthew Auld <matthew.auld@intel.com> Reviewed-by: Sourab Gupta <sourab.gupta@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/20161107194957.3385-2-robert@sixbynine.org
2016-11-07 12:49:47 -07:00
struct i915_perf_stream_ops {
/**
* @enable: Enables the collection of HW samples, either in response to
* `I915_PERF_IOCTL_ENABLE` or implicitly called when stream is opened
* without `I915_PERF_FLAG_DISABLED`.
drm/i915: Add i915 perf infrastructure Adds base i915 perf infrastructure for Gen performance metrics. This adds a DRM_IOCTL_I915_PERF_OPEN ioctl that takes an array of uint64 properties to configure a stream of metrics and returns a new fd usable with standard VFS system calls including read() to read typed and sized records; ioctl() to enable or disable capture and poll() to wait for data. A stream is opened something like: uint64_t properties[] = { /* Single context sampling */ DRM_I915_PERF_PROP_CTX_HANDLE, ctx_handle, /* Include OA reports in samples */ DRM_I915_PERF_PROP_SAMPLE_OA, true, /* OA unit configuration */ DRM_I915_PERF_PROP_OA_METRICS_SET, metrics_set_id, DRM_I915_PERF_PROP_OA_FORMAT, report_format, DRM_I915_PERF_PROP_OA_EXPONENT, period_exponent, }; struct drm_i915_perf_open_param parm = { .flags = I915_PERF_FLAG_FD_CLOEXEC | I915_PERF_FLAG_FD_NONBLOCK | I915_PERF_FLAG_DISABLED, .properties_ptr = (uint64_t)properties, .num_properties = sizeof(properties) / 16, }; int fd = drmIoctl(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param); Records read all start with a common { type, size } header with DRM_I915_PERF_RECORD_SAMPLE being of most interest. Sample records contain an extensible number of fields and it's the DRM_I915_PERF_PROP_SAMPLE_xyz properties given when opening that determine what's included in every sample. No specific streams are supported yet so any attempt to open a stream will return an error. v2: use i915_gem_context_get() - Chris Wilson v3: update read() interface to avoid passing state struct - Chris Wilson fix some rebase fallout, with i915-perf init/deinit v4: s/DRM_IORW/DRM_IOW/ - Emil Velikov Signed-off-by: Robert Bragg <robert@sixbynine.org> Reviewed-by: Matthew Auld <matthew.auld@intel.com> Reviewed-by: Sourab Gupta <sourab.gupta@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/20161107194957.3385-2-robert@sixbynine.org
2016-11-07 12:49:47 -07:00
*/
void (*enable)(struct i915_perf_stream *stream);
/**
* @disable: Disables the collection of HW samples, either in response
* to `I915_PERF_IOCTL_DISABLE` or implicitly called before destroying
* the stream.
drm/i915: Add i915 perf infrastructure Adds base i915 perf infrastructure for Gen performance metrics. This adds a DRM_IOCTL_I915_PERF_OPEN ioctl that takes an array of uint64 properties to configure a stream of metrics and returns a new fd usable with standard VFS system calls including read() to read typed and sized records; ioctl() to enable or disable capture and poll() to wait for data. A stream is opened something like: uint64_t properties[] = { /* Single context sampling */ DRM_I915_PERF_PROP_CTX_HANDLE, ctx_handle, /* Include OA reports in samples */ DRM_I915_PERF_PROP_SAMPLE_OA, true, /* OA unit configuration */ DRM_I915_PERF_PROP_OA_METRICS_SET, metrics_set_id, DRM_I915_PERF_PROP_OA_FORMAT, report_format, DRM_I915_PERF_PROP_OA_EXPONENT, period_exponent, }; struct drm_i915_perf_open_param parm = { .flags = I915_PERF_FLAG_FD_CLOEXEC | I915_PERF_FLAG_FD_NONBLOCK | I915_PERF_FLAG_DISABLED, .properties_ptr = (uint64_t)properties, .num_properties = sizeof(properties) / 16, }; int fd = drmIoctl(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param); Records read all start with a common { type, size } header with DRM_I915_PERF_RECORD_SAMPLE being of most interest. Sample records contain an extensible number of fields and it's the DRM_I915_PERF_PROP_SAMPLE_xyz properties given when opening that determine what's included in every sample. No specific streams are supported yet so any attempt to open a stream will return an error. v2: use i915_gem_context_get() - Chris Wilson v3: update read() interface to avoid passing state struct - Chris Wilson fix some rebase fallout, with i915-perf init/deinit v4: s/DRM_IORW/DRM_IOW/ - Emil Velikov Signed-off-by: Robert Bragg <robert@sixbynine.org> Reviewed-by: Matthew Auld <matthew.auld@intel.com> Reviewed-by: Sourab Gupta <sourab.gupta@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/20161107194957.3385-2-robert@sixbynine.org
2016-11-07 12:49:47 -07:00
*/
void (*disable)(struct i915_perf_stream *stream);
/**
* @poll_wait: Call poll_wait, passing a wait queue that will be woken
drm/i915: Add i915 perf infrastructure Adds base i915 perf infrastructure for Gen performance metrics. This adds a DRM_IOCTL_I915_PERF_OPEN ioctl that takes an array of uint64 properties to configure a stream of metrics and returns a new fd usable with standard VFS system calls including read() to read typed and sized records; ioctl() to enable or disable capture and poll() to wait for data. A stream is opened something like: uint64_t properties[] = { /* Single context sampling */ DRM_I915_PERF_PROP_CTX_HANDLE, ctx_handle, /* Include OA reports in samples */ DRM_I915_PERF_PROP_SAMPLE_OA, true, /* OA unit configuration */ DRM_I915_PERF_PROP_OA_METRICS_SET, metrics_set_id, DRM_I915_PERF_PROP_OA_FORMAT, report_format, DRM_I915_PERF_PROP_OA_EXPONENT, period_exponent, }; struct drm_i915_perf_open_param parm = { .flags = I915_PERF_FLAG_FD_CLOEXEC | I915_PERF_FLAG_FD_NONBLOCK | I915_PERF_FLAG_DISABLED, .properties_ptr = (uint64_t)properties, .num_properties = sizeof(properties) / 16, }; int fd = drmIoctl(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param); Records read all start with a common { type, size } header with DRM_I915_PERF_RECORD_SAMPLE being of most interest. Sample records contain an extensible number of fields and it's the DRM_I915_PERF_PROP_SAMPLE_xyz properties given when opening that determine what's included in every sample. No specific streams are supported yet so any attempt to open a stream will return an error. v2: use i915_gem_context_get() - Chris Wilson v3: update read() interface to avoid passing state struct - Chris Wilson fix some rebase fallout, with i915-perf init/deinit v4: s/DRM_IORW/DRM_IOW/ - Emil Velikov Signed-off-by: Robert Bragg <robert@sixbynine.org> Reviewed-by: Matthew Auld <matthew.auld@intel.com> Reviewed-by: Sourab Gupta <sourab.gupta@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/20161107194957.3385-2-robert@sixbynine.org
2016-11-07 12:49:47 -07:00
* once there is something ready to read() for the stream
*/
void (*poll_wait)(struct i915_perf_stream *stream,
struct file *file,
poll_table *wait);
/**
* @wait_unlocked: For handling a blocking read, wait until there is
* something to ready to read() for the stream. E.g. wait on the same
* wait queue that would be passed to poll_wait().
drm/i915: Add i915 perf infrastructure Adds base i915 perf infrastructure for Gen performance metrics. This adds a DRM_IOCTL_I915_PERF_OPEN ioctl that takes an array of uint64 properties to configure a stream of metrics and returns a new fd usable with standard VFS system calls including read() to read typed and sized records; ioctl() to enable or disable capture and poll() to wait for data. A stream is opened something like: uint64_t properties[] = { /* Single context sampling */ DRM_I915_PERF_PROP_CTX_HANDLE, ctx_handle, /* Include OA reports in samples */ DRM_I915_PERF_PROP_SAMPLE_OA, true, /* OA unit configuration */ DRM_I915_PERF_PROP_OA_METRICS_SET, metrics_set_id, DRM_I915_PERF_PROP_OA_FORMAT, report_format, DRM_I915_PERF_PROP_OA_EXPONENT, period_exponent, }; struct drm_i915_perf_open_param parm = { .flags = I915_PERF_FLAG_FD_CLOEXEC | I915_PERF_FLAG_FD_NONBLOCK | I915_PERF_FLAG_DISABLED, .properties_ptr = (uint64_t)properties, .num_properties = sizeof(properties) / 16, }; int fd = drmIoctl(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param); Records read all start with a common { type, size } header with DRM_I915_PERF_RECORD_SAMPLE being of most interest. Sample records contain an extensible number of fields and it's the DRM_I915_PERF_PROP_SAMPLE_xyz properties given when opening that determine what's included in every sample. No specific streams are supported yet so any attempt to open a stream will return an error. v2: use i915_gem_context_get() - Chris Wilson v3: update read() interface to avoid passing state struct - Chris Wilson fix some rebase fallout, with i915-perf init/deinit v4: s/DRM_IORW/DRM_IOW/ - Emil Velikov Signed-off-by: Robert Bragg <robert@sixbynine.org> Reviewed-by: Matthew Auld <matthew.auld@intel.com> Reviewed-by: Sourab Gupta <sourab.gupta@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/20161107194957.3385-2-robert@sixbynine.org
2016-11-07 12:49:47 -07:00
*/
int (*wait_unlocked)(struct i915_perf_stream *stream);
/**
* @read: Copy buffered metrics as records to userspace
* **buf**: the userspace, destination buffer
* **count**: the number of bytes to copy, requested by userspace
* **offset**: zero at the start of the read, updated as the read
* proceeds, it represents how many bytes have been copied so far and
* the buffer offset for copying the next record.
drm/i915: Add i915 perf infrastructure Adds base i915 perf infrastructure for Gen performance metrics. This adds a DRM_IOCTL_I915_PERF_OPEN ioctl that takes an array of uint64 properties to configure a stream of metrics and returns a new fd usable with standard VFS system calls including read() to read typed and sized records; ioctl() to enable or disable capture and poll() to wait for data. A stream is opened something like: uint64_t properties[] = { /* Single context sampling */ DRM_I915_PERF_PROP_CTX_HANDLE, ctx_handle, /* Include OA reports in samples */ DRM_I915_PERF_PROP_SAMPLE_OA, true, /* OA unit configuration */ DRM_I915_PERF_PROP_OA_METRICS_SET, metrics_set_id, DRM_I915_PERF_PROP_OA_FORMAT, report_format, DRM_I915_PERF_PROP_OA_EXPONENT, period_exponent, }; struct drm_i915_perf_open_param parm = { .flags = I915_PERF_FLAG_FD_CLOEXEC | I915_PERF_FLAG_FD_NONBLOCK | I915_PERF_FLAG_DISABLED, .properties_ptr = (uint64_t)properties, .num_properties = sizeof(properties) / 16, }; int fd = drmIoctl(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param); Records read all start with a common { type, size } header with DRM_I915_PERF_RECORD_SAMPLE being of most interest. Sample records contain an extensible number of fields and it's the DRM_I915_PERF_PROP_SAMPLE_xyz properties given when opening that determine what's included in every sample. No specific streams are supported yet so any attempt to open a stream will return an error. v2: use i915_gem_context_get() - Chris Wilson v3: update read() interface to avoid passing state struct - Chris Wilson fix some rebase fallout, with i915-perf init/deinit v4: s/DRM_IORW/DRM_IOW/ - Emil Velikov Signed-off-by: Robert Bragg <robert@sixbynine.org> Reviewed-by: Matthew Auld <matthew.auld@intel.com> Reviewed-by: Sourab Gupta <sourab.gupta@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/20161107194957.3385-2-robert@sixbynine.org
2016-11-07 12:49:47 -07:00
*
* Copy as many buffered i915 perf samples and records for this stream
* to userspace as will fit in the given buffer.
drm/i915: Add i915 perf infrastructure Adds base i915 perf infrastructure for Gen performance metrics. This adds a DRM_IOCTL_I915_PERF_OPEN ioctl that takes an array of uint64 properties to configure a stream of metrics and returns a new fd usable with standard VFS system calls including read() to read typed and sized records; ioctl() to enable or disable capture and poll() to wait for data. A stream is opened something like: uint64_t properties[] = { /* Single context sampling */ DRM_I915_PERF_PROP_CTX_HANDLE, ctx_handle, /* Include OA reports in samples */ DRM_I915_PERF_PROP_SAMPLE_OA, true, /* OA unit configuration */ DRM_I915_PERF_PROP_OA_METRICS_SET, metrics_set_id, DRM_I915_PERF_PROP_OA_FORMAT, report_format, DRM_I915_PERF_PROP_OA_EXPONENT, period_exponent, }; struct drm_i915_perf_open_param parm = { .flags = I915_PERF_FLAG_FD_CLOEXEC | I915_PERF_FLAG_FD_NONBLOCK | I915_PERF_FLAG_DISABLED, .properties_ptr = (uint64_t)properties, .num_properties = sizeof(properties) / 16, }; int fd = drmIoctl(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param); Records read all start with a common { type, size } header with DRM_I915_PERF_RECORD_SAMPLE being of most interest. Sample records contain an extensible number of fields and it's the DRM_I915_PERF_PROP_SAMPLE_xyz properties given when opening that determine what's included in every sample. No specific streams are supported yet so any attempt to open a stream will return an error. v2: use i915_gem_context_get() - Chris Wilson v3: update read() interface to avoid passing state struct - Chris Wilson fix some rebase fallout, with i915-perf init/deinit v4: s/DRM_IORW/DRM_IOW/ - Emil Velikov Signed-off-by: Robert Bragg <robert@sixbynine.org> Reviewed-by: Matthew Auld <matthew.auld@intel.com> Reviewed-by: Sourab Gupta <sourab.gupta@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/20161107194957.3385-2-robert@sixbynine.org
2016-11-07 12:49:47 -07:00
*
* Only write complete records; returning -%ENOSPC if there isn't room
* for a complete record.
drm/i915: Add i915 perf infrastructure Adds base i915 perf infrastructure for Gen performance metrics. This adds a DRM_IOCTL_I915_PERF_OPEN ioctl that takes an array of uint64 properties to configure a stream of metrics and returns a new fd usable with standard VFS system calls including read() to read typed and sized records; ioctl() to enable or disable capture and poll() to wait for data. A stream is opened something like: uint64_t properties[] = { /* Single context sampling */ DRM_I915_PERF_PROP_CTX_HANDLE, ctx_handle, /* Include OA reports in samples */ DRM_I915_PERF_PROP_SAMPLE_OA, true, /* OA unit configuration */ DRM_I915_PERF_PROP_OA_METRICS_SET, metrics_set_id, DRM_I915_PERF_PROP_OA_FORMAT, report_format, DRM_I915_PERF_PROP_OA_EXPONENT, period_exponent, }; struct drm_i915_perf_open_param parm = { .flags = I915_PERF_FLAG_FD_CLOEXEC | I915_PERF_FLAG_FD_NONBLOCK | I915_PERF_FLAG_DISABLED, .properties_ptr = (uint64_t)properties, .num_properties = sizeof(properties) / 16, }; int fd = drmIoctl(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param); Records read all start with a common { type, size } header with DRM_I915_PERF_RECORD_SAMPLE being of most interest. Sample records contain an extensible number of fields and it's the DRM_I915_PERF_PROP_SAMPLE_xyz properties given when opening that determine what's included in every sample. No specific streams are supported yet so any attempt to open a stream will return an error. v2: use i915_gem_context_get() - Chris Wilson v3: update read() interface to avoid passing state struct - Chris Wilson fix some rebase fallout, with i915-perf init/deinit v4: s/DRM_IORW/DRM_IOW/ - Emil Velikov Signed-off-by: Robert Bragg <robert@sixbynine.org> Reviewed-by: Matthew Auld <matthew.auld@intel.com> Reviewed-by: Sourab Gupta <sourab.gupta@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/20161107194957.3385-2-robert@sixbynine.org
2016-11-07 12:49:47 -07:00
*
* Return any error condition that results in a short read such as
* -%ENOSPC or -%EFAULT, even though these may be squashed before
* returning to userspace.
drm/i915: Add i915 perf infrastructure Adds base i915 perf infrastructure for Gen performance metrics. This adds a DRM_IOCTL_I915_PERF_OPEN ioctl that takes an array of uint64 properties to configure a stream of metrics and returns a new fd usable with standard VFS system calls including read() to read typed and sized records; ioctl() to enable or disable capture and poll() to wait for data. A stream is opened something like: uint64_t properties[] = { /* Single context sampling */ DRM_I915_PERF_PROP_CTX_HANDLE, ctx_handle, /* Include OA reports in samples */ DRM_I915_PERF_PROP_SAMPLE_OA, true, /* OA unit configuration */ DRM_I915_PERF_PROP_OA_METRICS_SET, metrics_set_id, DRM_I915_PERF_PROP_OA_FORMAT, report_format, DRM_I915_PERF_PROP_OA_EXPONENT, period_exponent, }; struct drm_i915_perf_open_param parm = { .flags = I915_PERF_FLAG_FD_CLOEXEC | I915_PERF_FLAG_FD_NONBLOCK | I915_PERF_FLAG_DISABLED, .properties_ptr = (uint64_t)properties, .num_properties = sizeof(properties) / 16, }; int fd = drmIoctl(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param); Records read all start with a common { type, size } header with DRM_I915_PERF_RECORD_SAMPLE being of most interest. Sample records contain an extensible number of fields and it's the DRM_I915_PERF_PROP_SAMPLE_xyz properties given when opening that determine what's included in every sample. No specific streams are supported yet so any attempt to open a stream will return an error. v2: use i915_gem_context_get() - Chris Wilson v3: update read() interface to avoid passing state struct - Chris Wilson fix some rebase fallout, with i915-perf init/deinit v4: s/DRM_IORW/DRM_IOW/ - Emil Velikov Signed-off-by: Robert Bragg <robert@sixbynine.org> Reviewed-by: Matthew Auld <matthew.auld@intel.com> Reviewed-by: Sourab Gupta <sourab.gupta@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/20161107194957.3385-2-robert@sixbynine.org
2016-11-07 12:49:47 -07:00
*/
int (*read)(struct i915_perf_stream *stream,
char __user *buf,
size_t count,
size_t *offset);
/**
* @destroy: Cleanup any stream specific resources.
drm/i915: Add i915 perf infrastructure Adds base i915 perf infrastructure for Gen performance metrics. This adds a DRM_IOCTL_I915_PERF_OPEN ioctl that takes an array of uint64 properties to configure a stream of metrics and returns a new fd usable with standard VFS system calls including read() to read typed and sized records; ioctl() to enable or disable capture and poll() to wait for data. A stream is opened something like: uint64_t properties[] = { /* Single context sampling */ DRM_I915_PERF_PROP_CTX_HANDLE, ctx_handle, /* Include OA reports in samples */ DRM_I915_PERF_PROP_SAMPLE_OA, true, /* OA unit configuration */ DRM_I915_PERF_PROP_OA_METRICS_SET, metrics_set_id, DRM_I915_PERF_PROP_OA_FORMAT, report_format, DRM_I915_PERF_PROP_OA_EXPONENT, period_exponent, }; struct drm_i915_perf_open_param parm = { .flags = I915_PERF_FLAG_FD_CLOEXEC | I915_PERF_FLAG_FD_NONBLOCK | I915_PERF_FLAG_DISABLED, .properties_ptr = (uint64_t)properties, .num_properties = sizeof(properties) / 16, }; int fd = drmIoctl(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param); Records read all start with a common { type, size } header with DRM_I915_PERF_RECORD_SAMPLE being of most interest. Sample records contain an extensible number of fields and it's the DRM_I915_PERF_PROP_SAMPLE_xyz properties given when opening that determine what's included in every sample. No specific streams are supported yet so any attempt to open a stream will return an error. v2: use i915_gem_context_get() - Chris Wilson v3: update read() interface to avoid passing state struct - Chris Wilson fix some rebase fallout, with i915-perf init/deinit v4: s/DRM_IORW/DRM_IOW/ - Emil Velikov Signed-off-by: Robert Bragg <robert@sixbynine.org> Reviewed-by: Matthew Auld <matthew.auld@intel.com> Reviewed-by: Sourab Gupta <sourab.gupta@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/20161107194957.3385-2-robert@sixbynine.org
2016-11-07 12:49:47 -07:00
*
* The stream will always be disabled before this is called.
*/
void (*destroy)(struct i915_perf_stream *stream);
};
/**
* struct i915_perf_stream - state for a single open stream FD
*/
drm/i915: Add i915 perf infrastructure Adds base i915 perf infrastructure for Gen performance metrics. This adds a DRM_IOCTL_I915_PERF_OPEN ioctl that takes an array of uint64 properties to configure a stream of metrics and returns a new fd usable with standard VFS system calls including read() to read typed and sized records; ioctl() to enable or disable capture and poll() to wait for data. A stream is opened something like: uint64_t properties[] = { /* Single context sampling */ DRM_I915_PERF_PROP_CTX_HANDLE, ctx_handle, /* Include OA reports in samples */ DRM_I915_PERF_PROP_SAMPLE_OA, true, /* OA unit configuration */ DRM_I915_PERF_PROP_OA_METRICS_SET, metrics_set_id, DRM_I915_PERF_PROP_OA_FORMAT, report_format, DRM_I915_PERF_PROP_OA_EXPONENT, period_exponent, }; struct drm_i915_perf_open_param parm = { .flags = I915_PERF_FLAG_FD_CLOEXEC | I915_PERF_FLAG_FD_NONBLOCK | I915_PERF_FLAG_DISABLED, .properties_ptr = (uint64_t)properties, .num_properties = sizeof(properties) / 16, }; int fd = drmIoctl(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param); Records read all start with a common { type, size } header with DRM_I915_PERF_RECORD_SAMPLE being of most interest. Sample records contain an extensible number of fields and it's the DRM_I915_PERF_PROP_SAMPLE_xyz properties given when opening that determine what's included in every sample. No specific streams are supported yet so any attempt to open a stream will return an error. v2: use i915_gem_context_get() - Chris Wilson v3: update read() interface to avoid passing state struct - Chris Wilson fix some rebase fallout, with i915-perf init/deinit v4: s/DRM_IORW/DRM_IOW/ - Emil Velikov Signed-off-by: Robert Bragg <robert@sixbynine.org> Reviewed-by: Matthew Auld <matthew.auld@intel.com> Reviewed-by: Sourab Gupta <sourab.gupta@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/20161107194957.3385-2-robert@sixbynine.org
2016-11-07 12:49:47 -07:00
struct i915_perf_stream {
/**
* @dev_priv: i915 drm device
*/
drm/i915: Add i915 perf infrastructure Adds base i915 perf infrastructure for Gen performance metrics. This adds a DRM_IOCTL_I915_PERF_OPEN ioctl that takes an array of uint64 properties to configure a stream of metrics and returns a new fd usable with standard VFS system calls including read() to read typed and sized records; ioctl() to enable or disable capture and poll() to wait for data. A stream is opened something like: uint64_t properties[] = { /* Single context sampling */ DRM_I915_PERF_PROP_CTX_HANDLE, ctx_handle, /* Include OA reports in samples */ DRM_I915_PERF_PROP_SAMPLE_OA, true, /* OA unit configuration */ DRM_I915_PERF_PROP_OA_METRICS_SET, metrics_set_id, DRM_I915_PERF_PROP_OA_FORMAT, report_format, DRM_I915_PERF_PROP_OA_EXPONENT, period_exponent, }; struct drm_i915_perf_open_param parm = { .flags = I915_PERF_FLAG_FD_CLOEXEC | I915_PERF_FLAG_FD_NONBLOCK | I915_PERF_FLAG_DISABLED, .properties_ptr = (uint64_t)properties, .num_properties = sizeof(properties) / 16, }; int fd = drmIoctl(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param); Records read all start with a common { type, size } header with DRM_I915_PERF_RECORD_SAMPLE being of most interest. Sample records contain an extensible number of fields and it's the DRM_I915_PERF_PROP_SAMPLE_xyz properties given when opening that determine what's included in every sample. No specific streams are supported yet so any attempt to open a stream will return an error. v2: use i915_gem_context_get() - Chris Wilson v3: update read() interface to avoid passing state struct - Chris Wilson fix some rebase fallout, with i915-perf init/deinit v4: s/DRM_IORW/DRM_IOW/ - Emil Velikov Signed-off-by: Robert Bragg <robert@sixbynine.org> Reviewed-by: Matthew Auld <matthew.auld@intel.com> Reviewed-by: Sourab Gupta <sourab.gupta@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/20161107194957.3385-2-robert@sixbynine.org
2016-11-07 12:49:47 -07:00
struct drm_i915_private *dev_priv;
/**
* @link: Links the stream into ``&drm_i915_private->streams``
*/
drm/i915: Add i915 perf infrastructure Adds base i915 perf infrastructure for Gen performance metrics. This adds a DRM_IOCTL_I915_PERF_OPEN ioctl that takes an array of uint64 properties to configure a stream of metrics and returns a new fd usable with standard VFS system calls including read() to read typed and sized records; ioctl() to enable or disable capture and poll() to wait for data. A stream is opened something like: uint64_t properties[] = { /* Single context sampling */ DRM_I915_PERF_PROP_CTX_HANDLE, ctx_handle, /* Include OA reports in samples */ DRM_I915_PERF_PROP_SAMPLE_OA, true, /* OA unit configuration */ DRM_I915_PERF_PROP_OA_METRICS_SET, metrics_set_id, DRM_I915_PERF_PROP_OA_FORMAT, report_format, DRM_I915_PERF_PROP_OA_EXPONENT, period_exponent, }; struct drm_i915_perf_open_param parm = { .flags = I915_PERF_FLAG_FD_CLOEXEC | I915_PERF_FLAG_FD_NONBLOCK | I915_PERF_FLAG_DISABLED, .properties_ptr = (uint64_t)properties, .num_properties = sizeof(properties) / 16, }; int fd = drmIoctl(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param); Records read all start with a common { type, size } header with DRM_I915_PERF_RECORD_SAMPLE being of most interest. Sample records contain an extensible number of fields and it's the DRM_I915_PERF_PROP_SAMPLE_xyz properties given when opening that determine what's included in every sample. No specific streams are supported yet so any attempt to open a stream will return an error. v2: use i915_gem_context_get() - Chris Wilson v3: update read() interface to avoid passing state struct - Chris Wilson fix some rebase fallout, with i915-perf init/deinit v4: s/DRM_IORW/DRM_IOW/ - Emil Velikov Signed-off-by: Robert Bragg <robert@sixbynine.org> Reviewed-by: Matthew Auld <matthew.auld@intel.com> Reviewed-by: Sourab Gupta <sourab.gupta@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/20161107194957.3385-2-robert@sixbynine.org
2016-11-07 12:49:47 -07:00
struct list_head link;
/**
* @sample_flags: Flags representing the `DRM_I915_PERF_PROP_SAMPLE_*`
* properties given when opening a stream, representing the contents
* of a single sample as read() by userspace.
*/
drm/i915: Add i915 perf infrastructure Adds base i915 perf infrastructure for Gen performance metrics. This adds a DRM_IOCTL_I915_PERF_OPEN ioctl that takes an array of uint64 properties to configure a stream of metrics and returns a new fd usable with standard VFS system calls including read() to read typed and sized records; ioctl() to enable or disable capture and poll() to wait for data. A stream is opened something like: uint64_t properties[] = { /* Single context sampling */ DRM_I915_PERF_PROP_CTX_HANDLE, ctx_handle, /* Include OA reports in samples */ DRM_I915_PERF_PROP_SAMPLE_OA, true, /* OA unit configuration */ DRM_I915_PERF_PROP_OA_METRICS_SET, metrics_set_id, DRM_I915_PERF_PROP_OA_FORMAT, report_format, DRM_I915_PERF_PROP_OA_EXPONENT, period_exponent, }; struct drm_i915_perf_open_param parm = { .flags = I915_PERF_FLAG_FD_CLOEXEC | I915_PERF_FLAG_FD_NONBLOCK | I915_PERF_FLAG_DISABLED, .properties_ptr = (uint64_t)properties, .num_properties = sizeof(properties) / 16, }; int fd = drmIoctl(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param); Records read all start with a common { type, size } header with DRM_I915_PERF_RECORD_SAMPLE being of most interest. Sample records contain an extensible number of fields and it's the DRM_I915_PERF_PROP_SAMPLE_xyz properties given when opening that determine what's included in every sample. No specific streams are supported yet so any attempt to open a stream will return an error. v2: use i915_gem_context_get() - Chris Wilson v3: update read() interface to avoid passing state struct - Chris Wilson fix some rebase fallout, with i915-perf init/deinit v4: s/DRM_IORW/DRM_IOW/ - Emil Velikov Signed-off-by: Robert Bragg <robert@sixbynine.org> Reviewed-by: Matthew Auld <matthew.auld@intel.com> Reviewed-by: Sourab Gupta <sourab.gupta@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/20161107194957.3385-2-robert@sixbynine.org
2016-11-07 12:49:47 -07:00
u32 sample_flags;
/**
* @sample_size: Considering the configured contents of a sample
* combined with the required header size, this is the total size
* of a single sample record.
*/
int sample_size;
drm/i915: Add i915 perf infrastructure Adds base i915 perf infrastructure for Gen performance metrics. This adds a DRM_IOCTL_I915_PERF_OPEN ioctl that takes an array of uint64 properties to configure a stream of metrics and returns a new fd usable with standard VFS system calls including read() to read typed and sized records; ioctl() to enable or disable capture and poll() to wait for data. A stream is opened something like: uint64_t properties[] = { /* Single context sampling */ DRM_I915_PERF_PROP_CTX_HANDLE, ctx_handle, /* Include OA reports in samples */ DRM_I915_PERF_PROP_SAMPLE_OA, true, /* OA unit configuration */ DRM_I915_PERF_PROP_OA_METRICS_SET, metrics_set_id, DRM_I915_PERF_PROP_OA_FORMAT, report_format, DRM_I915_PERF_PROP_OA_EXPONENT, period_exponent, }; struct drm_i915_perf_open_param parm = { .flags = I915_PERF_FLAG_FD_CLOEXEC | I915_PERF_FLAG_FD_NONBLOCK | I915_PERF_FLAG_DISABLED, .properties_ptr = (uint64_t)properties, .num_properties = sizeof(properties) / 16, }; int fd = drmIoctl(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param); Records read all start with a common { type, size } header with DRM_I915_PERF_RECORD_SAMPLE being of most interest. Sample records contain an extensible number of fields and it's the DRM_I915_PERF_PROP_SAMPLE_xyz properties given when opening that determine what's included in every sample. No specific streams are supported yet so any attempt to open a stream will return an error. v2: use i915_gem_context_get() - Chris Wilson v3: update read() interface to avoid passing state struct - Chris Wilson fix some rebase fallout, with i915-perf init/deinit v4: s/DRM_IORW/DRM_IOW/ - Emil Velikov Signed-off-by: Robert Bragg <robert@sixbynine.org> Reviewed-by: Matthew Auld <matthew.auld@intel.com> Reviewed-by: Sourab Gupta <sourab.gupta@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/20161107194957.3385-2-robert@sixbynine.org
2016-11-07 12:49:47 -07:00
/**
* @ctx: %NULL if measuring system-wide across all contexts or a
* specific context that is being monitored.
*/
drm/i915: Add i915 perf infrastructure Adds base i915 perf infrastructure for Gen performance metrics. This adds a DRM_IOCTL_I915_PERF_OPEN ioctl that takes an array of uint64 properties to configure a stream of metrics and returns a new fd usable with standard VFS system calls including read() to read typed and sized records; ioctl() to enable or disable capture and poll() to wait for data. A stream is opened something like: uint64_t properties[] = { /* Single context sampling */ DRM_I915_PERF_PROP_CTX_HANDLE, ctx_handle, /* Include OA reports in samples */ DRM_I915_PERF_PROP_SAMPLE_OA, true, /* OA unit configuration */ DRM_I915_PERF_PROP_OA_METRICS_SET, metrics_set_id, DRM_I915_PERF_PROP_OA_FORMAT, report_format, DRM_I915_PERF_PROP_OA_EXPONENT, period_exponent, }; struct drm_i915_perf_open_param parm = { .flags = I915_PERF_FLAG_FD_CLOEXEC | I915_PERF_FLAG_FD_NONBLOCK | I915_PERF_FLAG_DISABLED, .properties_ptr = (uint64_t)properties, .num_properties = sizeof(properties) / 16, }; int fd = drmIoctl(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param); Records read all start with a common { type, size } header with DRM_I915_PERF_RECORD_SAMPLE being of most interest. Sample records contain an extensible number of fields and it's the DRM_I915_PERF_PROP_SAMPLE_xyz properties given when opening that determine what's included in every sample. No specific streams are supported yet so any attempt to open a stream will return an error. v2: use i915_gem_context_get() - Chris Wilson v3: update read() interface to avoid passing state struct - Chris Wilson fix some rebase fallout, with i915-perf init/deinit v4: s/DRM_IORW/DRM_IOW/ - Emil Velikov Signed-off-by: Robert Bragg <robert@sixbynine.org> Reviewed-by: Matthew Auld <matthew.auld@intel.com> Reviewed-by: Sourab Gupta <sourab.gupta@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/20161107194957.3385-2-robert@sixbynine.org
2016-11-07 12:49:47 -07:00
struct i915_gem_context *ctx;
/**
* @enabled: Whether the stream is currently enabled, considering
* whether the stream was opened in a disabled state and based
* on `I915_PERF_IOCTL_ENABLE` and `I915_PERF_IOCTL_DISABLE` calls.
*/
drm/i915: Add i915 perf infrastructure Adds base i915 perf infrastructure for Gen performance metrics. This adds a DRM_IOCTL_I915_PERF_OPEN ioctl that takes an array of uint64 properties to configure a stream of metrics and returns a new fd usable with standard VFS system calls including read() to read typed and sized records; ioctl() to enable or disable capture and poll() to wait for data. A stream is opened something like: uint64_t properties[] = { /* Single context sampling */ DRM_I915_PERF_PROP_CTX_HANDLE, ctx_handle, /* Include OA reports in samples */ DRM_I915_PERF_PROP_SAMPLE_OA, true, /* OA unit configuration */ DRM_I915_PERF_PROP_OA_METRICS_SET, metrics_set_id, DRM_I915_PERF_PROP_OA_FORMAT, report_format, DRM_I915_PERF_PROP_OA_EXPONENT, period_exponent, }; struct drm_i915_perf_open_param parm = { .flags = I915_PERF_FLAG_FD_CLOEXEC | I915_PERF_FLAG_FD_NONBLOCK | I915_PERF_FLAG_DISABLED, .properties_ptr = (uint64_t)properties, .num_properties = sizeof(properties) / 16, }; int fd = drmIoctl(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param); Records read all start with a common { type, size } header with DRM_I915_PERF_RECORD_SAMPLE being of most interest. Sample records contain an extensible number of fields and it's the DRM_I915_PERF_PROP_SAMPLE_xyz properties given when opening that determine what's included in every sample. No specific streams are supported yet so any attempt to open a stream will return an error. v2: use i915_gem_context_get() - Chris Wilson v3: update read() interface to avoid passing state struct - Chris Wilson fix some rebase fallout, with i915-perf init/deinit v4: s/DRM_IORW/DRM_IOW/ - Emil Velikov Signed-off-by: Robert Bragg <robert@sixbynine.org> Reviewed-by: Matthew Auld <matthew.auld@intel.com> Reviewed-by: Sourab Gupta <sourab.gupta@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/20161107194957.3385-2-robert@sixbynine.org
2016-11-07 12:49:47 -07:00
bool enabled;
/**
* @ops: The callbacks providing the implementation of this specific
* type of configured stream.
*/
const struct i915_perf_stream_ops *ops;
/**
* @oa_config: The OA configuration used by the stream.
*/
struct i915_oa_config *oa_config;
};
/**
* struct i915_oa_ops - Gen specific implementation of an OA unit stream
*/
struct i915_oa_ops {
drm/i915/perf: Implement I915_PERF_ADD/REMOVE_CONFIG interface The motivation behind this new interface is expose at runtime the creation of new OA configs which can be used as part of the i915 perf open interface. This will enable the kernel to learn new configs which may be experimental, or otherwise not part of the core set currently available through the i915 perf interface. v2: Drop DRM_ERROR for userspace errors (Matthew) Add padding to userspace structure (Matthew) s/guid/uuid/ (Matthew) v3: Use u32 instead of int to iterate through registers (Matthew) v4: Lock access to dynamic config list (Lionel) v5: by Matthew: Fix uninitialized error values Fix incorrect unwiding when opening perf stream Use kmalloc_array() to store register Use uuid_is_valid() to valid config uuids Declare ioctls as write only Check padding members are set to 0 by Lionel: Return ENOENT rather than EINVAL when trying to remove non existing config v6: by Chris: Use ref counts for OA configs Store UUID in drm_i915_perf_oa_config rather then using pointer Shuffle fields of drm_i915_perf_oa_config to avoid padding v7: by Chris Rename uapi pointers fields to end with '_ptr' v8: by Andrzej, Marek, Sebastian Update register whitelisting by Lionel Add more register names for documentation Allow configuration programming in non-paranoid mode Add support for value filter for a couple of registers already programmed in other part of the kernel v9: Documentation fix (Lionel) Allow writing WAIT_FOR_RC6_EXIT only on Gen8+ (Andrzej) v10: Perform read access_ok() on register pointers (Lionel) Signed-off-by: Matthew Auld <matthew.auld@intel.com> Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com> Signed-off-by: Andrzej Datczuk <andrzej.datczuk@intel.com> Reviewed-by: Andrzej Datczuk <andrzej.datczuk@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20170803165812.2373-2-lionel.g.landwerlin@intel.com
2017-08-03 11:05:50 -06:00
/**
* @is_valid_b_counter_reg: Validates register's address for
* programming boolean counters for a particular platform.
*/
bool (*is_valid_b_counter_reg)(struct drm_i915_private *dev_priv,
u32 addr);
/**
* @is_valid_mux_reg: Validates register's address for programming mux
* for a particular platform.
*/
bool (*is_valid_mux_reg)(struct drm_i915_private *dev_priv, u32 addr);
/**
* @is_valid_flex_reg: Validates register's address for programming
* flex EU filtering for a particular platform.
*/
bool (*is_valid_flex_reg)(struct drm_i915_private *dev_priv, u32 addr);
/**
* @init_oa_buffer: Resets the head and tail pointers of the
* circular buffer for periodic OA reports.
*
* Called when first opening a stream for OA metrics, but also may be
* called in response to an OA buffer overflow or other error
* condition.
*
* Note it may be necessary to clear the full OA buffer here as part of
* maintaining the invariable that new reports must be written to
* zeroed memory for us to be able to reliable detect if an expected
* report has not yet landed in memory. (At least on Haswell the OA
* buffer tail pointer is not synchronized with reports being visible
* to the CPU)
*/
void (*init_oa_buffer)(struct drm_i915_private *dev_priv);
drm/i915/perf: Add OA unit support for Gen 8+ Enables access to OA unit metrics for BDW, CHV, SKL and BXT which all share (more-or-less) the same OA unit design. Of particular note in comparison to Haswell: some OA unit HW config state has become per-context state and as a consequence it is somewhat more complicated to manage synchronous state changes from the cpu while there's no guarantee of what context (if any) is currently actively running on the gpu. The periodic sampling frequency which can be particularly useful for system-wide analysis (as opposed to command stream synchronised MI_REPORT_PERF_COUNT commands) is perhaps the most surprising state to have become per-context save and restored (while the OABUFFER destination is still a shared, system-wide resource). This support for gen8+ takes care to consider a number of timing challenges involved in synchronously updating per-context state primarily by programming all config state from the cpu and updating all current and saved contexts synchronously while the OA unit is still disabled. The driver intentionally avoids depending on command streamer programming to update OA state considering the lack of synchronization between the automatic loading of OACTXCONTROL state (that includes the periodic sampling state and enable state) on context restore and the parsing of any general purpose BB the driver can control. I.e. this implementation is careful to avoid the possibility of a context restore temporarily enabling any out-of-date periodic sampling state. In addition to the risk of transiently-out-of-date state being loaded automatically; there are also internal HW latencies involved in the loading of MUX configurations which would be difficult to account for from the command streamer (and we only want to enable the unit when once the MUX configuration is complete). Since the Gen8+ OA unit design no longer supports clock gating the unit off for a single given context (which effectively stopped any progress of counters while any other context was running) and instead supports tagging OA reports with a context ID for filtering on the CPU, it means we can no longer hide the system-wide progress of counters from a non-privileged application only interested in metrics for its own context. Although we could theoretically try and subtract the progress of other contexts before forwarding reports via read() we aren't in a position to filter reports captured via MI_REPORT_PERF_COUNT commands. As a result, for Gen8+, we always require the dev.i915.perf_stream_paranoid to be unset for any access to OA metrics if not root. v5: Drain submitted requests when enabling metric set to ensure no lite-restore erases the context image we just updated (Lionel) v6: In addition to drain, switch to kernel context & update all context in place (Chris) v7: Add missing mutex_unlock() if switching to kernel context fails (Matthew) v8: Simplify OA period/flex-eu-counters programming by using the batchbuffer instead of modifying ctx-image (Lionel) v9: Back to updating the context image (due to erroneous testing, batchbuffer programming the OA unit doesn't actually work) (Lionel) Pin context before updating context image (Chris) Drop MMIO programming now that we switch to a kernel context with right values in initial context image (Chris) v10: Just pin_map the contexts we want to modify or let the configuration happen on first use (Chris) v11: Update kernel context OA config through the batchbuffer rather than on the fly ctx-image update (Lionel) v12: Rework OA context registers update again by swithing away from user contexts and reconfiguring the kernel context through the batchbuffer and updating all the other contexts' context image. Also take care to lock slice/subslice configuration when OA is on. (Lionel) v13: Request rpcs updates on all engine when updating the OA config (Lionel) v14: Drop any kind of rpcs management now that we monitor sseu configuration changes in a later patch (Lionel) Remove usleep after programming the NOA configs on Gen8+, this doesn't seem to be needed (Lionel) v15: Respect coding style for block comments (Chris) v16: Add missing i915_add_request() in case we fail to emit OA configuration (Matthew) Signed-off-by: Robert Bragg <robert@sixbynine.org> Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com> Reviewed-by: Matthew Auld <matthew.auld@intel.com> \o/ Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
2017-06-13 05:23:03 -06:00
/**
* @enable_metric_set: Selects and applies any MUX configuration to set
* up the Boolean and Custom (B/C) counters that are part of the
* counter reports being sampled. May apply system constraints such as
* disabling EU clock gating as required.
*/
int (*enable_metric_set)(struct drm_i915_private *dev_priv,
const struct i915_oa_config *oa_config);
/**
* @disable_metric_set: Remove system constraints associated with using
* the OA unit.
*/
void (*disable_metric_set)(struct drm_i915_private *dev_priv);
/**
* @oa_enable: Enable periodic sampling
*/
void (*oa_enable)(struct drm_i915_private *dev_priv);
/**
* @oa_disable: Disable periodic sampling
*/
void (*oa_disable)(struct drm_i915_private *dev_priv);
/**
* @read: Copy data from the circular OA buffer into a given userspace
* buffer.
*/
int (*read)(struct i915_perf_stream *stream,
char __user *buf,
size_t count,
size_t *offset);
/**
drm/i915/perf: Add OA unit support for Gen 8+ Enables access to OA unit metrics for BDW, CHV, SKL and BXT which all share (more-or-less) the same OA unit design. Of particular note in comparison to Haswell: some OA unit HW config state has become per-context state and as a consequence it is somewhat more complicated to manage synchronous state changes from the cpu while there's no guarantee of what context (if any) is currently actively running on the gpu. The periodic sampling frequency which can be particularly useful for system-wide analysis (as opposed to command stream synchronised MI_REPORT_PERF_COUNT commands) is perhaps the most surprising state to have become per-context save and restored (while the OABUFFER destination is still a shared, system-wide resource). This support for gen8+ takes care to consider a number of timing challenges involved in synchronously updating per-context state primarily by programming all config state from the cpu and updating all current and saved contexts synchronously while the OA unit is still disabled. The driver intentionally avoids depending on command streamer programming to update OA state considering the lack of synchronization between the automatic loading of OACTXCONTROL state (that includes the periodic sampling state and enable state) on context restore and the parsing of any general purpose BB the driver can control. I.e. this implementation is careful to avoid the possibility of a context restore temporarily enabling any out-of-date periodic sampling state. In addition to the risk of transiently-out-of-date state being loaded automatically; there are also internal HW latencies involved in the loading of MUX configurations which would be difficult to account for from the command streamer (and we only want to enable the unit when once the MUX configuration is complete). Since the Gen8+ OA unit design no longer supports clock gating the unit off for a single given context (which effectively stopped any progress of counters while any other context was running) and instead supports tagging OA reports with a context ID for filtering on the CPU, it means we can no longer hide the system-wide progress of counters from a non-privileged application only interested in metrics for its own context. Although we could theoretically try and subtract the progress of other contexts before forwarding reports via read() we aren't in a position to filter reports captured via MI_REPORT_PERF_COUNT commands. As a result, for Gen8+, we always require the dev.i915.perf_stream_paranoid to be unset for any access to OA metrics if not root. v5: Drain submitted requests when enabling metric set to ensure no lite-restore erases the context image we just updated (Lionel) v6: In addition to drain, switch to kernel context & update all context in place (Chris) v7: Add missing mutex_unlock() if switching to kernel context fails (Matthew) v8: Simplify OA period/flex-eu-counters programming by using the batchbuffer instead of modifying ctx-image (Lionel) v9: Back to updating the context image (due to erroneous testing, batchbuffer programming the OA unit doesn't actually work) (Lionel) Pin context before updating context image (Chris) Drop MMIO programming now that we switch to a kernel context with right values in initial context image (Chris) v10: Just pin_map the contexts we want to modify or let the configuration happen on first use (Chris) v11: Update kernel context OA config through the batchbuffer rather than on the fly ctx-image update (Lionel) v12: Rework OA context registers update again by swithing away from user contexts and reconfiguring the kernel context through the batchbuffer and updating all the other contexts' context image. Also take care to lock slice/subslice configuration when OA is on. (Lionel) v13: Request rpcs updates on all engine when updating the OA config (Lionel) v14: Drop any kind of rpcs management now that we monitor sseu configuration changes in a later patch (Lionel) Remove usleep after programming the NOA configs on Gen8+, this doesn't seem to be needed (Lionel) v15: Respect coding style for block comments (Chris) v16: Add missing i915_add_request() in case we fail to emit OA configuration (Matthew) Signed-off-by: Robert Bragg <robert@sixbynine.org> Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com> Reviewed-by: Matthew Auld <matthew.auld@intel.com> \o/ Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
2017-06-13 05:23:03 -06:00
* @oa_hw_tail_read: read the OA tail pointer register
*
drm/i915/perf: Add OA unit support for Gen 8+ Enables access to OA unit metrics for BDW, CHV, SKL and BXT which all share (more-or-less) the same OA unit design. Of particular note in comparison to Haswell: some OA unit HW config state has become per-context state and as a consequence it is somewhat more complicated to manage synchronous state changes from the cpu while there's no guarantee of what context (if any) is currently actively running on the gpu. The periodic sampling frequency which can be particularly useful for system-wide analysis (as opposed to command stream synchronised MI_REPORT_PERF_COUNT commands) is perhaps the most surprising state to have become per-context save and restored (while the OABUFFER destination is still a shared, system-wide resource). This support for gen8+ takes care to consider a number of timing challenges involved in synchronously updating per-context state primarily by programming all config state from the cpu and updating all current and saved contexts synchronously while the OA unit is still disabled. The driver intentionally avoids depending on command streamer programming to update OA state considering the lack of synchronization between the automatic loading of OACTXCONTROL state (that includes the periodic sampling state and enable state) on context restore and the parsing of any general purpose BB the driver can control. I.e. this implementation is careful to avoid the possibility of a context restore temporarily enabling any out-of-date periodic sampling state. In addition to the risk of transiently-out-of-date state being loaded automatically; there are also internal HW latencies involved in the loading of MUX configurations which would be difficult to account for from the command streamer (and we only want to enable the unit when once the MUX configuration is complete). Since the Gen8+ OA unit design no longer supports clock gating the unit off for a single given context (which effectively stopped any progress of counters while any other context was running) and instead supports tagging OA reports with a context ID for filtering on the CPU, it means we can no longer hide the system-wide progress of counters from a non-privileged application only interested in metrics for its own context. Although we could theoretically try and subtract the progress of other contexts before forwarding reports via read() we aren't in a position to filter reports captured via MI_REPORT_PERF_COUNT commands. As a result, for Gen8+, we always require the dev.i915.perf_stream_paranoid to be unset for any access to OA metrics if not root. v5: Drain submitted requests when enabling metric set to ensure no lite-restore erases the context image we just updated (Lionel) v6: In addition to drain, switch to kernel context & update all context in place (Chris) v7: Add missing mutex_unlock() if switching to kernel context fails (Matthew) v8: Simplify OA period/flex-eu-counters programming by using the batchbuffer instead of modifying ctx-image (Lionel) v9: Back to updating the context image (due to erroneous testing, batchbuffer programming the OA unit doesn't actually work) (Lionel) Pin context before updating context image (Chris) Drop MMIO programming now that we switch to a kernel context with right values in initial context image (Chris) v10: Just pin_map the contexts we want to modify or let the configuration happen on first use (Chris) v11: Update kernel context OA config through the batchbuffer rather than on the fly ctx-image update (Lionel) v12: Rework OA context registers update again by swithing away from user contexts and reconfiguring the kernel context through the batchbuffer and updating all the other contexts' context image. Also take care to lock slice/subslice configuration when OA is on. (Lionel) v13: Request rpcs updates on all engine when updating the OA config (Lionel) v14: Drop any kind of rpcs management now that we monitor sseu configuration changes in a later patch (Lionel) Remove usleep after programming the NOA configs on Gen8+, this doesn't seem to be needed (Lionel) v15: Respect coding style for block comments (Chris) v16: Add missing i915_add_request() in case we fail to emit OA configuration (Matthew) Signed-off-by: Robert Bragg <robert@sixbynine.org> Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com> Reviewed-by: Matthew Auld <matthew.auld@intel.com> \o/ Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
2017-06-13 05:23:03 -06:00
* In particular this enables us to share all the fiddly code for
* handling the OA unit tail pointer race that affects multiple
* generations.
*/
drm/i915/perf: Add OA unit support for Gen 8+ Enables access to OA unit metrics for BDW, CHV, SKL and BXT which all share (more-or-less) the same OA unit design. Of particular note in comparison to Haswell: some OA unit HW config state has become per-context state and as a consequence it is somewhat more complicated to manage synchronous state changes from the cpu while there's no guarantee of what context (if any) is currently actively running on the gpu. The periodic sampling frequency which can be particularly useful for system-wide analysis (as opposed to command stream synchronised MI_REPORT_PERF_COUNT commands) is perhaps the most surprising state to have become per-context save and restored (while the OABUFFER destination is still a shared, system-wide resource). This support for gen8+ takes care to consider a number of timing challenges involved in synchronously updating per-context state primarily by programming all config state from the cpu and updating all current and saved contexts synchronously while the OA unit is still disabled. The driver intentionally avoids depending on command streamer programming to update OA state considering the lack of synchronization between the automatic loading of OACTXCONTROL state (that includes the periodic sampling state and enable state) on context restore and the parsing of any general purpose BB the driver can control. I.e. this implementation is careful to avoid the possibility of a context restore temporarily enabling any out-of-date periodic sampling state. In addition to the risk of transiently-out-of-date state being loaded automatically; there are also internal HW latencies involved in the loading of MUX configurations which would be difficult to account for from the command streamer (and we only want to enable the unit when once the MUX configuration is complete). Since the Gen8+ OA unit design no longer supports clock gating the unit off for a single given context (which effectively stopped any progress of counters while any other context was running) and instead supports tagging OA reports with a context ID for filtering on the CPU, it means we can no longer hide the system-wide progress of counters from a non-privileged application only interested in metrics for its own context. Although we could theoretically try and subtract the progress of other contexts before forwarding reports via read() we aren't in a position to filter reports captured via MI_REPORT_PERF_COUNT commands. As a result, for Gen8+, we always require the dev.i915.perf_stream_paranoid to be unset for any access to OA metrics if not root. v5: Drain submitted requests when enabling metric set to ensure no lite-restore erases the context image we just updated (Lionel) v6: In addition to drain, switch to kernel context & update all context in place (Chris) v7: Add missing mutex_unlock() if switching to kernel context fails (Matthew) v8: Simplify OA period/flex-eu-counters programming by using the batchbuffer instead of modifying ctx-image (Lionel) v9: Back to updating the context image (due to erroneous testing, batchbuffer programming the OA unit doesn't actually work) (Lionel) Pin context before updating context image (Chris) Drop MMIO programming now that we switch to a kernel context with right values in initial context image (Chris) v10: Just pin_map the contexts we want to modify or let the configuration happen on first use (Chris) v11: Update kernel context OA config through the batchbuffer rather than on the fly ctx-image update (Lionel) v12: Rework OA context registers update again by swithing away from user contexts and reconfiguring the kernel context through the batchbuffer and updating all the other contexts' context image. Also take care to lock slice/subslice configuration when OA is on. (Lionel) v13: Request rpcs updates on all engine when updating the OA config (Lionel) v14: Drop any kind of rpcs management now that we monitor sseu configuration changes in a later patch (Lionel) Remove usleep after programming the NOA configs on Gen8+, this doesn't seem to be needed (Lionel) v15: Respect coding style for block comments (Chris) v16: Add missing i915_add_request() in case we fail to emit OA configuration (Matthew) Signed-off-by: Robert Bragg <robert@sixbynine.org> Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com> Reviewed-by: Matthew Auld <matthew.auld@intel.com> \o/ Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
2017-06-13 05:23:03 -06:00
u32 (*oa_hw_tail_read)(struct drm_i915_private *dev_priv);
drm/i915: Add i915 perf infrastructure Adds base i915 perf infrastructure for Gen performance metrics. This adds a DRM_IOCTL_I915_PERF_OPEN ioctl that takes an array of uint64 properties to configure a stream of metrics and returns a new fd usable with standard VFS system calls including read() to read typed and sized records; ioctl() to enable or disable capture and poll() to wait for data. A stream is opened something like: uint64_t properties[] = { /* Single context sampling */ DRM_I915_PERF_PROP_CTX_HANDLE, ctx_handle, /* Include OA reports in samples */ DRM_I915_PERF_PROP_SAMPLE_OA, true, /* OA unit configuration */ DRM_I915_PERF_PROP_OA_METRICS_SET, metrics_set_id, DRM_I915_PERF_PROP_OA_FORMAT, report_format, DRM_I915_PERF_PROP_OA_EXPONENT, period_exponent, }; struct drm_i915_perf_open_param parm = { .flags = I915_PERF_FLAG_FD_CLOEXEC | I915_PERF_FLAG_FD_NONBLOCK | I915_PERF_FLAG_DISABLED, .properties_ptr = (uint64_t)properties, .num_properties = sizeof(properties) / 16, }; int fd = drmIoctl(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param); Records read all start with a common { type, size } header with DRM_I915_PERF_RECORD_SAMPLE being of most interest. Sample records contain an extensible number of fields and it's the DRM_I915_PERF_PROP_SAMPLE_xyz properties given when opening that determine what's included in every sample. No specific streams are supported yet so any attempt to open a stream will return an error. v2: use i915_gem_context_get() - Chris Wilson v3: update read() interface to avoid passing state struct - Chris Wilson fix some rebase fallout, with i915-perf init/deinit v4: s/DRM_IORW/DRM_IOW/ - Emil Velikov Signed-off-by: Robert Bragg <robert@sixbynine.org> Reviewed-by: Matthew Auld <matthew.auld@intel.com> Reviewed-by: Sourab Gupta <sourab.gupta@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/20161107194957.3385-2-robert@sixbynine.org
2016-11-07 12:49:47 -07:00
};
struct intel_cdclk_state {
unsigned int cdclk, vco, ref;
};
struct drm_i915_private {
struct drm_device drm;
struct kmem_cache *objects;
struct kmem_cache *vmas;
struct kmem_cache *luts;
struct kmem_cache *requests;
struct kmem_cache *dependencies;
struct kmem_cache *priorities;
const struct intel_device_info info;
void __iomem *regs;
struct intel_uncore uncore;
drm/i915: Introduce a PV INFO page structure for Intel GVT-g. Introduce a PV INFO structure, to facilitate the Intel GVT-g technology, which is a GPU virtualization solution with mediated pass-through. This page contains the shared information between i915 driver and the host emulator. For now, this structure utilizes an area of 4K bytes on HSW GPU's unused MMIO space. Future hardware will have the reserved window architecturally defined, and layout of the page will be added in future BSpec. The i915 driver load routine detects if it is running in a VM by reading the contents of this PV INFO page. Thereafter a flag, vgpu.active is set, and intel_vgpu_active() is used by checking this flag to conclude if GPU is virtualized with Intel GVT-g. By now, intel_vgpu_active() will return true, only when the driver is running as a guest in the Intel GVT-g enhanced environment on HSW platform. v2: take Chris' comments: - call the i915_check_vgpu() in intel_uncore_init() - sanitize i915_check_vgpu() by adding BUILD_BUG_ON() and debug info take Daniel's comments: - put the definition of PV INFO into a new header - i915_vgt_if.h other changes: - access mmio regs by readq/readw in i915_check_vgpu() v3: take Daniel's comments: - move the i915/vgt interfaces into a new i915_vgpu.c - update makefile - add kerneldoc to functions which are non-static - add a DOC: section describing some of the high-level design - update drm docbook other changes: - rename i915_vgt_if.h to i915_vgpu.h v4: take Tvrtko's comments: - fix a typo in commit message - add debug message when vgt version mismatches - rename low_gmadr/high_gmadr to mappable/non-mappable in PV INFO structure Signed-off-by: Yu Zhang <yu.c.zhang@linux.intel.com> Signed-off-by: Jike Song <jike.song@intel.com> Signed-off-by: Eddie Dong <eddie.dong@intel.com> Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2015-02-10 04:05:47 -07:00
struct i915_virtual_gpu vgpu;
struct intel_gvt *gvt;
drm/i915: gvt: Introduce the basic architecture of GVT-g This patch introduces the very basic framework of GVT-g device model, includes basic prototypes, definitions, initialization. v12: - Call intel_gvt_init() in driver early initialization stage. (Chris) v8: - Remove the GVT idr and mutex in intel_gvt_host. (Joonas) v7: - Refine the URL link in Kconfig. (Joonas) - Refine the introduction of GVT-g host support in Kconfig. (Joonas) - Remove the macro GVT_ALIGN(), use round_down() instead. (Joonas) - Make "struct intel_gvt" a data member in struct drm_i915_private.(Joonas) - Remove {alloc, free}_gvt_device() - Rename intel_gvt_{create, destroy}_gvt_device() - Expost intel_gvt_init_host() - Remove the dummy "struct intel_gvt" declaration in intel_gvt.h (Joonas) v6: - Refine introduction in Kconfig. (Chris) - The exposed API functions will take struct intel_gvt * instead of void *. (Chris/Tvrtko) - Remove most memebers of strct intel_gvt_device_info. Will add them in the device model patches.(Chris) - Remove gvt_info() and gvt_err() in debug.h. (Chris) - Move GVT kernel parameter into i915_params. (Chris) - Remove include/drm/i915_gvt.h, as GVT-g will be built within i915. - Remove the redundant struct i915_gvt *, as the functions in i915 will directly take struct intel_gvt *. - Add more comments for reviewer. v5: Take Tvrtko's comments: - Fix the misspelled words in Kconfig - Let functions take drm_i915_private * instead of struct drm_device * - Remove redundant prints/local varible initialization v3: Take Joonas' comments: - Change file name i915_gvt.* to intel_gvt.* - Move GVT kernel parameter into intel_gvt.c - Remove redundant debug macros - Change error handling style - Add introductions for some stub functions - Introduce drm/i915_gvt.h. Take Kevin's comments: - Move GVT-g host/guest check into intel_vgt_balloon in i915_gem_gtt.c v2: - Introduce i915_gvt.c. It's necessary to introduce the stubs between i915 driver and GVT-g host, as GVT-g components is configurable in kernel config. When disabled, the stubs here do nothing. Take Joonas' comments: - Replace boolean return value with int. - Replace customized info/warn/debug macros with DRM macros. - Document all non-static functions like i915. - Remove empty and unused functions. - Replace magic number with marcos. - Set GVT-g in kernel config to "n" by default. Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> Cc: Kevin Tian <kevin.tian@intel.com> Signed-off-by: Zhi Wang <zhi.a.wang@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1466078825-6662-5-git-send-email-zhi.a.wang@intel.com Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
2016-06-16 06:07:00 -06:00
drm/i915/huc: Add HuC fw loading support The HuC loading process is similar to GuC. The intel_uc_fw_fetch() is used for both cases. HuC loading needs to be before GuC loading. The WOPCM setting must be done early before loading any of them. v2: rebased on-top of drm-intel-nightly. removed if(HAS_GUC()) before the guc call. (D.Gordon) update huc_version number of format. v3: rebased to drm-intel-nightly, changed the file name format to match the one in the huc package. Changed dev->dev_private to to_i915() v4: moved function back to where it was. change wait_for_atomic to wait_for. v5: rebased. Changed the year in the copyright message to reflect the right year.Correct the comments,remove the unwanted WARN message, replace drm_gem_object_unreference() with i915_gem_object_put().Make the prototypes in intel_huc.h non-extern. v6: rebased. Update the file construction done by HuC. It is similar to GuC.Adopted the approach used in- https://patchwork.freedesktop.org/patch/104355/ <Tvrtko Ursulin> v7: Change dev to dev_priv in macro definition. Corrected comments. v8: rebased on top of drm-tip. Updated functions intel_huc_load(), intel_huc_init() and intel_uc_fw_fetch() to accept dev_priv instead of dev. Moved contents of intel_huc.h to intel_uc.h. v9: change SKL_FW_ to SKL_HUC_FW_. Add intel_ prefix to guc_wopcm_size(). Remove unwanted checks in intel_uc.h. Rename huc_fw in struct intel_huc to simply fw to avoid redundency. v10: rebased. Correct comments. Make intel_huc_fini() accept dev_priv instead of dev like intel_huc_init() and intel_huc_load().Move definition to i915_guc_reg.h from intel_uc.h. Clean DMA_CTRL bits after HuC DMA transfer in huc_ucode_xfer() instead of guc_ucode_xfer(). Add suitable WARNs to give extra info. v11: rebased. Add proper bias for HuC and make sure there are asserts on failure by using guc_ggtt_offset_vma(). Introduce intel_huc.c and remove intel_huc_loader.c since it has functions that do more than just loading.Correct year in copyright. v12: remove invalidates that are not required anymore. Cc: Arkadiusz Hiler <arkadiusz.hiler@intel.com> Cc: Michal Wajdeczko <michal.wajdeczko@intel.com> Tested-by: Xiang Haihao <haihao.xiang@intel.com> Signed-off-by: Anusha Srivatsa <anusha.srivatsa@intel.com> Signed-off-by: Alex Dai <yu.dai@intel.com> Signed-off-by: Peter Antoine <peter.antoine@intel.com> Reviewed-by: Michal Wajdeczko <michal.wajdeczko@intel.com> Signed-off-by: Jani Nikula <jani.nikula@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1484755558-1234-1-git-send-email-anusha.srivatsa@intel.com
2017-01-18 09:05:53 -07:00
struct intel_huc huc;
struct intel_guc guc;
drm/i915/skl: Add support to load SKL CSR firmware. Display Context Save and Restore support is needed for various SKL Display C states like DC5, DC6. This implementation is added based on first version of DMC CSR program that we received from h/w team. Here we are using request_firmware based design. Finally this firmware should end up in linux-firmware tree. For SKL platform its mandatory to ensure that we load this csr program before enabling DC states like DC5/DC6. As CSR program gets reset on various conditions, we should ensure to load it during boot and in future change to be added to load this system resume sequence too. v1: Initial relese as RFC patch v2: Design change as per Daniel, Damien and Shobit's review comments request firmware method followed. v3: Some optimization and functional changes. Pulled register defines into drivers/gpu/drm/i915/i915_reg.h Used kmemdup to allocate and duplicate firmware content. Ensured to free allocated buffer. v4: Modified as per review comments from Satheesh and Daniel Removed temporary buffer. Optimized number of writes by replacing I915_WRITE with I915_WRITE64. v5: Modified as per review comemnts from Damien. - Changed name for functions and firmware. - Introduced HAS_CSR. - Reverted back previous change and used csr_buf with u8 size. - Using cpu_to_be64 for endianness change. Modified as per review comments from Imre. - Modified registers and macro names to be a bit closer to bspec terminology and the existing register naming in the driver. - Early return for non SKL platforms in intel_load_csr_program function. - Added locking around CSR program load function as it may be called concurrently during system/runtime resume. - Releasing the fw before loading the program for consistency - Handled error path during f/w load. v6: Modified as per review comments from Imre. - Corrected out_freecsr sequence. v7: Modified as per review comments from Imre. Fail loading fw if fw->size%8!=0. v8: Rebase to latest. v9: Rebase on top of -nightly (Damien) v10: Enabled support for dmc firmware ver 1.0. According to ver 1.0 in a single binary package all the firmware's that are required for different stepping's of the product will be stored. The package contains the css header, followed by the package header and the actual dmc firmwares. Package header contains the firmware/stepping mapping table and the corresponding firmware offsets to the individual binaries, within the package. Each individual program binary contains the header and the payload sections whose size is specified in the header section. This changes are done to extract the specific firmaware from the package. (Animesh) v11: Modified as per review comemnts from Imre. - Added code comment from bpec for header structure elements. - Added __packed to avoid structure padding. - Added helper functions for stepping and substepping info. - Added code comment for CSR_MAX_FW_SIZE. - Disabled BXT firmware loading, will be enabled with dmc 1.0 support. - Changed skl_stepping_info based on bspec, earlier used from config DB. - Removed duplicate call of cpu_to_be* from intel_csr_load_program function. - Used cpu_to_be32 instead of cpu_to_be64 as firmware binary in dword aligned. - Added sanity check for header length. - Added sanity check for mmio address got from firmware binary. - kmalloc done separately for dmc header and dmc firmware. (Animesh) v12: Modified as per review comemnts from Imre. - Corrected the typo error in skl stepping info structure. - Added out-of-bound access for skl_stepping_info. - Sanity check for mmio address modified. - Sanity check added for stepping and substeppig. - Modified the intel_dmc_info structure, cache only the required header info. (Animesh) v13: clarify firmware load error message. The reason for a firmware loading failure can be obscure if the driver is built-in. Provide an explanation to the user about the likely reason for the failure and how to resolve it. (Imre) v14: Suggested by Jani. - fix s/I915/CONFIG_DRM_I915/ typo - add fw_path to the firmware object instead of using a static ptr (Jani) v15: 1) Changed the firmware name as dmc_gen9.bin, everytime for a new firmware version a symbolic link with same name will help not to build kernel again. 2) Changes done as per review comments from Imre. - Error check removed for intel_csr_ucode_init. - Moved csr-specific data structure to intel_csr.h and optimization done on structure definition. - fw->data used directly for parsing the header info & memory allocation only done separately for payload. (Animesh) v16: - No need for out_regs label in i915_driver_load(), so removed it. - Changed the firmware name as skl_dmc_ver1.bin, followed naming convention <platform>_dmc_<api-version>.bin (Animesh) Issue: VIZ-2569 Signed-off-by: A.Sunil Kamath <sunil.kamath@intel.com> Signed-off-by: Damien Lespiau <damien.lespiau@intel.com> Signed-off-by: Animesh Manna <animesh.manna@intel.com> Signed-off-by: Imre Deak <imre.deak@intel.com> Reviewed-by: Imre Deak <imre.deak@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2015-05-04 06:58:44 -06:00
struct intel_csr csr;
struct intel_gmbus gmbus[GMBUS_NUM_PINS];
drm/i915: use the gmbus irq for waits We need two special things to properly wire this up: - Add another argument to gmbus_wait_hw_status to pass in the correct interrupt bit in gmbus4. - Since we can only get an irq for one of the two events we want, hand-roll the wait_event_timeout code so that we wake up every jiffie and can check for NAKs. This way we also subsume gmbus support for platforms without interrupts (or where those are not yet enabled). The important bit really is to only enable one gmbus interrupt source at the same time - with that piece of lore figured out, this seems to work flawlessly. Ben Widawsky rightfully complained the lack of measurements for the claimed benefits (especially since the first version was actually broken and fell back to bit-banging). Previously reading the 256 byte hdmi EDID takes about 72 ms here. With this patch it's down to 33 ms. Given that transfering the 256 bytes over i2c at wire speed takes 20.5ms alone, the reduction in additional overhead is rather nice. v2: Chris Wilson wondered whether GMBUS4 might contain some set bits when booting up an hence result in some spurious interrupts. Since we clear GMBUS4 after every wait and we do gmbus transfer really early in the setup sequence to detect displays the window is small, but still be paranoid and clear it properly. v3: Clarify the comment that gmbus irq generation can only support one kind of event, why it bothers us and how we work around that limit. Cc: Daniel Kurtz <djkurtz@chromium.org> Reviewed-by: Imre Deak <imre.deak@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2012-12-01 05:53:45 -07:00
/** gmbus_mutex protects against concurrent usage of the single hw gmbus
* controller on different i2c buses. */
struct mutex gmbus_mutex;
/**
* Base address of the gmbus and gpio block.
*/
uint32_t gpio_mmio_base;
/* MMIO base address for MIPI regs */
uint32_t mipi_mmio_base;
uint32_t psr_mmio_base;
uint32_t pps_mmio_base;
drm/i915: use the gmbus irq for waits We need two special things to properly wire this up: - Add another argument to gmbus_wait_hw_status to pass in the correct interrupt bit in gmbus4. - Since we can only get an irq for one of the two events we want, hand-roll the wait_event_timeout code so that we wake up every jiffie and can check for NAKs. This way we also subsume gmbus support for platforms without interrupts (or where those are not yet enabled). The important bit really is to only enable one gmbus interrupt source at the same time - with that piece of lore figured out, this seems to work flawlessly. Ben Widawsky rightfully complained the lack of measurements for the claimed benefits (especially since the first version was actually broken and fell back to bit-banging). Previously reading the 256 byte hdmi EDID takes about 72 ms here. With this patch it's down to 33 ms. Given that transfering the 256 bytes over i2c at wire speed takes 20.5ms alone, the reduction in additional overhead is rather nice. v2: Chris Wilson wondered whether GMBUS4 might contain some set bits when booting up an hence result in some spurious interrupts. Since we clear GMBUS4 after every wait and we do gmbus transfer really early in the setup sequence to detect displays the window is small, but still be paranoid and clear it properly. v3: Clarify the comment that gmbus irq generation can only support one kind of event, why it bothers us and how we work around that limit. Cc: Daniel Kurtz <djkurtz@chromium.org> Reviewed-by: Imre Deak <imre.deak@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2012-12-01 05:53:45 -07:00
wait_queue_head_t gmbus_wait_queue;
struct pci_dev *bridge_dev;
struct i915_gem_context *kernel_context;
drm/i915: Allocate intel_engine_cs structure only for the enabled engines With the possibility of addition of many more number of rings in future, the drm_i915_private structure could bloat as an array, of type intel_engine_cs, is embedded inside it. struct intel_engine_cs engine[I915_NUM_ENGINES]; Though this is still fine as generally there is only a single instance of drm_i915_private structure used, but not all of the possible rings would be enabled or active on most of the platforms. Some memory can be saved by allocating intel_engine_cs structure only for the enabled/active engines. Currently the engine/ring ID is kept static and dev_priv->engine[] is simply indexed using the enums defined in intel_engine_id. To save memory and continue using the static engine/ring IDs, 'engine' is defined as an array of pointers. struct intel_engine_cs *engine[I915_NUM_ENGINES]; dev_priv->engine[engine_ID] will be NULL for disabled engine instances. There is a text size reduction of 928 bytes, from 1028200 to 1027272, for i915.o file (but for i915.ko file text size remain same as 1193131 bytes). v2: - Remove the engine iterator field added in drm_i915_private structure, instead pass a local iterator variable to the for_each_engine** macros. (Chris) - Do away with intel_engine_initialized() and instead directly use the NULL pointer check on engine pointer. (Chris) v3: - Remove for_each_engine_id() macro, as the updated macro for_each_engine() can be used in place of it. (Chris) - Protect the access to Render engine Fault register with a NULL check, as engine specific init is done later in Driver load sequence. v4: - Use !!dev_priv->engine[VCS] style for the engine check in getparam. (Chris) - Kill the superfluous init_engine_lists(). v5: - Cleanup the intel_engines_init() & intel_engines_setup(), with respect to allocation of intel_engine_cs structure. (Chris) v6: - Rebase. v7: - Optimize the for_each_engine_masked() macro. (Chris) - Change the type of 'iter' local variable to enum intel_engine_id. (Chris) - Rebase. v8: Rebase. v9: Rebase. v10: - For index calculation use engine ID instead of pointer based arithmetic in intel_engine_sync_index() as engine pointers are not contiguous now (Chris) - For appropriateness, rename local enum variable 'iter' to 'id'. (Joonas) - Use for_each_engine macro for cleanup in intel_engines_init() and remove check for NULL engine pointer in cleanup() routines. (Joonas) v11: Rebase. Cc: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Akash Goel <akash.goel@intel.com> Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1476378888-7372-1-git-send-email-akash.goel@intel.com
2016-10-13 11:14:48 -06:00
struct intel_engine_cs *engine[I915_NUM_ENGINES];
struct i915_vma *semaphore;
struct drm_dma_handle *status_page_dmah;
struct resource mch_res;
/* protects the irq masks */
spinlock_t irq_lock;
bool display_irqs_enabled;
drm/i915: irq-drive the dp aux communication At least on the platforms that have a dp aux irq and also have it enabled - vlvhsw should have one, too. But I don't have a machine to test this on. Judging from docs there's no dp aux interrupt for gm45. Also, I only have an ivb cpu edp machine, so the dp aux A code for snb/ilk is untested. For dpcd probing when nothing is connected it slashes about 5ms of cpu time (cpu time is now negligible), which agrees with 3 * 5 400 usec timeouts. A previous version of this patch increases the time required to go through the dp_detect cycle (which includes reading the edid) from around 33 ms to around 40 ms. Experiments indicated that this is purely due to the irq latency - the hw doesn't allow us to queue up dp aux transactions and hence irq latency directly affects throughput. gmbus is much better, there we have a 8 byte buffer, and we get the irq once another 4 bytes can be queued up. But by using the pm_qos interface to request the lowest possible cpu wake-up latency this slowdown completely disappeared. Since all our output detection logic is single-threaded with the mode_config mutex right now anyway, I've decide not ot play fancy and to just reuse the gmbus wait queue. But this would definitely prep the way to run dp detection on different ports in parallel v2: Add a timeout for dp aux transfers when using interrupts - the hw _does_ prevent this with the hw-based 400 usec timeout, but if the irq somehow doesn't arrive we're screwed. Lesson learned while developing this ;-) v3: While at it also convert the busy-loop to wait_for_atomic, so that we don't run the risk of an infinite loop any more. v4: Ensure we have the smallest possible irq latency by using the pm_qos interface. v5: Add a comment to the code to explain why we frob pm_qos. Suggested by Chris Wilson. v6: Disable dp irq for vlv, that's easier than trying to get at docs and hw. v7: Squash in a fix for Haswell that Paulo Zanoni tracked down - the dp aux registers aren't at a fixed offset any more, but can be on the PCH while the DP port is on the cpu die. Reviewed-by: Imre Deak <imre.deak@intel.com> (v6) Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2012-12-01 05:53:48 -07:00
/* To control wakeup latency, e.g. for irq-driven dp aux transfers. */
struct pm_qos_request pm_qos;
/* Sideband mailbox protection */
struct mutex sb_lock;
/** Cached value of IMR to avoid reads in updating the bitfield */
drm/i915/bdw: Implement interrupt changes The interrupt handling implementation remains the same as previous generations with the 4 types of registers, status, identity, mask, and enable. However the layout of where the bits go have changed entirely. To address these changes, all of the interrupt vfuncs needed special gen8 code. The way it works is there is a top level status register now which informs the interrupt service routine which unit caused the interrupt, and therefore which interrupt registers to read to process the interrupt. For display the division is quite logical, a set of interrupt registers for each pipe, and in addition to those, a set each for "misc" and port. For GT the things get a bit hairy, as seen by the code. Each of the GT units has it's own bits defined. They all look *very similar* and resides in 16 bits of a GT register. As an example, RCS and BCS share register 0. To compact the code a bit, at a slight expense to complexity, this is exactly how the code works as well. 2 structures are added to the ring buffer so that our ring buffer interrupt handling code knows which ring shares the interrupt registers, and a shift value (ie. the top or bottom 16 bits of the register). The above allows us to kept the interrupt register caching scheme, the per interrupt enables, and the code to mask and unmask interrupts relatively clean (again at the cost of some more complexity). Most of the GT units mentioned above are command streamers, and so the symmetry should work quite well for even the yet to be implemented rings which Broadwell adds. v2: Fixes up a couple of bugs, and is more verbose about errors in the Broadwell interrupt handler. v3: fix DE_MISC IER offset v4: Simplify interrupts: I totally misread the docs the first time I implemented interrupts, and so this should greatly simplify the mess. Unlike GEN6, we never touch the regular mask registers in irq_get/put. v5: Rebased on to of recent pch hotplug setup changes. v6: Fixup on top of moving num_pipes to intel_info. v7: Rebased on top of Egbert Eich's hpd irq handling rework. Also wired up ibx_hpd_irq_setup for gen8. v8: Rebase on top of Jani's asle handling rework. v9: Rebase on top of Ben's VECS enabling for Haswell, where he unfortunately went OCD on the gt irq #defines. Not that they're still not yet fully consistent: - Used the GT_RENDER_ #defines + bdw shifts. - Dropped the shift from the L3_PARITY stuff, seemed clearer. - s/irq_refcount/irq_refcount.gt/ v10: Squash in VECS enabling patches and the gen8_gt_irq_handler refactoring from Zhao Yakui <yakui.zhao@intel.com> v11: Rebase on top of the interrupt cleanups in upstream. v12: Rebase on top of Ben's DPF changes in upstream. v13: Drop bdw from the HAS_L3_DPF feature flag for now, it's unclear what exactly needs to be done. Requested by Ben. v14: Fix the patch. - Drop the mask of reserved bits and assorted logic, it doesn't match the spec. - Do the posting read inconditionally instead of commenting it out. - Add a GEN8_MASTER_IRQ_CONTROL definition and use it. - Fix up the GEN8_PIPE interrupt defines and give the GEN8_ prefixes - we actually will need to use them. - Enclose macros in do {} while (0) (checkpatch). - Clear DE_MISC interrupt bits only after having processed them. - Fix whitespace fail (checkpatch). - Fix overtly long lines where appropriate (checkpatch). - Don't use typedef'ed private_t (maintainer-scripts). - Align the function parameter list correctly. Signed-off-by: Ben Widawsky <ben@bwidawsk.net> (v4) Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> bikeshed
2013-11-02 22:07:09 -06:00
union {
u32 irq_mask;
u32 de_irq_mask[I915_MAX_PIPES];
};
u32 gt_irq_mask;
u32 pm_imr;
u32 pm_ier;
u32 pm_rps_events;
drm/i915: Support for GuC interrupts There are certain types of interrupts which Host can receive from GuC. GuC ukernel sends an interrupt to Host for certain events, like for example retrieve/consume the logs generated by ukernel. This patch adds support to receive interrupts from GuC but currently enables & partially handles only the interrupt sent by GuC ukernel. Future patches will add support for handling other interrupt types. v2: - Use common low level routines for PM IER/IIR programming (Chris) - Rename interrupt functions to gen9_xxx from gen8_xxx (Chris) - Replace disabling of wake ref asserts with rpm get/put (Chris) v3: - Update comments for more clarity. (Tvrtko) - Remove the masking of GuC interrupt, which was kept masked till the start of bottom half, its not really needed as there is only a single instance of work item & wq is ordered. (Tvrtko) v4: - Rebase. - Rename guc_events to pm_guc_events so as to be indicative of the register/control block it is associated with. (Chris) - Add handling for back to back log buffer flush interrupts. v5: - Move the read & clearing of register, containing Guc2Host message bits, outside the irq spinlock. (Tvrtko) v6: - Move the log buffer flush interrupt related stuff to the following patch so as to do only generic bits in this patch. (Tvrtko) - Rebase. v7: - Remove the interrupts_enabled check from gen9_guc_irq_handler, want to process that last interrupt also before disabling the interrupt, sync against the work queued by irq handler will be done by caller disabling the interrupt. Signed-off-by: Sagar Arun Kamble <sagar.a.kamble@intel.com> Signed-off-by: Akash Goel <akash.goel@intel.com> Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
2016-10-12 10:24:31 -06:00
u32 pm_guc_events;
u32 pipestat_irq_mask[I915_MAX_PIPES];
struct i915_hotplug hotplug;
struct intel_fbc fbc;
drm/i915: Add support for DRRS to switch RR This patch computes and stored 2nd M/N/TU for switching to different refresh rate dynamically. PIPECONF_EDP_RR_MODE_SWITCH bit helps toggle between alternate refresh rates programmed in 2nd M/N/TU registers. v2: Daniel's review comments Computing M2/N2 in compute_config and storing it in crtc_config v3: Modified reference to edp_downclock and edp_downclock_avail based on the changes made to move them from dev_private to intel_panel. v4: Modified references to is_drrs_supported based on the changes made to rename it to drrs_support. v5: Jani's review comments Removed superfluous return statements. Changed support for Gen 7 and above. Corrected indentation. Re-structured the code which finds crtc and connector from encoder. Changed some logs to be less verbose. v6: Modifying i915_drrs to include only intel connector as intel_dp can be derived from intel connector when required. v7: As per internal review comments, acquiring mutex just before accessing drrs RR. As per Chris's review comments, added documentation about the use of locking in the function. v8: Incorporated Jani's review comments. Removed reference to edp_downclock. v9: Jani's review comments. Modified comment in set_drrs. Changed index to type edp_drrs_refresh_rate_type. Check if PSR is enabled before setting registers fo DRRS. Signed-off-by: Pradeep Bhat <pradeep.bhat@intel.com> Signed-off-by: Vandana Kannan <vandana.kannan@intel.com> Cc: Jani Nikula <jani.nikula@linux.intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2014-04-05 00:43:28 -06:00
struct i915_drrs drrs;
struct intel_opregion opregion;
struct intel_vbt_data vbt;
bool preserve_bios_swizzle;
/* overlay */
struct intel_overlay *overlay;
/* backlight registers and fields in struct intel_panel */
struct mutex backlight_lock;
/* LVDS info */
bool no_aux_handshake;
/* protects panel power sequencer state */
struct mutex pps_mutex;
struct drm_i915_fence_reg fence_regs[I915_MAX_NUM_FENCES]; /* assume 965 */
int num_fence_regs; /* 8 on pre-965, 16 otherwise */
unsigned int fsb_freq, mem_freq, is_ddr3;
unsigned int skl_preferred_vco_freq;
unsigned int max_cdclk_freq;
unsigned int max_dotclk_freq;
unsigned int rawclk_freq;
unsigned int hpll_freq;
unsigned int czclk_freq;
struct {
/*
* The current logical cdclk state.
* See intel_atomic_state.cdclk.logical
*
* For reading holding any crtc lock is sufficient,
* for writing must hold all of them.
*/
struct intel_cdclk_state logical;
/*
* The current actual cdclk state.
* See intel_atomic_state.cdclk.actual
*/
struct intel_cdclk_state actual;
/* The current hardware cdclk state */
struct intel_cdclk_state hw;
} cdclk;
drm/i915: fix hpd work vs. flush_work in the pageflip code deadlock Historically we've run our own driver hotplug handling in our own work-queue, which then launched the drm core hotplug handling in the system workqueue. This is important since we flush our own driver workqueue in the pageflip code while hodling modeset locks, and only the drm hotplug code grabbed these locks. But with commit 69787f7da6b2adc4054357a661aaa1701a9ca76f Author: Daniel Vetter <daniel.vetter@ffwll.ch> Date: Tue Oct 23 18:23:34 2012 +0000 drm: run the hpd irq event code directly this was changed and now we could deadlock in our flip handler if there's a hotplug work blocking the progress of the crucial unpin works. So this broke the careful deadlock avoidance implemented in commit b4a98e57fc27854b5938fc8b08b68e5e68b91e1f Author: Chris Wilson <chris@chris-wilson.co.uk> Date: Thu Nov 1 09:26:26 2012 +0000 drm/i915: Flush outstanding unpin tasks before pageflipping Since the rule thus far has been that work items on our own workqueue may never grab modeset locks simply restore that rule again. v2: Add a comment to the declaration of dev_priv->wq to warn readers about the tricky implications of using it. Suggested by Chris Wilson. Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Stuart Abercrombie <sabercrombie@chromium.org> Reported-by: Stuart Abercrombie <sabercrombie@chromium.org> References: http://permalink.gmane.org/gmane.comp.freedesktop.xorg.drivers.intel/26239 Cc: stable@vger.kernel.org Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> [danvet: Squash in a comment at the place where we schedule the work. Requested after-the-fact by Chris on irc since the hpd work isn't the only place we botch this.] Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2013-09-02 08:22:25 -06:00
/**
* wq - Driver workqueue for GEM.
*
* NOTE: Work items scheduled here are not allowed to grab any modeset
* locks, for otherwise the flushing done in the pageflip code will
* result in deadlocks.
*/
struct workqueue_struct *wq;
/* Display functions */
struct drm_i915_display_funcs display;
/* PCH chipset type */
enum intel_pch pch_type;
unsigned short pch_id;
unsigned long quirks;
i915: ignore lid open event when resuming i915 driver needs to do modeset when 1. system resumes from sleep 2. lid is opened In PM_SUSPEND_MEM state, all the GPEs are cleared when system resumes, thus it is the i915_resume code does the modeset rather than intel_lid_notify(). But in PM_SUSPEND_FREEZE state, this will be broken because system is still responsive to the lid events. 1. When we close the lid in Freeze state, intel_lid_notify() sets modeset_on_lid. 2. When we reopen the lid, intel_lid_notify() will do a modeset, before the system is resumed. here is the error log, [92146.548074] WARNING: at drivers/gpu/drm/i915/intel_display.c:1028 intel_wait_for_pipe_off+0x184/0x190 [i915]() [92146.548076] Hardware name: VGN-Z540N [92146.548078] pipe_off wait timed out [92146.548167] Modules linked in: hid_generic usbhid hid snd_hda_codec_realtek snd_hda_intel snd_hda_codec parport_pc snd_hwdep ppdev snd_pcm_oss i915 snd_mixer_oss snd_pcm arc4 iwldvm snd_seq_dummy mac80211 snd_seq_oss snd_seq_midi fbcon tileblit font bitblit softcursor drm_kms_helper snd_rawmidi snd_seq_midi_event coretemp drm snd_seq kvm btusb bluetooth snd_timer iwlwifi pcmcia tpm_infineon i2c_algo_bit joydev snd_seq_device intel_agp cfg80211 snd intel_gtt yenta_socket pcmcia_rsrc sony_laptop agpgart microcode psmouse tpm_tis serio_raw mxm_wmi soundcore snd_page_alloc tpm acpi_cpufreq lpc_ich pcmcia_core tpm_bios mperf processor lp parport firewire_ohci firewire_core crc_itu_t sdhci_pci sdhci thermal e1000e [92146.548173] Pid: 4304, comm: kworker/0:0 Tainted: G W 3.8.0-rc3-s0i3-v3-test+ #9 [92146.548175] Call Trace: [92146.548189] [<c10378e2>] warn_slowpath_common+0x72/0xa0 [92146.548227] [<f86398b4>] ? intel_wait_for_pipe_off+0x184/0x190 [i915] [92146.548263] [<f86398b4>] ? intel_wait_for_pipe_off+0x184/0x190 [i915] [92146.548270] [<c10379b3>] warn_slowpath_fmt+0x33/0x40 [92146.548307] [<f86398b4>] intel_wait_for_pipe_off+0x184/0x190 [i915] [92146.548344] [<f86399c2>] intel_disable_pipe+0x102/0x190 [i915] [92146.548380] [<f8639ea4>] ? intel_disable_plane+0x64/0x80 [i915] [92146.548417] [<f8639f7c>] i9xx_crtc_disable+0xbc/0x150 [i915] [92146.548456] [<f863ebee>] intel_crtc_update_dpms+0x5e/0x90 [i915] [92146.548493] [<f86437cf>] intel_modeset_setup_hw_state+0x42f/0x8f0 [i915] [92146.548535] [<f8645b0b>] intel_lid_notify+0x9b/0xc0 [i915] [92146.548543] [<c15610d3>] notifier_call_chain+0x43/0x60 [92146.548550] [<c105d1e1>] __blocking_notifier_call_chain+0x41/0x80 [92146.548556] [<c105d23f>] blocking_notifier_call_chain+0x1f/0x30 [92146.548563] [<c131a684>] acpi_lid_send_state+0x78/0xa4 [92146.548569] [<c131aa9e>] acpi_button_notify+0x3b/0xf1 [92146.548577] [<c12df56a>] ? acpi_os_execute+0x17/0x19 [92146.548582] [<c12e591a>] ? acpi_ec_sync_query+0xa5/0xbc [92146.548589] [<c12e2b82>] acpi_device_notify+0x16/0x18 [92146.548595] [<c12f4904>] acpi_ev_notify_dispatch+0x38/0x4f [92146.548600] [<c12df0e8>] acpi_os_execute_deferred+0x20/0x2b [92146.548607] [<c1051208>] process_one_work+0x128/0x3f0 [92146.548613] [<c1564f73>] ? common_interrupt+0x33/0x38 [92146.548618] [<c104f8c0>] ? wake_up_worker+0x30/0x30 [92146.548624] [<c12df0c8>] ? acpi_os_wait_events_complete+0x1e/0x1e [92146.548629] [<c10524f9>] worker_thread+0x119/0x3b0 [92146.548634] [<c10523e0>] ? manage_workers+0x240/0x240 [92146.548640] [<c1056e84>] kthread+0x94/0xa0 [92146.548647] [<c1060000>] ? ftrace_raw_output_sched_stat_runtime+0x70/0xf0 [92146.548652] [<c15649b7>] ret_from_kernel_thread+0x1b/0x28 [92146.548658] [<c1056df0>] ? kthread_create_on_node+0xc0/0xc0 three different modeset flags are introduced in this patch MODESET_ON_LID_OPEN: do modeset on next lid open event MODESET_DONE: modeset already done MODESET_SUSPENDED: suspended, only do modeset when system is resumed In this way, 1. when lid is closed, MODESET_ON_LID_OPEN is set so that we'll do modeset on next lid open event. 2. when lid is opened, MODESET_DONE is set so that duplicate lid open events will be ignored. 3. when system suspends, MODESET_SUSPENDED is set. In this case, we will not do modeset on any lid events. Plus, locking mechanism is also introduced to avoid racing. Signed-off-by: Zhang Rui <rui.zhang@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2013-02-05 00:41:53 -07:00
enum modeset_restore modeset_restore;
struct mutex modeset_restore_lock;
struct drm_atomic_state *modeset_restore_state;
struct drm_modeset_acquire_ctx reset_ctx;
struct list_head vm_list; /* Global list of all address spaces */
struct i915_ggtt ggtt; /* VM representing the global address space */
struct i915_gem_mm mm;
drm/i915: Prevent recursive deadlock on releasing a busy userptr During release of the GEM object we hold the struct_mutex. As the object may be holding onto the last reference for the task->mm, calling mmput() may trigger exit_mmap() which close the vma which will call drm_gem_vm_close() and attempt to reacquire the struct_mutex. In order to avoid that recursion, we have to defer the mmput() until after we drop the struct_mutex, i.e. we need to schedule a worker to do the clean up. A further issue spotted by Tvrtko was caused when we took a GTT mmapping of a userptr buffer object. In that case, we would never call mmput as the object would be cyclically referenced by the GTT mmapping and not freed upon process exit - keeping the entire process mm alive after the process task was reaped. The fix employed is to replace the mm_users/mmput() reference handling to mm_count/mmdrop() for the shared i915_mm_struct. INFO: task test_surfaces:1632 blocked for more than 120 seconds.       Tainted: GF          O 3.14.5+ #1 "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. test_surfaces   D 0000000000000000     0  1632   1590 0x00000082  ffff88014914baa8 0000000000000046 0000000000000000 ffff88014914a010  0000000000012c40 0000000000012c40 ffff8800a0058210 ffff88014784b010  ffff88014914a010 ffff880037b1c820 ffff8800a0058210 ffff880037b1c824 Call Trace:  [<ffffffff81582499>] schedule+0x29/0x70  [<ffffffff815825fe>] schedule_preempt_disabled+0xe/0x10  [<ffffffff81583b93>] __mutex_lock_slowpath+0x183/0x220  [<ffffffff81583c53>] mutex_lock+0x23/0x40  [<ffffffffa005c2a3>] drm_gem_vm_close+0x33/0x70 [drm]  [<ffffffff8115a483>] remove_vma+0x33/0x70  [<ffffffff8115a5dc>] exit_mmap+0x11c/0x170  [<ffffffff8104d6eb>] mmput+0x6b/0x100  [<ffffffffa00f44b9>] i915_gem_userptr_release+0x89/0xc0 [i915]  [<ffffffffa00e6706>] i915_gem_free_object+0x126/0x250 [i915]  [<ffffffffa005c06a>] drm_gem_object_free+0x2a/0x40 [drm]  [<ffffffffa005cc32>] drm_gem_object_handle_unreference_unlocked+0xe2/0x120 [drm]  [<ffffffffa005ccd4>] drm_gem_object_release_handle+0x64/0x90 [drm]  [<ffffffff8127ffeb>] idr_for_each+0xab/0x100  [<ffffffffa005cc70>] ? drm_gem_object_handle_unreference_unlocked+0x120/0x120 [drm]  [<ffffffff81583c46>] ? mutex_lock+0x16/0x40  [<ffffffffa005c354>] drm_gem_release+0x24/0x40 [drm]  [<ffffffffa005b82b>] drm_release+0x3fb/0x480 [drm]  [<ffffffff8118d482>] __fput+0xb2/0x260  [<ffffffff8118d6de>] ____fput+0xe/0x10  [<ffffffff8106f27f>] task_work_run+0x8f/0xf0  [<ffffffff81052228>] do_exit+0x1a8/0x480  [<ffffffff81052551>] do_group_exit+0x51/0xc0  [<ffffffff810525d7>] SyS_exit_group+0x17/0x20  [<ffffffff8158e092>] system_call_fastpath+0x16/0x1b v2: Incorporate feedback from Tvrtko and remove the unnessary mm referencing when creating the i915_mm_struct and improve some of the function names and comments. Reported-by: Jacek Danecki <jacek.danecki@intel.com> Test-case: igt/gem_userptr_blits/process-exit* Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Tested-by: "Gong, Zhipeng" <zhipeng.gong@intel.com> Cc: Jacek Danecki <jacek.danecki@intel.com> Cc: "Ursulin, Tvrtko" <tvrtko.ursulin@intel.com> Reviewed-by: "Ursulin, Tvrtko" <tvrtko.ursulin@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: stable@vger.kernel.org # hold off until 3.17 ships for additional testing Signed-off-by: Jani Nikula <jani.nikula@intel.com>
2014-08-07 07:20:40 -06:00
DECLARE_HASHTABLE(mm_structs, 7);
struct mutex mm_lock;
/* Kernel Modesetting */
struct intel_crtc *plane_to_crtc_mapping[I915_MAX_PIPES];
struct intel_crtc *pipe_to_crtc_mapping[I915_MAX_PIPES];
#ifdef CONFIG_DEBUG_FS
struct intel_pipe_crc pipe_crc[I915_MAX_PIPES];
#endif
/* dpll and cdclk state is protected by connection_mutex */
int num_shared_dpll;
struct intel_shared_dpll shared_dplls[I915_NUM_PLLS];
const struct intel_dpll_mgr *dpll_mgr;
/*
* dpll_lock serializes intel_{prepare,enable,disable}_shared_dpll.
* Must be global rather than per dpll, because on some platforms
* plls share registers.
*/
struct mutex dpll_lock;
unsigned int active_crtcs;
unsigned int min_pixclk[I915_MAX_PIPES];
int dpio_phy_iosf_port[I915_NUM_PHYS_VLV];
struct i915_workarounds workarounds;
drm/i915: Track frontbuffer invalidation/flushing So these are the guts of the new beast. This tracks when a frontbuffer gets invalidated (due to frontbuffer rendering) and hence should be constantly scaned out, and when it's flushed again and can be compressed/one-shot-upload. Rules for flushing are simple: The frontbuffer needs one more full upload starting from the next vblank. Which means that the flushing can _only_ be called once the frontbuffer update has been latched. But this poses a problem for pageflips: We can't just delay the flushing until the pageflip is latched, since that would pose the risk that we override frontbuffer rendering that has been scheduled in-between the pageflip ioctl and the actual latching. To handle this track asynchronous invalidations (and also pageflip) state per-ring and delay any in-between flushing until the rendering has completed. And also cancel any delayed flushing if we get a new invalidation request (whether delayed or not). Also call intel_mark_fb_busy in both cases in all cases to make sure that we keep the screen at the highest refresh rate both on flips, synchronous plane updates and for frontbuffer rendering. v2: Lots of improvements Suggestions from Chris: - Move invalidate/flush in flush_*_domain and set_to_*_domain. - Drop the flush in busy_ioctl since it's redundant. Was a leftover from an earlier concept to track flips/delayed flushes. - Don't forget about the initial modeset enable/final disable. Suggested by Chris. Track flips accurately, too. Since flips complete independently of rendering we need to track pending flips in a separate mask. Again if an invalidate happens we need to cancel the evenutal flush to avoid races. v3: Provide correct header declarations for flip functions. Currently not needed outside of intel_display.c, but part of the proper interface. v4: Add proper domain management to fbcon so that the fbcon buffer is also tracked correctly. v5: Fixup locking around the fbcon set_to_gtt_domain call. v6: More comments from Chris: - Split out fbcon changes. - Drop superflous checks for potential scanout before calling intel_fb functions - we can micro-optimize this later. - s/intel_fb_/intel_fb_obj_/ to make it clear that this deals in gem object. We already have precedence for fb_obj in the pin_and_fence functions. v7: Clarify the semantics of the flip flush handling by renaming things a bit: - Don't go through a gem object but take the relevant frontbuffer bits directly. These functions center on the plane, the actual object is irrelevant - even a flip to the same object as already active should cause a flush. - Add a new intel_frontbuffer_flip for synchronous plane updates. It currently just calls intel_frontbuffer_flush since the implemenation differs. This way we achieve a clear split between one-shot update events on one side and frontbuffer rendering with potentially a very long delay between the invalidate and flush. Chris and I also had some discussions about mark_busy and whether it is appropriate to call from flush. But mark busy is a state which should be derived from the 3 events (invalidate, flush, flip) we now have by the users, like psr does by tracking relevant information in psr.busy_frontbuffer_bits. DRRS (the only real use of mark_busy for frontbuffer) needs to have similar logic. With that the overall mark_busy in the core could be removed. v8: Only when retiring gpu buffers only flush frontbuffer bits we actually invalidated in a batch. Just for safety since before any additional usage/invalidate we should always retire current rendering. Suggested by Chris Wilson. v9: Actually use intel_frontbuffer_flip in all appropriate places. Spotted by Chris. v10: Address more comments from Chris: - Don't call _flip in set_base when the crtc is inactive, avoids redunancy in the modeset case with the initial enabling of all planes. - Add comments explaining that the initial/final plane enable/disable still has work left to do before it's fully generic. v11: Only invalidate for gtt/cpu access when writing. Spotted by Chris. v12: s/_flush/_flip/ in intel_overlay.c per Chris' comment. Cc: Rodrigo Vivi <rodrigo.vivi@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2014-06-19 08:01:59 -06:00
struct i915_frontbuffer_tracking fb_tracking;
drm/i915: Move atomic state free from out of fence release Fences are required to support being released from under an atomic context. The drm_atomic_state struct may take a mutex when being released and so we cannot drop a reference to the drm_atomic_state from the fence release path directly, and so we need to defer that unreference to a worker. [ 326.576697] WARNING: CPU: 2 PID: 366 at kernel/sched/core.c:7737 __might_sleep+0x5d/0x80 [ 326.576816] do not call blocking ops when !TASK_RUNNING; state=1 set at [<ffffffffc0359549>] intel_breadcrumbs_signaler+0x59/0x270 [i915] [ 326.576818] Modules linked in: rfcomm fuse snd_hda_codec_hdmi bnep snd_hda_codec_realtek snd_hda_codec_generic snd_hda_intel snd_hda_codec snd_hwdep snd_hda_core snd_pcm snd_seq_midi snd_seq_midi_event snd_rawmidi snd_seq snd_seq_device snd_timer input_leds led_class snd punit_atom_debug btusb btrtl btbcm btintel intel_rapl bluetooth i915 drm_kms_helper syscopyarea sysfillrect iwlwifi sysimgblt soundcore fb_sys_fops mei_txe cfg80211 drm pwm_lpss_platform pwm_lpss pinctrl_cherryview fjes acpi_pad parport_pc ppdev parport autofs4 [ 326.576899] CPU: 2 PID: 366 Comm: i915/signal:0 Tainted: G U 4.10.0-rc3-patser+ #5030 [ 326.576902] Hardware name: /NUC5PPYB, BIOS PYBSWCEL.86A.0031.2015.0601.1712 06/01/2015 [ 326.576905] Call Trace: [ 326.576920] dump_stack+0x4d/0x6d [ 326.576926] __warn+0xc0/0xe0 [ 326.576931] warn_slowpath_fmt+0x5a/0x80 [ 326.577004] ? intel_breadcrumbs_signaler+0x59/0x270 [i915] [ 326.577075] ? intel_breadcrumbs_signaler+0x59/0x270 [i915] [ 326.577079] __might_sleep+0x5d/0x80 [ 326.577087] mutex_lock+0x1b/0x40 [ 326.577133] drm_property_free_blob+0x1e/0x80 [drm] [ 326.577167] ? drm_property_destroy+0xe0/0xe0 [drm] [ 326.577200] drm_mode_object_unreference+0x5c/0x70 [drm] [ 326.577233] drm_property_unreference_blob+0xe/0x10 [drm] [ 326.577260] __drm_atomic_helper_crtc_destroy_state+0x14/0x40 [drm_kms_helper] [ 326.577278] drm_atomic_helper_crtc_destroy_state+0x10/0x20 [drm_kms_helper] [ 326.577352] intel_crtc_destroy_state+0x9/0x10 [i915] [ 326.577388] drm_atomic_state_default_clear+0xea/0x1d0 [drm] [ 326.577462] intel_atomic_state_clear+0xd/0x20 [i915] [ 326.577497] drm_atomic_state_clear+0x1a/0x30 [drm] [ 326.577532] __drm_atomic_state_free+0x13/0x60 [drm] [ 326.577607] intel_atomic_commit_ready+0x6f/0x78 [i915] [ 326.577670] i915_sw_fence_release+0x3a/0x50 [i915] [ 326.577733] dma_i915_sw_fence_wake+0x39/0x80 [i915] [ 326.577741] dma_fence_signal+0xda/0x120 [ 326.577812] ? intel_breadcrumbs_signaler+0x59/0x270 [i915] [ 326.577884] intel_breadcrumbs_signaler+0xb1/0x270 [i915] [ 326.577889] kthread+0x127/0x130 [ 326.577961] ? intel_engine_remove_wait+0x1a0/0x1a0 [i915] [ 326.577964] ? kthread_stop+0x120/0x120 [ 326.577970] ret_from_fork+0x22/0x30 Fixes: c004a90b7263 ("drm/i915: Restore nonblocking awaits for modesetting") Reported-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/20170123212939.30345-1-chris@chris-wilson.co.uk Cc: <drm-intel-fixes@lists.freedesktop.org> # v4.10-rc1+ Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
2017-01-23 14:29:39 -07:00
struct intel_atomic_helper {
struct llist_head free_list;
struct work_struct free_work;
} atomic_helper;
u16 orig_clock;
bool mchbar_need_disable;
struct intel_l3_parity l3_parity;
/* Cannot be determined by PCIID. You must always read a register. */
u32 edram_cap;
/* gen6+ rps state */
struct intel_gen6_power_mgmt rps;
/* ilk-only ips/rps state. Everything in here is protected by the global
* mchdev_lock in intel_pm.c */
struct intel_ilk_power_mgmt ips;
struct i915_power_domains power_domains;
struct i915_psr psr;
struct i915_gpu_error gpu_error;
struct drm_i915_gem_object *vlv_pctx;
/* list of fbdev register on this device */
struct intel_fbdev *fbdev;
struct work_struct fbdev_suspend_work;
struct drm_property *broadcast_rgb_property;
struct drm_property *force_audio_property;
/* hda/i915 audio component */
struct i915_audio_component *audio_component;
bool audio_component_registered;
/**
* av_mutex - mutex for audio/video sync
*
*/
struct mutex av_mutex;
struct {
struct list_head list;
struct llist_head free_list;
struct work_struct free_work;
/* The hw wants to have a stable context identifier for the
* lifetime of the context (for OA, PASID, faults, etc).
* This is limited in execlists to 21 bits.
*/
struct ida hw_ida;
#define MAX_CONTEXT_HW_ID (1<<21) /* exclusive */
} contexts;
u32 fdi_rx_config;
drm/i915: Implement WaPixelRepeatModeFixForC0:chv DPLL_MD(PIPE_C) is AWOL on CHV. Instead of fixing it someone added chicken bits to propagate the pixel multiplier from DPLL_MD(PIPE_B) to either pipe B or C. So do that to make pixel repeat work on pipes B and C. Pipe A is fine without any tricks. Fortunately the pixel repeat propagation appears to be a oneshot operation, so once the value has been written we can clear the chicken bits. So it is still possible to drive pipe B and C with different pixel multipliers simultaneosly. Looks like DPLL_VGA_MODE_DIS must also be set in DPLL(PIPE_B) for this to work. But since we keep that bit always set in all DPLLs there's no problem. This of course means we can't reliably read out the pixel multiplier for pipes B and C. That would make the state checker unhappy, so I added shadow copies of those registers in to dev_priv. The other option would have been to skip pixel multiplier, dpll_md an dotclock checks entirely on CHV, but that feels like a serious loss of cross checking, so just pretending that we have working DPLL MD registers seemed better. Obviously with the shadow copies we can't detect if the pixel multiplier was properly configured, nor can we take over its state from the BIOS, but hopefully people won't have displays that would be limitd to such crappy modes. There is one strange flicker still remaining. It's visible on pipe C/HDMID when HDMIB is enabled while driven by pipe B. It doesn't occur if pipe A drives HDMIB, nor is there any glitch on pipe B/HDMIB when port C/HDMID starts up. I don't have a board with HDMIC so not sure if it happens there too. So I'm not sure if it's somehow tied in with this strange linkage between pipe B and C. Sadly I was unable to find an enable sequence that would avoid the glitch, but at least it's not fatal ie. the output recovers afterwards. Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1458052809-23426-4-git-send-email-ville.syrjala@linux.intel.com Reviewed-by: Jani Nikula <jani.nikula@intel.com>
2016-03-15 08:39:56 -06:00
/* Shadow for DISPLAY_PHY_CONTROL which can't be safely read */
drm/i915: Work around DISPLAY_PHY_CONTROL register corruption on CHV Sometimes (exactly when is a bit unclear) DISPLAY_PHY_CONTROL appears to get corrupted. The values I've managed to read from it seem to have some pattern but vary quite a lot. The corruption doesn't seem to just happen when the register is accessed, but can also happen spontaneosly during modeset. When this happens during a modeset things go south and the display doesn't light up. I've managed to hit the problemn when toggling HDMI on port D on and off. When things get corrupted the display doesn't light up, but as soon as I manually write the correct value to the register the display comes up. First I was suspicious that we ourselves accidentally overwrite it with garbage, but didn't catch anything with the reg_rw tracepoint. Also I sprinkled check all over the modeset path to see exactly when the corruption happens, and eg. the read back value was fine just before intel_dp_set_m(), and corrupted immediately after it. I also made my check function repair the register value whenever it was wrong, and with this approach the corruption repeated several times during the modeset operation, always seeming to trigger in the same exact calls to the check function, while other calls to the function never caught anything. So far I've not seen this problem occurring when carefully avoiding all read accesses to DISPLAY_PHY_CONTROL. Not sure if that's just pure luck or an actual workaround, but we can hope it works. So let's avoid reading the register and instead track the desired value of the register in dev_priv. v2: Read out the power well state to determine initial register value v3: Use DPIO_CHx names instead of raw numbers Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Reviewed-by: Deepak S <deepak.s@linux.intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2015-04-10 09:21:28 -06:00
u32 chv_phy_control;
drm/i915: Implement WaPixelRepeatModeFixForC0:chv DPLL_MD(PIPE_C) is AWOL on CHV. Instead of fixing it someone added chicken bits to propagate the pixel multiplier from DPLL_MD(PIPE_B) to either pipe B or C. So do that to make pixel repeat work on pipes B and C. Pipe A is fine without any tricks. Fortunately the pixel repeat propagation appears to be a oneshot operation, so once the value has been written we can clear the chicken bits. So it is still possible to drive pipe B and C with different pixel multipliers simultaneosly. Looks like DPLL_VGA_MODE_DIS must also be set in DPLL(PIPE_B) for this to work. But since we keep that bit always set in all DPLLs there's no problem. This of course means we can't reliably read out the pixel multiplier for pipes B and C. That would make the state checker unhappy, so I added shadow copies of those registers in to dev_priv. The other option would have been to skip pixel multiplier, dpll_md an dotclock checks entirely on CHV, but that feels like a serious loss of cross checking, so just pretending that we have working DPLL MD registers seemed better. Obviously with the shadow copies we can't detect if the pixel multiplier was properly configured, nor can we take over its state from the BIOS, but hopefully people won't have displays that would be limitd to such crappy modes. There is one strange flicker still remaining. It's visible on pipe C/HDMID when HDMIB is enabled while driven by pipe B. It doesn't occur if pipe A drives HDMIB, nor is there any glitch on pipe B/HDMIB when port C/HDMID starts up. I don't have a board with HDMIC so not sure if it happens there too. So I'm not sure if it's somehow tied in with this strange linkage between pipe B and C. Sadly I was unable to find an enable sequence that would avoid the glitch, but at least it's not fatal ie. the output recovers afterwards. Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1458052809-23426-4-git-send-email-ville.syrjala@linux.intel.com Reviewed-by: Jani Nikula <jani.nikula@intel.com>
2016-03-15 08:39:56 -06:00
/*
* Shadows for CHV DPLL_MD regs to keep the state
* checker somewhat working in the presence hardware
* crappiness (can't read out DPLL_MD for pipes B & C).
*/
u32 chv_dpll_md[I915_MAX_PIPES];
u32 bxt_phy_grc;
drm/i915: Work around DISPLAY_PHY_CONTROL register corruption on CHV Sometimes (exactly when is a bit unclear) DISPLAY_PHY_CONTROL appears to get corrupted. The values I've managed to read from it seem to have some pattern but vary quite a lot. The corruption doesn't seem to just happen when the register is accessed, but can also happen spontaneosly during modeset. When this happens during a modeset things go south and the display doesn't light up. I've managed to hit the problemn when toggling HDMI on port D on and off. When things get corrupted the display doesn't light up, but as soon as I manually write the correct value to the register the display comes up. First I was suspicious that we ourselves accidentally overwrite it with garbage, but didn't catch anything with the reg_rw tracepoint. Also I sprinkled check all over the modeset path to see exactly when the corruption happens, and eg. the read back value was fine just before intel_dp_set_m(), and corrupted immediately after it. I also made my check function repair the register value whenever it was wrong, and with this approach the corruption repeated several times during the modeset operation, always seeming to trigger in the same exact calls to the check function, while other calls to the function never caught anything. So far I've not seen this problem occurring when carefully avoiding all read accesses to DISPLAY_PHY_CONTROL. Not sure if that's just pure luck or an actual workaround, but we can hope it works. So let's avoid reading the register and instead track the desired value of the register in dev_priv. v2: Read out the power well state to determine initial register value v3: Use DPIO_CHx names instead of raw numbers Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Reviewed-by: Deepak S <deepak.s@linux.intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2015-04-10 09:21:28 -06:00
u32 suspend_count;
bool suspended_to_idle;
struct i915_suspend_saved_registers regfile;
struct vlv_s0ix_state vlv_s0ix_state;
drm/i915/skl: Add support for the SAGV, fix underrun hangs Since the watermark calculations for Skylake are still broken, we're apt to hitting underruns very easily under multi-monitor configurations. While it would be lovely if this was fixed, it's not. Another problem that's been coming from this however, is the mysterious issue of underruns causing full system hangs. An easy way to reproduce this with a skylake system: - Get a laptop with a skylake GPU, and hook up two external monitors to it - Move the cursor from the built-in LCD to one of the external displays as quickly as you can - You'll get a few pipe underruns, and eventually the entire system will just freeze. After doing a lot of investigation and reading through the bspec, I found the existence of the SAGV, which is responsible for adjusting the system agent voltage and clock frequencies depending on how much power we need. According to the bspec: "The display engine access to system memory is blocked during the adjustment time. SAGV defaults to enabled. Software must use the GT-driver pcode mailbox to disable SAGV when the display engine is not able to tolerate the blocking time." The rest of the bspec goes on to explain that software can simply leave the SAGV enabled, and disable it when we use interlaced pipes/have more then one pipe active. Sure enough, with this patchset the system hangs resulting from pipe underruns on Skylake have completely vanished on my T460s. Additionally, the bspec mentions turning off the SAGV with more then one pipe enabled as a workaround for display underruns. While this patch doesn't entirely fix that, it looks like it does improve the situation a little bit so it's likely this is going to be required to make watermarks on Skylake fully functional. This will still need additional work in the future: we shouldn't be enabling the SAGV if any of the currently enabled planes can't enable WM levels that introduce latencies >= 30 µs. Changes since v11: - Add skl_can_enable_sagv() - Make sure we don't enable SAGV when not all planes can enable watermarks >= the SAGV engine block time. I was originally going to save this for later, but I recently managed to run into a machine that was having problems with a single pipe configuration + SAGV. - Make comparisons to I915_SKL_SAGV_NOT_CONTROLLED explicit - Change I915_SAGV_DYNAMIC_FREQ to I915_SAGV_ENABLE - Move printks outside of mutexes - Don't print error messages twice Changes since v10: - Apparently sandybridge_pcode_read actually writes values and reads them back, despite it's misleading function name. This means we've been doing this mostly wrong and have been writing garbage to the SAGV control. Because of this, we no longer attempt to read the SAGV status during initialization (since there are no helpers for this). - mlankhorst noticed that this patch was breaking on some very early pre-release Skylake machines, which apparently don't allow you to disable the SAGV. To prevent machines from failing tests due to SAGV errors, if the first time we try to control the SAGV results in the mailbox indicating an invalid command, we just disable future attempts to control the SAGV state by setting dev_priv->skl_sagv_status to I915_SKL_SAGV_NOT_CONTROLLED and make a note of it in dmesg. - Move mutex_unlock() a little higher in skl_enable_sagv(). This doesn't actually fix anything, but lets us release the lock a little sooner since we're finished with it. Changes since v9: - Only enable/disable sagv on Skylake Changes since v8: - Add intel_state->modeset guard to the conditional for skl_enable_sagv() Changes since v7: - Remove GEN9_SAGV_LOW_FREQ, replace with GEN9_SAGV_IS_ENABLED (that's all we use it for anyway) - Use GEN9_SAGV_IS_ENABLED instead of 0x1 for clarification - Fix a styling error that snuck past me Changes since v6: - Protect skl_enable_sagv() with intel_state->modeset conditional in intel_atomic_commit_tail() Changes since v5: - Don't use is_power_of_2. Makes things confusing - Don't use the old state to figure out whether or not to enable/disable the sagv, use the new one - Split the loop in skl_disable_sagv into it's own function - Move skl_sagv_enable/disable() calls into intel_atomic_commit_tail() Changes since v4: - Use is_power_of_2 against active_crtcs to check whether we have > 1 pipe enabled - Fix skl_sagv_get_hw_state(): (temp & 0x1) indicates disabled, 0x0 enabled - Call skl_sagv_enable/disable() from pre/post-plane updates Changes since v3: - Use time_before() to compare timeout to jiffies Changes since v2: - Really apply minor style nitpicks to patch this time Changes since v1: - Added comments about this probably being one of the requirements to fixing Skylake's watermark issues - Minor style nitpicks from Matt Roper - Disable these functions on Broxton, since it doesn't have an SAGV Signed-off-by: Lyude <cpaul@redhat.com> Cc: Matt Roper <matthew.d.roper@intel.com> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Ville Syrjälä <ville.syrjala@linux.intel.com> Cc: stable@vger.kernel.org Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1471463761-26796-3-git-send-email-cpaul@redhat.com [mlankhorst: ENOSYS -> ENXIO, whitespace fixes]
2016-08-17 13:55:54 -06:00
enum {
I915_SAGV_UNKNOWN = 0,
I915_SAGV_DISABLED,
I915_SAGV_ENABLED,
I915_SAGV_NOT_CONTROLLED
} sagv_status;
drm/i915/skl: Add support for the SAGV, fix underrun hangs Since the watermark calculations for Skylake are still broken, we're apt to hitting underruns very easily under multi-monitor configurations. While it would be lovely if this was fixed, it's not. Another problem that's been coming from this however, is the mysterious issue of underruns causing full system hangs. An easy way to reproduce this with a skylake system: - Get a laptop with a skylake GPU, and hook up two external monitors to it - Move the cursor from the built-in LCD to one of the external displays as quickly as you can - You'll get a few pipe underruns, and eventually the entire system will just freeze. After doing a lot of investigation and reading through the bspec, I found the existence of the SAGV, which is responsible for adjusting the system agent voltage and clock frequencies depending on how much power we need. According to the bspec: "The display engine access to system memory is blocked during the adjustment time. SAGV defaults to enabled. Software must use the GT-driver pcode mailbox to disable SAGV when the display engine is not able to tolerate the blocking time." The rest of the bspec goes on to explain that software can simply leave the SAGV enabled, and disable it when we use interlaced pipes/have more then one pipe active. Sure enough, with this patchset the system hangs resulting from pipe underruns on Skylake have completely vanished on my T460s. Additionally, the bspec mentions turning off the SAGV with more then one pipe enabled as a workaround for display underruns. While this patch doesn't entirely fix that, it looks like it does improve the situation a little bit so it's likely this is going to be required to make watermarks on Skylake fully functional. This will still need additional work in the future: we shouldn't be enabling the SAGV if any of the currently enabled planes can't enable WM levels that introduce latencies >= 30 µs. Changes since v11: - Add skl_can_enable_sagv() - Make sure we don't enable SAGV when not all planes can enable watermarks >= the SAGV engine block time. I was originally going to save this for later, but I recently managed to run into a machine that was having problems with a single pipe configuration + SAGV. - Make comparisons to I915_SKL_SAGV_NOT_CONTROLLED explicit - Change I915_SAGV_DYNAMIC_FREQ to I915_SAGV_ENABLE - Move printks outside of mutexes - Don't print error messages twice Changes since v10: - Apparently sandybridge_pcode_read actually writes values and reads them back, despite it's misleading function name. This means we've been doing this mostly wrong and have been writing garbage to the SAGV control. Because of this, we no longer attempt to read the SAGV status during initialization (since there are no helpers for this). - mlankhorst noticed that this patch was breaking on some very early pre-release Skylake machines, which apparently don't allow you to disable the SAGV. To prevent machines from failing tests due to SAGV errors, if the first time we try to control the SAGV results in the mailbox indicating an invalid command, we just disable future attempts to control the SAGV state by setting dev_priv->skl_sagv_status to I915_SKL_SAGV_NOT_CONTROLLED and make a note of it in dmesg. - Move mutex_unlock() a little higher in skl_enable_sagv(). This doesn't actually fix anything, but lets us release the lock a little sooner since we're finished with it. Changes since v9: - Only enable/disable sagv on Skylake Changes since v8: - Add intel_state->modeset guard to the conditional for skl_enable_sagv() Changes since v7: - Remove GEN9_SAGV_LOW_FREQ, replace with GEN9_SAGV_IS_ENABLED (that's all we use it for anyway) - Use GEN9_SAGV_IS_ENABLED instead of 0x1 for clarification - Fix a styling error that snuck past me Changes since v6: - Protect skl_enable_sagv() with intel_state->modeset conditional in intel_atomic_commit_tail() Changes since v5: - Don't use is_power_of_2. Makes things confusing - Don't use the old state to figure out whether or not to enable/disable the sagv, use the new one - Split the loop in skl_disable_sagv into it's own function - Move skl_sagv_enable/disable() calls into intel_atomic_commit_tail() Changes since v4: - Use is_power_of_2 against active_crtcs to check whether we have > 1 pipe enabled - Fix skl_sagv_get_hw_state(): (temp & 0x1) indicates disabled, 0x0 enabled - Call skl_sagv_enable/disable() from pre/post-plane updates Changes since v3: - Use time_before() to compare timeout to jiffies Changes since v2: - Really apply minor style nitpicks to patch this time Changes since v1: - Added comments about this probably being one of the requirements to fixing Skylake's watermark issues - Minor style nitpicks from Matt Roper - Disable these functions on Broxton, since it doesn't have an SAGV Signed-off-by: Lyude <cpaul@redhat.com> Cc: Matt Roper <matthew.d.roper@intel.com> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Ville Syrjälä <ville.syrjala@linux.intel.com> Cc: stable@vger.kernel.org Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1471463761-26796-3-git-send-email-cpaul@redhat.com [mlankhorst: ENOSYS -> ENXIO, whitespace fixes]
2016-08-17 13:55:54 -06:00
struct {
/*
* Raw watermark latency values:
* in 0.1us units for WM0,
* in 0.5us units for WM1+.
*/
/* primary */
uint16_t pri_latency[5];
/* sprite */
uint16_t spr_latency[5];
/* cursor */
uint16_t cur_latency[5];
/*
* Raw watermark memory latency values
* for SKL for all 8 levels
* in 1us units.
*/
uint16_t skl_latency[8];
/* current hardware state */
drm/i915/skl: SKL Watermark Computation This patch implements the watermark algorithm and its necessary functions. Two function pointers skl_update_wm and skl_update_sprite_wm are provided. The skl_update_wm will update the watermarks for the crtc provided as an argument and then checks for change in DDB allocation for other active pipes and recomputes the watermarks for those Pipes and planes as well. Finally it does the register programming for all dirty pipes. The trigger of the Watermark double buffer registers will have to be once the plane configurations are done by the caller. v2: fixed the divide-by-0 error in the results computation func. Also reworked the PLANE_WM register values computation func to make it more compact. Incorporated all other review comments from Damien. v3: Changed the skl_compute_plane_wm function to now return success or failure. Also the result blocks and lines are computed here instead of in skl_compute_wm_results function. v4: Adjust skl_ddb_alloc_changed() to the new planes/cursor split (Damien) v5: Reworked the affected functions to implement new plane/cursor split. v6: Rework the logic that triggers the DDB allocation and WM computation of skl_update_other_pipe_wm() to not depend on non-computed DDB values. Always give a valid cursor_width (at boot it's 0) to keep the invariant that we consider the cursor plane always enabled. Otherwise we end up dividing by 0 in skl_compute_plane_wm() (Damien Lespiau) v7: Spell out allocation skl_ddb_ functions should have the ddb as first argument Make the skl_ddb_alloc_changed() parameters const (Damien) v8: Rebase on top of the crtc->primary changes v9: Split the staging results structure to not exceed the 1Kb stack allocation in skl_update_wm() v10: Make skl_pipe_pixel_rate() take a pointer to the pipe config Add a comment about overflow considerations for skl_wm_method1() Various additions of const Various use of sizeof(variable) instead of sizeof(type) Various move of variable definitons to a narrower scope Zero initialize some stack allocated structures to make sure we don't have garbage in case we don't write all the values (Ville) v11: Remove non-necessary default number of blocks/lines when the plane is disabled (Ville) Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Signed-off-by: Pradeep Bhat <pradeep.bhat@intel.com> Signed-off-by: Damien Lespiau <damien.lespiau@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2014-11-04 10:06:42 -07:00
union {
struct ilk_wm_values hw;
struct skl_wm_values skl_hw;
struct vlv_wm_values vlv;
drm/i915: Two stage watermarks for g4x Implement proper two stage watermark programming for g4x. As with other pre-SKL platforms, the watermark registers aren't double buffered on g4x. Hence we must sequence the watermark update carefully around plane updates. The code is quite heavily modelled on the VLV/CHV code, with some fairly significant differences due to the different hardware architecture: * g4x doesn't use inverted watermark values * CxSR actually affects the watermarks since it controls memory self refresh in addition to the max FIFO mode * A further HPLL SR mode is possible with higher memory wakeup latency * g4x has FBC2 and so it also has FBC watermarks * max FIFO mode for primary plane only (cursor is allowed, sprite is not) * g4x has no manual FIFO repartitioning * some TLB miss related workarounds are needed for the watermarks Actually the hardware is quite similar to ILK+ in many ways. The most visible differences are in the actual watermakr register layout. ILK revamped that part quite heavily whereas g4x is still using the layout inherited from earlier platforms. Note that we didn't previously enable the HPLL SR on g4x. So in order to not introduce too many functional changes in this patch I've not actually enabled it here either, even though the code is now fully ready for it. We'll enable it separately later on. Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/20170421181432.15216-13-ville.syrjala@linux.intel.com Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
2017-04-21 12:14:29 -06:00
struct g4x_wm_values g4x;
drm/i915/skl: SKL Watermark Computation This patch implements the watermark algorithm and its necessary functions. Two function pointers skl_update_wm and skl_update_sprite_wm are provided. The skl_update_wm will update the watermarks for the crtc provided as an argument and then checks for change in DDB allocation for other active pipes and recomputes the watermarks for those Pipes and planes as well. Finally it does the register programming for all dirty pipes. The trigger of the Watermark double buffer registers will have to be once the plane configurations are done by the caller. v2: fixed the divide-by-0 error in the results computation func. Also reworked the PLANE_WM register values computation func to make it more compact. Incorporated all other review comments from Damien. v3: Changed the skl_compute_plane_wm function to now return success or failure. Also the result blocks and lines are computed here instead of in skl_compute_wm_results function. v4: Adjust skl_ddb_alloc_changed() to the new planes/cursor split (Damien) v5: Reworked the affected functions to implement new plane/cursor split. v6: Rework the logic that triggers the DDB allocation and WM computation of skl_update_other_pipe_wm() to not depend on non-computed DDB values. Always give a valid cursor_width (at boot it's 0) to keep the invariant that we consider the cursor plane always enabled. Otherwise we end up dividing by 0 in skl_compute_plane_wm() (Damien Lespiau) v7: Spell out allocation skl_ddb_ functions should have the ddb as first argument Make the skl_ddb_alloc_changed() parameters const (Damien) v8: Rebase on top of the crtc->primary changes v9: Split the staging results structure to not exceed the 1Kb stack allocation in skl_update_wm() v10: Make skl_pipe_pixel_rate() take a pointer to the pipe config Add a comment about overflow considerations for skl_wm_method1() Various additions of const Various use of sizeof(variable) instead of sizeof(type) Various move of variable definitons to a narrower scope Zero initialize some stack allocated structures to make sure we don't have garbage in case we don't write all the values (Ville) v11: Remove non-necessary default number of blocks/lines when the plane is disabled (Ville) Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Signed-off-by: Pradeep Bhat <pradeep.bhat@intel.com> Signed-off-by: Damien Lespiau <damien.lespiau@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2014-11-04 10:06:42 -07:00
};
uint8_t max_level;
drm/i915: Add two-stage ILK-style watermark programming (v11) In addition to calculating final watermarks, let's also pre-calculate a set of intermediate watermark values at atomic check time. These intermediate watermarks are a combination of the watermarks for the old state and the new state; they should satisfy the requirements of both states which means they can be programmed immediately when we commit the atomic state (without waiting for a vblank). Once the vblank does happen, we can then re-program watermarks to the more optimal final value. v2: Significant rebasing/rewriting. v3: - Move 'need_postvbl_update' flag to CRTC state (Daniel) - Don't forget to check intermediate watermark values for validity (Maarten) - Don't due async watermark optimization; just do it at the end of the atomic transaction, after waiting for vblanks. We do want it to be async eventually, but adding that now will cause more trouble for Maarten's in-progress work. (Maarten) - Don't allocate space in crtc_state for intermediate watermarks on platforms that don't need it (gen9+). - Move WaCxSRDisabledForSpriteScaling:ivb into intel_begin_crtc_commit now that ilk_update_wm is gone. v4: - Add a wm_mutex to cover updates to intel_crtc->active and the need_postvbl_update flag. Since we don't have async yet it isn't terribly important yet, but might as well add it now. - Change interface to program watermarks. Platforms will now expose .initial_watermarks() and .optimize_watermarks() functions to do watermark programming. These should lock wm_mutex, copy the appropriate state values into intel_crtc->active, and then call the internal program watermarks function. v5: - Skip intermediate watermark calculation/check during initial hardware readout since we don't trust the existing HW values (and don't have valid values of our own yet). - Don't try to call .optimize_watermarks() on platforms that don't have atomic watermarks yet. (Maarten) v6: - Rebase v7: - Further rebase v8: - A few minor indentation and line length fixes v9: - Yet another rebase since Maarten's patches reworked a bunch of the code (wm_pre, wm_post, etc.) that this was previously based on. v10: - Move wm_mutex to dev_priv to protect against racing commits against disjoint CRTC sets. (Maarten) - Drop unnecessary clearing of cstate->wm.need_postvbl_update (Maarten) v11: - Now that we've moved to atomic watermark updates, make sure we call the proper function to program watermarks in {ironlake,haswell}_crtc_enable(); the failure to do so on the previous patch iteration led to us not actually programming the watermarks before turning on the CRTC, which was the cause of the underruns that the CI system was seeing. - Fix inverted logic for determining when to optimize watermarks. We were needlessly optimizing when the intermediate/optimal values were the same (harmless), but not actually optimizing when they differed (also harmless, but wasteful from a power/bandwidth perspective). Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Signed-off-by: Matt Roper <matthew.d.roper@intel.com> Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1456276813-5689-1-git-send-email-matthew.d.roper@intel.com
2016-02-23 18:20:13 -07:00
/*
* Should be held around atomic WM register writing; also
* protects * intel_crtc->wm.active and
* cstate->wm.need_postvbl_update.
*/
struct mutex wm_mutex;
/*
* Set during HW readout of watermarks/DDB. Some platforms
* need to know when we're still using BIOS-provided values
* (which we don't fully trust).
*/
bool distrust_bios_wm;
} wm;
struct i915_runtime_pm pm;
drm/i915: Add i915 perf infrastructure Adds base i915 perf infrastructure for Gen performance metrics. This adds a DRM_IOCTL_I915_PERF_OPEN ioctl that takes an array of uint64 properties to configure a stream of metrics and returns a new fd usable with standard VFS system calls including read() to read typed and sized records; ioctl() to enable or disable capture and poll() to wait for data. A stream is opened something like: uint64_t properties[] = { /* Single context sampling */ DRM_I915_PERF_PROP_CTX_HANDLE, ctx_handle, /* Include OA reports in samples */ DRM_I915_PERF_PROP_SAMPLE_OA, true, /* OA unit configuration */ DRM_I915_PERF_PROP_OA_METRICS_SET, metrics_set_id, DRM_I915_PERF_PROP_OA_FORMAT, report_format, DRM_I915_PERF_PROP_OA_EXPONENT, period_exponent, }; struct drm_i915_perf_open_param parm = { .flags = I915_PERF_FLAG_FD_CLOEXEC | I915_PERF_FLAG_FD_NONBLOCK | I915_PERF_FLAG_DISABLED, .properties_ptr = (uint64_t)properties, .num_properties = sizeof(properties) / 16, }; int fd = drmIoctl(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param); Records read all start with a common { type, size } header with DRM_I915_PERF_RECORD_SAMPLE being of most interest. Sample records contain an extensible number of fields and it's the DRM_I915_PERF_PROP_SAMPLE_xyz properties given when opening that determine what's included in every sample. No specific streams are supported yet so any attempt to open a stream will return an error. v2: use i915_gem_context_get() - Chris Wilson v3: update read() interface to avoid passing state struct - Chris Wilson fix some rebase fallout, with i915-perf init/deinit v4: s/DRM_IORW/DRM_IOW/ - Emil Velikov Signed-off-by: Robert Bragg <robert@sixbynine.org> Reviewed-by: Matthew Auld <matthew.auld@intel.com> Reviewed-by: Sourab Gupta <sourab.gupta@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/20161107194957.3385-2-robert@sixbynine.org
2016-11-07 12:49:47 -07:00
struct {
bool initialized;
struct kobject *metrics_kobj;
struct ctl_table_header *sysctl_header;
drm/i915/perf: Implement I915_PERF_ADD/REMOVE_CONFIG interface The motivation behind this new interface is expose at runtime the creation of new OA configs which can be used as part of the i915 perf open interface. This will enable the kernel to learn new configs which may be experimental, or otherwise not part of the core set currently available through the i915 perf interface. v2: Drop DRM_ERROR for userspace errors (Matthew) Add padding to userspace structure (Matthew) s/guid/uuid/ (Matthew) v3: Use u32 instead of int to iterate through registers (Matthew) v4: Lock access to dynamic config list (Lionel) v5: by Matthew: Fix uninitialized error values Fix incorrect unwiding when opening perf stream Use kmalloc_array() to store register Use uuid_is_valid() to valid config uuids Declare ioctls as write only Check padding members are set to 0 by Lionel: Return ENOENT rather than EINVAL when trying to remove non existing config v6: by Chris: Use ref counts for OA configs Store UUID in drm_i915_perf_oa_config rather then using pointer Shuffle fields of drm_i915_perf_oa_config to avoid padding v7: by Chris Rename uapi pointers fields to end with '_ptr' v8: by Andrzej, Marek, Sebastian Update register whitelisting by Lionel Add more register names for documentation Allow configuration programming in non-paranoid mode Add support for value filter for a couple of registers already programmed in other part of the kernel v9: Documentation fix (Lionel) Allow writing WAIT_FOR_RC6_EXIT only on Gen8+ (Andrzej) v10: Perform read access_ok() on register pointers (Lionel) Signed-off-by: Matthew Auld <matthew.auld@intel.com> Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com> Signed-off-by: Andrzej Datczuk <andrzej.datczuk@intel.com> Reviewed-by: Andrzej Datczuk <andrzej.datczuk@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20170803165812.2373-2-lionel.g.landwerlin@intel.com
2017-08-03 11:05:50 -06:00
/*
* Lock associated with adding/modifying/removing OA configs
* in dev_priv->perf.metrics_idr.
*/
struct mutex metrics_lock;
/*
* List of dynamic configurations, you need to hold
* dev_priv->perf.metrics_lock to access it.
*/
struct idr metrics_idr;
/*
* Lock associated with anything below within this structure
* except exclusive_stream.
*/
drm/i915: Add i915 perf infrastructure Adds base i915 perf infrastructure for Gen performance metrics. This adds a DRM_IOCTL_I915_PERF_OPEN ioctl that takes an array of uint64 properties to configure a stream of metrics and returns a new fd usable with standard VFS system calls including read() to read typed and sized records; ioctl() to enable or disable capture and poll() to wait for data. A stream is opened something like: uint64_t properties[] = { /* Single context sampling */ DRM_I915_PERF_PROP_CTX_HANDLE, ctx_handle, /* Include OA reports in samples */ DRM_I915_PERF_PROP_SAMPLE_OA, true, /* OA unit configuration */ DRM_I915_PERF_PROP_OA_METRICS_SET, metrics_set_id, DRM_I915_PERF_PROP_OA_FORMAT, report_format, DRM_I915_PERF_PROP_OA_EXPONENT, period_exponent, }; struct drm_i915_perf_open_param parm = { .flags = I915_PERF_FLAG_FD_CLOEXEC | I915_PERF_FLAG_FD_NONBLOCK | I915_PERF_FLAG_DISABLED, .properties_ptr = (uint64_t)properties, .num_properties = sizeof(properties) / 16, }; int fd = drmIoctl(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param); Records read all start with a common { type, size } header with DRM_I915_PERF_RECORD_SAMPLE being of most interest. Sample records contain an extensible number of fields and it's the DRM_I915_PERF_PROP_SAMPLE_xyz properties given when opening that determine what's included in every sample. No specific streams are supported yet so any attempt to open a stream will return an error. v2: use i915_gem_context_get() - Chris Wilson v3: update read() interface to avoid passing state struct - Chris Wilson fix some rebase fallout, with i915-perf init/deinit v4: s/DRM_IORW/DRM_IOW/ - Emil Velikov Signed-off-by: Robert Bragg <robert@sixbynine.org> Reviewed-by: Matthew Auld <matthew.auld@intel.com> Reviewed-by: Sourab Gupta <sourab.gupta@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/20161107194957.3385-2-robert@sixbynine.org
2016-11-07 12:49:47 -07:00
struct mutex lock;
struct list_head streams;
struct {
drm/i915/perf: Implement I915_PERF_ADD/REMOVE_CONFIG interface The motivation behind this new interface is expose at runtime the creation of new OA configs which can be used as part of the i915 perf open interface. This will enable the kernel to learn new configs which may be experimental, or otherwise not part of the core set currently available through the i915 perf interface. v2: Drop DRM_ERROR for userspace errors (Matthew) Add padding to userspace structure (Matthew) s/guid/uuid/ (Matthew) v3: Use u32 instead of int to iterate through registers (Matthew) v4: Lock access to dynamic config list (Lionel) v5: by Matthew: Fix uninitialized error values Fix incorrect unwiding when opening perf stream Use kmalloc_array() to store register Use uuid_is_valid() to valid config uuids Declare ioctls as write only Check padding members are set to 0 by Lionel: Return ENOENT rather than EINVAL when trying to remove non existing config v6: by Chris: Use ref counts for OA configs Store UUID in drm_i915_perf_oa_config rather then using pointer Shuffle fields of drm_i915_perf_oa_config to avoid padding v7: by Chris Rename uapi pointers fields to end with '_ptr' v8: by Andrzej, Marek, Sebastian Update register whitelisting by Lionel Add more register names for documentation Allow configuration programming in non-paranoid mode Add support for value filter for a couple of registers already programmed in other part of the kernel v9: Documentation fix (Lionel) Allow writing WAIT_FOR_RC6_EXIT only on Gen8+ (Andrzej) v10: Perform read access_ok() on register pointers (Lionel) Signed-off-by: Matthew Auld <matthew.auld@intel.com> Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com> Signed-off-by: Andrzej Datczuk <andrzej.datczuk@intel.com> Reviewed-by: Andrzej Datczuk <andrzej.datczuk@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20170803165812.2373-2-lionel.g.landwerlin@intel.com
2017-08-03 11:05:50 -06:00
/*
* The stream currently using the OA unit. If accessed
* outside a syscall associated to its file
* descriptor, you need to hold
* dev_priv->drm.struct_mutex.
*/
struct i915_perf_stream *exclusive_stream;
u32 specific_ctx_id;
struct hrtimer poll_check_timer;
wait_queue_head_t poll_wq;
bool pollin;
/**
* For rate limiting any notifications of spurious
* invalid OA reports
*/
struct ratelimit_state spurious_report_rs;
bool periodic;
int period_exponent;
int timestamp_frequency;
struct i915_oa_config test_config;
struct {
struct i915_vma *vma;
u8 *vaddr;
drm/i915/perf: Add OA unit support for Gen 8+ Enables access to OA unit metrics for BDW, CHV, SKL and BXT which all share (more-or-less) the same OA unit design. Of particular note in comparison to Haswell: some OA unit HW config state has become per-context state and as a consequence it is somewhat more complicated to manage synchronous state changes from the cpu while there's no guarantee of what context (if any) is currently actively running on the gpu. The periodic sampling frequency which can be particularly useful for system-wide analysis (as opposed to command stream synchronised MI_REPORT_PERF_COUNT commands) is perhaps the most surprising state to have become per-context save and restored (while the OABUFFER destination is still a shared, system-wide resource). This support for gen8+ takes care to consider a number of timing challenges involved in synchronously updating per-context state primarily by programming all config state from the cpu and updating all current and saved contexts synchronously while the OA unit is still disabled. The driver intentionally avoids depending on command streamer programming to update OA state considering the lack of synchronization between the automatic loading of OACTXCONTROL state (that includes the periodic sampling state and enable state) on context restore and the parsing of any general purpose BB the driver can control. I.e. this implementation is careful to avoid the possibility of a context restore temporarily enabling any out-of-date periodic sampling state. In addition to the risk of transiently-out-of-date state being loaded automatically; there are also internal HW latencies involved in the loading of MUX configurations which would be difficult to account for from the command streamer (and we only want to enable the unit when once the MUX configuration is complete). Since the Gen8+ OA unit design no longer supports clock gating the unit off for a single given context (which effectively stopped any progress of counters while any other context was running) and instead supports tagging OA reports with a context ID for filtering on the CPU, it means we can no longer hide the system-wide progress of counters from a non-privileged application only interested in metrics for its own context. Although we could theoretically try and subtract the progress of other contexts before forwarding reports via read() we aren't in a position to filter reports captured via MI_REPORT_PERF_COUNT commands. As a result, for Gen8+, we always require the dev.i915.perf_stream_paranoid to be unset for any access to OA metrics if not root. v5: Drain submitted requests when enabling metric set to ensure no lite-restore erases the context image we just updated (Lionel) v6: In addition to drain, switch to kernel context & update all context in place (Chris) v7: Add missing mutex_unlock() if switching to kernel context fails (Matthew) v8: Simplify OA period/flex-eu-counters programming by using the batchbuffer instead of modifying ctx-image (Lionel) v9: Back to updating the context image (due to erroneous testing, batchbuffer programming the OA unit doesn't actually work) (Lionel) Pin context before updating context image (Chris) Drop MMIO programming now that we switch to a kernel context with right values in initial context image (Chris) v10: Just pin_map the contexts we want to modify or let the configuration happen on first use (Chris) v11: Update kernel context OA config through the batchbuffer rather than on the fly ctx-image update (Lionel) v12: Rework OA context registers update again by swithing away from user contexts and reconfiguring the kernel context through the batchbuffer and updating all the other contexts' context image. Also take care to lock slice/subslice configuration when OA is on. (Lionel) v13: Request rpcs updates on all engine when updating the OA config (Lionel) v14: Drop any kind of rpcs management now that we monitor sseu configuration changes in a later patch (Lionel) Remove usleep after programming the NOA configs on Gen8+, this doesn't seem to be needed (Lionel) v15: Respect coding style for block comments (Chris) v16: Add missing i915_add_request() in case we fail to emit OA configuration (Matthew) Signed-off-by: Robert Bragg <robert@sixbynine.org> Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com> Reviewed-by: Matthew Auld <matthew.auld@intel.com> \o/ Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
2017-06-13 05:23:03 -06:00
u32 last_ctx_id;
int format;
int format_size;
drm/i915/perf: improve tail race workaround There's a HW race condition between OA unit tail pointer register updates and writes to memory whereby the tail pointer can sometimes get ahead of what's been written out to the OA buffer so far (in terms of what's visible to the CPU). Although this can be observed explicitly while copying reports to userspace by checking for a zeroed report-id field in tail reports, we want to account for this earlier, as part of the _oa_buffer_check to avoid lots of redundant read() attempts. Previously the driver used to define an effective tail pointer that lagged the real pointer by a 'tail margin' measured in bytes derived from OA_TAIL_MARGIN_NSEC and the configured sampling frequency. Unfortunately this was flawed considering that the OA unit may also automatically generate non-periodic reports (such as on context switch) or the OA unit may be enabled without any periodic sampling. This improves how we define a tail pointer for reading that lags the real tail pointer by at least %OA_TAIL_MARGIN_NSEC nanoseconds, which gives enough time for the corresponding reports to become visible to the CPU. The driver now maintains two tail pointers: 1) An 'aging' tail with an associated timestamp that is tracked until we can trust the corresponding data is visible to the CPU; at which point it is considered 'aged'. 2) An 'aged' tail that can be used for read()ing. The two separate pointers let us decouple read()s from tail pointer aging. The tail pointers are checked and updated at a limited rate within a hrtimer callback (the same callback that is used for delivering POLLIN events) and since we're now measuring the wall clock time elapsed since a given tail pointer was read the mechanism no longer cares about the OA unit's periodic sampling frequency. The natural place to handle the tail pointer updates was in gen7_oa_buffer_is_empty() which is called as part of blocking reads and the hrtimer callback used for polling, and so this was renamed to oa_buffer_check() considering the added side effect while checking whether the buffer contains data. Signed-off-by: Robert Bragg <robert@sixbynine.org> Reviewed-by: Matthew Auld <matthew.auld@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/20170511154345.962-6-lionel.g.landwerlin@intel.com Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
2017-05-11 09:43:28 -06:00
/**
* Locks reads and writes to all head/tail state
*
* Consider: the head and tail pointer state
* needs to be read consistently from a hrtimer
* callback (atomic context) and read() fop
* (user context) with tail pointer updates
* happening in atomic context and head updates
* in user context and the (unlikely)
* possibility of read() errors needing to
* reset all head/tail state.
*
* Note: Contention or performance aren't
* currently a significant concern here
* considering the relatively low frequency of
* hrtimer callbacks (5ms period) and that
* reads typically only happen in response to a
* hrtimer event and likely complete before the
* next callback.
*
* Note: This lock is not held *while* reading
* and copying data to userspace so the value
* of head observed in htrimer callbacks won't
* represent any partial consumption of data.
*/
spinlock_t ptr_lock;
/**
* One 'aging' tail pointer and one 'aged'
* tail pointer ready to used for reading.
*
* Initial values of 0xffffffff are invalid
* and imply that an update is required
* (and should be ignored by an attempted
* read)
*/
struct {
u32 offset;
} tails[2];
/**
* Index for the aged tail ready to read()
* data up to.
*/
unsigned int aged_tail_idx;
/**
* A monotonic timestamp for when the current
* aging tail pointer was read; used to
* determine when it is old enough to trust.
*/
u64 aging_timestamp;
/**
* Although we can always read back the head
* pointer register, we prefer to avoid
* trusting the HW state, just to avoid any
* risk that some hardware condition could
* somehow bump the head pointer unpredictably
* and cause us to forward the wrong OA buffer
* data to userspace.
*/
u32 head;
} oa_buffer;
u32 gen7_latched_oastatus1;
drm/i915/perf: Add OA unit support for Gen 8+ Enables access to OA unit metrics for BDW, CHV, SKL and BXT which all share (more-or-less) the same OA unit design. Of particular note in comparison to Haswell: some OA unit HW config state has become per-context state and as a consequence it is somewhat more complicated to manage synchronous state changes from the cpu while there's no guarantee of what context (if any) is currently actively running on the gpu. The periodic sampling frequency which can be particularly useful for system-wide analysis (as opposed to command stream synchronised MI_REPORT_PERF_COUNT commands) is perhaps the most surprising state to have become per-context save and restored (while the OABUFFER destination is still a shared, system-wide resource). This support for gen8+ takes care to consider a number of timing challenges involved in synchronously updating per-context state primarily by programming all config state from the cpu and updating all current and saved contexts synchronously while the OA unit is still disabled. The driver intentionally avoids depending on command streamer programming to update OA state considering the lack of synchronization between the automatic loading of OACTXCONTROL state (that includes the periodic sampling state and enable state) on context restore and the parsing of any general purpose BB the driver can control. I.e. this implementation is careful to avoid the possibility of a context restore temporarily enabling any out-of-date periodic sampling state. In addition to the risk of transiently-out-of-date state being loaded automatically; there are also internal HW latencies involved in the loading of MUX configurations which would be difficult to account for from the command streamer (and we only want to enable the unit when once the MUX configuration is complete). Since the Gen8+ OA unit design no longer supports clock gating the unit off for a single given context (which effectively stopped any progress of counters while any other context was running) and instead supports tagging OA reports with a context ID for filtering on the CPU, it means we can no longer hide the system-wide progress of counters from a non-privileged application only interested in metrics for its own context. Although we could theoretically try and subtract the progress of other contexts before forwarding reports via read() we aren't in a position to filter reports captured via MI_REPORT_PERF_COUNT commands. As a result, for Gen8+, we always require the dev.i915.perf_stream_paranoid to be unset for any access to OA metrics if not root. v5: Drain submitted requests when enabling metric set to ensure no lite-restore erases the context image we just updated (Lionel) v6: In addition to drain, switch to kernel context & update all context in place (Chris) v7: Add missing mutex_unlock() if switching to kernel context fails (Matthew) v8: Simplify OA period/flex-eu-counters programming by using the batchbuffer instead of modifying ctx-image (Lionel) v9: Back to updating the context image (due to erroneous testing, batchbuffer programming the OA unit doesn't actually work) (Lionel) Pin context before updating context image (Chris) Drop MMIO programming now that we switch to a kernel context with right values in initial context image (Chris) v10: Just pin_map the contexts we want to modify or let the configuration happen on first use (Chris) v11: Update kernel context OA config through the batchbuffer rather than on the fly ctx-image update (Lionel) v12: Rework OA context registers update again by swithing away from user contexts and reconfiguring the kernel context through the batchbuffer and updating all the other contexts' context image. Also take care to lock slice/subslice configuration when OA is on. (Lionel) v13: Request rpcs updates on all engine when updating the OA config (Lionel) v14: Drop any kind of rpcs management now that we monitor sseu configuration changes in a later patch (Lionel) Remove usleep after programming the NOA configs on Gen8+, this doesn't seem to be needed (Lionel) v15: Respect coding style for block comments (Chris) v16: Add missing i915_add_request() in case we fail to emit OA configuration (Matthew) Signed-off-by: Robert Bragg <robert@sixbynine.org> Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com> Reviewed-by: Matthew Auld <matthew.auld@intel.com> \o/ Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
2017-06-13 05:23:03 -06:00
u32 ctx_oactxctrl_offset;
u32 ctx_flexeu0_offset;
/**
* The RPT_ID/reason field for Gen8+ includes a bit
* to determine if the CTX ID in the report is valid
* but the specific bit differs between Gen 8 and 9
*/
u32 gen8_valid_ctx_bit;
struct i915_oa_ops ops;
const struct i915_oa_format *oa_formats;
} oa;
drm/i915: Add i915 perf infrastructure Adds base i915 perf infrastructure for Gen performance metrics. This adds a DRM_IOCTL_I915_PERF_OPEN ioctl that takes an array of uint64 properties to configure a stream of metrics and returns a new fd usable with standard VFS system calls including read() to read typed and sized records; ioctl() to enable or disable capture and poll() to wait for data. A stream is opened something like: uint64_t properties[] = { /* Single context sampling */ DRM_I915_PERF_PROP_CTX_HANDLE, ctx_handle, /* Include OA reports in samples */ DRM_I915_PERF_PROP_SAMPLE_OA, true, /* OA unit configuration */ DRM_I915_PERF_PROP_OA_METRICS_SET, metrics_set_id, DRM_I915_PERF_PROP_OA_FORMAT, report_format, DRM_I915_PERF_PROP_OA_EXPONENT, period_exponent, }; struct drm_i915_perf_open_param parm = { .flags = I915_PERF_FLAG_FD_CLOEXEC | I915_PERF_FLAG_FD_NONBLOCK | I915_PERF_FLAG_DISABLED, .properties_ptr = (uint64_t)properties, .num_properties = sizeof(properties) / 16, }; int fd = drmIoctl(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param); Records read all start with a common { type, size } header with DRM_I915_PERF_RECORD_SAMPLE being of most interest. Sample records contain an extensible number of fields and it's the DRM_I915_PERF_PROP_SAMPLE_xyz properties given when opening that determine what's included in every sample. No specific streams are supported yet so any attempt to open a stream will return an error. v2: use i915_gem_context_get() - Chris Wilson v3: update read() interface to avoid passing state struct - Chris Wilson fix some rebase fallout, with i915-perf init/deinit v4: s/DRM_IORW/DRM_IOW/ - Emil Velikov Signed-off-by: Robert Bragg <robert@sixbynine.org> Reviewed-by: Matthew Auld <matthew.auld@intel.com> Reviewed-by: Sourab Gupta <sourab.gupta@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/20161107194957.3385-2-robert@sixbynine.org
2016-11-07 12:49:47 -07:00
} perf;
/* Abstract the submission mechanism (legacy ringbuffer or execlists) away */
struct {
drm/i915: Update reset path to fix incomplete requests Update reset path in preparation for engine reset which requires identification of incomplete requests and associated context and fixing their state so that engine can resume correctly after reset. The request that caused the hang will be skipped and head is reset to the start of breadcrumb. This allows us to resume from where we left-off. Since this request didn't complete normally we also need to cleanup elsp queue manually. This is vital if we employ nonblocking request submission where we may have a web of dependencies upon the hung request and so advancing the seqno manually is no longer trivial. ABI: gem_reset_stats / DRM_IOCTL_I915_GET_RESET_STATS We change the way we count pending batches. Only the active context involved in the reset is marked as either innocent or guilty, and not mark the entire world as pending. By inspection this only affects igt/gem_reset_stats (which assumes implementation details) and not piglit. ARB_robustness gives this guide on how we expect the user of this interface to behave: * Provide a mechanism for an OpenGL application to learn about graphics resets that affect the context. When a graphics reset occurs, the OpenGL context becomes unusable and the application must create a new context to continue operation. Detecting a graphics reset happens through an inexpensive query. And with regards to the actual meaning of the reset values: Certain events can result in a reset of the GL context. Such a reset causes all context state to be lost. Recovery from such events requires recreation of all objects in the affected context. The current status of the graphics reset state is returned by enum GetGraphicsResetStatusARB(); The symbolic constant returned indicates if the GL context has been in a reset state at any point since the last call to GetGraphicsResetStatusARB. NO_ERROR indicates that the GL context has not been in a reset state since the last call. GUILTY_CONTEXT_RESET_ARB indicates that a reset has been detected that is attributable to the current GL context. INNOCENT_CONTEXT_RESET_ARB indicates a reset has been detected that is not attributable to the current GL context. UNKNOWN_CONTEXT_RESET_ARB indicates a detected graphics reset whose cause is unknown. The language here is explicit in that we must mark up the guilty batch, but is loose enough for us to relax the innocent (i.e. pending) accounting as only the active batches are involved with the reset. In the future, we are looking towards single engine resetting (with minimal locking), where it seems inappropriate to mark the entire world as innocent since the reset occurred on a different engine. Reducing the information available means we only have to encounter the pain once, and also reduces the information leaking from one context to another. v2: Legacy ringbuffer submission required a reset following hibernation, or else we restore stale values to the RING_HEAD and walked over stolen garbage. v3: GuC requires replaying the requests after a reset. v4: Restore engine IRQ after reset (so waiters will be woken!) Rearm hangcheck if resetting with a waiter. Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Cc: Mika Kuoppala <mika.kuoppala@intel.com> Cc: Arun Siluvery <arun.siluvery@linux.intel.com> Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/20160909131201.16673-13-chris@chris-wilson.co.uk
2016-09-09 07:11:53 -06:00
void (*resume)(struct drm_i915_private *);
void (*cleanup_engine)(struct intel_engine_cs *engine);
struct list_head timelines;
struct i915_gem_timeline global_timeline;
u32 active_requests;
/**
* Is the GPU currently considered idle, or busy executing
* userspace requests? Whilst idle, we allow runtime power
* management to power down the hardware and display clocks.
* In order to reduce the effect on performance, there
* is a slight delay before we do so.
*/
bool awake;
/**
* We leave the user IRQ off as much as possible,
* but this means that requests will finish and never
* be retired once the system goes idle. Set a timer to
* fire periodically while the ring is running. When it
* fires, go retire requests.
*/
struct delayed_work retire_work;
/**
* When we detect an idle GPU, we want to turn on
* powersaving features. So once we see that there
* are no more requests outstanding and no more
* arrive within a small period of time, we fire
* off the idle_work.
*/
struct delayed_work idle_work;
ktime_t last_init_time;
} gt;
/* perform PHY state sanity checks? */
bool chv_phy_assert[2];
bool ipc_enabled;
drm/i915/dp: DP audio API changes for MST DP MST provides the capability to send multiple video and audio streams through a single port. This requires the API's between i915 and audio drivers to distinguish between multiple audio capable displays that can be connected to a port. Currently only the port identity is shared in the APIs. This patch adds support for MST with an additional parameter 'int pipe'. The existing parameter 'port' does not change it's meaning. pipe = MST : display pipe that the stream originates from Non-MST : -1 Affected APIs: struct i915_audio_component_ops - int (*sync_audio_rate)(struct device *, int port, int rate); + int (*sync_audio_rate)(struct device *, int port, int pipe, + int rate); - int (*get_eld)(struct device *, int port, bool *enabled, - unsigned char *buf, int max_bytes); + int (*get_eld)(struct device *, int port, int pipe, + bool *enabled, unsigned char *buf, int max_bytes); struct i915_audio_component_audio_ops - void (*pin_eld_notify)(void *audio_ptr, int port); + void (*pin_eld_notify)(void *audio_ptr, int port, int pipe); This patch makes dummy changes in the audio drivers (thanks Libin) for build to succeed. The audio side drivers will send the right 'pipe' values for MST in patches that will follow. v2: Renamed the new API parameter from 'dev_id' to 'pipe'. (Jim, Ville) Included Asoc driver API compatibility changes from Jeeja. Added WARN_ON() for invalid pipe in get_saved_encoder(). (Takashi) Added comment for av_enc_map[] definition. (Takashi) v3: Fixed logic error introduced while renaming 'dev_id' as 'pipe' (Ville) Renamed get_saved_encoder() to get_saved_enc() to reduce line length v4: Rebased. Parameter check for pipe < -1 values in get_saved_enc() (Ville) Switched to for_each_pipe() in get_saved_enc() (Ville) Renamed 'pipe' to 'dev_id' in audio side code (Takashi) v5: Included a comment for the dev_id arg. (Libin) Signed-off-by: Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com> Reviewed-by: Takashi Iwai <tiwai@suse.de> Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1474488168-2343-1-git-send-email-dhinakaran.pandiyan@intel.com
2016-09-21 14:02:48 -06:00
/* Used to save the pipe-to-encoder mapping for audio */
struct intel_encoder *av_enc_map[I915_MAX_PIPES];
/* necessary resource sharing with HDMI LPE audio driver. */
struct {
struct platform_device *platdev;
int irq;
} lpe_audio;
/*
* NOTE: This is the dri1/ums dungeon, don't add stuff here. Your patch
* will be rejected. Instead look for a better place.
*/
};
static inline struct drm_i915_private *to_i915(const struct drm_device *dev)
{
return container_of(dev, struct drm_i915_private, drm);
}
static inline struct drm_i915_private *kdev_to_i915(struct device *kdev)
{
return to_i915(dev_get_drvdata(kdev));
}
static inline struct drm_i915_private *guc_to_i915(struct intel_guc *guc)
{
return container_of(guc, struct drm_i915_private, guc);
}
static inline struct drm_i915_private *huc_to_i915(struct intel_huc *huc)
{
return container_of(huc, struct drm_i915_private, huc);
}
/* Simple iterator over all initialised engines */
drm/i915: Allocate intel_engine_cs structure only for the enabled engines With the possibility of addition of many more number of rings in future, the drm_i915_private structure could bloat as an array, of type intel_engine_cs, is embedded inside it. struct intel_engine_cs engine[I915_NUM_ENGINES]; Though this is still fine as generally there is only a single instance of drm_i915_private structure used, but not all of the possible rings would be enabled or active on most of the platforms. Some memory can be saved by allocating intel_engine_cs structure only for the enabled/active engines. Currently the engine/ring ID is kept static and dev_priv->engine[] is simply indexed using the enums defined in intel_engine_id. To save memory and continue using the static engine/ring IDs, 'engine' is defined as an array of pointers. struct intel_engine_cs *engine[I915_NUM_ENGINES]; dev_priv->engine[engine_ID] will be NULL for disabled engine instances. There is a text size reduction of 928 bytes, from 1028200 to 1027272, for i915.o file (but for i915.ko file text size remain same as 1193131 bytes). v2: - Remove the engine iterator field added in drm_i915_private structure, instead pass a local iterator variable to the for_each_engine** macros. (Chris) - Do away with intel_engine_initialized() and instead directly use the NULL pointer check on engine pointer. (Chris) v3: - Remove for_each_engine_id() macro, as the updated macro for_each_engine() can be used in place of it. (Chris) - Protect the access to Render engine Fault register with a NULL check, as engine specific init is done later in Driver load sequence. v4: - Use !!dev_priv->engine[VCS] style for the engine check in getparam. (Chris) - Kill the superfluous init_engine_lists(). v5: - Cleanup the intel_engines_init() & intel_engines_setup(), with respect to allocation of intel_engine_cs structure. (Chris) v6: - Rebase. v7: - Optimize the for_each_engine_masked() macro. (Chris) - Change the type of 'iter' local variable to enum intel_engine_id. (Chris) - Rebase. v8: Rebase. v9: Rebase. v10: - For index calculation use engine ID instead of pointer based arithmetic in intel_engine_sync_index() as engine pointers are not contiguous now (Chris) - For appropriateness, rename local enum variable 'iter' to 'id'. (Joonas) - Use for_each_engine macro for cleanup in intel_engines_init() and remove check for NULL engine pointer in cleanup() routines. (Joonas) v11: Rebase. Cc: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Akash Goel <akash.goel@intel.com> Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1476378888-7372-1-git-send-email-akash.goel@intel.com
2016-10-13 11:14:48 -06:00
#define for_each_engine(engine__, dev_priv__, id__) \
for ((id__) = 0; \
(id__) < I915_NUM_ENGINES; \
(id__)++) \
for_each_if ((engine__) = (dev_priv__)->engine[(id__)])
/* Iterator over subset of engines selected by mask */
#define for_each_engine_masked(engine__, dev_priv__, mask__, tmp__) \
for (tmp__ = mask__ & INTEL_INFO(dev_priv__)->ring_mask; \
drm/i915: Allocate intel_engine_cs structure only for the enabled engines With the possibility of addition of many more number of rings in future, the drm_i915_private structure could bloat as an array, of type intel_engine_cs, is embedded inside it. struct intel_engine_cs engine[I915_NUM_ENGINES]; Though this is still fine as generally there is only a single instance of drm_i915_private structure used, but not all of the possible rings would be enabled or active on most of the platforms. Some memory can be saved by allocating intel_engine_cs structure only for the enabled/active engines. Currently the engine/ring ID is kept static and dev_priv->engine[] is simply indexed using the enums defined in intel_engine_id. To save memory and continue using the static engine/ring IDs, 'engine' is defined as an array of pointers. struct intel_engine_cs *engine[I915_NUM_ENGINES]; dev_priv->engine[engine_ID] will be NULL for disabled engine instances. There is a text size reduction of 928 bytes, from 1028200 to 1027272, for i915.o file (but for i915.ko file text size remain same as 1193131 bytes). v2: - Remove the engine iterator field added in drm_i915_private structure, instead pass a local iterator variable to the for_each_engine** macros. (Chris) - Do away with intel_engine_initialized() and instead directly use the NULL pointer check on engine pointer. (Chris) v3: - Remove for_each_engine_id() macro, as the updated macro for_each_engine() can be used in place of it. (Chris) - Protect the access to Render engine Fault register with a NULL check, as engine specific init is done later in Driver load sequence. v4: - Use !!dev_priv->engine[VCS] style for the engine check in getparam. (Chris) - Kill the superfluous init_engine_lists(). v5: - Cleanup the intel_engines_init() & intel_engines_setup(), with respect to allocation of intel_engine_cs structure. (Chris) v6: - Rebase. v7: - Optimize the for_each_engine_masked() macro. (Chris) - Change the type of 'iter' local variable to enum intel_engine_id. (Chris) - Rebase. v8: Rebase. v9: Rebase. v10: - For index calculation use engine ID instead of pointer based arithmetic in intel_engine_sync_index() as engine pointers are not contiguous now (Chris) - For appropriateness, rename local enum variable 'iter' to 'id'. (Joonas) - Use for_each_engine macro for cleanup in intel_engines_init() and remove check for NULL engine pointer in cleanup() routines. (Joonas) v11: Rebase. Cc: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Akash Goel <akash.goel@intel.com> Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1476378888-7372-1-git-send-email-akash.goel@intel.com
2016-10-13 11:14:48 -06:00
tmp__ ? (engine__ = (dev_priv__)->engine[__mask_next_bit(tmp__)]), 1 : 0; )
enum hdmi_force_audio {
HDMI_AUDIO_OFF_DVI = -2, /* no aux data for HDMI-DVI converter */
HDMI_AUDIO_OFF, /* force turn off HDMI audio */
HDMI_AUDIO_AUTO, /* trust EDID */
HDMI_AUDIO_ON, /* force turn on HDMI audio */
};
#define I915_GTT_OFFSET_NONE ((u32)-1)
drm/i915: Introduce accurate frontbuffer tracking So from just a quick look we seem to have enough information to accurately figure out whether a given gem bo is used as a frontbuffer and where exactly: We have obj->pin_count as a first check with no false negatives and only negligible false positives. And then we can just walk the modeset objects and figure out where exactly a buffer is used as scanout. Except that we can't due to locking order: If we already hold dev->struct_mutex we can't acquire any modeset locks, so could potential chase freed pointers and other evil stuff. So we need something else. For that introduce a new set of bits obj->frontbuffer_bits to track where a buffer object is used. That we can then chase without grabbing any modeset locks. Of course the consumers of this (DRRS, PSR, FBC, ...) still need to be able to do their magic both when called from modeset and from gem code. But that can be easily achieved by adding locks for these specific subsystems which always nest within either kms or gem locking. This patch just adds the relevant update code to all places. Note that if we ever support multi-planar scanout targets then we need one frontbuffer tracking bit per attachment point that we expose to userspace. v2: - Fix more oopsen. Oops. - WARN if we leak obj->frontbuffer_bits when freeing a gem buffer. Fix the bugs this brought to light. - s/update_frontbuffer_bits/update_fb_bits/. More consistent with the fb tracking functions (fb for gem object, frontbuffer for raw bits). And the function name was way too long. v3: Size obj->frontbuffer_bits correctly so that all pipes fit in. v4: Don't update fb bits in set_base on failure. Noticed by Chris. v5: s/i915_gem_update_fb_bits/i915_gem_track_fb/ Also remove a few local enum pipe variables which are now no longer needed to make the function arguments no drop over the 80 char limit. Cc: Rodrigo Vivi <rodrigo.vivi@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2014-06-18 15:28:09 -06:00
/*
* Frontbuffer tracking bits. Set in obj->frontbuffer_bits while a gem bo is
* considered to be the frontbuffer for the given plane interface-wise. This
drm/i915: Introduce accurate frontbuffer tracking So from just a quick look we seem to have enough information to accurately figure out whether a given gem bo is used as a frontbuffer and where exactly: We have obj->pin_count as a first check with no false negatives and only negligible false positives. And then we can just walk the modeset objects and figure out where exactly a buffer is used as scanout. Except that we can't due to locking order: If we already hold dev->struct_mutex we can't acquire any modeset locks, so could potential chase freed pointers and other evil stuff. So we need something else. For that introduce a new set of bits obj->frontbuffer_bits to track where a buffer object is used. That we can then chase without grabbing any modeset locks. Of course the consumers of this (DRRS, PSR, FBC, ...) still need to be able to do their magic both when called from modeset and from gem code. But that can be easily achieved by adding locks for these specific subsystems which always nest within either kms or gem locking. This patch just adds the relevant update code to all places. Note that if we ever support multi-planar scanout targets then we need one frontbuffer tracking bit per attachment point that we expose to userspace. v2: - Fix more oopsen. Oops. - WARN if we leak obj->frontbuffer_bits when freeing a gem buffer. Fix the bugs this brought to light. - s/update_frontbuffer_bits/update_fb_bits/. More consistent with the fb tracking functions (fb for gem object, frontbuffer for raw bits). And the function name was way too long. v3: Size obj->frontbuffer_bits correctly so that all pipes fit in. v4: Don't update fb bits in set_base on failure. Noticed by Chris. v5: s/i915_gem_update_fb_bits/i915_gem_track_fb/ Also remove a few local enum pipe variables which are now no longer needed to make the function arguments no drop over the 80 char limit. Cc: Rodrigo Vivi <rodrigo.vivi@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2014-06-18 15:28:09 -06:00
* doesn't mean that the hw necessarily already scans it out, but that any
* rendering (by the cpu or gpu) will land in the frontbuffer eventually.
*
* We have one bit per pipe and per scanout plane type.
*/
#define INTEL_MAX_SPRITE_BITS_PER_PIPE 5
#define INTEL_FRONTBUFFER_BITS_PER_PIPE 8
drm/i915: Introduce accurate frontbuffer tracking So from just a quick look we seem to have enough information to accurately figure out whether a given gem bo is used as a frontbuffer and where exactly: We have obj->pin_count as a first check with no false negatives and only negligible false positives. And then we can just walk the modeset objects and figure out where exactly a buffer is used as scanout. Except that we can't due to locking order: If we already hold dev->struct_mutex we can't acquire any modeset locks, so could potential chase freed pointers and other evil stuff. So we need something else. For that introduce a new set of bits obj->frontbuffer_bits to track where a buffer object is used. That we can then chase without grabbing any modeset locks. Of course the consumers of this (DRRS, PSR, FBC, ...) still need to be able to do their magic both when called from modeset and from gem code. But that can be easily achieved by adding locks for these specific subsystems which always nest within either kms or gem locking. This patch just adds the relevant update code to all places. Note that if we ever support multi-planar scanout targets then we need one frontbuffer tracking bit per attachment point that we expose to userspace. v2: - Fix more oopsen. Oops. - WARN if we leak obj->frontbuffer_bits when freeing a gem buffer. Fix the bugs this brought to light. - s/update_frontbuffer_bits/update_fb_bits/. More consistent with the fb tracking functions (fb for gem object, frontbuffer for raw bits). And the function name was way too long. v3: Size obj->frontbuffer_bits correctly so that all pipes fit in. v4: Don't update fb bits in set_base on failure. Noticed by Chris. v5: s/i915_gem_update_fb_bits/i915_gem_track_fb/ Also remove a few local enum pipe variables which are now no longer needed to make the function arguments no drop over the 80 char limit. Cc: Rodrigo Vivi <rodrigo.vivi@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2014-06-18 15:28:09 -06:00
#define INTEL_FRONTBUFFER_PRIMARY(pipe) \
(1 << (INTEL_FRONTBUFFER_BITS_PER_PIPE * (pipe)))
#define INTEL_FRONTBUFFER_CURSOR(pipe) \
(1 << (1 + (INTEL_FRONTBUFFER_BITS_PER_PIPE * (pipe))))
#define INTEL_FRONTBUFFER_SPRITE(pipe, plane) \
(1 << (2 + plane + (INTEL_FRONTBUFFER_BITS_PER_PIPE * (pipe))))
drm/i915: Introduce accurate frontbuffer tracking So from just a quick look we seem to have enough information to accurately figure out whether a given gem bo is used as a frontbuffer and where exactly: We have obj->pin_count as a first check with no false negatives and only negligible false positives. And then we can just walk the modeset objects and figure out where exactly a buffer is used as scanout. Except that we can't due to locking order: If we already hold dev->struct_mutex we can't acquire any modeset locks, so could potential chase freed pointers and other evil stuff. So we need something else. For that introduce a new set of bits obj->frontbuffer_bits to track where a buffer object is used. That we can then chase without grabbing any modeset locks. Of course the consumers of this (DRRS, PSR, FBC, ...) still need to be able to do their magic both when called from modeset and from gem code. But that can be easily achieved by adding locks for these specific subsystems which always nest within either kms or gem locking. This patch just adds the relevant update code to all places. Note that if we ever support multi-planar scanout targets then we need one frontbuffer tracking bit per attachment point that we expose to userspace. v2: - Fix more oopsen. Oops. - WARN if we leak obj->frontbuffer_bits when freeing a gem buffer. Fix the bugs this brought to light. - s/update_frontbuffer_bits/update_fb_bits/. More consistent with the fb tracking functions (fb for gem object, frontbuffer for raw bits). And the function name was way too long. v3: Size obj->frontbuffer_bits correctly so that all pipes fit in. v4: Don't update fb bits in set_base on failure. Noticed by Chris. v5: s/i915_gem_update_fb_bits/i915_gem_track_fb/ Also remove a few local enum pipe variables which are now no longer needed to make the function arguments no drop over the 80 char limit. Cc: Rodrigo Vivi <rodrigo.vivi@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2014-06-18 15:28:09 -06:00
#define INTEL_FRONTBUFFER_OVERLAY(pipe) \
(1 << (2 + INTEL_MAX_SPRITE_BITS_PER_PIPE + (INTEL_FRONTBUFFER_BITS_PER_PIPE * (pipe))))
#define INTEL_FRONTBUFFER_ALL_MASK(pipe) \
(0xff << (INTEL_FRONTBUFFER_BITS_PER_PIPE * (pipe)))
drm/i915: Introduce accurate frontbuffer tracking So from just a quick look we seem to have enough information to accurately figure out whether a given gem bo is used as a frontbuffer and where exactly: We have obj->pin_count as a first check with no false negatives and only negligible false positives. And then we can just walk the modeset objects and figure out where exactly a buffer is used as scanout. Except that we can't due to locking order: If we already hold dev->struct_mutex we can't acquire any modeset locks, so could potential chase freed pointers and other evil stuff. So we need something else. For that introduce a new set of bits obj->frontbuffer_bits to track where a buffer object is used. That we can then chase without grabbing any modeset locks. Of course the consumers of this (DRRS, PSR, FBC, ...) still need to be able to do their magic both when called from modeset and from gem code. But that can be easily achieved by adding locks for these specific subsystems which always nest within either kms or gem locking. This patch just adds the relevant update code to all places. Note that if we ever support multi-planar scanout targets then we need one frontbuffer tracking bit per attachment point that we expose to userspace. v2: - Fix more oopsen. Oops. - WARN if we leak obj->frontbuffer_bits when freeing a gem buffer. Fix the bugs this brought to light. - s/update_frontbuffer_bits/update_fb_bits/. More consistent with the fb tracking functions (fb for gem object, frontbuffer for raw bits). And the function name was way too long. v3: Size obj->frontbuffer_bits correctly so that all pipes fit in. v4: Don't update fb bits in set_base on failure. Noticed by Chris. v5: s/i915_gem_update_fb_bits/i915_gem_track_fb/ Also remove a few local enum pipe variables which are now no longer needed to make the function arguments no drop over the 80 char limit. Cc: Rodrigo Vivi <rodrigo.vivi@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2014-06-18 15:28:09 -06:00
/*
* Optimised SGL iterator for GEM objects
*/
static __always_inline struct sgt_iter {
struct scatterlist *sgp;
union {
unsigned long pfn;
dma_addr_t dma;
};
unsigned int curr;
unsigned int max;
} __sgt_iter(struct scatterlist *sgl, bool dma) {
struct sgt_iter s = { .sgp = sgl };
if (s.sgp) {
s.max = s.curr = s.sgp->offset;
s.max += s.sgp->length;
if (dma)
s.dma = sg_dma_address(s.sgp);
else
s.pfn = page_to_pfn(sg_page(s.sgp));
}
return s;
}
static inline struct scatterlist *____sg_next(struct scatterlist *sg)
{
++sg;
if (unlikely(sg_is_chain(sg)))
sg = sg_chain_ptr(sg);
return sg;
}
/**
* __sg_next - return the next scatterlist entry in a list
* @sg: The current sg entry
*
* Description:
* If the entry is the last, return NULL; otherwise, step to the next
* element in the array (@sg@+1). If that's a chain pointer, follow it;
* otherwise just return the pointer to the current element.
**/
static inline struct scatterlist *__sg_next(struct scatterlist *sg)
{
#ifdef CONFIG_DEBUG_SG
BUG_ON(sg->sg_magic != SG_MAGIC);
#endif
return sg_is_last(sg) ? NULL : ____sg_next(sg);
}
/**
* for_each_sgt_dma - iterate over the DMA addresses of the given sg_table
* @__dmap: DMA address (output)
* @__iter: 'struct sgt_iter' (iterator state, internal)
* @__sgt: sg_table to iterate over (input)
*/
#define for_each_sgt_dma(__dmap, __iter, __sgt) \
for ((__iter) = __sgt_iter((__sgt)->sgl, true); \
((__dmap) = (__iter).dma + (__iter).curr); \
(((__iter).curr += PAGE_SIZE) < (__iter).max) || \
((__iter) = __sgt_iter(__sg_next((__iter).sgp), true), 0))
/**
* for_each_sgt_page - iterate over the pages of the given sg_table
* @__pp: page pointer (output)
* @__iter: 'struct sgt_iter' (iterator state, internal)
* @__sgt: sg_table to iterate over (input)
*/
#define for_each_sgt_page(__pp, __iter, __sgt) \
for ((__iter) = __sgt_iter((__sgt)->sgl, false); \
((__pp) = (__iter).pfn == 0 ? NULL : \
pfn_to_page((__iter).pfn + ((__iter).curr >> PAGE_SHIFT))); \
(((__iter).curr += PAGE_SIZE) < (__iter).max) || \
((__iter) = __sgt_iter(__sg_next((__iter).sgp), false), 0))
drm/i915: Introduce accurate frontbuffer tracking So from just a quick look we seem to have enough information to accurately figure out whether a given gem bo is used as a frontbuffer and where exactly: We have obj->pin_count as a first check with no false negatives and only negligible false positives. And then we can just walk the modeset objects and figure out where exactly a buffer is used as scanout. Except that we can't due to locking order: If we already hold dev->struct_mutex we can't acquire any modeset locks, so could potential chase freed pointers and other evil stuff. So we need something else. For that introduce a new set of bits obj->frontbuffer_bits to track where a buffer object is used. That we can then chase without grabbing any modeset locks. Of course the consumers of this (DRRS, PSR, FBC, ...) still need to be able to do their magic both when called from modeset and from gem code. But that can be easily achieved by adding locks for these specific subsystems which always nest within either kms or gem locking. This patch just adds the relevant update code to all places. Note that if we ever support multi-planar scanout targets then we need one frontbuffer tracking bit per attachment point that we expose to userspace. v2: - Fix more oopsen. Oops. - WARN if we leak obj->frontbuffer_bits when freeing a gem buffer. Fix the bugs this brought to light. - s/update_frontbuffer_bits/update_fb_bits/. More consistent with the fb tracking functions (fb for gem object, frontbuffer for raw bits). And the function name was way too long. v3: Size obj->frontbuffer_bits correctly so that all pipes fit in. v4: Don't update fb bits in set_base on failure. Noticed by Chris. v5: s/i915_gem_update_fb_bits/i915_gem_track_fb/ Also remove a few local enum pipe variables which are now no longer needed to make the function arguments no drop over the 80 char limit. Cc: Rodrigo Vivi <rodrigo.vivi@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2014-06-18 15:28:09 -06:00
static inline const struct intel_device_info *
intel_info(const struct drm_i915_private *dev_priv)
{
return &dev_priv->info;
}
#define INTEL_INFO(dev_priv) intel_info((dev_priv))
#define INTEL_GEN(dev_priv) ((dev_priv)->info.gen)
#define INTEL_DEVID(dev_priv) ((dev_priv)->info.device_id)
#define REVID_FOREVER 0xff
#define INTEL_REVID(dev_priv) ((dev_priv)->drm.pdev->revision)
#define GEN_FOREVER (0)
/*
* Returns true if Gen is in inclusive range [Start, End].
*
* Use GEN_FOREVER for unbound start and or end.
*/
#define IS_GEN(dev_priv, s, e) ({ \
unsigned int __s = (s), __e = (e); \
BUILD_BUG_ON(!__builtin_constant_p(s)); \
BUILD_BUG_ON(!__builtin_constant_p(e)); \
if ((__s) != GEN_FOREVER) \
__s = (s) - 1; \
if ((__e) == GEN_FOREVER) \
__e = BITS_PER_LONG - 1; \
else \
__e = (e) - 1; \
!!((dev_priv)->info.gen_mask & GENMASK((__e), (__s))); \
})
/*
* Return true if revision is in range [since,until] inclusive.
*
* Use 0 for open-ended since, and REVID_FOREVER for open-ended until.
*/
#define IS_REVID(p, since, until) \
(INTEL_REVID(p) >= (since) && INTEL_REVID(p) <= (until))
#define IS_I830(dev_priv) ((dev_priv)->info.platform == INTEL_I830)
#define IS_I845G(dev_priv) ((dev_priv)->info.platform == INTEL_I845G)
#define IS_I85X(dev_priv) ((dev_priv)->info.platform == INTEL_I85X)
#define IS_I865G(dev_priv) ((dev_priv)->info.platform == INTEL_I865G)
#define IS_I915G(dev_priv) ((dev_priv)->info.platform == INTEL_I915G)
#define IS_I915GM(dev_priv) ((dev_priv)->info.platform == INTEL_I915GM)
#define IS_I945G(dev_priv) ((dev_priv)->info.platform == INTEL_I945G)
#define IS_I945GM(dev_priv) ((dev_priv)->info.platform == INTEL_I945GM)
#define IS_I965G(dev_priv) ((dev_priv)->info.platform == INTEL_I965G)
#define IS_I965GM(dev_priv) ((dev_priv)->info.platform == INTEL_I965GM)
#define IS_G45(dev_priv) ((dev_priv)->info.platform == INTEL_G45)
#define IS_GM45(dev_priv) ((dev_priv)->info.platform == INTEL_GM45)
#define IS_G4X(dev_priv) (IS_G45(dev_priv) || IS_GM45(dev_priv))
#define IS_PINEVIEW_G(dev_priv) (INTEL_DEVID(dev_priv) == 0xa001)
#define IS_PINEVIEW_M(dev_priv) (INTEL_DEVID(dev_priv) == 0xa011)
#define IS_PINEVIEW(dev_priv) ((dev_priv)->info.platform == INTEL_PINEVIEW)
#define IS_G33(dev_priv) ((dev_priv)->info.platform == INTEL_G33)
#define IS_IRONLAKE_M(dev_priv) (INTEL_DEVID(dev_priv) == 0x0046)
#define IS_IVYBRIDGE(dev_priv) ((dev_priv)->info.platform == INTEL_IVYBRIDGE)
#define IS_IVB_GT1(dev_priv) (INTEL_DEVID(dev_priv) == 0x0156 || \
INTEL_DEVID(dev_priv) == 0x0152 || \
INTEL_DEVID(dev_priv) == 0x015a)
#define IS_VALLEYVIEW(dev_priv) ((dev_priv)->info.platform == INTEL_VALLEYVIEW)
#define IS_CHERRYVIEW(dev_priv) ((dev_priv)->info.platform == INTEL_CHERRYVIEW)
#define IS_HASWELL(dev_priv) ((dev_priv)->info.platform == INTEL_HASWELL)
#define IS_BROADWELL(dev_priv) ((dev_priv)->info.platform == INTEL_BROADWELL)
#define IS_SKYLAKE(dev_priv) ((dev_priv)->info.platform == INTEL_SKYLAKE)
#define IS_BROXTON(dev_priv) ((dev_priv)->info.platform == INTEL_BROXTON)
#define IS_KABYLAKE(dev_priv) ((dev_priv)->info.platform == INTEL_KABYLAKE)
#define IS_GEMINILAKE(dev_priv) ((dev_priv)->info.platform == INTEL_GEMINILAKE)
#define IS_COFFEELAKE(dev_priv) ((dev_priv)->info.platform == INTEL_COFFEELAKE)
#define IS_CANNONLAKE(dev_priv) ((dev_priv)->info.platform == INTEL_CANNONLAKE)
#define IS_MOBILE(dev_priv) ((dev_priv)->info.is_mobile)
#define IS_HSW_EARLY_SDV(dev_priv) (IS_HASWELL(dev_priv) && \
(INTEL_DEVID(dev_priv) & 0xFF00) == 0x0C00)
#define IS_BDW_ULT(dev_priv) (IS_BROADWELL(dev_priv) && \
((INTEL_DEVID(dev_priv) & 0xf) == 0x6 || \
(INTEL_DEVID(dev_priv) & 0xf) == 0xb || \
(INTEL_DEVID(dev_priv) & 0xf) == 0xe))
/* ULX machines are also considered ULT. */
#define IS_BDW_ULX(dev_priv) (IS_BROADWELL(dev_priv) && \
(INTEL_DEVID(dev_priv) & 0xf) == 0xe)
#define IS_BDW_GT3(dev_priv) (IS_BROADWELL(dev_priv) && \
(INTEL_DEVID(dev_priv) & 0x00F0) == 0x0020)
#define IS_HSW_ULT(dev_priv) (IS_HASWELL(dev_priv) && \
(INTEL_DEVID(dev_priv) & 0xFF00) == 0x0A00)
#define IS_HSW_GT3(dev_priv) (IS_HASWELL(dev_priv) && \
(INTEL_DEVID(dev_priv) & 0x00F0) == 0x0020)
/* ULX machines are also considered ULT. */
#define IS_HSW_ULX(dev_priv) (INTEL_DEVID(dev_priv) == 0x0A0E || \
INTEL_DEVID(dev_priv) == 0x0A1E)
#define IS_SKL_ULT(dev_priv) (INTEL_DEVID(dev_priv) == 0x1906 || \
INTEL_DEVID(dev_priv) == 0x1913 || \
INTEL_DEVID(dev_priv) == 0x1916 || \
INTEL_DEVID(dev_priv) == 0x1921 || \
INTEL_DEVID(dev_priv) == 0x1926)
#define IS_SKL_ULX(dev_priv) (INTEL_DEVID(dev_priv) == 0x190E || \
INTEL_DEVID(dev_priv) == 0x1915 || \
INTEL_DEVID(dev_priv) == 0x191E)
#define IS_KBL_ULT(dev_priv) (INTEL_DEVID(dev_priv) == 0x5906 || \
INTEL_DEVID(dev_priv) == 0x5913 || \
INTEL_DEVID(dev_priv) == 0x5916 || \
INTEL_DEVID(dev_priv) == 0x5921 || \
INTEL_DEVID(dev_priv) == 0x5926)
#define IS_KBL_ULX(dev_priv) (INTEL_DEVID(dev_priv) == 0x590E || \
INTEL_DEVID(dev_priv) == 0x5915 || \
INTEL_DEVID(dev_priv) == 0x591E)
drm/i915/perf: Add OA unit support for Gen 8+ Enables access to OA unit metrics for BDW, CHV, SKL and BXT which all share (more-or-less) the same OA unit design. Of particular note in comparison to Haswell: some OA unit HW config state has become per-context state and as a consequence it is somewhat more complicated to manage synchronous state changes from the cpu while there's no guarantee of what context (if any) is currently actively running on the gpu. The periodic sampling frequency which can be particularly useful for system-wide analysis (as opposed to command stream synchronised MI_REPORT_PERF_COUNT commands) is perhaps the most surprising state to have become per-context save and restored (while the OABUFFER destination is still a shared, system-wide resource). This support for gen8+ takes care to consider a number of timing challenges involved in synchronously updating per-context state primarily by programming all config state from the cpu and updating all current and saved contexts synchronously while the OA unit is still disabled. The driver intentionally avoids depending on command streamer programming to update OA state considering the lack of synchronization between the automatic loading of OACTXCONTROL state (that includes the periodic sampling state and enable state) on context restore and the parsing of any general purpose BB the driver can control. I.e. this implementation is careful to avoid the possibility of a context restore temporarily enabling any out-of-date periodic sampling state. In addition to the risk of transiently-out-of-date state being loaded automatically; there are also internal HW latencies involved in the loading of MUX configurations which would be difficult to account for from the command streamer (and we only want to enable the unit when once the MUX configuration is complete). Since the Gen8+ OA unit design no longer supports clock gating the unit off for a single given context (which effectively stopped any progress of counters while any other context was running) and instead supports tagging OA reports with a context ID for filtering on the CPU, it means we can no longer hide the system-wide progress of counters from a non-privileged application only interested in metrics for its own context. Although we could theoretically try and subtract the progress of other contexts before forwarding reports via read() we aren't in a position to filter reports captured via MI_REPORT_PERF_COUNT commands. As a result, for Gen8+, we always require the dev.i915.perf_stream_paranoid to be unset for any access to OA metrics if not root. v5: Drain submitted requests when enabling metric set to ensure no lite-restore erases the context image we just updated (Lionel) v6: In addition to drain, switch to kernel context & update all context in place (Chris) v7: Add missing mutex_unlock() if switching to kernel context fails (Matthew) v8: Simplify OA period/flex-eu-counters programming by using the batchbuffer instead of modifying ctx-image (Lionel) v9: Back to updating the context image (due to erroneous testing, batchbuffer programming the OA unit doesn't actually work) (Lionel) Pin context before updating context image (Chris) Drop MMIO programming now that we switch to a kernel context with right values in initial context image (Chris) v10: Just pin_map the contexts we want to modify or let the configuration happen on first use (Chris) v11: Update kernel context OA config through the batchbuffer rather than on the fly ctx-image update (Lionel) v12: Rework OA context registers update again by swithing away from user contexts and reconfiguring the kernel context through the batchbuffer and updating all the other contexts' context image. Also take care to lock slice/subslice configuration when OA is on. (Lionel) v13: Request rpcs updates on all engine when updating the OA config (Lionel) v14: Drop any kind of rpcs management now that we monitor sseu configuration changes in a later patch (Lionel) Remove usleep after programming the NOA configs on Gen8+, this doesn't seem to be needed (Lionel) v15: Respect coding style for block comments (Chris) v16: Add missing i915_add_request() in case we fail to emit OA configuration (Matthew) Signed-off-by: Robert Bragg <robert@sixbynine.org> Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com> Reviewed-by: Matthew Auld <matthew.auld@intel.com> \o/ Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
2017-06-13 05:23:03 -06:00
#define IS_SKL_GT2(dev_priv) (IS_SKYLAKE(dev_priv) && \
(INTEL_DEVID(dev_priv) & 0x00F0) == 0x0010)
#define IS_SKL_GT3(dev_priv) (IS_SKYLAKE(dev_priv) && \
(INTEL_DEVID(dev_priv) & 0x00F0) == 0x0020)
#define IS_SKL_GT4(dev_priv) (IS_SKYLAKE(dev_priv) && \
(INTEL_DEVID(dev_priv) & 0x00F0) == 0x0030)
#define IS_KBL_GT2(dev_priv) (IS_KABYLAKE(dev_priv) && \
(INTEL_DEVID(dev_priv) & 0x00F0) == 0x0010)
#define IS_KBL_GT3(dev_priv) (IS_KABYLAKE(dev_priv) && \
(INTEL_DEVID(dev_priv) & 0x00F0) == 0x0020)
#define IS_CFL_ULT(dev_priv) (IS_COFFEELAKE(dev_priv) && \
(INTEL_DEVID(dev_priv) & 0x00F0) == 0x00A0)
#define IS_ALPHA_SUPPORT(intel_info) ((intel_info)->is_alpha_support)
#define SKL_REVID_A0 0x0
#define SKL_REVID_B0 0x1
#define SKL_REVID_C0 0x2
#define SKL_REVID_D0 0x3
#define SKL_REVID_E0 0x4
#define SKL_REVID_F0 0x5
#define SKL_REVID_G0 0x6
#define SKL_REVID_H0 0x7
#define IS_SKL_REVID(p, since, until) (IS_SKYLAKE(p) && IS_REVID(p, since, until))
#define BXT_REVID_A0 0x0
#define BXT_REVID_A1 0x1
#define BXT_REVID_B0 0x3
#define BXT_REVID_B_LAST 0x8
#define BXT_REVID_C0 0x9
#define IS_BXT_REVID(dev_priv, since, until) \
(IS_BROXTON(dev_priv) && IS_REVID(dev_priv, since, until))
#define KBL_REVID_A0 0x0
#define KBL_REVID_B0 0x1
#define KBL_REVID_C0 0x2
#define KBL_REVID_D0 0x3
#define KBL_REVID_E0 0x4
#define IS_KBL_REVID(dev_priv, since, until) \
(IS_KABYLAKE(dev_priv) && IS_REVID(dev_priv, since, until))
#define GLK_REVID_A0 0x0
#define GLK_REVID_A1 0x1
#define IS_GLK_REVID(dev_priv, since, until) \
(IS_GEMINILAKE(dev_priv) && IS_REVID(dev_priv, since, until))
#define CNL_REVID_A0 0x0
#define CNL_REVID_B0 0x1
#define IS_CNL_REVID(p, since, until) \
(IS_CANNONLAKE(p) && IS_REVID(p, since, until))
/*
* The genX designation typically refers to the render engine, so render
* capability related checks should use IS_GEN, while display and other checks
* have their own (e.g. HAS_PCH_SPLIT for ILK+ display, IS_foo for particular
* chips, etc.).
*/
#define IS_GEN2(dev_priv) (!!((dev_priv)->info.gen_mask & BIT(1)))
#define IS_GEN3(dev_priv) (!!((dev_priv)->info.gen_mask & BIT(2)))
#define IS_GEN4(dev_priv) (!!((dev_priv)->info.gen_mask & BIT(3)))
#define IS_GEN5(dev_priv) (!!((dev_priv)->info.gen_mask & BIT(4)))
#define IS_GEN6(dev_priv) (!!((dev_priv)->info.gen_mask & BIT(5)))
#define IS_GEN7(dev_priv) (!!((dev_priv)->info.gen_mask & BIT(6)))
#define IS_GEN8(dev_priv) (!!((dev_priv)->info.gen_mask & BIT(7)))
#define IS_GEN9(dev_priv) (!!((dev_priv)->info.gen_mask & BIT(8)))
#define IS_GEN10(dev_priv) (!!((dev_priv)->info.gen_mask & BIT(9)))
#define IS_LP(dev_priv) (INTEL_INFO(dev_priv)->is_lp)
#define IS_GEN9_LP(dev_priv) (IS_GEN9(dev_priv) && IS_LP(dev_priv))
#define IS_GEN9_BC(dev_priv) (IS_GEN9(dev_priv) && !IS_LP(dev_priv))
#define ENGINE_MASK(id) BIT(id)
#define RENDER_RING ENGINE_MASK(RCS)
#define BSD_RING ENGINE_MASK(VCS)
#define BLT_RING ENGINE_MASK(BCS)
#define VEBOX_RING ENGINE_MASK(VECS)
#define BSD2_RING ENGINE_MASK(VCS2)
#define ALL_ENGINES (~0)
#define HAS_ENGINE(dev_priv, id) \
(!!((dev_priv)->info.ring_mask & ENGINE_MASK(id)))
#define HAS_BSD(dev_priv) HAS_ENGINE(dev_priv, VCS)
#define HAS_BSD2(dev_priv) HAS_ENGINE(dev_priv, VCS2)
#define HAS_BLT(dev_priv) HAS_ENGINE(dev_priv, BCS)
#define HAS_VEBOX(dev_priv) HAS_ENGINE(dev_priv, VECS)
#define HAS_LLC(dev_priv) ((dev_priv)->info.has_llc)
#define HAS_SNOOP(dev_priv) ((dev_priv)->info.has_snoop)
#define HAS_EDRAM(dev_priv) (!!((dev_priv)->edram_cap & EDRAM_ENABLED))
#define HAS_WT(dev_priv) ((IS_HASWELL(dev_priv) || \
IS_BROADWELL(dev_priv)) && HAS_EDRAM(dev_priv))
#define HWS_NEEDS_PHYSICAL(dev_priv) ((dev_priv)->info.hws_needs_physical)
#define HAS_LOGICAL_RING_CONTEXTS(dev_priv) \
((dev_priv)->info.has_logical_ring_contexts)
#define USES_PPGTT(dev_priv) (i915.enable_ppgtt)
#define USES_FULL_PPGTT(dev_priv) (i915.enable_ppgtt >= 2)
#define USES_FULL_48BIT_PPGTT(dev_priv) (i915.enable_ppgtt == 3)
#define HAS_OVERLAY(dev_priv) ((dev_priv)->info.has_overlay)
#define OVERLAY_NEEDS_PHYSICAL(dev_priv) \
((dev_priv)->info.overlay_needs_physical)
/* Early gen2 have a totally busted CS tlb and require pinned batches. */
#define HAS_BROKEN_CS_TLB(dev_priv) (IS_I830(dev_priv) || IS_I845G(dev_priv))
/* WaRsDisableCoarsePowerGating:skl,bxt */
#define NEEDS_WaRsDisableCoarsePowerGating(dev_priv) \
(IS_SKL_GT3(dev_priv) || IS_SKL_GT4(dev_priv))
/*
* dp aux and gmbus irq on gen4 seems to be able to generate legacy interrupts
* even when in MSI mode. This results in spurious interrupt warnings if the
* legacy irq no. is shared with another device. The kernel then disables that
* interrupt source and so prevents the other device from working properly.
*/
#define HAS_AUX_IRQ(dev_priv) ((dev_priv)->info.gen >= 5)
#define HAS_GMBUS_IRQ(dev_priv) ((dev_priv)->info.has_gmbus_irq)
/* With the 945 and later, Y tiling got adjusted so that it was 32 128-byte
* rows, which changed the alignment requirements and fence programming.
*/
#define HAS_128_BYTE_Y_TILING(dev_priv) (!IS_GEN2(dev_priv) && \
!(IS_I915G(dev_priv) || \
IS_I915GM(dev_priv)))
#define SUPPORTS_TV(dev_priv) ((dev_priv)->info.supports_tv)
#define I915_HAS_HOTPLUG(dev_priv) ((dev_priv)->info.has_hotplug)
#define HAS_FW_BLC(dev_priv) (INTEL_GEN(dev_priv) > 2)
#define HAS_PIPE_CXSR(dev_priv) ((dev_priv)->info.has_pipe_cxsr)
#define HAS_FBC(dev_priv) ((dev_priv)->info.has_fbc)
#define HAS_CUR_FBC(dev_priv) (!HAS_GMCH_DISPLAY(dev_priv) && INTEL_INFO(dev_priv)->gen >= 7)
#define HAS_IPS(dev_priv) (IS_HSW_ULT(dev_priv) || IS_BROADWELL(dev_priv))
#define HAS_DP_MST(dev_priv) ((dev_priv)->info.has_dp_mst)
#define HAS_DDI(dev_priv) ((dev_priv)->info.has_ddi)
#define HAS_FPGA_DBG_UNCLAIMED(dev_priv) ((dev_priv)->info.has_fpga_dbg)
#define HAS_PSR(dev_priv) ((dev_priv)->info.has_psr)
#define HAS_RC6(dev_priv) ((dev_priv)->info.has_rc6)
#define HAS_RC6p(dev_priv) ((dev_priv)->info.has_rc6p)
#define HAS_CSR(dev_priv) ((dev_priv)->info.has_csr)
drm/i915/skl: Add support to load SKL CSR firmware. Display Context Save and Restore support is needed for various SKL Display C states like DC5, DC6. This implementation is added based on first version of DMC CSR program that we received from h/w team. Here we are using request_firmware based design. Finally this firmware should end up in linux-firmware tree. For SKL platform its mandatory to ensure that we load this csr program before enabling DC states like DC5/DC6. As CSR program gets reset on various conditions, we should ensure to load it during boot and in future change to be added to load this system resume sequence too. v1: Initial relese as RFC patch v2: Design change as per Daniel, Damien and Shobit's review comments request firmware method followed. v3: Some optimization and functional changes. Pulled register defines into drivers/gpu/drm/i915/i915_reg.h Used kmemdup to allocate and duplicate firmware content. Ensured to free allocated buffer. v4: Modified as per review comments from Satheesh and Daniel Removed temporary buffer. Optimized number of writes by replacing I915_WRITE with I915_WRITE64. v5: Modified as per review comemnts from Damien. - Changed name for functions and firmware. - Introduced HAS_CSR. - Reverted back previous change and used csr_buf with u8 size. - Using cpu_to_be64 for endianness change. Modified as per review comments from Imre. - Modified registers and macro names to be a bit closer to bspec terminology and the existing register naming in the driver. - Early return for non SKL platforms in intel_load_csr_program function. - Added locking around CSR program load function as it may be called concurrently during system/runtime resume. - Releasing the fw before loading the program for consistency - Handled error path during f/w load. v6: Modified as per review comments from Imre. - Corrected out_freecsr sequence. v7: Modified as per review comments from Imre. Fail loading fw if fw->size%8!=0. v8: Rebase to latest. v9: Rebase on top of -nightly (Damien) v10: Enabled support for dmc firmware ver 1.0. According to ver 1.0 in a single binary package all the firmware's that are required for different stepping's of the product will be stored. The package contains the css header, followed by the package header and the actual dmc firmwares. Package header contains the firmware/stepping mapping table and the corresponding firmware offsets to the individual binaries, within the package. Each individual program binary contains the header and the payload sections whose size is specified in the header section. This changes are done to extract the specific firmaware from the package. (Animesh) v11: Modified as per review comemnts from Imre. - Added code comment from bpec for header structure elements. - Added __packed to avoid structure padding. - Added helper functions for stepping and substepping info. - Added code comment for CSR_MAX_FW_SIZE. - Disabled BXT firmware loading, will be enabled with dmc 1.0 support. - Changed skl_stepping_info based on bspec, earlier used from config DB. - Removed duplicate call of cpu_to_be* from intel_csr_load_program function. - Used cpu_to_be32 instead of cpu_to_be64 as firmware binary in dword aligned. - Added sanity check for header length. - Added sanity check for mmio address got from firmware binary. - kmalloc done separately for dmc header and dmc firmware. (Animesh) v12: Modified as per review comemnts from Imre. - Corrected the typo error in skl stepping info structure. - Added out-of-bound access for skl_stepping_info. - Sanity check for mmio address modified. - Sanity check added for stepping and substeppig. - Modified the intel_dmc_info structure, cache only the required header info. (Animesh) v13: clarify firmware load error message. The reason for a firmware loading failure can be obscure if the driver is built-in. Provide an explanation to the user about the likely reason for the failure and how to resolve it. (Imre) v14: Suggested by Jani. - fix s/I915/CONFIG_DRM_I915/ typo - add fw_path to the firmware object instead of using a static ptr (Jani) v15: 1) Changed the firmware name as dmc_gen9.bin, everytime for a new firmware version a symbolic link with same name will help not to build kernel again. 2) Changes done as per review comments from Imre. - Error check removed for intel_csr_ucode_init. - Moved csr-specific data structure to intel_csr.h and optimization done on structure definition. - fw->data used directly for parsing the header info & memory allocation only done separately for payload. (Animesh) v16: - No need for out_regs label in i915_driver_load(), so removed it. - Changed the firmware name as skl_dmc_ver1.bin, followed naming convention <platform>_dmc_<api-version>.bin (Animesh) Issue: VIZ-2569 Signed-off-by: A.Sunil Kamath <sunil.kamath@intel.com> Signed-off-by: Damien Lespiau <damien.lespiau@intel.com> Signed-off-by: Animesh Manna <animesh.manna@intel.com> Signed-off-by: Imre Deak <imre.deak@intel.com> Reviewed-by: Imre Deak <imre.deak@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2015-05-04 06:58:44 -06:00
#define HAS_RUNTIME_PM(dev_priv) ((dev_priv)->info.has_runtime_pm)
#define HAS_64BIT_RELOC(dev_priv) ((dev_priv)->info.has_64bit_reloc)
/*
* For now, anything with a GuC requires uCode loading, and then supports
* command submission once loaded. But these are logically independent
* properties, so we have separate macros to test them.
*/
#define HAS_GUC(dev_priv) ((dev_priv)->info.has_guc)
#define HAS_GUC_CT(dev_priv) ((dev_priv)->info.has_guc_ct)
#define HAS_GUC_UCODE(dev_priv) (HAS_GUC(dev_priv))
#define HAS_GUC_SCHED(dev_priv) (HAS_GUC(dev_priv))
drm/i915/huc: Add HuC fw loading support The HuC loading process is similar to GuC. The intel_uc_fw_fetch() is used for both cases. HuC loading needs to be before GuC loading. The WOPCM setting must be done early before loading any of them. v2: rebased on-top of drm-intel-nightly. removed if(HAS_GUC()) before the guc call. (D.Gordon) update huc_version number of format. v3: rebased to drm-intel-nightly, changed the file name format to match the one in the huc package. Changed dev->dev_private to to_i915() v4: moved function back to where it was. change wait_for_atomic to wait_for. v5: rebased. Changed the year in the copyright message to reflect the right year.Correct the comments,remove the unwanted WARN message, replace drm_gem_object_unreference() with i915_gem_object_put().Make the prototypes in intel_huc.h non-extern. v6: rebased. Update the file construction done by HuC. It is similar to GuC.Adopted the approach used in- https://patchwork.freedesktop.org/patch/104355/ <Tvrtko Ursulin> v7: Change dev to dev_priv in macro definition. Corrected comments. v8: rebased on top of drm-tip. Updated functions intel_huc_load(), intel_huc_init() and intel_uc_fw_fetch() to accept dev_priv instead of dev. Moved contents of intel_huc.h to intel_uc.h. v9: change SKL_FW_ to SKL_HUC_FW_. Add intel_ prefix to guc_wopcm_size(). Remove unwanted checks in intel_uc.h. Rename huc_fw in struct intel_huc to simply fw to avoid redundency. v10: rebased. Correct comments. Make intel_huc_fini() accept dev_priv instead of dev like intel_huc_init() and intel_huc_load().Move definition to i915_guc_reg.h from intel_uc.h. Clean DMA_CTRL bits after HuC DMA transfer in huc_ucode_xfer() instead of guc_ucode_xfer(). Add suitable WARNs to give extra info. v11: rebased. Add proper bias for HuC and make sure there are asserts on failure by using guc_ggtt_offset_vma(). Introduce intel_huc.c and remove intel_huc_loader.c since it has functions that do more than just loading.Correct year in copyright. v12: remove invalidates that are not required anymore. Cc: Arkadiusz Hiler <arkadiusz.hiler@intel.com> Cc: Michal Wajdeczko <michal.wajdeczko@intel.com> Tested-by: Xiang Haihao <haihao.xiang@intel.com> Signed-off-by: Anusha Srivatsa <anusha.srivatsa@intel.com> Signed-off-by: Alex Dai <yu.dai@intel.com> Signed-off-by: Peter Antoine <peter.antoine@intel.com> Reviewed-by: Michal Wajdeczko <michal.wajdeczko@intel.com> Signed-off-by: Jani Nikula <jani.nikula@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1484755558-1234-1-git-send-email-anusha.srivatsa@intel.com
2017-01-18 09:05:53 -07:00
#define HAS_HUC_UCODE(dev_priv) (HAS_GUC(dev_priv))
#define HAS_RESOURCE_STREAMER(dev_priv) ((dev_priv)->info.has_resource_streamer)
#define HAS_POOLED_EU(dev_priv) ((dev_priv)->info.has_pooled_eu)
drm/i915:bxt: Enable Pooled EU support This mode allows to assign EUs to pools which can process work collectively. The command to enable this mode should be issued as part of context initialization. The pooled mode is global, once enabled it has to stay the same across all contexts until HW reset hence this is sent in auxiliary golden context batch. Thanks to Mika for the preliminary review and comments. v2: explain why this is enabled in golden context, use feature flag while enabling the support (Chris) v3: Include only kernel support as userspace support is not available yet. User space clients need to know when the pooled EU feature is present and enabled on the hardware so that they can adapt work submissions. Create a new device info flag for this purpose. Set has_pooled_eu to true in the Broxton static device info - Broxton supports the feature in hardware and the driver will enable it by default. We need to add getparam ioctls to enable userspace to query availability of this feature and to retrieve min. no of eus in a pool but we will expose them once userspace support is available. Opensource users for this feature are mesa, libva and beignet. Beignet team is currently working on adding userspace support. Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> (v2) Cc: Winiarski, Michal <michal.winiarski@intel.com> Cc: Zou, Nanhai <nanhai.zou@intel.com> Cc: Yang, Rong R <rong.r.yang@intel.com> Cc: Mika Kuoppala <mika.kuoppala@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Armin Reese <armin.c.reese@intel.com> Cc: Tim Gore <tim.gore@intel.com> Signed-off-by: Jeff McGee <jeff.mcgee@intel.com> Signed-off-by: Arun Siluvery <arun.siluvery@linux.intel.com> Reviewed-by: Michał Winiarski <michal.winiarski@intel.com> Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
2016-06-02 23:34:33 -06:00
#define INTEL_PCH_DEVICE_ID_MASK 0xff80
#define INTEL_PCH_IBX_DEVICE_ID_TYPE 0x3b00
#define INTEL_PCH_CPT_DEVICE_ID_TYPE 0x1c00
#define INTEL_PCH_PPT_DEVICE_ID_TYPE 0x1e00
#define INTEL_PCH_LPT_DEVICE_ID_TYPE 0x8c00
#define INTEL_PCH_LPT_LP_DEVICE_ID_TYPE 0x9c00
#define INTEL_PCH_WPT_DEVICE_ID_TYPE 0x8c80
#define INTEL_PCH_WPT_LP_DEVICE_ID_TYPE 0x9c80
#define INTEL_PCH_SPT_DEVICE_ID_TYPE 0xA100
#define INTEL_PCH_SPT_LP_DEVICE_ID_TYPE 0x9D00
#define INTEL_PCH_KBP_DEVICE_ID_TYPE 0xA280
#define INTEL_PCH_CNP_DEVICE_ID_TYPE 0xA300
#define INTEL_PCH_CNP_LP_DEVICE_ID_TYPE 0x9D80
#define INTEL_PCH_P2X_DEVICE_ID_TYPE 0x7100
#define INTEL_PCH_P3X_DEVICE_ID_TYPE 0x7000
#define INTEL_PCH_QEMU_DEVICE_ID_TYPE 0x2900 /* qemu q35 has 2918 */
#define INTEL_PCH_TYPE(dev_priv) ((dev_priv)->pch_type)
#define HAS_PCH_CNP(dev_priv) (INTEL_PCH_TYPE(dev_priv) == PCH_CNP)
#define HAS_PCH_CNP_LP(dev_priv) \
((dev_priv)->pch_id == INTEL_PCH_CNP_LP_DEVICE_ID_TYPE)
#define HAS_PCH_KBP(dev_priv) (INTEL_PCH_TYPE(dev_priv) == PCH_KBP)
#define HAS_PCH_SPT(dev_priv) (INTEL_PCH_TYPE(dev_priv) == PCH_SPT)
#define HAS_PCH_LPT(dev_priv) (INTEL_PCH_TYPE(dev_priv) == PCH_LPT)
#define HAS_PCH_LPT_LP(dev_priv) \
((dev_priv)->pch_id == INTEL_PCH_LPT_LP_DEVICE_ID_TYPE || \
(dev_priv)->pch_id == INTEL_PCH_WPT_LP_DEVICE_ID_TYPE)
#define HAS_PCH_LPT_H(dev_priv) \
((dev_priv)->pch_id == INTEL_PCH_LPT_DEVICE_ID_TYPE || \
(dev_priv)->pch_id == INTEL_PCH_WPT_DEVICE_ID_TYPE)
#define HAS_PCH_CPT(dev_priv) (INTEL_PCH_TYPE(dev_priv) == PCH_CPT)
#define HAS_PCH_IBX(dev_priv) (INTEL_PCH_TYPE(dev_priv) == PCH_IBX)
#define HAS_PCH_NOP(dev_priv) (INTEL_PCH_TYPE(dev_priv) == PCH_NOP)
#define HAS_PCH_SPLIT(dev_priv) (INTEL_PCH_TYPE(dev_priv) != PCH_NONE)
#define HAS_GMCH_DISPLAY(dev_priv) ((dev_priv)->info.has_gmch_display)
#define HAS_LSPCON(dev_priv) (INTEL_GEN(dev_priv) >= 9)
/* DPF == dynamic parity feature */
#define HAS_L3_DPF(dev_priv) ((dev_priv)->info.has_l3_dpf)
#define NUM_L3_SLICES(dev_priv) (IS_HSW_GT3(dev_priv) ? \
2 : HAS_L3_DPF(dev_priv))
#define GT_FREQUENCY_MULTIPLIER 50
#define GEN9_FREQ_SCALER 3
#include "i915_trace.h"
static inline bool intel_vtd_active(void)
{
#ifdef CONFIG_INTEL_IOMMU
if (intel_iommu_gfx_mapped)
return true;
#endif
return false;
}
static inline bool intel_scanout_needs_vtd_wa(struct drm_i915_private *dev_priv)
{
return INTEL_GEN(dev_priv) >= 6 && intel_vtd_active();
}
drm/i915: Serialize GTT/Aperture accesses on BXT BXT has a H/W issue with IOMMU which can lead to system hangs when Aperture accesses are queued within the GAM behind GTT Accesses. This patch avoids the condition by wrapping all GTT updates in stop_machine and using a flushing read prior to restarting the machine. The stop_machine guarantees no new Aperture accesses can begin while the PTE writes are being emmitted. The flushing read ensures that any following Aperture accesses cannot begin until the PTE writes have been cleared out of the GAM's fifo. Only FOLLOWING Aperture accesses need to be separated from in flight PTE updates. PTE Writes may follow tightly behind already in flight Aperture accesses, so no flushing read is required at the start of a PTE update sequence. This issue was reproduced by running igt/gem_readwrite and igt/gem_render_copy simultaneously from different processes, each in a tight loop, with INTEL_IOMMU enabled. This patch was originally published as: drm/i915: Serialize GTT Updates on BXT v2: Move bxt/iommu detection into static function Remove #ifdef CONFIG_INTEL_IOMMU protection Make function names more reflective of purpose Move flushing read into static function v3: Tidy up for checkpatch.pl Testcase: igt/gem_concurrent_blit Signed-off-by: Jon Bloomfield <jon.bloomfield@intel.com> Cc: John Harrison <john.C.Harrison@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Daniel Vetter <daniel.vetter@intel.com> Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1495641251-30022-1-git-send-email-jon.bloomfield@intel.com Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
2017-05-24 09:54:11 -06:00
static inline bool
intel_ggtt_update_needs_vtd_wa(struct drm_i915_private *dev_priv)
{
return IS_BROXTON(dev_priv) && intel_vtd_active();
drm/i915: Serialize GTT/Aperture accesses on BXT BXT has a H/W issue with IOMMU which can lead to system hangs when Aperture accesses are queued within the GAM behind GTT Accesses. This patch avoids the condition by wrapping all GTT updates in stop_machine and using a flushing read prior to restarting the machine. The stop_machine guarantees no new Aperture accesses can begin while the PTE writes are being emmitted. The flushing read ensures that any following Aperture accesses cannot begin until the PTE writes have been cleared out of the GAM's fifo. Only FOLLOWING Aperture accesses need to be separated from in flight PTE updates. PTE Writes may follow tightly behind already in flight Aperture accesses, so no flushing read is required at the start of a PTE update sequence. This issue was reproduced by running igt/gem_readwrite and igt/gem_render_copy simultaneously from different processes, each in a tight loop, with INTEL_IOMMU enabled. This patch was originally published as: drm/i915: Serialize GTT Updates on BXT v2: Move bxt/iommu detection into static function Remove #ifdef CONFIG_INTEL_IOMMU protection Make function names more reflective of purpose Move flushing read into static function v3: Tidy up for checkpatch.pl Testcase: igt/gem_concurrent_blit Signed-off-by: Jon Bloomfield <jon.bloomfield@intel.com> Cc: John Harrison <john.C.Harrison@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Daniel Vetter <daniel.vetter@intel.com> Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1495641251-30022-1-git-send-email-jon.bloomfield@intel.com Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
2017-05-24 09:54:11 -06:00
}
int intel_sanitize_enable_ppgtt(struct drm_i915_private *dev_priv,
int enable_ppgtt);
bool intel_sanitize_semaphores(struct drm_i915_private *dev_priv, int value);
/* i915_drv.c */
drm/i915: Tune down init error message due to failure injection Atm, in case failure injection forces an error the subsequent "*ERROR* failed to init modeset" error message will make automated tests (CI) report this event as a breakage even though the event is expected. To fix this print the error message with debug log level in this case. While at it print the error message for any init failure and change it to """ Device initialization failed (errno) Please file a bug at https://bugs.freedesktop.org/enter_bug.cgi?product=DRI against DRM/Intel providing the dmesg log by booting with drm.debug=0xf """ and export a helper printing error messages using this same format. A follow-up patch will convert all uses of DRM_ERROR reporting a user facing problem to use this new helper instead. v2: - Include the problematic error message in the commit log, add a request to file an fdo bug to the message (Chris) v3: - Include the new error message too in the commit log, make the fdo link more precise and print part of the message with info log level (Chris) v4: (Chris) - Use dev_printk instead of DRM_ERROR/INFO and use NOTICE instead of INFO loglevel - Export a helper for printing user facing error messages v5: - Keep the DRM_ERROR message prefix used by piglit-igt/CI to filter relevant dmesg lines - Use dev_notice(), instead of dev_printk(KERN_NOTICE,...) v6: - Print the fdo bug link only once (Chris) CC: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Imre Deak <imre.deak@intel.com> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Link: http://patchwork.freedesktop.org/patch/msgid/1458290770-15480-1-git-send-email-imre.deak@intel.com
2016-03-18 02:46:10 -06:00
void __printf(3, 4)
__i915_printk(struct drm_i915_private *dev_priv, const char *level,
const char *fmt, ...);
#define i915_report_error(dev_priv, fmt, ...) \
__i915_printk(dev_priv, KERN_ERR, fmt, ##__VA_ARGS__)
#ifdef CONFIG_COMPAT
extern long i915_compat_ioctl(struct file *filp, unsigned int cmd,
unsigned long arg);
#else
#define i915_compat_ioctl NULL
#endif
extern const struct dev_pm_ops i915_pm_ops;
extern int i915_driver_load(struct pci_dev *pdev,
const struct pci_device_id *ent);
extern void i915_driver_unload(struct drm_device *dev);
extern int intel_gpu_reset(struct drm_i915_private *dev_priv, u32 engine_mask);
extern bool intel_has_gpu_reset(struct drm_i915_private *dev_priv);
#define I915_RESET_QUIET BIT(0)
extern void i915_reset(struct drm_i915_private *i915, unsigned int flags);
extern int i915_reset_engine(struct intel_engine_cs *engine,
unsigned int flags);
drm/i915: Modify error handler for per engine hang recovery This is a preparatory patch which modifies error handler to do per engine hang recovery. The actual patch which implements this sequence follows later in the series. The aim is to prepare existing recovery function to adapt to this new function where applicable (which fails at this point because core implementation is lacking) and continue recovery using legacy full gpu reset. A helper function is also added to query the availability of engine reset. A subsequent patch will add the capability to query which type of reset is present (engine -> full -> no-reset) via the get-param ioctl. It has been decided that the error events that are used to notify user of reset will only be sent in case if full chip reset. In case of just single (or multiple) engine resets, userspace won't be notified by these events. Note that this implementation of engine reset is for i915 directly submitting to the ELSP, where the driver manages the hang detection, recovery and resubmission. With GuC submission these tasks are shared between driver and firmware; i915 will still responsible for detecting a hang, and when it does it will have to request GuC to reset that Engine and remind the firmware about the outstanding submissions. This will be added in different patch. v2: rebase, advertise engine reset availability in platform definition, add note about GuC submission. v3: s/*engine_reset*/*reset_engine*/. (Chris) Handle reset as 2 level resets, by first going to engine only and fall backing to full/chip reset as needed, i.e. reset_engine will need the struct_mutex. v4: Pass the engine mask to i915_reset. (Chris) v5: Rebase, update selftests. v6: Rebase, prepare for mutex-less reset engine. v7: Pass reset_engine mask as a function parameter, and iterate over the engine mask for reset_engine. (Chris) v8: Use i915.reset >=2 in has_reset_engine; remove redundant reset logging; add a reset-engine-in-progress flag to prevent concurrent resets, and avoid dual purposing of reset-backoff. (Chris) v9: Support reset of different engines in parallel (Chris) v10: Handle reset-engine flag locking better (Chris) v11: Squash in reporting of per-engine-reset availability. Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com> Signed-off-by: Ian Lister <ian.lister@intel.com> Signed-off-by: Tomas Elf <tomas.elf@intel.com> Signed-off-by: Arun Siluvery <arun.siluvery@linux.intel.com> Signed-off-by: Michel Thierry <michel.thierry@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/20170615201828.23144-4-michel.thierry@intel.com Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Link: http://patchwork.freedesktop.org/patch/msgid/20170620095751.13127-5-chris@chris-wilson.co.uk
2017-06-20 03:57:46 -06:00
extern bool intel_has_reset_engine(struct drm_i915_private *dev_priv);
extern int intel_guc_reset(struct drm_i915_private *dev_priv);
extern void intel_engine_init_hangcheck(struct intel_engine_cs *engine);
extern void intel_hangcheck_init(struct drm_i915_private *dev_priv);
extern unsigned long i915_chipset_val(struct drm_i915_private *dev_priv);
extern unsigned long i915_mch_val(struct drm_i915_private *dev_priv);
extern unsigned long i915_gfx_val(struct drm_i915_private *dev_priv);
extern void i915_update_gfx_val(struct drm_i915_private *dev_priv);
int vlv_force_gfx_clock(struct drm_i915_private *dev_priv, bool on);
int intel_engines_init_mmio(struct drm_i915_private *dev_priv);
int intel_engines_init(struct drm_i915_private *dev_priv);
/* intel_hotplug.c */
drm/i915: Small display interrupt handlers tidy I have noticed some of our interrupt handlers use both dev and dev_priv while they could get away with only dev_priv in the huge majority of cases. Tidying that up had a cascading effect on changing functions prototypes, so relatively big churn factor, but I think it is for the better. For example even where changes cascade out of i915_irq.c, for functions prefixed with intel_, genX_ or <plat>_, it makes more sense to take dev_priv directly anyway. This allows us to eliminate local variables and intermixed usage of dev and dev_priv where only one is good enough. End result is shrinkage of both source and the resulting binary. i915.ko: - .text 000b0899 + .text 000b0619 Or if we look at the Gen8 display irq chain: -00000000000006ad t gen8_irq_handler +0000000000000663 t gen8_irq_handler -0000000000000028 T intel_opregion_asle_intr +0000000000000024 T intel_opregion_asle_intr -000000000000008c t ilk_hpd_irq_handler +000000000000007f t ilk_hpd_irq_handler -0000000000000116 T intel_check_page_flip +0000000000000112 T intel_check_page_flip -000000000000011a T intel_prepare_page_flip +0000000000000119 T intel_prepare_page_flip -0000000000000014 T intel_finish_page_flip_plane +0000000000000013 T intel_finish_page_flip_plane -0000000000000053 t hsw_pipe_crc_irq_handler +000000000000004c t hsw_pipe_crc_irq_handler -000000000000022e t cpt_irq_handler +0000000000000213 t cpt_irq_handler So small shrinkage but it is all fast paths so doesn't harm. Situation is similar in other interrupt handlers as well. v2: Tidy intel_queue_rps_boost_for_request as well. (Chris Wilson) Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
2016-05-06 07:48:28 -06:00
void intel_hpd_irq_handler(struct drm_i915_private *dev_priv,
u32 pin_mask, u32 long_mask);
void intel_hpd_init(struct drm_i915_private *dev_priv);
void intel_hpd_init_work(struct drm_i915_private *dev_priv);
void intel_hpd_cancel_work(struct drm_i915_private *dev_priv);
enum port intel_hpd_pin_to_port(enum hpd_pin pin);
enum hpd_pin intel_hpd_pin(enum port port);
bool intel_hpd_disable(struct drm_i915_private *dev_priv, enum hpd_pin pin);
void intel_hpd_enable(struct drm_i915_private *dev_priv, enum hpd_pin pin);
/* i915_irq.c */
static inline void i915_queue_hangcheck(struct drm_i915_private *dev_priv)
{
unsigned long delay;
if (unlikely(!i915.enable_hangcheck))
return;
/* Don't continually defer the hangcheck so that it is always run at
* least once after work has been scheduled on any ring. Otherwise,
* we will ignore a hung ring if a second ring is kept busy.
*/
delay = round_jiffies_up_relative(DRM_I915_HANGCHECK_JIFFIES);
queue_delayed_work(system_long_wq,
&dev_priv->gpu_error.hangcheck_work, delay);
}
__printf(3, 4)
void i915_handle_error(struct drm_i915_private *dev_priv,
u32 engine_mask,
const char *fmt, ...);
extern void intel_irq_init(struct drm_i915_private *dev_priv);
extern void intel_irq_fini(struct drm_i915_private *dev_priv);
int intel_irq_install(struct drm_i915_private *dev_priv);
void intel_irq_uninstall(struct drm_i915_private *dev_priv);
drm/i915: gvt: Introduce the basic architecture of GVT-g This patch introduces the very basic framework of GVT-g device model, includes basic prototypes, definitions, initialization. v12: - Call intel_gvt_init() in driver early initialization stage. (Chris) v8: - Remove the GVT idr and mutex in intel_gvt_host. (Joonas) v7: - Refine the URL link in Kconfig. (Joonas) - Refine the introduction of GVT-g host support in Kconfig. (Joonas) - Remove the macro GVT_ALIGN(), use round_down() instead. (Joonas) - Make "struct intel_gvt" a data member in struct drm_i915_private.(Joonas) - Remove {alloc, free}_gvt_device() - Rename intel_gvt_{create, destroy}_gvt_device() - Expost intel_gvt_init_host() - Remove the dummy "struct intel_gvt" declaration in intel_gvt.h (Joonas) v6: - Refine introduction in Kconfig. (Chris) - The exposed API functions will take struct intel_gvt * instead of void *. (Chris/Tvrtko) - Remove most memebers of strct intel_gvt_device_info. Will add them in the device model patches.(Chris) - Remove gvt_info() and gvt_err() in debug.h. (Chris) - Move GVT kernel parameter into i915_params. (Chris) - Remove include/drm/i915_gvt.h, as GVT-g will be built within i915. - Remove the redundant struct i915_gvt *, as the functions in i915 will directly take struct intel_gvt *. - Add more comments for reviewer. v5: Take Tvrtko's comments: - Fix the misspelled words in Kconfig - Let functions take drm_i915_private * instead of struct drm_device * - Remove redundant prints/local varible initialization v3: Take Joonas' comments: - Change file name i915_gvt.* to intel_gvt.* - Move GVT kernel parameter into intel_gvt.c - Remove redundant debug macros - Change error handling style - Add introductions for some stub functions - Introduce drm/i915_gvt.h. Take Kevin's comments: - Move GVT-g host/guest check into intel_vgt_balloon in i915_gem_gtt.c v2: - Introduce i915_gvt.c. It's necessary to introduce the stubs between i915 driver and GVT-g host, as GVT-g components is configurable in kernel config. When disabled, the stubs here do nothing. Take Joonas' comments: - Replace boolean return value with int. - Replace customized info/warn/debug macros with DRM macros. - Document all non-static functions like i915. - Remove empty and unused functions. - Replace magic number with marcos. - Set GVT-g in kernel config to "n" by default. Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> Cc: Kevin Tian <kevin.tian@intel.com> Signed-off-by: Zhi Wang <zhi.a.wang@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1466078825-6662-5-git-send-email-zhi.a.wang@intel.com Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
2016-06-16 06:07:00 -06:00
static inline bool intel_gvt_active(struct drm_i915_private *dev_priv)
{
return dev_priv->gvt;
drm/i915: gvt: Introduce the basic architecture of GVT-g This patch introduces the very basic framework of GVT-g device model, includes basic prototypes, definitions, initialization. v12: - Call intel_gvt_init() in driver early initialization stage. (Chris) v8: - Remove the GVT idr and mutex in intel_gvt_host. (Joonas) v7: - Refine the URL link in Kconfig. (Joonas) - Refine the introduction of GVT-g host support in Kconfig. (Joonas) - Remove the macro GVT_ALIGN(), use round_down() instead. (Joonas) - Make "struct intel_gvt" a data member in struct drm_i915_private.(Joonas) - Remove {alloc, free}_gvt_device() - Rename intel_gvt_{create, destroy}_gvt_device() - Expost intel_gvt_init_host() - Remove the dummy "struct intel_gvt" declaration in intel_gvt.h (Joonas) v6: - Refine introduction in Kconfig. (Chris) - The exposed API functions will take struct intel_gvt * instead of void *. (Chris/Tvrtko) - Remove most memebers of strct intel_gvt_device_info. Will add them in the device model patches.(Chris) - Remove gvt_info() and gvt_err() in debug.h. (Chris) - Move GVT kernel parameter into i915_params. (Chris) - Remove include/drm/i915_gvt.h, as GVT-g will be built within i915. - Remove the redundant struct i915_gvt *, as the functions in i915 will directly take struct intel_gvt *. - Add more comments for reviewer. v5: Take Tvrtko's comments: - Fix the misspelled words in Kconfig - Let functions take drm_i915_private * instead of struct drm_device * - Remove redundant prints/local varible initialization v3: Take Joonas' comments: - Change file name i915_gvt.* to intel_gvt.* - Move GVT kernel parameter into intel_gvt.c - Remove redundant debug macros - Change error handling style - Add introductions for some stub functions - Introduce drm/i915_gvt.h. Take Kevin's comments: - Move GVT-g host/guest check into intel_vgt_balloon in i915_gem_gtt.c v2: - Introduce i915_gvt.c. It's necessary to introduce the stubs between i915 driver and GVT-g host, as GVT-g components is configurable in kernel config. When disabled, the stubs here do nothing. Take Joonas' comments: - Replace boolean return value with int. - Replace customized info/warn/debug macros with DRM macros. - Document all non-static functions like i915. - Remove empty and unused functions. - Replace magic number with marcos. - Set GVT-g in kernel config to "n" by default. Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> Cc: Kevin Tian <kevin.tian@intel.com> Signed-off-by: Zhi Wang <zhi.a.wang@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1466078825-6662-5-git-send-email-zhi.a.wang@intel.com Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
2016-06-16 06:07:00 -06:00
}
static inline bool intel_vgpu_active(struct drm_i915_private *dev_priv)
drm/i915: Introduce a PV INFO page structure for Intel GVT-g. Introduce a PV INFO structure, to facilitate the Intel GVT-g technology, which is a GPU virtualization solution with mediated pass-through. This page contains the shared information between i915 driver and the host emulator. For now, this structure utilizes an area of 4K bytes on HSW GPU's unused MMIO space. Future hardware will have the reserved window architecturally defined, and layout of the page will be added in future BSpec. The i915 driver load routine detects if it is running in a VM by reading the contents of this PV INFO page. Thereafter a flag, vgpu.active is set, and intel_vgpu_active() is used by checking this flag to conclude if GPU is virtualized with Intel GVT-g. By now, intel_vgpu_active() will return true, only when the driver is running as a guest in the Intel GVT-g enhanced environment on HSW platform. v2: take Chris' comments: - call the i915_check_vgpu() in intel_uncore_init() - sanitize i915_check_vgpu() by adding BUILD_BUG_ON() and debug info take Daniel's comments: - put the definition of PV INFO into a new header - i915_vgt_if.h other changes: - access mmio regs by readq/readw in i915_check_vgpu() v3: take Daniel's comments: - move the i915/vgt interfaces into a new i915_vgpu.c - update makefile - add kerneldoc to functions which are non-static - add a DOC: section describing some of the high-level design - update drm docbook other changes: - rename i915_vgt_if.h to i915_vgpu.h v4: take Tvrtko's comments: - fix a typo in commit message - add debug message when vgt version mismatches - rename low_gmadr/high_gmadr to mappable/non-mappable in PV INFO structure Signed-off-by: Yu Zhang <yu.c.zhang@linux.intel.com> Signed-off-by: Jike Song <jike.song@intel.com> Signed-off-by: Eddie Dong <eddie.dong@intel.com> Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2015-02-10 04:05:47 -07:00
{
return dev_priv->vgpu.active;
drm/i915: Introduce a PV INFO page structure for Intel GVT-g. Introduce a PV INFO structure, to facilitate the Intel GVT-g technology, which is a GPU virtualization solution with mediated pass-through. This page contains the shared information between i915 driver and the host emulator. For now, this structure utilizes an area of 4K bytes on HSW GPU's unused MMIO space. Future hardware will have the reserved window architecturally defined, and layout of the page will be added in future BSpec. The i915 driver load routine detects if it is running in a VM by reading the contents of this PV INFO page. Thereafter a flag, vgpu.active is set, and intel_vgpu_active() is used by checking this flag to conclude if GPU is virtualized with Intel GVT-g. By now, intel_vgpu_active() will return true, only when the driver is running as a guest in the Intel GVT-g enhanced environment on HSW platform. v2: take Chris' comments: - call the i915_check_vgpu() in intel_uncore_init() - sanitize i915_check_vgpu() by adding BUILD_BUG_ON() and debug info take Daniel's comments: - put the definition of PV INFO into a new header - i915_vgt_if.h other changes: - access mmio regs by readq/readw in i915_check_vgpu() v3: take Daniel's comments: - move the i915/vgt interfaces into a new i915_vgpu.c - update makefile - add kerneldoc to functions which are non-static - add a DOC: section describing some of the high-level design - update drm docbook other changes: - rename i915_vgt_if.h to i915_vgpu.h v4: take Tvrtko's comments: - fix a typo in commit message - add debug message when vgt version mismatches - rename low_gmadr/high_gmadr to mappable/non-mappable in PV INFO structure Signed-off-by: Yu Zhang <yu.c.zhang@linux.intel.com> Signed-off-by: Jike Song <jike.song@intel.com> Signed-off-by: Eddie Dong <eddie.dong@intel.com> Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2015-02-10 04:05:47 -07:00
}
void
i915_enable_pipestat(struct drm_i915_private *dev_priv, enum pipe pipe,
u32 status_mask);
void
i915_disable_pipestat(struct drm_i915_private *dev_priv, enum pipe pipe,
u32 status_mask);
void valleyview_enable_display_irqs(struct drm_i915_private *dev_priv);
void valleyview_disable_display_irqs(struct drm_i915_private *dev_priv);
void i915_hotplug_interrupt_update(struct drm_i915_private *dev_priv,
uint32_t mask,
uint32_t bits);
void ilk_update_display_irq(struct drm_i915_private *dev_priv,
uint32_t interrupt_mask,
uint32_t enabled_irq_mask);
static inline void
ilk_enable_display_irq(struct drm_i915_private *dev_priv, uint32_t bits)
{
ilk_update_display_irq(dev_priv, bits, bits);
}
static inline void
ilk_disable_display_irq(struct drm_i915_private *dev_priv, uint32_t bits)
{
ilk_update_display_irq(dev_priv, bits, 0);
}
void bdw_update_pipe_irq(struct drm_i915_private *dev_priv,
enum pipe pipe,
uint32_t interrupt_mask,
uint32_t enabled_irq_mask);
static inline void bdw_enable_pipe_irq(struct drm_i915_private *dev_priv,
enum pipe pipe, uint32_t bits)
{
bdw_update_pipe_irq(dev_priv, pipe, bits, bits);
}
static inline void bdw_disable_pipe_irq(struct drm_i915_private *dev_priv,
enum pipe pipe, uint32_t bits)
{
bdw_update_pipe_irq(dev_priv, pipe, bits, 0);
}
void ibx_display_interrupt_update(struct drm_i915_private *dev_priv,
uint32_t interrupt_mask,
uint32_t enabled_irq_mask);
static inline void
ibx_enable_display_interrupt(struct drm_i915_private *dev_priv, uint32_t bits)
{
ibx_display_interrupt_update(dev_priv, bits, bits);
}
static inline void
ibx_disable_display_interrupt(struct drm_i915_private *dev_priv, uint32_t bits)
{
ibx_display_interrupt_update(dev_priv, bits, 0);
}
/* i915_gem.c */
int i915_gem_create_ioctl(struct drm_device *dev, void *data,
struct drm_file *file_priv);
int i915_gem_pread_ioctl(struct drm_device *dev, void *data,
struct drm_file *file_priv);
int i915_gem_pwrite_ioctl(struct drm_device *dev, void *data,
struct drm_file *file_priv);
int i915_gem_mmap_ioctl(struct drm_device *dev, void *data,
struct drm_file *file_priv);
int i915_gem_mmap_gtt_ioctl(struct drm_device *dev, void *data,
struct drm_file *file_priv);
int i915_gem_set_domain_ioctl(struct drm_device *dev, void *data,
struct drm_file *file_priv);
int i915_gem_sw_finish_ioctl(struct drm_device *dev, void *data,
struct drm_file *file_priv);
int i915_gem_execbuffer(struct drm_device *dev, void *data,
struct drm_file *file_priv);
int i915_gem_execbuffer2(struct drm_device *dev, void *data,
struct drm_file *file_priv);
int i915_gem_busy_ioctl(struct drm_device *dev, void *data,
struct drm_file *file_priv);
int i915_gem_get_caching_ioctl(struct drm_device *dev, void *data,
struct drm_file *file);
int i915_gem_set_caching_ioctl(struct drm_device *dev, void *data,
struct drm_file *file);
int i915_gem_throttle_ioctl(struct drm_device *dev, void *data,
struct drm_file *file_priv);
int i915_gem_madvise_ioctl(struct drm_device *dev, void *data,
struct drm_file *file_priv);
int i915_gem_set_tiling_ioctl(struct drm_device *dev, void *data,
struct drm_file *file_priv);
int i915_gem_get_tiling_ioctl(struct drm_device *dev, void *data,
struct drm_file *file_priv);
int i915_gem_init_userptr(struct drm_i915_private *dev_priv);
void i915_gem_cleanup_userptr(struct drm_i915_private *dev_priv);
drm/i915: Introduce mapping of user pages into video memory (userptr) ioctl By exporting the ability to map user address and inserting PTEs representing their backing pages into the GTT, we can exploit UMA in order to utilize normal application data as a texture source or even as a render target (depending upon the capabilities of the chipset). This has a number of uses, with zero-copy downloads to the GPU and efficient readback making the intermixed streaming of CPU and GPU operations fairly efficient. This ability has many widespread implications from faster rendering of client-side software rasterisers (chromium), mitigation of stalls due to read back (firefox) and to faster pipelining of texture data (such as pixel buffer objects in GL or data blobs in CL). v2: Compile with CONFIG_MMU_NOTIFIER v3: We can sleep while performing invalidate-range, which we can utilise to drop our page references prior to the kernel manipulating the vma (for either discard or cloning) and so protect normal users. v4: Only run the invalidate notifier if the range intercepts the bo. v5: Prevent userspace from attempting to GTT mmap non-page aligned buffers v6: Recheck after reacquire mutex for lost mmu. v7: Fix implicit padding of ioctl struct by rounding to next 64bit boundary. v8: Fix rebasing error after forwarding porting the back port. v9: Limit the userptr to page aligned entries. We now expect userspace to handle all the offset-in-page adjustments itself. v10: Prevent vma from being copied across fork to avoid issues with cow. v11: Drop vma behaviour changes -- locking is nigh on impossible. Use a worker to load user pages to avoid lock inversions. v12: Use get_task_mm()/mmput() for correct refcounting of mm. v13: Use a worker to release the mmu_notifier to avoid lock inversion v14: Decouple mmu_notifier from struct_mutex using a custom mmu_notifer with its own locking and tree of objects for each mm/mmu_notifier. v15: Prevent overlapping userptr objects, and invalidate all objects within the mmu_notifier range v16: Fix a typo for iterating over multiple objects in the range and rearrange error path to destroy the mmu_notifier locklessly. Also close a race between invalidate_range and the get_pages_worker. v17: Close a race between get_pages_worker/invalidate_range and fresh allocations of the same userptr range - and notice that struct_mutex was presumed to be held when during creation it wasn't. v18: Sigh. Fix the refactor of st_set_pages() to allocate enough memory for the struct sg_table and to clear it before reporting an error. v19: Always error out on read-only userptr requests as we don't have the hardware infrastructure to support them at the moment. v20: Refuse to implement read-only support until we have the required infrastructure - but reserve the bit in flags for future use. v21: use_mm() is not required for get_user_pages(). It is only meant to be used to fix up the kernel thread's current->mm for use with copy_user(). v22: Use sg_alloc_table_from_pages for that chunky feeling v23: Export a function for sanity checking dma-buf rather than encode userptr details elsewhere, and clean up comments based on suggestions by Bradley. Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> Cc: "Gong, Zhipeng" <zhipeng.gong@intel.com> Cc: Akash Goel <akash.goel@intel.com> Cc: "Volkin, Bradley D" <bradley.d.volkin@intel.com> Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> Reviewed-by: Brad Volkin <bradley.d.volkin@intel.com> [danvet: Frob ioctl allocation to pick the next one - will cause a bit of fuss with create2 apparently, but such are the rules.] [danvet2: oops, forgot to git add after manual patch application] [danvet3: Appease sparse.] Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2014-05-16 07:22:37 -06:00
int i915_gem_userptr_ioctl(struct drm_device *dev, void *data,
struct drm_file *file);
int i915_gem_get_aperture_ioctl(struct drm_device *dev, void *data,
struct drm_file *file_priv);
drm/i915: wait render timeout ioctl This helps implement GL_ARB_sync but stops short of allowing full blown sync objects. Finally we can use the new timed seqno waiting function to allow userspace to wait on a buffer object with a timeout. This implements that interface. The IOCTL will take as input a buffer object handle, and a timeout in nanoseconds (flags is currently optional but will likely be used for permutations of flush operations). Users may specify 0 nanoseconds to instantly check. The wait ioctl with a timeout of 0 reimplements the busy ioctl. With any non-zero timeout parameter the wait ioctl will wait for the given number of nanoseconds on an object becoming unbusy. Since the wait itself does so holding struct_mutex the object may become re-busied before this completes. A similar but shorter race condition exists in the busy ioctl. v2: ETIME/ERESTARTSYS instead of changing to EBUSY, and EGAIN (Chris) Flush the object from the gpu write domain (Chris + Daniel) Fix leaked refcount in good case (Chris) Naturally align ioctl struct (Chris) v3: Drop lock after getting seqno to avoid ugly dance (Chris) v4: check for 0 timeout after olr check to allow polling (Chris) v5: Updated the comment. (Chris) v6: Return -ETIME instead of -EBUSY when timeout_ns is 0 (Daniel) Fix the commit message comment to be less ugly (Ben) Add a warning to check the return timespec (Ben) v7: Use DRM_AUTH for the ioctl. (Eugeni) Signed-off-by: Ben Widawsky <ben@bwidawsk.net> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2012-05-24 16:03:10 -06:00
int i915_gem_wait_ioctl(struct drm_device *dev, void *data,
struct drm_file *file_priv);
void i915_gem_sanitize(struct drm_i915_private *i915);
int i915_gem_load_init(struct drm_i915_private *dev_priv);
void i915_gem_load_cleanup(struct drm_i915_private *dev_priv);
void i915_gem_load_init_fences(struct drm_i915_private *dev_priv);
drm/i915: Only shrink the unbound objects during freeze At the point of creating the hibernation image, the runtime power manage core is disabled - and using the rpm functions triggers a warn. i915_gem_shrink_all() tries to unbind objects, which requires device access and so tries to how an rpm reference triggering a warning: [ 44.235420] ------------[ cut here ]------------ [ 44.235424] WARNING: CPU: 2 PID: 2199 at drivers/gpu/drm/i915/intel_runtime_pm.c:2688 intel_runtime_pm_get_if_in_use+0xe6/0xf0 [ 44.235426] WARN_ON_ONCE(ret < 0) [ 44.235445] Modules linked in: ctr ccm arc4 rt2800usb rt2x00usb rt2800lib rt2x00lib crc_ccitt mac80211 cmac cfg80211 btusb rfcomm bnep btrtl btbcm btintel bluetooth dcdbas x86_pkg_temp_thermal intel_powerclamp coretemp snd_hda_codec_realtek crct10dif_pclmul crc32_pclmul ghash_clmulni_intel snd_hda_codec_generic aesni_intel snd_hda_codec_hdmi aes_x86_64 lrw gf128mul snd_hda_intel glue_helper ablk_helper cryptd snd_hda_codec hid_multitouch joydev snd_hda_core binfmt_misc i2c_hid serio_raw snd_pcm acpi_pad snd_timer snd i2c_designware_platform 8250_dw nls_iso8859_1 i2c_designware_core lpc_ich mfd_core soundcore usbhid hid psmouse ahci libahci [ 44.235447] CPU: 2 PID: 2199 Comm: kworker/u8:8 Not tainted 4.8.0-rc5+ #130 [ 44.235447] Hardware name: Dell Inc. XPS 13 9343/0310JH, BIOS A07 11/11/2015 [ 44.235450] Workqueue: events_unbound async_run_entry_fn [ 44.235453] 0000000000000000 ffff8801b2f7fb98 ffffffff81306c2f ffff8801b2f7fbe8 [ 44.235454] 0000000000000000 ffff8801b2f7fbd8 ffffffff81056c01 00000a801f50ecc0 [ 44.235456] ffff88020ce50000 ffff88020ce59b60 ffffffff81a60b5c ffffffff81414840 [ 44.235456] Call Trace: [ 44.235459] [<ffffffff81306c2f>] dump_stack+0x4d/0x6e [ 44.235461] [<ffffffff81056c01>] __warn+0xd1/0xf0 [ 44.235464] [<ffffffff81414840>] ? i915_pm_suspend_late+0x30/0x30 [ 44.235465] [<ffffffff81056c6f>] warn_slowpath_fmt+0x4f/0x60 [ 44.235468] [<ffffffff814e73ce>] ? pm_runtime_get_if_in_use+0x6e/0xa0 [ 44.235469] [<ffffffff81433526>] intel_runtime_pm_get_if_in_use+0xe6/0xf0 [ 44.235471] [<ffffffff81458a26>] i915_gem_shrink+0x306/0x360 [ 44.235473] [<ffffffff81343fd4>] ? pci_platform_power_transition+0x24/0x90 [ 44.235475] [<ffffffff81414840>] ? i915_pm_suspend_late+0x30/0x30 [ 44.235476] [<ffffffff81458dfb>] i915_gem_shrink_all+0x1b/0x30 [ 44.235478] [<ffffffff814560b3>] i915_gem_freeze_late+0x33/0x90 [ 44.235479] [<ffffffff81414877>] i915_pm_freeze_late+0x37/0x40 [ 44.235481] [<ffffffff814e9b8e>] dpm_run_callback+0x4e/0x130 [ 44.235483] [<ffffffff814ea5db>] __device_suspend_late+0xdb/0x1f0 [ 44.235484] [<ffffffff814ea70f>] async_suspend_late+0x1f/0xa0 [ 44.235486] [<ffffffff81077557>] async_run_entry_fn+0x37/0x150 [ 44.235488] [<ffffffff8106f518>] process_one_work+0x148/0x3f0 [ 44.235490] [<ffffffff8106f8eb>] worker_thread+0x12b/0x490 [ 44.235491] [<ffffffff8106f7c0>] ? process_one_work+0x3f0/0x3f0 [ 44.235492] [<ffffffff81074d09>] kthread+0xc9/0xe0 [ 44.235495] [<ffffffff816e257f>] ret_from_fork+0x1f/0x40 [ 44.235496] [<ffffffff81074c40>] ? kthread_park+0x60/0x60 [ 44.235497] ---[ end trace e438706b97c7f132 ]--- Alternatively, to actually shrink everything we have to do so slightly earlier in the hibernation process. To keep lockdep silent, we need to take struct_mutex for the shrinker even though we know that we are the only user during the freeze. Fixes: 7aab2d534e35 ("drm/i915: Shrink objects prior to hibernation") Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/20160921135108.29574-2-chris@chris-wilson.co.uk
2016-09-21 07:51:07 -06:00
int i915_gem_freeze(struct drm_i915_private *dev_priv);
int i915_gem_freeze_late(struct drm_i915_private *dev_priv);
void *i915_gem_object_alloc(struct drm_i915_private *dev_priv);
void i915_gem_object_free(struct drm_i915_gem_object *obj);
void i915_gem_object_init(struct drm_i915_gem_object *obj,
const struct drm_i915_gem_object_ops *ops);
struct drm_i915_gem_object *
i915_gem_object_create(struct drm_i915_private *dev_priv, u64 size);
struct drm_i915_gem_object *
i915_gem_object_create_from_data(struct drm_i915_private *dev_priv,
const void *data, size_t size);
void i915_gem_close_object(struct drm_gem_object *gem, struct drm_file *file);
void i915_gem_free_object(struct drm_gem_object *obj);
static inline void i915_gem_drain_freed_objects(struct drm_i915_private *i915)
{
/* A single pass should suffice to release all the freed objects (along
* most call paths) , but be a little more paranoid in that freeing
* the objects does take a little amount of time, during which the rcu
* callbacks could have added new objects into the freed list, and
* armed the work again.
*/
do {
rcu_barrier();
} while (flush_work(&i915->mm.free_work));
}
static inline void i915_gem_drain_workqueue(struct drm_i915_private *i915)
{
/*
* Similar to objects above (see i915_gem_drain_freed-objects), in
* general we have workers that are armed by RCU and then rearm
* themselves in their callbacks. To be paranoid, we need to
* drain the workqueue a second time after waiting for the RCU
* grace period so that we catch work queued via RCU from the first
* pass. As neither drain_workqueue() nor flush_workqueue() report
* a result, we make an assumption that we only don't require more
* than 2 passes to catch all recursive RCU delayed work.
*
*/
int pass = 2;
do {
rcu_barrier();
drain_workqueue(i915->wq);
} while (--pass);
}
struct i915_vma * __must_check
i915_gem_object_ggtt_pin(struct drm_i915_gem_object *obj,
const struct i915_ggtt_view *view,
u64 size,
u64 alignment,
u64 flags);
drm/i915: Infrastructure for supporting different GGTT views per object Things like reliable GGTT mappings and mirrored 2d-on-3d display will need to map objects into the same address space multiple times. Added a GGTT view concept and linked it with the VMA to distinguish between multiple instances per address space. New objects and GEM functions which do not take this new view as a parameter assume the default of zero (I915_GGTT_VIEW_NORMAL) which preserves the previous behaviour. This now means that objects can have multiple VMA entries so the code which assumed there will only be one also had to be modified. Alternative GGTT views are supposed to borrow DMA addresses from obj->pages which is DMA mapped on first VMA instantiation and unmapped on the last one going away. v2: * Removed per view special casing in i915_gem_ggtt_prepare / finish_object in favour of creating and destroying DMA mappings on first VMA instantiation and last VMA destruction. (Daniel Vetter) * Simplified i915_vma_unbind which does not need to count the GGTT views. (Daniel Vetter) * Also moved obj->map_and_fenceable reset under the same check. * Checkpatch cleanups. v3: * Only retire objects once the last VMA is unbound. v4: * Keep scatter-gather table for alternative views persistent for the lifetime of the VMA. * Propagate binding errors to callers and handle appropriately. v5: * Explicitly look for normal GGTT view in i915_gem_obj_bound to align usage in i915_gem_object_ggtt_unpin. (Michel Thierry) * Change to single if statement in i915_gem_obj_to_ggtt. (Michel Thierry) * Removed stray semi-colon in i915_gem_object_set_cache_level. For: VIZ-4544 Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Michel Thierry <michel.thierry@intel.com> [danvet: Drop hunk from i915_gem_shrink since it's just prettification but upsets a __must_check warning.] Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2014-12-10 10:27:58 -07:00
int i915_gem_object_unbind(struct drm_i915_gem_object *obj);
void i915_gem_release_mmap(struct drm_i915_gem_object *obj);
void i915_gem_runtime_suspend(struct drm_i915_private *dev_priv);
static inline int __sg_page_count(const struct scatterlist *sg)
{
return sg->length >> PAGE_SHIFT;
}
struct scatterlist *
i915_gem_object_get_sg(struct drm_i915_gem_object *obj,
unsigned int n, unsigned int *offset);
struct page *
i915_gem_object_get_page(struct drm_i915_gem_object *obj,
unsigned int n);
struct page *
i915_gem_object_get_dirty_page(struct drm_i915_gem_object *obj,
unsigned int n);
dma_addr_t
i915_gem_object_get_dma_address(struct drm_i915_gem_object *obj,
unsigned long n);
void __i915_gem_object_set_pages(struct drm_i915_gem_object *obj,
struct sg_table *pages);
int __i915_gem_object_get_pages(struct drm_i915_gem_object *obj);
static inline int __must_check
i915_gem_object_pin_pages(struct drm_i915_gem_object *obj)
{
might_lock(&obj->mm.lock);
if (atomic_inc_not_zero(&obj->mm.pages_pin_count))
return 0;
return __i915_gem_object_get_pages(obj);
}
static inline void
__i915_gem_object_pin_pages(struct drm_i915_gem_object *obj)
{
GEM_BUG_ON(!obj->mm.pages);
atomic_inc(&obj->mm.pages_pin_count);
}
static inline bool
i915_gem_object_has_pinned_pages(struct drm_i915_gem_object *obj)
{
return atomic_read(&obj->mm.pages_pin_count);
}
static inline void
__i915_gem_object_unpin_pages(struct drm_i915_gem_object *obj)
{
GEM_BUG_ON(!i915_gem_object_has_pinned_pages(obj));
GEM_BUG_ON(!obj->mm.pages);
atomic_dec(&obj->mm.pages_pin_count);
}
static inline void
i915_gem_object_unpin_pages(struct drm_i915_gem_object *obj)
{
__i915_gem_object_unpin_pages(obj);
}
enum i915_mm_subclass { /* lockdep subclass for obj->mm.lock */
I915_MM_NORMAL = 0,
I915_MM_SHRINKER
};
void __i915_gem_object_put_pages(struct drm_i915_gem_object *obj,
enum i915_mm_subclass subclass);
void __i915_gem_object_invalidate(struct drm_i915_gem_object *obj);
drm/i915: Support for creating write combined type vmaps vmaps has a provision for controlling the page protection bits, with which we can use to control the mapping type, e.g. WB, WC, UC or even WT. To allow the caller to choose their mapping type, we add a parameter to i915_gem_object_pin_map - but we still only allow one vmap to be cached per object. If the object is currently not pinned, then we recreate the previous vmap with the new access type, but if it was pinned we report an error. This effectively limits the access via i915_gem_object_pin_map to a single mapping type for the lifetime of the object. Not usually a problem, but something to be aware of when setting up the object's vmap. We will want to vary the access type to enable WC mappings of ringbuffer and context objects on !llc platforms, as well as other objects where we need coherent access to the GPU's pages without going through the GTT v2: Remove the redundant braces around pin count check and fix the marker in documentation (Chris) v3: - Add a new enum for the vmalloc mapping type & pass that as an argument to i915_object_pin_map. (Tvrtko) - Use PAGE_MASK to extract or filter the mapping type info and remove a superfluous BUG_ON.(Tvrtko) v4: - Rename the enums and clean up the pin_map function. (Chris) v5: Drop the VM_NO_GUARD, minor cosmetics. Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Akash Goel <akash.goel@intel.com> Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1471001999-17787-1-git-send-email-chris@chris-wilson.co.uk
2016-08-12 05:39:58 -06:00
enum i915_map_type {
I915_MAP_WB = 0,
I915_MAP_WC,
};
/**
* i915_gem_object_pin_map - return a contiguous mapping of the entire object
* @obj: the object to map into kernel address space
* @type: the type of mapping, used to select pgprot_t
*
* Calls i915_gem_object_pin_pages() to prevent reaping of the object's
* pages and then returns a contiguous mapping of the backing storage into
drm/i915: Support for creating write combined type vmaps vmaps has a provision for controlling the page protection bits, with which we can use to control the mapping type, e.g. WB, WC, UC or even WT. To allow the caller to choose their mapping type, we add a parameter to i915_gem_object_pin_map - but we still only allow one vmap to be cached per object. If the object is currently not pinned, then we recreate the previous vmap with the new access type, but if it was pinned we report an error. This effectively limits the access via i915_gem_object_pin_map to a single mapping type for the lifetime of the object. Not usually a problem, but something to be aware of when setting up the object's vmap. We will want to vary the access type to enable WC mappings of ringbuffer and context objects on !llc platforms, as well as other objects where we need coherent access to the GPU's pages without going through the GTT v2: Remove the redundant braces around pin count check and fix the marker in documentation (Chris) v3: - Add a new enum for the vmalloc mapping type & pass that as an argument to i915_object_pin_map. (Tvrtko) - Use PAGE_MASK to extract or filter the mapping type info and remove a superfluous BUG_ON.(Tvrtko) v4: - Rename the enums and clean up the pin_map function. (Chris) v5: Drop the VM_NO_GUARD, minor cosmetics. Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Akash Goel <akash.goel@intel.com> Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1471001999-17787-1-git-send-email-chris@chris-wilson.co.uk
2016-08-12 05:39:58 -06:00
* the kernel address space. Based on the @type of mapping, the PTE will be
* set to either WriteBack or WriteCombine (via pgprot_t).
*
* The caller is responsible for calling i915_gem_object_unpin_map() when the
* mapping is no longer required.
*
* Returns the pointer through which to access the mapped object, or an
* ERR_PTR() on error.
*/
drm/i915: Support for creating write combined type vmaps vmaps has a provision for controlling the page protection bits, with which we can use to control the mapping type, e.g. WB, WC, UC or even WT. To allow the caller to choose their mapping type, we add a parameter to i915_gem_object_pin_map - but we still only allow one vmap to be cached per object. If the object is currently not pinned, then we recreate the previous vmap with the new access type, but if it was pinned we report an error. This effectively limits the access via i915_gem_object_pin_map to a single mapping type for the lifetime of the object. Not usually a problem, but something to be aware of when setting up the object's vmap. We will want to vary the access type to enable WC mappings of ringbuffer and context objects on !llc platforms, as well as other objects where we need coherent access to the GPU's pages without going through the GTT v2: Remove the redundant braces around pin count check and fix the marker in documentation (Chris) v3: - Add a new enum for the vmalloc mapping type & pass that as an argument to i915_object_pin_map. (Tvrtko) - Use PAGE_MASK to extract or filter the mapping type info and remove a superfluous BUG_ON.(Tvrtko) v4: - Rename the enums and clean up the pin_map function. (Chris) v5: Drop the VM_NO_GUARD, minor cosmetics. Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Akash Goel <akash.goel@intel.com> Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1471001999-17787-1-git-send-email-chris@chris-wilson.co.uk
2016-08-12 05:39:58 -06:00
void *__must_check i915_gem_object_pin_map(struct drm_i915_gem_object *obj,
enum i915_map_type type);
/**
* i915_gem_object_unpin_map - releases an earlier mapping
* @obj: the object to unmap
*
* After pinning the object and mapping its pages, once you are finished
* with your access, call i915_gem_object_unpin_map() to release the pin
* upon the mapping. Once the pin count reaches zero, that mapping may be
* removed.
*/
static inline void i915_gem_object_unpin_map(struct drm_i915_gem_object *obj)
{
i915_gem_object_unpin_pages(obj);
}
int i915_gem_obj_prepare_shmem_read(struct drm_i915_gem_object *obj,
unsigned int *needs_clflush);
int i915_gem_obj_prepare_shmem_write(struct drm_i915_gem_object *obj,
unsigned int *needs_clflush);
#define CLFLUSH_BEFORE BIT(0)
#define CLFLUSH_AFTER BIT(1)
#define CLFLUSH_FLAGS (CLFLUSH_BEFORE | CLFLUSH_AFTER)
static inline void
i915_gem_obj_finish_shmem_access(struct drm_i915_gem_object *obj)
{
i915_gem_object_unpin_pages(obj);
}
int __must_check i915_mutex_lock_interruptible(struct drm_device *dev);
void i915_vma_move_to_active(struct i915_vma *vma,
struct drm_i915_gem_request *req,
unsigned int flags);
int i915_gem_dumb_create(struct drm_file *file_priv,
struct drm_device *dev,
struct drm_mode_create_dumb *args);
int i915_gem_mmap_gtt(struct drm_file *file_priv, struct drm_device *dev,
uint32_t handle, uint64_t *offset);
int i915_gem_mmap_gtt_version(void);
void i915_gem_track_fb(struct drm_i915_gem_object *old,
struct drm_i915_gem_object *new,
unsigned frontbuffer_bits);
int __must_check i915_gem_set_global_seqno(struct drm_device *dev, u32 seqno);
drm/i915: Rely on accurate request tracking for finding hung batches In the past, it was possible to have multiple batches per request due to a stray signal or ENOMEM. As a result we had to scan each active object (filtered by those having the COMMAND domain) for the one that contained the ACTHD pointer. This was then made more complicated by the introduction of ppgtt, whereby ACTHD then pointed into the address space of the context and so also needed to be taken into account. This is a fairly robust approach (though the implementation is a little fragile and depends upon the per-generation setup, registers and parameters). However, due to the requirements for hangstats, we needed a robust method for associating batches with a particular request and having that we can rely upon it for finding the associated batch object for error capture. If the batch buffer tracking is not robust enough, that should become apparent quite quickly through an erroneous error capture. That should also help to make sure that the runtime reporting to userspace is robust. It also means that we then report the oldest incomplete batch on each ring, which can be useful for determining the state of userspace at the time of a hang. v2: Use i915_gem_find_active_request (Mika) v3: remove check for ring->get_seqno, split long lines (Ben) v4: check that context is available (Chris) checkpatch warnings fixed Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> (v1) Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com> (v3) Cc: Ben Widawsky <benjamin.widawsky@intel.com> Reviewed-by: Ben Widawsky <ben@bwidawsk.net> (v3) Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2014-02-25 08:11:23 -07:00
struct drm_i915_gem_request *
i915_gem_find_active_request(struct intel_engine_cs *engine);
drm/i915: Rely on accurate request tracking for finding hung batches In the past, it was possible to have multiple batches per request due to a stray signal or ENOMEM. As a result we had to scan each active object (filtered by those having the COMMAND domain) for the one that contained the ACTHD pointer. This was then made more complicated by the introduction of ppgtt, whereby ACTHD then pointed into the address space of the context and so also needed to be taken into account. This is a fairly robust approach (though the implementation is a little fragile and depends upon the per-generation setup, registers and parameters). However, due to the requirements for hangstats, we needed a robust method for associating batches with a particular request and having that we can rely upon it for finding the associated batch object for error capture. If the batch buffer tracking is not robust enough, that should become apparent quite quickly through an erroneous error capture. That should also help to make sure that the runtime reporting to userspace is robust. It also means that we then report the oldest incomplete batch on each ring, which can be useful for determining the state of userspace at the time of a hang. v2: Use i915_gem_find_active_request (Mika) v3: remove check for ring->get_seqno, split long lines (Ben) v4: check that context is available (Chris) checkpatch warnings fixed Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> (v1) Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com> (v3) Cc: Ben Widawsky <benjamin.widawsky@intel.com> Reviewed-by: Ben Widawsky <ben@bwidawsk.net> (v3) Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2014-02-25 08:11:23 -07:00
void i915_gem_retire_requests(struct drm_i915_private *dev_priv);
drm/i915: Replaced Blitter ring based flips with MMIO flips This patch enables the framework for using MMIO based flip calls, in contrast with the CS based flip calls which are being used currently. MMIO based flip calls can be enabled on architectures where Render and Blitter engines reside in different power wells. The decision to use MMIO flips can be made based on workloads to give 100% residency for Media power well. v2: The MMIO flips now use the interrupt driven mechanism for issuing the flips when target seqno is reached. (Incorporating Ville's idea) v3: Rebasing on latest code. Code restructuring after incorporating Damien's comments v4: Addressing Ville's review comments -general cleanup -updating only base addr instead of calling update_primary_plane -extending patch for gen5+ platforms v5: Addressed Ville's review comments -Making mmio flip vs cs flip selection based on module parameter -Adding check for DRIVER_MODESET feature in notify_ring before calling notify mmio flip. -Other changes mostly in function arguments v6: -Having a seperate function to check condition for using mmio flips (Ville) -propogating error code from i915_gem_check_olr (Ville) v7: -Adding __must_check with i915_gem_check_olr (Chris) -Renaming mmio_flip_data to mmio_flip (Chris) -Rebasing on latest nightly v8: -Rebasing on latest code -squash 3rd patch in series(mmio setbase vs page flip race) with this patch -Added new tiling mode update in intel_do_mmio_flip (Chris) v9: -check for obj->last_write_seqno being 0 instead of obj->ring being NULL in intel_postpone_flip, as this is a more restrictive condition (Chris) v10: -Applied Chris's suggestions for squashing patches 2,3 into this patch. These patches make the selection of CS vs MMIO flip at the page flip time, and make the module parameter for using mmio flips as tristate, the states being 'force CS flips', 'force mmio flips', 'driver discretion'. Changed the logic for driver discretion (Chris) v11: Minor code cleanup(better readability, fixing whitespace errors, using lockdep to check mutex locked status in postpone_flip, removal of __must_check in function definition) (Chris) Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Sourab Gupta <sourab.gupta@intel.com> Signed-off-by: Akash Goel <akash.goel@intel.com> Tested-by: Chris Wilson <chris@chris-wilson.co.uk> # snb, ivb [danvet: Fix up parameter alignement checkpatch spotted.] Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2014-06-02 05:17:17 -06:00
static inline bool i915_reset_backoff(struct i915_gpu_error *error)
{
return unlikely(test_bit(I915_RESET_BACKOFF, &error->flags));
}
static inline bool i915_reset_handoff(struct i915_gpu_error *error)
drm/i915: clear up wedged transitions We have two important transitions of the wedged state in the current code: - 0 -> 1: This means a hang has been detected, and signals to everyone that they please get of any locks, so that the reset work item can do its job. - 1 -> 0: The reset handler has completed. Now the last transition mixes up two states: "Reset completed and successful" and "Reset failed". To distinguish these two we do some tricks with the reset completion, but I simply could not convince myself that this doesn't race under odd circumstances. Hence split this up, and add a new terminal state indicating that the hw is gone for good. Also add explicit #defines for both states, update comments. v2: Split out the reset handling bugfix for the throttle ioctl. v3: s/tmp/wedged/ sugested by Chris Wilson. Also fixup up a rebase error which prevented this patch from actually compiling. v4: To unify the wedged state with the reset counter, keep the reset-in-progress state just as a flag. The terminally-wedged state is now denoted with a big number. v5: Add a comment to the reset_counter special values explaining that WEDGED & RESET_IN_PROGRESS needs to be true for the code to be correct. v6: Fixup logic errors introduced with the wedged+reset_counter unification. Since WEDGED implies reset-in-progress (in a way we're terminally stuck in the dead-but-reset-not-completed state), we need ensure that we check for this everywhere. The specific bug was in wait_for_error, which would simply have timed out. v7: Extract an inline i915_reset_in_progress helper to make the code more readable. Also annote the reset-in-progress case with an unlikely, to help the compiler optimize the fastpath. Do the same for the terminally wedged case with i915_terminally_wedged. Reviewed-by: Damien Lespiau <damien.lespiau@intel.com> Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2012-11-15 09:17:22 -07:00
{
return unlikely(test_bit(I915_RESET_HANDOFF, &error->flags));
}
static inline bool i915_terminally_wedged(struct i915_gpu_error *error)
{
return unlikely(test_bit(I915_WEDGED, &error->flags));
drm/i915: clear up wedged transitions We have two important transitions of the wedged state in the current code: - 0 -> 1: This means a hang has been detected, and signals to everyone that they please get of any locks, so that the reset work item can do its job. - 1 -> 0: The reset handler has completed. Now the last transition mixes up two states: "Reset completed and successful" and "Reset failed". To distinguish these two we do some tricks with the reset completion, but I simply could not convince myself that this doesn't race under odd circumstances. Hence split this up, and add a new terminal state indicating that the hw is gone for good. Also add explicit #defines for both states, update comments. v2: Split out the reset handling bugfix for the throttle ioctl. v3: s/tmp/wedged/ sugested by Chris Wilson. Also fixup up a rebase error which prevented this patch from actually compiling. v4: To unify the wedged state with the reset counter, keep the reset-in-progress state just as a flag. The terminally-wedged state is now denoted with a big number. v5: Add a comment to the reset_counter special values explaining that WEDGED & RESET_IN_PROGRESS needs to be true for the code to be correct. v6: Fixup logic errors introduced with the wedged+reset_counter unification. Since WEDGED implies reset-in-progress (in a way we're terminally stuck in the dead-but-reset-not-completed state), we need ensure that we check for this everywhere. The specific bug was in wait_for_error, which would simply have timed out. v7: Extract an inline i915_reset_in_progress helper to make the code more readable. Also annote the reset-in-progress case with an unlikely, to help the compiler optimize the fastpath. Do the same for the terminally wedged case with i915_terminally_wedged. Reviewed-by: Damien Lespiau <damien.lespiau@intel.com> Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2012-11-15 09:17:22 -07:00
}
static inline bool i915_reset_backoff_or_wedged(struct i915_gpu_error *error)
drm/i915: clear up wedged transitions We have two important transitions of the wedged state in the current code: - 0 -> 1: This means a hang has been detected, and signals to everyone that they please get of any locks, so that the reset work item can do its job. - 1 -> 0: The reset handler has completed. Now the last transition mixes up two states: "Reset completed and successful" and "Reset failed". To distinguish these two we do some tricks with the reset completion, but I simply could not convince myself that this doesn't race under odd circumstances. Hence split this up, and add a new terminal state indicating that the hw is gone for good. Also add explicit #defines for both states, update comments. v2: Split out the reset handling bugfix for the throttle ioctl. v3: s/tmp/wedged/ sugested by Chris Wilson. Also fixup up a rebase error which prevented this patch from actually compiling. v4: To unify the wedged state with the reset counter, keep the reset-in-progress state just as a flag. The terminally-wedged state is now denoted with a big number. v5: Add a comment to the reset_counter special values explaining that WEDGED & RESET_IN_PROGRESS needs to be true for the code to be correct. v6: Fixup logic errors introduced with the wedged+reset_counter unification. Since WEDGED implies reset-in-progress (in a way we're terminally stuck in the dead-but-reset-not-completed state), we need ensure that we check for this everywhere. The specific bug was in wait_for_error, which would simply have timed out. v7: Extract an inline i915_reset_in_progress helper to make the code more readable. Also annote the reset-in-progress case with an unlikely, to help the compiler optimize the fastpath. Do the same for the terminally wedged case with i915_terminally_wedged. Reviewed-by: Damien Lespiau <damien.lespiau@intel.com> Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2012-11-15 09:17:22 -07:00
{
return i915_reset_backoff(error) | i915_terminally_wedged(error);
}
static inline u32 i915_reset_count(struct i915_gpu_error *error)
{
return READ_ONCE(error->reset_count);
drm/i915: clear up wedged transitions We have two important transitions of the wedged state in the current code: - 0 -> 1: This means a hang has been detected, and signals to everyone that they please get of any locks, so that the reset work item can do its job. - 1 -> 0: The reset handler has completed. Now the last transition mixes up two states: "Reset completed and successful" and "Reset failed". To distinguish these two we do some tricks with the reset completion, but I simply could not convince myself that this doesn't race under odd circumstances. Hence split this up, and add a new terminal state indicating that the hw is gone for good. Also add explicit #defines for both states, update comments. v2: Split out the reset handling bugfix for the throttle ioctl. v3: s/tmp/wedged/ sugested by Chris Wilson. Also fixup up a rebase error which prevented this patch from actually compiling. v4: To unify the wedged state with the reset counter, keep the reset-in-progress state just as a flag. The terminally-wedged state is now denoted with a big number. v5: Add a comment to the reset_counter special values explaining that WEDGED & RESET_IN_PROGRESS needs to be true for the code to be correct. v6: Fixup logic errors introduced with the wedged+reset_counter unification. Since WEDGED implies reset-in-progress (in a way we're terminally stuck in the dead-but-reset-not-completed state), we need ensure that we check for this everywhere. The specific bug was in wait_for_error, which would simply have timed out. v7: Extract an inline i915_reset_in_progress helper to make the code more readable. Also annote the reset-in-progress case with an unlikely, to help the compiler optimize the fastpath. Do the same for the terminally wedged case with i915_terminally_wedged. Reviewed-by: Damien Lespiau <damien.lespiau@intel.com> Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2012-11-15 09:17:22 -07:00
}
drm/i915: Record the tail at each request and use it to estimate the head By recording the location of every request in the ringbuffer, we know that in order to retire the request the GPU must have finished reading it and so the GPU head is now beyond the tail of the request. We can therefore provide a conservative estimate of where the GPU is reading from in order to avoid having to read back the ring buffer registers when polling for space upon starting a new write into the ringbuffer. A secondary effect is that this allows us to convert intel_ring_buffer_wait() to use i915_wait_request() and so consolidate upon the single function to handle the complicated task of waiting upon the GPU. A necessary precaution is that we need to make that wait uninterruptible to match the existing conditions as all the callers of intel_ring_begin() have not been audited to handle ERESTARTSYS correctly. By using a conservative estimate for the head, and always processing all outstanding requests first, we prevent a race condition between using the estimate and direct reads of I915_RING_HEAD which could result in the value of the head going backwards, and the tail overflowing once again. We are also careful to mark any request that we skip over in order to free space in ring as consumed which provides a self-consistency check. Given sufficient abuse, such as a set of unthrottled GPU bound cairo-traces, avoiding the use of I915_RING_HEAD gives a 10-20% boost on Sandy Bridge (i5-2520m): firefox-paintball 18927ms -> 15646ms: 1.21x speedup firefox-fishtank 12563ms -> 11278ms: 1.11x speedup which is a mild consolation for the performance those traces achieved from exploiting the buggy autoreported head. v2: Add a few more comments and make request->tail a conservative estimate as suggested by Daniel Vetter. Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> [danvet: resolve conflicts with retirement defering and the lack of the autoreport head removal (that will go in through -fixes).] Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2012-02-15 04:25:36 -07:00
static inline u32 i915_reset_engine_count(struct i915_gpu_error *error,
struct intel_engine_cs *engine)
{
return READ_ONCE(error->reset_engine_count[engine->id]);
}
drm/i915: Add support for per engine reset recovery This change implements support for per-engine reset as an initial, less intrusive hang recovery option to be attempted before falling back to the legacy full GPU reset recovery mode if necessary. This is only supported from Gen8 onwards. Hangchecker determines which engines are hung and invokes error handler to recover from it. Error handler schedules recovery for each of those engines that are hung. The recovery procedure is as follows, - identifies the request that caused the hang and it is dropped - force engine to idle: this is done by issuing a reset request - reset the engine - re-init the engine to resume submissions. If engine reset fails then we fall back to heavy weight full gpu reset which resets all engines and reinitiazes complete state of HW and SW. v2: Rebase. v3: s/*engine_reset*/*reset_engine*/; freeze engine and irqs before calling i915_gem_reset_engine (Chris). v4: Rebase, modify i915_gem_reset_prepare to use a ring mask and reuse the function for reset_engine. v5: intel_reset_engine_start/cancel instead of request/unrequest_reset. v6: Clean up reset_engine function to not require mutex, i.e. no need to call revoke/restore_fences and _retire_requests (Chris). v7: Remove leftovers from v5, i.e. no need to disable irq, hold forcewake or wakeup the handoff bit (Chris). v8: engine_retire_requests should be (and it was) static; explain that we have to re-init the engine after reset, which is why the init_hw call is needed; check reset-in-progress flag (Chris). v9: Rebase, include code to pass the active request to gem_reset_engine (as it is already done in full reset). Remove unnecessary intel_reset_engine_start/cancel, these are executed as part of the reset. v10: Rebase, use the right I915_RESET_ENGINE flag. v11: Fixup to call reset_finish_engine even on error. Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com> Signed-off-by: Tomas Elf <tomas.elf@intel.com> Signed-off-by: Arun Siluvery <arun.siluvery@linux.intel.com> Signed-off-by: Michel Thierry <michel.thierry@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/20170615201828.23144-6-michel.thierry@intel.com Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Link: http://patchwork.freedesktop.org/patch/msgid/20170620095751.13127-6-chris@chris-wilson.co.uk
2017-06-20 03:57:47 -06:00
struct drm_i915_gem_request *
i915_gem_reset_prepare_engine(struct intel_engine_cs *engine);
int i915_gem_reset_prepare(struct drm_i915_private *dev_priv);
void i915_gem_reset(struct drm_i915_private *dev_priv);
drm/i915: Add support for per engine reset recovery This change implements support for per-engine reset as an initial, less intrusive hang recovery option to be attempted before falling back to the legacy full GPU reset recovery mode if necessary. This is only supported from Gen8 onwards. Hangchecker determines which engines are hung and invokes error handler to recover from it. Error handler schedules recovery for each of those engines that are hung. The recovery procedure is as follows, - identifies the request that caused the hang and it is dropped - force engine to idle: this is done by issuing a reset request - reset the engine - re-init the engine to resume submissions. If engine reset fails then we fall back to heavy weight full gpu reset which resets all engines and reinitiazes complete state of HW and SW. v2: Rebase. v3: s/*engine_reset*/*reset_engine*/; freeze engine and irqs before calling i915_gem_reset_engine (Chris). v4: Rebase, modify i915_gem_reset_prepare to use a ring mask and reuse the function for reset_engine. v5: intel_reset_engine_start/cancel instead of request/unrequest_reset. v6: Clean up reset_engine function to not require mutex, i.e. no need to call revoke/restore_fences and _retire_requests (Chris). v7: Remove leftovers from v5, i.e. no need to disable irq, hold forcewake or wakeup the handoff bit (Chris). v8: engine_retire_requests should be (and it was) static; explain that we have to re-init the engine after reset, which is why the init_hw call is needed; check reset-in-progress flag (Chris). v9: Rebase, include code to pass the active request to gem_reset_engine (as it is already done in full reset). Remove unnecessary intel_reset_engine_start/cancel, these are executed as part of the reset. v10: Rebase, use the right I915_RESET_ENGINE flag. v11: Fixup to call reset_finish_engine even on error. Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com> Signed-off-by: Tomas Elf <tomas.elf@intel.com> Signed-off-by: Arun Siluvery <arun.siluvery@linux.intel.com> Signed-off-by: Michel Thierry <michel.thierry@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/20170615201828.23144-6-michel.thierry@intel.com Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Link: http://patchwork.freedesktop.org/patch/msgid/20170620095751.13127-6-chris@chris-wilson.co.uk
2017-06-20 03:57:47 -06:00
void i915_gem_reset_finish_engine(struct intel_engine_cs *engine);
void i915_gem_reset_finish(struct drm_i915_private *dev_priv);
drm/i915: Update reset path to fix incomplete requests Update reset path in preparation for engine reset which requires identification of incomplete requests and associated context and fixing their state so that engine can resume correctly after reset. The request that caused the hang will be skipped and head is reset to the start of breadcrumb. This allows us to resume from where we left-off. Since this request didn't complete normally we also need to cleanup elsp queue manually. This is vital if we employ nonblocking request submission where we may have a web of dependencies upon the hung request and so advancing the seqno manually is no longer trivial. ABI: gem_reset_stats / DRM_IOCTL_I915_GET_RESET_STATS We change the way we count pending batches. Only the active context involved in the reset is marked as either innocent or guilty, and not mark the entire world as pending. By inspection this only affects igt/gem_reset_stats (which assumes implementation details) and not piglit. ARB_robustness gives this guide on how we expect the user of this interface to behave: * Provide a mechanism for an OpenGL application to learn about graphics resets that affect the context. When a graphics reset occurs, the OpenGL context becomes unusable and the application must create a new context to continue operation. Detecting a graphics reset happens through an inexpensive query. And with regards to the actual meaning of the reset values: Certain events can result in a reset of the GL context. Such a reset causes all context state to be lost. Recovery from such events requires recreation of all objects in the affected context. The current status of the graphics reset state is returned by enum GetGraphicsResetStatusARB(); The symbolic constant returned indicates if the GL context has been in a reset state at any point since the last call to GetGraphicsResetStatusARB. NO_ERROR indicates that the GL context has not been in a reset state since the last call. GUILTY_CONTEXT_RESET_ARB indicates that a reset has been detected that is attributable to the current GL context. INNOCENT_CONTEXT_RESET_ARB indicates a reset has been detected that is not attributable to the current GL context. UNKNOWN_CONTEXT_RESET_ARB indicates a detected graphics reset whose cause is unknown. The language here is explicit in that we must mark up the guilty batch, but is loose enough for us to relax the innocent (i.e. pending) accounting as only the active batches are involved with the reset. In the future, we are looking towards single engine resetting (with minimal locking), where it seems inappropriate to mark the entire world as innocent since the reset occurred on a different engine. Reducing the information available means we only have to encounter the pain once, and also reduces the information leaking from one context to another. v2: Legacy ringbuffer submission required a reset following hibernation, or else we restore stale values to the RING_HEAD and walked over stolen garbage. v3: GuC requires replaying the requests after a reset. v4: Restore engine IRQ after reset (so waiters will be woken!) Rearm hangcheck if resetting with a waiter. Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Cc: Mika Kuoppala <mika.kuoppala@intel.com> Cc: Arun Siluvery <arun.siluvery@linux.intel.com> Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/20160909131201.16673-13-chris@chris-wilson.co.uk
2016-09-09 07:11:53 -06:00
void i915_gem_set_wedged(struct drm_i915_private *dev_priv);
bool i915_gem_unset_wedged(struct drm_i915_private *dev_priv);
drm/i915: Add support for per engine reset recovery This change implements support for per-engine reset as an initial, less intrusive hang recovery option to be attempted before falling back to the legacy full GPU reset recovery mode if necessary. This is only supported from Gen8 onwards. Hangchecker determines which engines are hung and invokes error handler to recover from it. Error handler schedules recovery for each of those engines that are hung. The recovery procedure is as follows, - identifies the request that caused the hang and it is dropped - force engine to idle: this is done by issuing a reset request - reset the engine - re-init the engine to resume submissions. If engine reset fails then we fall back to heavy weight full gpu reset which resets all engines and reinitiazes complete state of HW and SW. v2: Rebase. v3: s/*engine_reset*/*reset_engine*/; freeze engine and irqs before calling i915_gem_reset_engine (Chris). v4: Rebase, modify i915_gem_reset_prepare to use a ring mask and reuse the function for reset_engine. v5: intel_reset_engine_start/cancel instead of request/unrequest_reset. v6: Clean up reset_engine function to not require mutex, i.e. no need to call revoke/restore_fences and _retire_requests (Chris). v7: Remove leftovers from v5, i.e. no need to disable irq, hold forcewake or wakeup the handoff bit (Chris). v8: engine_retire_requests should be (and it was) static; explain that we have to re-init the engine after reset, which is why the init_hw call is needed; check reset-in-progress flag (Chris). v9: Rebase, include code to pass the active request to gem_reset_engine (as it is already done in full reset). Remove unnecessary intel_reset_engine_start/cancel, these are executed as part of the reset. v10: Rebase, use the right I915_RESET_ENGINE flag. v11: Fixup to call reset_finish_engine even on error. Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com> Signed-off-by: Tomas Elf <tomas.elf@intel.com> Signed-off-by: Arun Siluvery <arun.siluvery@linux.intel.com> Signed-off-by: Michel Thierry <michel.thierry@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/20170615201828.23144-6-michel.thierry@intel.com Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Link: http://patchwork.freedesktop.org/patch/msgid/20170620095751.13127-6-chris@chris-wilson.co.uk
2017-06-20 03:57:47 -06:00
void i915_gem_reset_engine(struct intel_engine_cs *engine,
struct drm_i915_gem_request *request);
void i915_gem_init_mmio(struct drm_i915_private *i915);
int __must_check i915_gem_init(struct drm_i915_private *dev_priv);
int __must_check i915_gem_init_hw(struct drm_i915_private *dev_priv);
void i915_gem_init_swizzling(struct drm_i915_private *dev_priv);
void i915_gem_cleanup_engines(struct drm_i915_private *dev_priv);
int i915_gem_wait_for_idle(struct drm_i915_private *dev_priv,
unsigned int flags);
int __must_check i915_gem_suspend(struct drm_i915_private *dev_priv);
void i915_gem_resume(struct drm_i915_private *dev_priv);
int i915_gem_fault(struct vm_fault *vmf);
int i915_gem_object_wait(struct drm_i915_gem_object *obj,
unsigned int flags,
long timeout,
struct intel_rps_client *rps);
int i915_gem_object_wait_priority(struct drm_i915_gem_object *obj,
unsigned int flags,
int priority);
#define I915_PRIORITY_DISPLAY I915_PRIORITY_MAX
int __must_check
i915_gem_object_set_to_wc_domain(struct drm_i915_gem_object *obj, bool write);
int __must_check
i915_gem_object_set_to_gtt_domain(struct drm_i915_gem_object *obj, bool write);
int __must_check
i915_gem_object_set_to_cpu_domain(struct drm_i915_gem_object *obj, bool write);
struct i915_vma * __must_check
i915_gem_object_pin_to_display_plane(struct drm_i915_gem_object *obj,
u32 alignment,
const struct i915_ggtt_view *view);
void i915_gem_object_unpin_from_display_plane(struct i915_vma *vma);
int i915_gem_object_attach_phys(struct drm_i915_gem_object *obj,
int align);
int i915_gem_open(struct drm_i915_private *i915, struct drm_file *file);
void i915_gem_release(struct drm_device *dev, struct drm_file *file);
int i915_gem_object_set_cache_level(struct drm_i915_gem_object *obj,
enum i915_cache_level cache_level);
i915: add dmabuf/prime buffer sharing support. This adds handle->fd and fd->handle support to i915, this is to allow for offloading of rendering in one direction and outputs in the other. v2 from Daniel Vetter: - fixup conflicts with the prepare/finish gtt prep work. - implement ppgtt binding support. Note that we have squat i-g-t testcoverage for any of the lifetime and access rules dma_buf/prime support brings along. And there are quite a few intricate situations here. Also note that the integration with the existing code is a bit hackish, especially around get_gtt_pages and put_gtt_pages. It imo would be easier with the prep code from Chris Wilson's unbound series, but that is for 3.6. Also note that I didn't bother to put the new prepare/finish gtt hooks to good use by moving the dma_buf_map/unmap_attachment calls in there (like we've originally planned for). Last but not least this patch is only compile-tested, but I've changed very little compared to Dave Airlie's version. So there's a decent chance v2 on drm-next works as well as v1 on 3.4-rc. v3: Right when I've hit sent I've noticed that I've screwed up one obj->sg_list (for dmar support) and obj->sg_table (for prime support) disdinction. We should be able to merge these 2 paths, but that's material for another patch. v4: fix the error reporting bugs pointed out by ickle. v5: fix another error, and stop non-gtt mmaps on shared objects stop pread/pwrite on imported objects, add fake kmap Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2012-05-10 07:25:09 -06:00
struct drm_gem_object *i915_gem_prime_import(struct drm_device *dev,
struct dma_buf *dma_buf);
struct dma_buf *i915_gem_prime_export(struct drm_device *dev,
struct drm_gem_object *gem_obj, int flags);
static inline struct i915_hw_ppgtt *
i915_vm_to_ppgtt(struct i915_address_space *vm)
{
return container_of(vm, struct i915_hw_ppgtt, base);
}
/* i915_gem_fence_reg.c */
int __must_check i915_vma_get_fence(struct i915_vma *vma);
int __must_check i915_vma_put_fence(struct i915_vma *vma);
void i915_gem_revoke_fences(struct drm_i915_private *dev_priv);
void i915_gem_restore_fences(struct drm_i915_private *dev_priv);
void i915_gem_detect_bit_6_swizzle(struct drm_i915_private *dev_priv);
void i915_gem_object_do_bit_17_swizzle(struct drm_i915_gem_object *obj,
struct sg_table *pages);
void i915_gem_object_save_bit_17_swizzle(struct drm_i915_gem_object *obj,
struct sg_table *pages);
static inline struct i915_gem_context *
__i915_gem_context_lookup_rcu(struct drm_i915_file_private *file_priv, u32 id)
{
return idr_find(&file_priv->context_idr, id);
}
static inline struct i915_gem_context *
i915_gem_context_lookup(struct drm_i915_file_private *file_priv, u32 id)
{
struct i915_gem_context *ctx;
rcu_read_lock();
ctx = __i915_gem_context_lookup_rcu(file_priv, id);
if (ctx && !kref_get_unless_zero(&ctx->ref))
ctx = NULL;
rcu_read_unlock();
return ctx;
}
static inline struct intel_timeline *
i915_gem_context_lookup_timeline(struct i915_gem_context *ctx,
struct intel_engine_cs *engine)
{
struct i915_address_space *vm;
vm = ctx->ppgtt ? &ctx->ppgtt->base : &ctx->i915->ggtt.base;
return &vm->timeline.engine[engine->id];
}
drm/i915: Add i915 perf infrastructure Adds base i915 perf infrastructure for Gen performance metrics. This adds a DRM_IOCTL_I915_PERF_OPEN ioctl that takes an array of uint64 properties to configure a stream of metrics and returns a new fd usable with standard VFS system calls including read() to read typed and sized records; ioctl() to enable or disable capture and poll() to wait for data. A stream is opened something like: uint64_t properties[] = { /* Single context sampling */ DRM_I915_PERF_PROP_CTX_HANDLE, ctx_handle, /* Include OA reports in samples */ DRM_I915_PERF_PROP_SAMPLE_OA, true, /* OA unit configuration */ DRM_I915_PERF_PROP_OA_METRICS_SET, metrics_set_id, DRM_I915_PERF_PROP_OA_FORMAT, report_format, DRM_I915_PERF_PROP_OA_EXPONENT, period_exponent, }; struct drm_i915_perf_open_param parm = { .flags = I915_PERF_FLAG_FD_CLOEXEC | I915_PERF_FLAG_FD_NONBLOCK | I915_PERF_FLAG_DISABLED, .properties_ptr = (uint64_t)properties, .num_properties = sizeof(properties) / 16, }; int fd = drmIoctl(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param); Records read all start with a common { type, size } header with DRM_I915_PERF_RECORD_SAMPLE being of most interest. Sample records contain an extensible number of fields and it's the DRM_I915_PERF_PROP_SAMPLE_xyz properties given when opening that determine what's included in every sample. No specific streams are supported yet so any attempt to open a stream will return an error. v2: use i915_gem_context_get() - Chris Wilson v3: update read() interface to avoid passing state struct - Chris Wilson fix some rebase fallout, with i915-perf init/deinit v4: s/DRM_IORW/DRM_IOW/ - Emil Velikov Signed-off-by: Robert Bragg <robert@sixbynine.org> Reviewed-by: Matthew Auld <matthew.auld@intel.com> Reviewed-by: Sourab Gupta <sourab.gupta@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/20161107194957.3385-2-robert@sixbynine.org
2016-11-07 12:49:47 -07:00
int i915_perf_open_ioctl(struct drm_device *dev, void *data,
struct drm_file *file);
drm/i915/perf: Implement I915_PERF_ADD/REMOVE_CONFIG interface The motivation behind this new interface is expose at runtime the creation of new OA configs which can be used as part of the i915 perf open interface. This will enable the kernel to learn new configs which may be experimental, or otherwise not part of the core set currently available through the i915 perf interface. v2: Drop DRM_ERROR for userspace errors (Matthew) Add padding to userspace structure (Matthew) s/guid/uuid/ (Matthew) v3: Use u32 instead of int to iterate through registers (Matthew) v4: Lock access to dynamic config list (Lionel) v5: by Matthew: Fix uninitialized error values Fix incorrect unwiding when opening perf stream Use kmalloc_array() to store register Use uuid_is_valid() to valid config uuids Declare ioctls as write only Check padding members are set to 0 by Lionel: Return ENOENT rather than EINVAL when trying to remove non existing config v6: by Chris: Use ref counts for OA configs Store UUID in drm_i915_perf_oa_config rather then using pointer Shuffle fields of drm_i915_perf_oa_config to avoid padding v7: by Chris Rename uapi pointers fields to end with '_ptr' v8: by Andrzej, Marek, Sebastian Update register whitelisting by Lionel Add more register names for documentation Allow configuration programming in non-paranoid mode Add support for value filter for a couple of registers already programmed in other part of the kernel v9: Documentation fix (Lionel) Allow writing WAIT_FOR_RC6_EXIT only on Gen8+ (Andrzej) v10: Perform read access_ok() on register pointers (Lionel) Signed-off-by: Matthew Auld <matthew.auld@intel.com> Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com> Signed-off-by: Andrzej Datczuk <andrzej.datczuk@intel.com> Reviewed-by: Andrzej Datczuk <andrzej.datczuk@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20170803165812.2373-2-lionel.g.landwerlin@intel.com
2017-08-03 11:05:50 -06:00
int i915_perf_add_config_ioctl(struct drm_device *dev, void *data,
struct drm_file *file);
int i915_perf_remove_config_ioctl(struct drm_device *dev, void *data,
struct drm_file *file);
drm/i915/perf: Add OA unit support for Gen 8+ Enables access to OA unit metrics for BDW, CHV, SKL and BXT which all share (more-or-less) the same OA unit design. Of particular note in comparison to Haswell: some OA unit HW config state has become per-context state and as a consequence it is somewhat more complicated to manage synchronous state changes from the cpu while there's no guarantee of what context (if any) is currently actively running on the gpu. The periodic sampling frequency which can be particularly useful for system-wide analysis (as opposed to command stream synchronised MI_REPORT_PERF_COUNT commands) is perhaps the most surprising state to have become per-context save and restored (while the OABUFFER destination is still a shared, system-wide resource). This support for gen8+ takes care to consider a number of timing challenges involved in synchronously updating per-context state primarily by programming all config state from the cpu and updating all current and saved contexts synchronously while the OA unit is still disabled. The driver intentionally avoids depending on command streamer programming to update OA state considering the lack of synchronization between the automatic loading of OACTXCONTROL state (that includes the periodic sampling state and enable state) on context restore and the parsing of any general purpose BB the driver can control. I.e. this implementation is careful to avoid the possibility of a context restore temporarily enabling any out-of-date periodic sampling state. In addition to the risk of transiently-out-of-date state being loaded automatically; there are also internal HW latencies involved in the loading of MUX configurations which would be difficult to account for from the command streamer (and we only want to enable the unit when once the MUX configuration is complete). Since the Gen8+ OA unit design no longer supports clock gating the unit off for a single given context (which effectively stopped any progress of counters while any other context was running) and instead supports tagging OA reports with a context ID for filtering on the CPU, it means we can no longer hide the system-wide progress of counters from a non-privileged application only interested in metrics for its own context. Although we could theoretically try and subtract the progress of other contexts before forwarding reports via read() we aren't in a position to filter reports captured via MI_REPORT_PERF_COUNT commands. As a result, for Gen8+, we always require the dev.i915.perf_stream_paranoid to be unset for any access to OA metrics if not root. v5: Drain submitted requests when enabling metric set to ensure no lite-restore erases the context image we just updated (Lionel) v6: In addition to drain, switch to kernel context & update all context in place (Chris) v7: Add missing mutex_unlock() if switching to kernel context fails (Matthew) v8: Simplify OA period/flex-eu-counters programming by using the batchbuffer instead of modifying ctx-image (Lionel) v9: Back to updating the context image (due to erroneous testing, batchbuffer programming the OA unit doesn't actually work) (Lionel) Pin context before updating context image (Chris) Drop MMIO programming now that we switch to a kernel context with right values in initial context image (Chris) v10: Just pin_map the contexts we want to modify or let the configuration happen on first use (Chris) v11: Update kernel context OA config through the batchbuffer rather than on the fly ctx-image update (Lionel) v12: Rework OA context registers update again by swithing away from user contexts and reconfiguring the kernel context through the batchbuffer and updating all the other contexts' context image. Also take care to lock slice/subslice configuration when OA is on. (Lionel) v13: Request rpcs updates on all engine when updating the OA config (Lionel) v14: Drop any kind of rpcs management now that we monitor sseu configuration changes in a later patch (Lionel) Remove usleep after programming the NOA configs on Gen8+, this doesn't seem to be needed (Lionel) v15: Respect coding style for block comments (Chris) v16: Add missing i915_add_request() in case we fail to emit OA configuration (Matthew) Signed-off-by: Robert Bragg <robert@sixbynine.org> Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com> Reviewed-by: Matthew Auld <matthew.auld@intel.com> \o/ Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
2017-06-13 05:23:03 -06:00
void i915_oa_init_reg_state(struct intel_engine_cs *engine,
struct i915_gem_context *ctx,
uint32_t *reg_state);
drm/i915: Add i915 perf infrastructure Adds base i915 perf infrastructure for Gen performance metrics. This adds a DRM_IOCTL_I915_PERF_OPEN ioctl that takes an array of uint64 properties to configure a stream of metrics and returns a new fd usable with standard VFS system calls including read() to read typed and sized records; ioctl() to enable or disable capture and poll() to wait for data. A stream is opened something like: uint64_t properties[] = { /* Single context sampling */ DRM_I915_PERF_PROP_CTX_HANDLE, ctx_handle, /* Include OA reports in samples */ DRM_I915_PERF_PROP_SAMPLE_OA, true, /* OA unit configuration */ DRM_I915_PERF_PROP_OA_METRICS_SET, metrics_set_id, DRM_I915_PERF_PROP_OA_FORMAT, report_format, DRM_I915_PERF_PROP_OA_EXPONENT, period_exponent, }; struct drm_i915_perf_open_param parm = { .flags = I915_PERF_FLAG_FD_CLOEXEC | I915_PERF_FLAG_FD_NONBLOCK | I915_PERF_FLAG_DISABLED, .properties_ptr = (uint64_t)properties, .num_properties = sizeof(properties) / 16, }; int fd = drmIoctl(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param); Records read all start with a common { type, size } header with DRM_I915_PERF_RECORD_SAMPLE being of most interest. Sample records contain an extensible number of fields and it's the DRM_I915_PERF_PROP_SAMPLE_xyz properties given when opening that determine what's included in every sample. No specific streams are supported yet so any attempt to open a stream will return an error. v2: use i915_gem_context_get() - Chris Wilson v3: update read() interface to avoid passing state struct - Chris Wilson fix some rebase fallout, with i915-perf init/deinit v4: s/DRM_IORW/DRM_IOW/ - Emil Velikov Signed-off-by: Robert Bragg <robert@sixbynine.org> Reviewed-by: Matthew Auld <matthew.auld@intel.com> Reviewed-by: Sourab Gupta <sourab.gupta@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/20161107194957.3385-2-robert@sixbynine.org
2016-11-07 12:49:47 -07:00
/* i915_gem_evict.c */
int __must_check i915_gem_evict_something(struct i915_address_space *vm,
u64 min_size, u64 alignment,
unsigned cache_level,
u64 start, u64 end,
unsigned flags);
int __must_check i915_gem_evict_for_node(struct i915_address_space *vm,
struct drm_mm_node *node,
unsigned int flags);
drm/i915: Eliminate lots of iterations over the execobjects array The major scaling bottleneck in execbuffer is the processing of the execobjects. Creating an auxiliary list is inefficient when compared to using the execobject array we already have allocated. Reservation is then split into phases. As we lookup up the VMA, we try and bind it back into active location. Only if that fails, do we add it to the unbound list for phase 2. In phase 2, we try and add all those objects that could not fit into their previous location, with fallback to retrying all objects and evicting the VM in case of severe fragmentation. (This is the same as before, except that phase 1 is now done inline with looking up the VMA to avoid an iteration over the execobject array. In the ideal case, we eliminate the separate reservation phase). During the reservation phase, we only evict from the VM between passes (rather than currently as we try to fit every new VMA). In testing with Unreal Engine's Atlantis demo which stresses the eviction logic on gen7 class hardware, this speed up the framerate by a factor of 2. The second loop amalgamation is between move_to_gpu and move_to_active. As we always submit the request, even if incomplete, we can use the current request to track active VMA as we perform the flushes and synchronisation required. The next big advancement is to avoid copying back to the user any execobjects and relocations that are not changed. v2: Add a Theory of Operation spiel. v3: Fall back to slow relocations in preparation for flushing userptrs. v4: Document struct members, factor out eb_validate_vma(), add a few more comments to explain some magic and hide other magic behind macros. Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
2017-06-16 08:05:19 -06:00
int i915_gem_evict_vm(struct i915_address_space *vm);
/* belongs in i915_gem_gtt.h */
static inline void i915_gem_chipset_flush(struct drm_i915_private *dev_priv)
drm/i915: Stop using AGP layer for GEN6+ As a quick hack we make the old intel_gtt structure mutable so we can fool a bunch of the existing code which depends on elements in that data structure. We can/should try to remove this in a subsequent patch. This should preserve the old gtt init behavior which upon writing these patches seems incorrect. The next patch will fix these things. The one exception is VLV which doesn't have the preserved flush control write behavior. Since we want to do that for all GEN6+ stuff, we'll handle that in a later patch. Mainstream VLV support doesn't actually exist yet anyway. v2: Update the comment to remove the "voodoo" Check that the last pte written matches what we readback v3: actually kill cache_level_to_agp_type since most of the flags will disappear in an upcoming patch v4: v3 was actually not what we wanted (Daniel) Make the ggtt bind assertions better and stricter (Chris) Fix some uncaught errors at gtt init (Chris) Some other random stuff that Chris wanted v5: check for i==0 in gen6_ggtt_bind_object to shut up gcc (Ben) Signed-off-by: Ben Widawsky <ben@bwidawsk.net> Reviewed-by [v4]: Chris Wilson <chris@chris-wilson.co.uk> [danvet: Make the cache_level -> agp_flags conversion for pre-gen6 a tad more robust by mapping everything != CACHE_NONE to the cached agp flag - we have a 1:1 uncached mapping, but different modes of cacheable (at least on later generations). Suggested by Chris Wilson.] Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2012-11-04 10:21:27 -07:00
{
wmb();
if (INTEL_GEN(dev_priv) < 6)
drm/i915: Stop using AGP layer for GEN6+ As a quick hack we make the old intel_gtt structure mutable so we can fool a bunch of the existing code which depends on elements in that data structure. We can/should try to remove this in a subsequent patch. This should preserve the old gtt init behavior which upon writing these patches seems incorrect. The next patch will fix these things. The one exception is VLV which doesn't have the preserved flush control write behavior. Since we want to do that for all GEN6+ stuff, we'll handle that in a later patch. Mainstream VLV support doesn't actually exist yet anyway. v2: Update the comment to remove the "voodoo" Check that the last pte written matches what we readback v3: actually kill cache_level_to_agp_type since most of the flags will disappear in an upcoming patch v4: v3 was actually not what we wanted (Daniel) Make the ggtt bind assertions better and stricter (Chris) Fix some uncaught errors at gtt init (Chris) Some other random stuff that Chris wanted v5: check for i==0 in gen6_ggtt_bind_object to shut up gcc (Ben) Signed-off-by: Ben Widawsky <ben@bwidawsk.net> Reviewed-by [v4]: Chris Wilson <chris@chris-wilson.co.uk> [danvet: Make the cache_level -> agp_flags conversion for pre-gen6 a tad more robust by mapping everything != CACHE_NONE to the cached agp flag - we have a 1:1 uncached mapping, but different modes of cacheable (at least on later generations). Suggested by Chris Wilson.] Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2012-11-04 10:21:27 -07:00
intel_gtt_chipset_flush();
}
/* i915_gem_stolen.c */
int i915_gem_stolen_insert_node(struct drm_i915_private *dev_priv,
struct drm_mm_node *node, u64 size,
unsigned alignment);
int i915_gem_stolen_insert_node_in_range(struct drm_i915_private *dev_priv,
struct drm_mm_node *node, u64 size,
unsigned alignment, u64 start,
u64 end);
void i915_gem_stolen_remove_node(struct drm_i915_private *dev_priv,
struct drm_mm_node *node);
int i915_gem_init_stolen(struct drm_i915_private *dev_priv);
void i915_gem_cleanup_stolen(struct drm_device *dev);
struct drm_i915_gem_object *
i915_gem_object_create_stolen(struct drm_i915_private *dev_priv, u32 size);
struct drm_i915_gem_object *
i915_gem_object_create_stolen_for_preallocated(struct drm_i915_private *dev_priv,
u32 stolen_offset,
u32 gtt_offset,
u32 size);
drm/i915: Introduce an internal allocator for disposable private objects Quite a few of our objects used for internal hardware programming do not benefit from being swappable or from being zero initialised. As such they do not benefit from using a shmemfs backing storage and since they are internal and never directly exposed to the user, we do not need to worry about providing a filp. For these we can use an drm_i915_gem_object wrapper around a sg_table of plain struct page. They are not swap backed and not automatically pinned. If they are reaped by the shrinker, the pages are released and the contents discarded. For the internal use case, this is fine as for example, ringbuffers are pinned from being written by a request to be read by the hardware. Once they are idle, they can be discarded entirely. As such they are a good match for execlist ringbuffers and a small variety of other internal objects. In the first iteration, this is limited to the scratch batch buffers we use (for command parsing and state initialisation). v2: Allocate physically contiguous pages, where possible. v3: Reduce maximum order on subsequent requests following an allocation failure. v4: Fix up mismatch between swiotlb segment size and page count (it counts in 2k units, not 4k pages) Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/20161028125858.23563-7-chris@chris-wilson.co.uk
2016-10-28 06:58:30 -06:00
/* i915_gem_internal.c */
struct drm_i915_gem_object *
i915_gem_object_create_internal(struct drm_i915_private *dev_priv,
phys_addr_t size);
drm/i915: Introduce an internal allocator for disposable private objects Quite a few of our objects used for internal hardware programming do not benefit from being swappable or from being zero initialised. As such they do not benefit from using a shmemfs backing storage and since they are internal and never directly exposed to the user, we do not need to worry about providing a filp. For these we can use an drm_i915_gem_object wrapper around a sg_table of plain struct page. They are not swap backed and not automatically pinned. If they are reaped by the shrinker, the pages are released and the contents discarded. For the internal use case, this is fine as for example, ringbuffers are pinned from being written by a request to be read by the hardware. Once they are idle, they can be discarded entirely. As such they are a good match for execlist ringbuffers and a small variety of other internal objects. In the first iteration, this is limited to the scratch batch buffers we use (for command parsing and state initialisation). v2: Allocate physically contiguous pages, where possible. v3: Reduce maximum order on subsequent requests following an allocation failure. v4: Fix up mismatch between swiotlb segment size and page count (it counts in 2k units, not 4k pages) Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/20161028125858.23563-7-chris@chris-wilson.co.uk
2016-10-28 06:58:30 -06:00
/* i915_gem_shrinker.c */
unsigned long i915_gem_shrink(struct drm_i915_private *dev_priv,
unsigned long target,
unsigned flags);
#define I915_SHRINK_PURGEABLE 0x1
#define I915_SHRINK_UNBOUND 0x2
#define I915_SHRINK_BOUND 0x4
#define I915_SHRINK_ACTIVE 0x8
#define I915_SHRINK_VMAPS 0x10
unsigned long i915_gem_shrink_all(struct drm_i915_private *dev_priv);
void i915_gem_shrinker_init(struct drm_i915_private *dev_priv);
void i915_gem_shrinker_cleanup(struct drm_i915_private *dev_priv);
/* i915_gem_tiling.c */
static inline bool i915_gem_object_needs_bit17_swizzle(struct drm_i915_gem_object *obj)
{
struct drm_i915_private *dev_priv = to_i915(obj->base.dev);
return dev_priv->mm.bit_6_swizzle_x == I915_BIT_6_SWIZZLE_9_10_17 &&
i915_gem_object_is_tiled(obj);
}
u32 i915_gem_fence_size(struct drm_i915_private *dev_priv, u32 size,
unsigned int tiling, unsigned int stride);
u32 i915_gem_fence_alignment(struct drm_i915_private *dev_priv, u32 size,
unsigned int tiling, unsigned int stride);
/* i915_debugfs.c */
#ifdef CONFIG_DEBUG_FS
int i915_debugfs_register(struct drm_i915_private *dev_priv);
int i915_debugfs_connector_add(struct drm_connector *connector);
void intel_display_crc_init(struct drm_i915_private *dev_priv);
#else
static inline int i915_debugfs_register(struct drm_i915_private *dev_priv) {return 0;}
static inline int i915_debugfs_connector_add(struct drm_connector *connector)
{ return 0; }
static inline void intel_display_crc_init(struct drm_i915_private *dev_priv) {}
#endif
/* i915_gpu_error.c */
#if IS_ENABLED(CONFIG_DRM_I915_CAPTURE_ERROR)
__printf(2, 3)
void i915_error_printf(struct drm_i915_error_state_buf *e, const char *f, ...);
int i915_error_state_to_str(struct drm_i915_error_state_buf *estr,
const struct i915_gpu_state *gpu);
int i915_error_state_buf_init(struct drm_i915_error_state_buf *eb,
struct drm_i915_private *i915,
size_t count, loff_t pos);
static inline void i915_error_state_buf_release(
struct drm_i915_error_state_buf *eb)
{
kfree(eb->buf);
}
struct i915_gpu_state *i915_capture_gpu_state(struct drm_i915_private *i915);
void i915_capture_error_state(struct drm_i915_private *dev_priv,
u32 engine_mask,
const char *error_msg);
static inline struct i915_gpu_state *
i915_gpu_state_get(struct i915_gpu_state *gpu)
{
kref_get(&gpu->ref);
return gpu;
}
void __i915_gpu_state_free(struct kref *kref);
static inline void i915_gpu_state_put(struct i915_gpu_state *gpu)
{
if (gpu)
kref_put(&gpu->ref, __i915_gpu_state_free);
}
struct i915_gpu_state *i915_first_error_state(struct drm_i915_private *i915);
void i915_reset_error_state(struct drm_i915_private *i915);
#else
static inline void i915_capture_error_state(struct drm_i915_private *dev_priv,
u32 engine_mask,
const char *error_msg)
{
}
static inline struct i915_gpu_state *
i915_first_error_state(struct drm_i915_private *i915)
{
return NULL;
}
static inline void i915_reset_error_state(struct drm_i915_private *i915)
{
}
#endif
const char *i915_cache_level_str(struct drm_i915_private *i915, int type);
/* i915_cmd_parser.c */
int i915_cmd_parser_get_version(struct drm_i915_private *dev_priv);
void intel_engine_init_cmd_parser(struct intel_engine_cs *engine);
void intel_engine_cleanup_cmd_parser(struct intel_engine_cs *engine);
int intel_engine_cmd_parser(struct intel_engine_cs *engine,
struct drm_i915_gem_object *batch_obj,
struct drm_i915_gem_object *shadow_batch_obj,
u32 batch_start_offset,
u32 batch_len,
bool is_master);
drm/i915: Add i915 perf infrastructure Adds base i915 perf infrastructure for Gen performance metrics. This adds a DRM_IOCTL_I915_PERF_OPEN ioctl that takes an array of uint64 properties to configure a stream of metrics and returns a new fd usable with standard VFS system calls including read() to read typed and sized records; ioctl() to enable or disable capture and poll() to wait for data. A stream is opened something like: uint64_t properties[] = { /* Single context sampling */ DRM_I915_PERF_PROP_CTX_HANDLE, ctx_handle, /* Include OA reports in samples */ DRM_I915_PERF_PROP_SAMPLE_OA, true, /* OA unit configuration */ DRM_I915_PERF_PROP_OA_METRICS_SET, metrics_set_id, DRM_I915_PERF_PROP_OA_FORMAT, report_format, DRM_I915_PERF_PROP_OA_EXPONENT, period_exponent, }; struct drm_i915_perf_open_param parm = { .flags = I915_PERF_FLAG_FD_CLOEXEC | I915_PERF_FLAG_FD_NONBLOCK | I915_PERF_FLAG_DISABLED, .properties_ptr = (uint64_t)properties, .num_properties = sizeof(properties) / 16, }; int fd = drmIoctl(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param); Records read all start with a common { type, size } header with DRM_I915_PERF_RECORD_SAMPLE being of most interest. Sample records contain an extensible number of fields and it's the DRM_I915_PERF_PROP_SAMPLE_xyz properties given when opening that determine what's included in every sample. No specific streams are supported yet so any attempt to open a stream will return an error. v2: use i915_gem_context_get() - Chris Wilson v3: update read() interface to avoid passing state struct - Chris Wilson fix some rebase fallout, with i915-perf init/deinit v4: s/DRM_IORW/DRM_IOW/ - Emil Velikov Signed-off-by: Robert Bragg <robert@sixbynine.org> Reviewed-by: Matthew Auld <matthew.auld@intel.com> Reviewed-by: Sourab Gupta <sourab.gupta@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/20161107194957.3385-2-robert@sixbynine.org
2016-11-07 12:49:47 -07:00
/* i915_perf.c */
extern void i915_perf_init(struct drm_i915_private *dev_priv);
extern void i915_perf_fini(struct drm_i915_private *dev_priv);
extern void i915_perf_register(struct drm_i915_private *dev_priv);
extern void i915_perf_unregister(struct drm_i915_private *dev_priv);
drm/i915: Add i915 perf infrastructure Adds base i915 perf infrastructure for Gen performance metrics. This adds a DRM_IOCTL_I915_PERF_OPEN ioctl that takes an array of uint64 properties to configure a stream of metrics and returns a new fd usable with standard VFS system calls including read() to read typed and sized records; ioctl() to enable or disable capture and poll() to wait for data. A stream is opened something like: uint64_t properties[] = { /* Single context sampling */ DRM_I915_PERF_PROP_CTX_HANDLE, ctx_handle, /* Include OA reports in samples */ DRM_I915_PERF_PROP_SAMPLE_OA, true, /* OA unit configuration */ DRM_I915_PERF_PROP_OA_METRICS_SET, metrics_set_id, DRM_I915_PERF_PROP_OA_FORMAT, report_format, DRM_I915_PERF_PROP_OA_EXPONENT, period_exponent, }; struct drm_i915_perf_open_param parm = { .flags = I915_PERF_FLAG_FD_CLOEXEC | I915_PERF_FLAG_FD_NONBLOCK | I915_PERF_FLAG_DISABLED, .properties_ptr = (uint64_t)properties, .num_properties = sizeof(properties) / 16, }; int fd = drmIoctl(drm_fd, DRM_IOCTL_I915_PERF_OPEN, &param); Records read all start with a common { type, size } header with DRM_I915_PERF_RECORD_SAMPLE being of most interest. Sample records contain an extensible number of fields and it's the DRM_I915_PERF_PROP_SAMPLE_xyz properties given when opening that determine what's included in every sample. No specific streams are supported yet so any attempt to open a stream will return an error. v2: use i915_gem_context_get() - Chris Wilson v3: update read() interface to avoid passing state struct - Chris Wilson fix some rebase fallout, with i915-perf init/deinit v4: s/DRM_IORW/DRM_IOW/ - Emil Velikov Signed-off-by: Robert Bragg <robert@sixbynine.org> Reviewed-by: Matthew Auld <matthew.auld@intel.com> Reviewed-by: Sourab Gupta <sourab.gupta@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/20161107194957.3385-2-robert@sixbynine.org
2016-11-07 12:49:47 -07:00
/* i915_suspend.c */
extern int i915_save_state(struct drm_i915_private *dev_priv);
extern int i915_restore_state(struct drm_i915_private *dev_priv);
drm/i915: rc6 in sysfs Merge rc6 information into the power group for our device. Until now the i915 driver has not had any sysfs entries (aside from the connector stuff enabled by drm core). Since it seems like we're likely to have more in the future I created a new file for sysfs stubs, as well as the rc6 sysfs functions which don't really belong elsewhere (perhaps i915_suspend, but most of the stuff is in intel_display,c). displays rc6 modes enabled (as a hex mask): cat /sys/class/drm/card0/power/rc6_enable displays #ms GPU has been in rc6 since boot: cat /sys/class/drm/card0/power/rc6_residency_ms displays #ms GPU has been in deep rc6 since boot: cat /sys/class/drm/card0/power/rc6p_residency_ms displays #ms GPU has been in deepest rc6 since boot: cat /sys/class/drm/card0/power/rc6pp_residency_ms Important note: I've seen on SNB that even when RC6 is *not* enabled the rc6 register seems to have a random value in it. I can only guess at the reason reason for this. Those writing tools that utilize this value need to be careful and probably want to scrutinize the value very carefully. v2: use common rc6 residency units to milliseconds for the other RC6 types v3: don't create sysfs files for GEN <= 5 add a rc6_enable to show a mask of enabled rc6 types use unmerge instead of remove for sysfs group squash intel_enable_rc6() extraction into this patch v4: rename sysfs files (Chris) CC: Chris Wilson <chris@chris-wilson.co.uk> CC: Daniel Vetter <daniel.vetter@ffwll.ch>f CC: Arjan van de Ven <arjan@linux.intel.com> Signed-off-by: Ben Widawsky <benjamin.widawsky@intel.com> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> [danvet: squash in the 64bit division fix by Chris Wilson.] Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2012-04-10 22:17:01 -06:00
/* i915_sysfs.c */
void i915_setup_sysfs(struct drm_i915_private *dev_priv);
void i915_teardown_sysfs(struct drm_i915_private *dev_priv);
drm/i915: rc6 in sysfs Merge rc6 information into the power group for our device. Until now the i915 driver has not had any sysfs entries (aside from the connector stuff enabled by drm core). Since it seems like we're likely to have more in the future I created a new file for sysfs stubs, as well as the rc6 sysfs functions which don't really belong elsewhere (perhaps i915_suspend, but most of the stuff is in intel_display,c). displays rc6 modes enabled (as a hex mask): cat /sys/class/drm/card0/power/rc6_enable displays #ms GPU has been in rc6 since boot: cat /sys/class/drm/card0/power/rc6_residency_ms displays #ms GPU has been in deep rc6 since boot: cat /sys/class/drm/card0/power/rc6p_residency_ms displays #ms GPU has been in deepest rc6 since boot: cat /sys/class/drm/card0/power/rc6pp_residency_ms Important note: I've seen on SNB that even when RC6 is *not* enabled the rc6 register seems to have a random value in it. I can only guess at the reason reason for this. Those writing tools that utilize this value need to be careful and probably want to scrutinize the value very carefully. v2: use common rc6 residency units to milliseconds for the other RC6 types v3: don't create sysfs files for GEN <= 5 add a rc6_enable to show a mask of enabled rc6 types use unmerge instead of remove for sysfs group squash intel_enable_rc6() extraction into this patch v4: rename sysfs files (Chris) CC: Chris Wilson <chris@chris-wilson.co.uk> CC: Daniel Vetter <daniel.vetter@ffwll.ch>f CC: Arjan van de Ven <arjan@linux.intel.com> Signed-off-by: Ben Widawsky <benjamin.widawsky@intel.com> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> [danvet: squash in the 64bit division fix by Chris Wilson.] Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2012-04-10 22:17:01 -06:00
/* intel_lpe_audio.c */
int intel_lpe_audio_init(struct drm_i915_private *dev_priv);
void intel_lpe_audio_teardown(struct drm_i915_private *dev_priv);
void intel_lpe_audio_irq_handler(struct drm_i915_private *dev_priv);
void intel_lpe_audio_notify(struct drm_i915_private *dev_priv,
enum pipe pipe, enum port port,
const void *eld, int ls_clock, bool dp_output);
/* intel_i2c.c */
extern int intel_setup_gmbus(struct drm_i915_private *dev_priv);
extern void intel_teardown_gmbus(struct drm_i915_private *dev_priv);
extern bool intel_gmbus_is_valid_pin(struct drm_i915_private *dev_priv,
unsigned int pin);
extern struct i2c_adapter *
intel_gmbus_get_adapter(struct drm_i915_private *dev_priv, unsigned int pin);
extern void intel_gmbus_set_speed(struct i2c_adapter *adapter, int speed);
extern void intel_gmbus_force_bit(struct i2c_adapter *adapter, bool force_bit);
static inline bool intel_gmbus_is_forced_bit(struct i2c_adapter *adapter)
{
return container_of(adapter, struct intel_gmbus, adapter)->force_bit;
}
extern void intel_i2c_reset(struct drm_i915_private *dev_priv);
/* intel_bios.c */
void intel_bios_init(struct drm_i915_private *dev_priv);
bool intel_bios_is_valid_vbt(const void *buf, size_t size);
bool intel_bios_is_tv_present(struct drm_i915_private *dev_priv);
bool intel_bios_is_lvds_present(struct drm_i915_private *dev_priv, u8 *i2c_pin);
bool intel_bios_is_port_present(struct drm_i915_private *dev_priv, enum port port);
bool intel_bios_is_port_edp(struct drm_i915_private *dev_priv, enum port port);
bool intel_bios_is_port_dp_dual_mode(struct drm_i915_private *dev_priv, enum port port);
bool intel_bios_is_dsi_present(struct drm_i915_private *dev_priv, enum port *port);
bool intel_bios_is_port_hpd_inverted(struct drm_i915_private *dev_priv,
enum port port);
bool intel_bios_is_lspcon_present(struct drm_i915_private *dev_priv,
enum port port);
/* intel_opregion.c */
#ifdef CONFIG_ACPI
extern int intel_opregion_setup(struct drm_i915_private *dev_priv);
extern void intel_opregion_register(struct drm_i915_private *dev_priv);
extern void intel_opregion_unregister(struct drm_i915_private *dev_priv);
drm/i915: Small display interrupt handlers tidy I have noticed some of our interrupt handlers use both dev and dev_priv while they could get away with only dev_priv in the huge majority of cases. Tidying that up had a cascading effect on changing functions prototypes, so relatively big churn factor, but I think it is for the better. For example even where changes cascade out of i915_irq.c, for functions prefixed with intel_, genX_ or <plat>_, it makes more sense to take dev_priv directly anyway. This allows us to eliminate local variables and intermixed usage of dev and dev_priv where only one is good enough. End result is shrinkage of both source and the resulting binary. i915.ko: - .text 000b0899 + .text 000b0619 Or if we look at the Gen8 display irq chain: -00000000000006ad t gen8_irq_handler +0000000000000663 t gen8_irq_handler -0000000000000028 T intel_opregion_asle_intr +0000000000000024 T intel_opregion_asle_intr -000000000000008c t ilk_hpd_irq_handler +000000000000007f t ilk_hpd_irq_handler -0000000000000116 T intel_check_page_flip +0000000000000112 T intel_check_page_flip -000000000000011a T intel_prepare_page_flip +0000000000000119 T intel_prepare_page_flip -0000000000000014 T intel_finish_page_flip_plane +0000000000000013 T intel_finish_page_flip_plane -0000000000000053 t hsw_pipe_crc_irq_handler +000000000000004c t hsw_pipe_crc_irq_handler -000000000000022e t cpt_irq_handler +0000000000000213 t cpt_irq_handler So small shrinkage but it is all fast paths so doesn't harm. Situation is similar in other interrupt handlers as well. v2: Tidy intel_queue_rps_boost_for_request as well. (Chris Wilson) Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
2016-05-06 07:48:28 -06:00
extern void intel_opregion_asle_intr(struct drm_i915_private *dev_priv);
extern int intel_opregion_notify_encoder(struct intel_encoder *intel_encoder,
bool enable);
extern int intel_opregion_notify_adapter(struct drm_i915_private *dev_priv,
pci_power_t state);
extern int intel_opregion_get_panel_type(struct drm_i915_private *dev_priv);
#else
static inline int intel_opregion_setup(struct drm_i915_private *dev) { return 0; }
static inline void intel_opregion_register(struct drm_i915_private *dev_priv) { }
static inline void intel_opregion_unregister(struct drm_i915_private *dev_priv) { }
drm/i915: Small display interrupt handlers tidy I have noticed some of our interrupt handlers use both dev and dev_priv while they could get away with only dev_priv in the huge majority of cases. Tidying that up had a cascading effect on changing functions prototypes, so relatively big churn factor, but I think it is for the better. For example even where changes cascade out of i915_irq.c, for functions prefixed with intel_, genX_ or <plat>_, it makes more sense to take dev_priv directly anyway. This allows us to eliminate local variables and intermixed usage of dev and dev_priv where only one is good enough. End result is shrinkage of both source and the resulting binary. i915.ko: - .text 000b0899 + .text 000b0619 Or if we look at the Gen8 display irq chain: -00000000000006ad t gen8_irq_handler +0000000000000663 t gen8_irq_handler -0000000000000028 T intel_opregion_asle_intr +0000000000000024 T intel_opregion_asle_intr -000000000000008c t ilk_hpd_irq_handler +000000000000007f t ilk_hpd_irq_handler -0000000000000116 T intel_check_page_flip +0000000000000112 T intel_check_page_flip -000000000000011a T intel_prepare_page_flip +0000000000000119 T intel_prepare_page_flip -0000000000000014 T intel_finish_page_flip_plane +0000000000000013 T intel_finish_page_flip_plane -0000000000000053 t hsw_pipe_crc_irq_handler +000000000000004c t hsw_pipe_crc_irq_handler -000000000000022e t cpt_irq_handler +0000000000000213 t cpt_irq_handler So small shrinkage but it is all fast paths so doesn't harm. Situation is similar in other interrupt handlers as well. v2: Tidy intel_queue_rps_boost_for_request as well. (Chris Wilson) Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
2016-05-06 07:48:28 -06:00
static inline void intel_opregion_asle_intr(struct drm_i915_private *dev_priv)
{
}
static inline int
intel_opregion_notify_encoder(struct intel_encoder *intel_encoder, bool enable)
{
return 0;
}
static inline int
intel_opregion_notify_adapter(struct drm_i915_private *dev, pci_power_t state)
{
return 0;
}
static inline int intel_opregion_get_panel_type(struct drm_i915_private *dev)
drm/i915: Get panel_type from OpRegion panel details We've had problems on several occasions with using the panel type from the VBT block 40. Usually it seems to be 2, which often doesn't give us the correct timings for the panel. After some more digging I found a way to get a panel type via the OpRegion SWSCI GBDA "Get Panel Details" method. Let's try to use it. The spec has this to say about the output: "Bits [15:8] - Panel Type Bits contain the panel type user setting from CMOS 00h = Not Valid, use default Panel Type & Timings from VBT 01h - 0Fh = Panel Number" Another version of the spec lists the valid range as 1-16, which makes more sense since VBT supports 16 panels. Based on actual results from Rob's G45, 1-16 is what we need to accept. The other bits in the output don't look relevant for the problem at hand. The input is specified as: "Bits [31:4] - Reserved Reserved (must be zero) Bits [3:0] - Panel Number These bits contain the sequential index of Panel, starting at 0 and counting upwards from the first integrated Internal Flat-Panel Display Encoder present, and then from the first external Display Encoder (e.g., S/DVO-B then S/DVO-C) which supports Internal Flat-Panels. 0h - 0Fh = Panel number" For now I've just hardcoded the input panel number as 0. That would seem like a decent choise for LVDS. Not so sure about eDP when port != A. v2: Accept values 1-16 Filter out bogus results in opregion code (Jani) Add debug logging for all the different branches (Jani) Cc: Jani Nikula <jani.nikula@linux.intel.com> Cc: Rob Kramer <rob@solution-space.com> Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=94825 Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1460359431-11003-1-git-send-email-ville.syrjala@linux.intel.com Reviewed-by: Jani Nikula <jani.nikula@intel.com> Tested-by: Rob Kramer <rob@solution-space.com>
2016-04-11 01:23:51 -06:00
{
return -ENODEV;
}
#endif
/* intel_acpi.c */
#ifdef CONFIG_ACPI
extern void intel_register_dsm_handler(void);
extern void intel_unregister_dsm_handler(void);
#else
static inline void intel_register_dsm_handler(void) { return; }
static inline void intel_unregister_dsm_handler(void) { return; }
#endif /* CONFIG_ACPI */
/* intel_device_info.c */
static inline struct intel_device_info *
mkwrite_device_info(struct drm_i915_private *dev_priv)
{
return (struct intel_device_info *)&dev_priv->info;
}
const char *intel_platform_name(enum intel_platform platform);
void intel_device_info_runtime_init(struct drm_i915_private *dev_priv);
void intel_device_info_dump(struct drm_i915_private *dev_priv);
/* modesetting */
extern void intel_modeset_init_hw(struct drm_device *dev);
extern int intel_modeset_init(struct drm_device *dev);
extern void intel_modeset_gem_init(struct drm_device *dev);
extern void intel_modeset_cleanup(struct drm_device *dev);
extern int intel_connector_register(struct drm_connector *);
extern void intel_connector_unregister(struct drm_connector *);
extern int intel_modeset_vga_set_state(struct drm_i915_private *dev_priv,
bool state);
extern void intel_display_resume(struct drm_device *dev);
extern void i915_redisable_vga(struct drm_i915_private *dev_priv);
extern void i915_redisable_vga_power_on(struct drm_i915_private *dev_priv);
drm/i915: Small display interrupt handlers tidy I have noticed some of our interrupt handlers use both dev and dev_priv while they could get away with only dev_priv in the huge majority of cases. Tidying that up had a cascading effect on changing functions prototypes, so relatively big churn factor, but I think it is for the better. For example even where changes cascade out of i915_irq.c, for functions prefixed with intel_, genX_ or <plat>_, it makes more sense to take dev_priv directly anyway. This allows us to eliminate local variables and intermixed usage of dev and dev_priv where only one is good enough. End result is shrinkage of both source and the resulting binary. i915.ko: - .text 000b0899 + .text 000b0619 Or if we look at the Gen8 display irq chain: -00000000000006ad t gen8_irq_handler +0000000000000663 t gen8_irq_handler -0000000000000028 T intel_opregion_asle_intr +0000000000000024 T intel_opregion_asle_intr -000000000000008c t ilk_hpd_irq_handler +000000000000007f t ilk_hpd_irq_handler -0000000000000116 T intel_check_page_flip +0000000000000112 T intel_check_page_flip -000000000000011a T intel_prepare_page_flip +0000000000000119 T intel_prepare_page_flip -0000000000000014 T intel_finish_page_flip_plane +0000000000000013 T intel_finish_page_flip_plane -0000000000000053 t hsw_pipe_crc_irq_handler +000000000000004c t hsw_pipe_crc_irq_handler -000000000000022e t cpt_irq_handler +0000000000000213 t cpt_irq_handler So small shrinkage but it is all fast paths so doesn't harm. Situation is similar in other interrupt handlers as well. v2: Tidy intel_queue_rps_boost_for_request as well. (Chris Wilson) Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
2016-05-06 07:48:28 -06:00
extern bool ironlake_set_drps(struct drm_i915_private *dev_priv, u8 val);
extern void intel_init_pch_refclk(struct drm_i915_private *dev_priv);
extern int intel_set_rps(struct drm_i915_private *dev_priv, u8 val);
extern bool intel_set_memory_cxsr(struct drm_i915_private *dev_priv,
bool enable);
int i915_reg_read_ioctl(struct drm_device *dev, void *data,
struct drm_file *file);
/* overlay */
extern struct intel_overlay_error_state *
intel_overlay_capture_error_state(struct drm_i915_private *dev_priv);
extern void intel_overlay_print_error_state(struct drm_i915_error_state_buf *e,
struct intel_overlay_error_state *error);
extern struct intel_display_error_state *
intel_display_capture_error_state(struct drm_i915_private *dev_priv);
extern void intel_display_print_error_state(struct drm_i915_error_state_buf *e,
struct intel_display_error_state *error);
int sandybridge_pcode_read(struct drm_i915_private *dev_priv, u32 mbox, u32 *val);
int sandybridge_pcode_write(struct drm_i915_private *dev_priv, u32 mbox, u32 val);
drm/i915/gen9: Fix PCODE polling during CDCLK change notification commit 848496e5902833600f7992f4faa82dc1546051ba Author: Ville Syrjälä <ville.syrjala@linux.intel.com> Date: Wed Jul 13 16:32:03 2016 +0300 drm/i915: Wait up to 3ms for the pcu to ack the cdclk change request on SKL increased the timeout to match the spec, but we still see a timeout on at least one SKL. A CDCLK change request following the failed one will succeed nevertheless. I could reproduce this problem easily by running kms_pipe_crc_basic in a loop. In all failure cases _wait_for() was pre-empted for >3ms and so in the worst case - when the pre-emption happened right after calculating timeout__ in _wait_for() - we called skl_cdclk_wait_for_pcu_ready() only once which failed and so _wait_for() timed out. As opposed to this the spec says to keep retrying the request for at most a 3ms period. To fix this send the first request explicitly to guarantee that there is 3ms between the first and last request. Though this matches the spec, I noticed that in rare cases this can still time out if we sent only a few requests (in the worst case 2) _and_ PCODE is busy for some reason even after a previous request and a 3ms delay. To work around this retry the polling with pre-emption disabled to maximize the number of requests. Also increase the timeout to 10ms to account for interrupts that could reduce the number of requests. With this change I couldn't trigger the problem. v2: - Use 1ms poll period instead of 10us. (Chris) v3: - Poll with pre-emption disabled to increase the number of request attempts. (Ville, Chris) - Factor out a helper to poll, it's also needed by the next patch. v4: - Pass reply_mask, reply to skl_pcode_request(), instead of assuming the reply is generic. (Ville) v5: - List the request specific timeout values as code comment. (Ville) v6: - Try the poll first with preemption enabled. - Add code comment about first request being queued by PCODE. (Art) - Add timeout_base_ms argument. (Ville) v7: - Clarify code comment about first queued request. (Chris) Cc: Ville Syrjälä <ville.syrjala@linux.intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Art Runyan <arthur.j.runyan@intel.com> Cc: <stable@vger.kernel.org> # v4.2- : 3b2c171 : drm/i915: Wait up to 3ms Cc: <stable@vger.kernel.org> # v4.2- Fixes: 5d96d8afcfbb ("drm/i915/skl: Deinit/init the display at suspend/resume") Reference: https://bugs.freedesktop.org/show_bug.cgi?id=97929 Testcase: igt/kms_pipe_crc_basic/suspend-read-crc-pipe-B Signed-off-by: Imre Deak <imre.deak@intel.com> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Link: http://patchwork.freedesktop.org/patch/msgid/1480955258-26311-1-git-send-email-imre.deak@intel.com
2016-12-05 09:27:37 -07:00
int skl_pcode_request(struct drm_i915_private *dev_priv, u32 mbox, u32 request,
u32 reply_mask, u32 reply, int timeout_base_ms);
/* intel_sideband.c */
u32 vlv_punit_read(struct drm_i915_private *dev_priv, u32 addr);
int vlv_punit_write(struct drm_i915_private *dev_priv, u32 addr, u32 val);
u32 vlv_nc_read(struct drm_i915_private *dev_priv, u8 addr);
u32 vlv_iosf_sb_read(struct drm_i915_private *dev_priv, u8 port, u32 reg);
void vlv_iosf_sb_write(struct drm_i915_private *dev_priv, u8 port, u32 reg, u32 val);
u32 vlv_cck_read(struct drm_i915_private *dev_priv, u32 reg);
void vlv_cck_write(struct drm_i915_private *dev_priv, u32 reg, u32 val);
u32 vlv_ccu_read(struct drm_i915_private *dev_priv, u32 reg);
void vlv_ccu_write(struct drm_i915_private *dev_priv, u32 reg, u32 val);
u32 vlv_bunit_read(struct drm_i915_private *dev_priv, u32 reg);
void vlv_bunit_write(struct drm_i915_private *dev_priv, u32 reg, u32 val);
u32 vlv_dpio_read(struct drm_i915_private *dev_priv, enum pipe pipe, int reg);
void vlv_dpio_write(struct drm_i915_private *dev_priv, enum pipe pipe, int reg, u32 val);
u32 intel_sbi_read(struct drm_i915_private *dev_priv, u16 reg,
enum intel_sbi_destination destination);
void intel_sbi_write(struct drm_i915_private *dev_priv, u16 reg, u32 value,
enum intel_sbi_destination destination);
u32 vlv_flisdsi_read(struct drm_i915_private *dev_priv, u32 reg);
void vlv_flisdsi_write(struct drm_i915_private *dev_priv, u32 reg, u32 val);
/* intel_dpio_phy.c */
void bxt_port_to_phy_channel(struct drm_i915_private *dev_priv, enum port port,
enum dpio_phy *phy, enum dpio_channel *ch);
void bxt_ddi_phy_set_signal_level(struct drm_i915_private *dev_priv,
enum port port, u32 margin, u32 scale,
u32 enable, u32 deemphasis);
void bxt_ddi_phy_init(struct drm_i915_private *dev_priv, enum dpio_phy phy);
void bxt_ddi_phy_uninit(struct drm_i915_private *dev_priv, enum dpio_phy phy);
bool bxt_ddi_phy_is_enabled(struct drm_i915_private *dev_priv,
enum dpio_phy phy);
bool bxt_ddi_phy_verify_state(struct drm_i915_private *dev_priv,
enum dpio_phy phy);
uint8_t bxt_ddi_phy_calc_lane_lat_optim_mask(struct intel_encoder *encoder,
uint8_t lane_count);
void bxt_ddi_phy_set_lane_optim_mask(struct intel_encoder *encoder,
uint8_t lane_lat_optim_mask);
uint8_t bxt_ddi_phy_get_lane_lat_optim_mask(struct intel_encoder *encoder);
void chv_set_phy_signal_level(struct intel_encoder *encoder,
u32 deemph_reg_value, u32 margin_reg_value,
bool uniq_trans_scale);
void chv_data_lane_soft_reset(struct intel_encoder *encoder,
bool reset);
void chv_phy_pre_pll_enable(struct intel_encoder *encoder);
void chv_phy_pre_encoder_enable(struct intel_encoder *encoder);
void chv_phy_release_cl2_override(struct intel_encoder *encoder);
void chv_phy_post_pll_disable(struct intel_encoder *encoder);
void vlv_set_phy_signal_level(struct intel_encoder *encoder,
u32 demph_reg_value, u32 preemph_reg_value,
u32 uniqtranscale_reg_value, u32 tx3_demph);
void vlv_phy_pre_pll_enable(struct intel_encoder *encoder);
void vlv_phy_pre_encoder_enable(struct intel_encoder *encoder);
void vlv_phy_reset_lanes(struct intel_encoder *encoder);
int intel_gpu_freq(struct drm_i915_private *dev_priv, int val);
int intel_freq_opcode(struct drm_i915_private *dev_priv, int val);
u64 intel_rc6_residency_us(struct drm_i915_private *dev_priv,
const i915_reg_t reg);
#define I915_READ8(reg) dev_priv->uncore.funcs.mmio_readb(dev_priv, (reg), true)
#define I915_WRITE8(reg, val) dev_priv->uncore.funcs.mmio_writeb(dev_priv, (reg), (val), true)
#define I915_READ16(reg) dev_priv->uncore.funcs.mmio_readw(dev_priv, (reg), true)
#define I915_WRITE16(reg, val) dev_priv->uncore.funcs.mmio_writew(dev_priv, (reg), (val), true)
#define I915_READ16_NOTRACE(reg) dev_priv->uncore.funcs.mmio_readw(dev_priv, (reg), false)
#define I915_WRITE16_NOTRACE(reg, val) dev_priv->uncore.funcs.mmio_writew(dev_priv, (reg), (val), false)
#define I915_READ(reg) dev_priv->uncore.funcs.mmio_readl(dev_priv, (reg), true)
#define I915_WRITE(reg, val) dev_priv->uncore.funcs.mmio_writel(dev_priv, (reg), (val), true)
#define I915_READ_NOTRACE(reg) dev_priv->uncore.funcs.mmio_readl(dev_priv, (reg), false)
#define I915_WRITE_NOTRACE(reg, val) dev_priv->uncore.funcs.mmio_writel(dev_priv, (reg), (val), false)
/* Be very careful with read/write 64-bit values. On 32-bit machines, they
* will be implemented using 2 32-bit writes in an arbitrary order with
* an arbitrary delay between them. This can cause the hardware to
* act upon the intermediate value, possibly leading to corruption and
* machine death. For this reason we do not support I915_WRITE64, or
* dev_priv->uncore.funcs.mmio_writeq.
*
* When reading a 64-bit value as two 32-bit values, the delay may cause
* the two reads to mismatch, e.g. a timestamp overflowing. Also note that
* occasionally a 64-bit register does not actualy support a full readq
* and must be read using two 32-bit reads.
*
* You have been warned.
*/
#define I915_READ64(reg) dev_priv->uncore.funcs.mmio_readq(dev_priv, (reg), true)
#define I915_READ64_2x32(lower_reg, upper_reg) ({ \
u32 upper, lower, old_upper, loop = 0; \
upper = I915_READ(upper_reg); \
do { \
old_upper = upper; \
lower = I915_READ(lower_reg); \
upper = I915_READ(upper_reg); \
} while (upper != old_upper && loop++ < 2); \
(u64)upper << 32 | lower; })
#define POSTING_READ(reg) (void)I915_READ_NOTRACE(reg)
#define POSTING_READ16(reg) (void)I915_READ16_NOTRACE(reg)
#define __raw_read(x, s) \
static inline uint##x##_t __raw_i915_read##x(const struct drm_i915_private *dev_priv, \
drm/i915: Type safe register read/write Make I915_READ and I915_WRITE more type safe by wrapping the register offset in a struct. This should eliminate most of the fumbles we've had with misplaced parens. This only takes care of normal mmio registers. We could extend the idea to other register types and define each with its own struct. That way you wouldn't be able to accidentally pass the wrong thing to a specific register access function. The gpio_reg setup is probably the ugliest thing left. But I figure I'd just leave it for now, and wait for some divine inspiration to strike before making it nice. As for the generated code, it's actually a bit better sometimes. Eg. looking at i915_irq_handler(), we can see the following change: lea 0x70024(%rdx,%rax,1),%r9d mov $0x1,%edx - movslq %r9d,%r9 - mov %r9,%rsi - mov %r9,-0x58(%rbp) - callq *0xd8(%rbx) + mov %r9d,%esi + mov %r9d,-0x48(%rbp) callq *0xd8(%rbx) So previously gcc thought the register offset might be signed and decided to sign extend it, just in case. The rest appears to be mostly just minor shuffling of instructions. v2: i915_mmio_reg_{offset,equal,valid}() helpers added s/_REG/_MMIO/ in the register defines mo more switch statements left to worry about ring_emit stuff got sorted in a prep patch cmd parser, lrc context and w/a batch buildup also in prep patch vgpu stuff cleaned up and moved to a prep patch all other unrelated changes split out v3: Rebased due to BXT DSI/BLC, MOCS, etc. v4: Rebased due to churn, s/i915_mmio_reg_t/i915_reg_t/ Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Link: http://patchwork.freedesktop.org/patch/msgid/1447853606-2751-1-git-send-email-ville.syrjala@linux.intel.com
2015-11-18 06:33:26 -07:00
i915_reg_t reg) \
{ \
drm/i915: Type safe register read/write Make I915_READ and I915_WRITE more type safe by wrapping the register offset in a struct. This should eliminate most of the fumbles we've had with misplaced parens. This only takes care of normal mmio registers. We could extend the idea to other register types and define each with its own struct. That way you wouldn't be able to accidentally pass the wrong thing to a specific register access function. The gpio_reg setup is probably the ugliest thing left. But I figure I'd just leave it for now, and wait for some divine inspiration to strike before making it nice. As for the generated code, it's actually a bit better sometimes. Eg. looking at i915_irq_handler(), we can see the following change: lea 0x70024(%rdx,%rax,1),%r9d mov $0x1,%edx - movslq %r9d,%r9 - mov %r9,%rsi - mov %r9,-0x58(%rbp) - callq *0xd8(%rbx) + mov %r9d,%esi + mov %r9d,-0x48(%rbp) callq *0xd8(%rbx) So previously gcc thought the register offset might be signed and decided to sign extend it, just in case. The rest appears to be mostly just minor shuffling of instructions. v2: i915_mmio_reg_{offset,equal,valid}() helpers added s/_REG/_MMIO/ in the register defines mo more switch statements left to worry about ring_emit stuff got sorted in a prep patch cmd parser, lrc context and w/a batch buildup also in prep patch vgpu stuff cleaned up and moved to a prep patch all other unrelated changes split out v3: Rebased due to BXT DSI/BLC, MOCS, etc. v4: Rebased due to churn, s/i915_mmio_reg_t/i915_reg_t/ Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Link: http://patchwork.freedesktop.org/patch/msgid/1447853606-2751-1-git-send-email-ville.syrjala@linux.intel.com
2015-11-18 06:33:26 -07:00
return read##s(dev_priv->regs + i915_mmio_reg_offset(reg)); \
}
#define __raw_write(x, s) \
static inline void __raw_i915_write##x(const struct drm_i915_private *dev_priv, \
drm/i915: Type safe register read/write Make I915_READ and I915_WRITE more type safe by wrapping the register offset in a struct. This should eliminate most of the fumbles we've had with misplaced parens. This only takes care of normal mmio registers. We could extend the idea to other register types and define each with its own struct. That way you wouldn't be able to accidentally pass the wrong thing to a specific register access function. The gpio_reg setup is probably the ugliest thing left. But I figure I'd just leave it for now, and wait for some divine inspiration to strike before making it nice. As for the generated code, it's actually a bit better sometimes. Eg. looking at i915_irq_handler(), we can see the following change: lea 0x70024(%rdx,%rax,1),%r9d mov $0x1,%edx - movslq %r9d,%r9 - mov %r9,%rsi - mov %r9,-0x58(%rbp) - callq *0xd8(%rbx) + mov %r9d,%esi + mov %r9d,-0x48(%rbp) callq *0xd8(%rbx) So previously gcc thought the register offset might be signed and decided to sign extend it, just in case. The rest appears to be mostly just minor shuffling of instructions. v2: i915_mmio_reg_{offset,equal,valid}() helpers added s/_REG/_MMIO/ in the register defines mo more switch statements left to worry about ring_emit stuff got sorted in a prep patch cmd parser, lrc context and w/a batch buildup also in prep patch vgpu stuff cleaned up and moved to a prep patch all other unrelated changes split out v3: Rebased due to BXT DSI/BLC, MOCS, etc. v4: Rebased due to churn, s/i915_mmio_reg_t/i915_reg_t/ Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Link: http://patchwork.freedesktop.org/patch/msgid/1447853606-2751-1-git-send-email-ville.syrjala@linux.intel.com
2015-11-18 06:33:26 -07:00
i915_reg_t reg, uint##x##_t val) \
{ \
drm/i915: Type safe register read/write Make I915_READ and I915_WRITE more type safe by wrapping the register offset in a struct. This should eliminate most of the fumbles we've had with misplaced parens. This only takes care of normal mmio registers. We could extend the idea to other register types and define each with its own struct. That way you wouldn't be able to accidentally pass the wrong thing to a specific register access function. The gpio_reg setup is probably the ugliest thing left. But I figure I'd just leave it for now, and wait for some divine inspiration to strike before making it nice. As for the generated code, it's actually a bit better sometimes. Eg. looking at i915_irq_handler(), we can see the following change: lea 0x70024(%rdx,%rax,1),%r9d mov $0x1,%edx - movslq %r9d,%r9 - mov %r9,%rsi - mov %r9,-0x58(%rbp) - callq *0xd8(%rbx) + mov %r9d,%esi + mov %r9d,-0x48(%rbp) callq *0xd8(%rbx) So previously gcc thought the register offset might be signed and decided to sign extend it, just in case. The rest appears to be mostly just minor shuffling of instructions. v2: i915_mmio_reg_{offset,equal,valid}() helpers added s/_REG/_MMIO/ in the register defines mo more switch statements left to worry about ring_emit stuff got sorted in a prep patch cmd parser, lrc context and w/a batch buildup also in prep patch vgpu stuff cleaned up and moved to a prep patch all other unrelated changes split out v3: Rebased due to BXT DSI/BLC, MOCS, etc. v4: Rebased due to churn, s/i915_mmio_reg_t/i915_reg_t/ Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Link: http://patchwork.freedesktop.org/patch/msgid/1447853606-2751-1-git-send-email-ville.syrjala@linux.intel.com
2015-11-18 06:33:26 -07:00
write##s(val, dev_priv->regs + i915_mmio_reg_offset(reg)); \
}
__raw_read(8, b)
__raw_read(16, w)
__raw_read(32, l)
__raw_read(64, q)
__raw_write(8, b)
__raw_write(16, w)
__raw_write(32, l)
__raw_write(64, q)
#undef __raw_read
#undef __raw_write
/* These are untraced mmio-accessors that are only valid to be used inside
* critical sections, such as inside IRQ handlers, where forcewake is explicitly
* controlled.
*
* Think twice, and think again, before using these.
*
* As an example, these accessors can possibly be used between:
*
* spin_lock_irq(&dev_priv->uncore.lock);
* intel_uncore_forcewake_get__locked();
*
* and
*
* intel_uncore_forcewake_put__locked();
* spin_unlock_irq(&dev_priv->uncore.lock);
*
*
* Note: some registers may not need forcewake held, so
* intel_uncore_forcewake_{get,put} can be omitted, see
* intel_uncore_forcewake_for_reg().
*
* Certain architectures will die if the same cacheline is concurrently accessed
* by different clients (e.g. on Ivybridge). Access to registers should
* therefore generally be serialised, by either the dev_priv->uncore.lock or
* a more localised lock guarding all access to that bank of registers.
*/
#define I915_READ_FW(reg__) __raw_i915_read32(dev_priv, (reg__))
#define I915_WRITE_FW(reg__, val__) __raw_i915_write32(dev_priv, (reg__), (val__))
#define I915_WRITE64_FW(reg__, val__) __raw_i915_write64(dev_priv, (reg__), (val__))
#define POSTING_READ_FW(reg__) (void)I915_READ_FW(reg__)
/* "Broadcast RGB" property */
#define INTEL_BROADCAST_RGB_AUTO 0
#define INTEL_BROADCAST_RGB_FULL 1
#define INTEL_BROADCAST_RGB_LIMITED 2
static inline i915_reg_t i915_vgacntrl_reg(struct drm_i915_private *dev_priv)
{
if (IS_VALLEYVIEW(dev_priv) || IS_CHERRYVIEW(dev_priv))
return VLV_VGACNTRL;
else if (INTEL_GEN(dev_priv) >= 5)
return CPU_VGACNTRL;
else
return VGACNTRL;
}
static inline unsigned long msecs_to_jiffies_timeout(const unsigned int m)
{
unsigned long j = msecs_to_jiffies(m);
return min_t(unsigned long, MAX_JIFFY_OFFSET, j + 1);
}
static inline unsigned long nsecs_to_jiffies_timeout(const u64 n)
{
/* nsecs_to_jiffies64() does not guard against overflow */
if (NSEC_PER_SEC % HZ &&
div_u64(n, NSEC_PER_SEC) >= MAX_JIFFY_OFFSET / HZ)
return MAX_JIFFY_OFFSET;
return min_t(u64, MAX_JIFFY_OFFSET, nsecs_to_jiffies64(n) + 1);
}
static inline unsigned long
timespec_to_jiffies_timeout(const struct timespec *value)
{
unsigned long j = timespec_to_jiffies(value);
return min_t(unsigned long, MAX_JIFFY_OFFSET, j + 1);
}
/*
* If you need to wait X milliseconds between events A and B, but event B
* doesn't happen exactly after event A, you record the timestamp (jiffies) of
* when event A happened, then just before event B you call this function and
* pass the timestamp as the first argument, and X as the second argument.
*/
static inline void
wait_remaining_ms_from_jiffies(unsigned long timestamp_jiffies, int to_wait_ms)
{
unsigned long target_jiffies, tmp_jiffies, remaining_jiffies;
/*
* Don't re-read the value of "jiffies" every time since it may change
* behind our back and break the math.
*/
tmp_jiffies = jiffies;
target_jiffies = timestamp_jiffies +
msecs_to_jiffies_timeout(to_wait_ms);
if (time_after(target_jiffies, tmp_jiffies)) {
remaining_jiffies = target_jiffies - tmp_jiffies;
while (remaining_jiffies)
remaining_jiffies =
schedule_timeout_uninterruptible(remaining_jiffies);
}
}
static inline bool
__i915_request_irq_complete(const struct drm_i915_gem_request *req)
drm/i915: Slaughter the thundering i915_wait_request herd One particularly stressful scenario consists of many independent tasks all competing for GPU time and waiting upon the results (e.g. realtime transcoding of many, many streams). One bottleneck in particular is that each client waits on its own results, but every client is woken up after every batchbuffer - hence the thunder of hooves as then every client must do its heavyweight dance to read a coherent seqno to see if it is the lucky one. Ideally, we only want one client to wake up after the interrupt and check its request for completion. Since the requests must retire in order, we can select the first client on the oldest request to be woken. Once that client has completed his wait, we can then wake up the next client and so on. However, all clients then incur latency as every process in the chain may be delayed for scheduling - this may also then cause some priority inversion. To reduce the latency, when a client is added or removed from the list, we scan the tree for completed seqno and wake up all the completed waiters in parallel. Using igt/benchmarks/gem_latency, we can demonstrate this effect. The benchmark measures the number of GPU cycles between completion of a batch and the client waking up from a call to wait-ioctl. With many concurrent waiters, with each on a different request, we observe that the wakeup latency before the patch scales nearly linearly with the number of waiters (before external factors kick in making the scaling much worse). After applying the patch, we can see that only the single waiter for the request is being woken up, providing a constant wakeup latency for every operation. However, the situation is not quite as rosy for many waiters on the same request, though to the best of my knowledge this is much less likely in practice. Here, we can observe that the concurrent waiters incur extra latency from being woken up by the solitary bottom-half, rather than directly by the interrupt. This appears to be scheduler induced (having discounted adverse effects from having a rbtree walk/erase in the wakeup path), each additional wake_up_process() costs approximately 1us on big core. Another effect of performing the secondary wakeups from the first bottom-half is the incurred delay this imposes on high priority threads - rather than immediately returning to userspace and leaving the interrupt handler to wake the others. To offset the delay incurred with additional waiters on a request, we could use a hybrid scheme that did a quick read in the interrupt handler and dequeued all the completed waiters (incurring the overhead in the interrupt handler, not the best plan either as we then incur GPU submission latency) but we would still have to wake up the bottom-half every time to do the heavyweight slow read. Or we could only kick the waiters on the seqno with the same priority as the current task (i.e. in the realtime waiter scenario, only it is woken up immediately by the interrupt and simply queues the next waiter before returning to userspace, minimising its delay at the expense of the chain, and also reducing contention on its scheduler runqueue). This is effective at avoid long pauses in the interrupt handler and at avoiding the extra latency in realtime/high-priority waiters. v2: Convert from a kworker per engine into a dedicated kthread for the bottom-half. v3: Rename request members and tweak comments. v4: Use a per-engine spinlock in the breadcrumbs bottom-half. v5: Fix race in locklessly checking waiter status and kicking the task on adding a new waiter. v6: Fix deciding when to force the timer to hide missing interrupts. v7: Move the bottom-half from the kthread to the first client process. v8: Reword a few comments v9: Break the busy loop when the interrupt is unmasked or has fired. v10: Comments, unnecessary churn, better debugging from Tvrtko v11: Wake all completed waiters on removing the current bottom-half to reduce the latency of waking up a herd of clients all waiting on the same request. v12: Rearrange missed-interrupt fault injection so that it works with igt/drv_missed_irq_hang v13: Rename intel_breadcrumb and friends to intel_wait in preparation for signal handling. v14: RCU commentary, assert_spin_locked v15: Hide BUG_ON behind the compiler; report on gem_latency findings. v16: Sort seqno-groups by priority so that first-waiter has the highest task priority (and so avoid priority inversion). v17: Add waiters to post-mortem GPU hang state. v18: Return early for a completed wait after acquiring the spinlock. Avoids adding ourselves to the tree if the is already complete, and skips the awkward question of why we don't do completion wakeups for waits earlier than or equal to ourselves. v19: Prepare for init_breadcrumbs to fail. Later patches may want to allocate during init, so be prepared to propagate back the error code. Testcase: igt/gem_concurrent_blit Testcase: igt/benchmarks/gem_latency Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: "Rogozhkin, Dmitry V" <dmitry.v.rogozhkin@intel.com> Cc: "Gong, Zhipeng" <zhipeng.gong@intel.com> Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> Cc: Dave Gordon <david.s.gordon@intel.com> Cc: "Goel, Akash" <akash.goel@intel.com> Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> #v18 Link: http://patchwork.freedesktop.org/patch/msgid/1467390209-3576-6-git-send-email-chris@chris-wilson.co.uk
2016-07-01 10:23:15 -06:00
{
struct intel_engine_cs *engine = req->engine;
u32 seqno;
/* Note that the engine may have wrapped around the seqno, and
* so our request->global_seqno will be ahead of the hardware,
* even though it completed the request before wrapping. We catch
* this by kicking all the waiters before resetting the seqno
* in hardware, and also signal the fence.
*/
if (test_bit(DMA_FENCE_FLAG_SIGNALED_BIT, &req->fence.flags))
return true;
/* The request was dequeued before we were awoken. We check after
* inspecting the hw to confirm that this was the same request
* that generated the HWS update. The memory barriers within
* the request execution are sufficient to ensure that a check
* after reading the value from hw matches this request.
*/
seqno = i915_gem_request_global_seqno(req);
if (!seqno)
return false;
/* Before we do the heavier coherent read of the seqno,
* check the value (hopefully) in the CPU cacheline.
*/
if (__i915_gem_request_completed(req, seqno))
return true;
drm/i915: Slaughter the thundering i915_wait_request herd One particularly stressful scenario consists of many independent tasks all competing for GPU time and waiting upon the results (e.g. realtime transcoding of many, many streams). One bottleneck in particular is that each client waits on its own results, but every client is woken up after every batchbuffer - hence the thunder of hooves as then every client must do its heavyweight dance to read a coherent seqno to see if it is the lucky one. Ideally, we only want one client to wake up after the interrupt and check its request for completion. Since the requests must retire in order, we can select the first client on the oldest request to be woken. Once that client has completed his wait, we can then wake up the next client and so on. However, all clients then incur latency as every process in the chain may be delayed for scheduling - this may also then cause some priority inversion. To reduce the latency, when a client is added or removed from the list, we scan the tree for completed seqno and wake up all the completed waiters in parallel. Using igt/benchmarks/gem_latency, we can demonstrate this effect. The benchmark measures the number of GPU cycles between completion of a batch and the client waking up from a call to wait-ioctl. With many concurrent waiters, with each on a different request, we observe that the wakeup latency before the patch scales nearly linearly with the number of waiters (before external factors kick in making the scaling much worse). After applying the patch, we can see that only the single waiter for the request is being woken up, providing a constant wakeup latency for every operation. However, the situation is not quite as rosy for many waiters on the same request, though to the best of my knowledge this is much less likely in practice. Here, we can observe that the concurrent waiters incur extra latency from being woken up by the solitary bottom-half, rather than directly by the interrupt. This appears to be scheduler induced (having discounted adverse effects from having a rbtree walk/erase in the wakeup path), each additional wake_up_process() costs approximately 1us on big core. Another effect of performing the secondary wakeups from the first bottom-half is the incurred delay this imposes on high priority threads - rather than immediately returning to userspace and leaving the interrupt handler to wake the others. To offset the delay incurred with additional waiters on a request, we could use a hybrid scheme that did a quick read in the interrupt handler and dequeued all the completed waiters (incurring the overhead in the interrupt handler, not the best plan either as we then incur GPU submission latency) but we would still have to wake up the bottom-half every time to do the heavyweight slow read. Or we could only kick the waiters on the seqno with the same priority as the current task (i.e. in the realtime waiter scenario, only it is woken up immediately by the interrupt and simply queues the next waiter before returning to userspace, minimising its delay at the expense of the chain, and also reducing contention on its scheduler runqueue). This is effective at avoid long pauses in the interrupt handler and at avoiding the extra latency in realtime/high-priority waiters. v2: Convert from a kworker per engine into a dedicated kthread for the bottom-half. v3: Rename request members and tweak comments. v4: Use a per-engine spinlock in the breadcrumbs bottom-half. v5: Fix race in locklessly checking waiter status and kicking the task on adding a new waiter. v6: Fix deciding when to force the timer to hide missing interrupts. v7: Move the bottom-half from the kthread to the first client process. v8: Reword a few comments v9: Break the busy loop when the interrupt is unmasked or has fired. v10: Comments, unnecessary churn, better debugging from Tvrtko v11: Wake all completed waiters on removing the current bottom-half to reduce the latency of waking up a herd of clients all waiting on the same request. v12: Rearrange missed-interrupt fault injection so that it works with igt/drv_missed_irq_hang v13: Rename intel_breadcrumb and friends to intel_wait in preparation for signal handling. v14: RCU commentary, assert_spin_locked v15: Hide BUG_ON behind the compiler; report on gem_latency findings. v16: Sort seqno-groups by priority so that first-waiter has the highest task priority (and so avoid priority inversion). v17: Add waiters to post-mortem GPU hang state. v18: Return early for a completed wait after acquiring the spinlock. Avoids adding ourselves to the tree if the is already complete, and skips the awkward question of why we don't do completion wakeups for waits earlier than or equal to ourselves. v19: Prepare for init_breadcrumbs to fail. Later patches may want to allocate during init, so be prepared to propagate back the error code. Testcase: igt/gem_concurrent_blit Testcase: igt/benchmarks/gem_latency Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: "Rogozhkin, Dmitry V" <dmitry.v.rogozhkin@intel.com> Cc: "Gong, Zhipeng" <zhipeng.gong@intel.com> Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> Cc: Dave Gordon <david.s.gordon@intel.com> Cc: "Goel, Akash" <akash.goel@intel.com> Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> #v18 Link: http://patchwork.freedesktop.org/patch/msgid/1467390209-3576-6-git-send-email-chris@chris-wilson.co.uk
2016-07-01 10:23:15 -06:00
/* Ensure our read of the seqno is coherent so that we
* do not "miss an interrupt" (i.e. if this is the last
* request and the seqno write from the GPU is not visible
* by the time the interrupt fires, we will see that the
* request is incomplete and go back to sleep awaiting
* another interrupt that will never come.)
*
* Strictly, we only need to do this once after an interrupt,
* but it is easier and safer to do it every time the waiter
* is woken.
*/
if (engine->irq_seqno_barrier &&
test_and_clear_bit(ENGINE_IRQ_BREADCRUMB, &engine->irq_posted)) {
struct intel_breadcrumbs *b = &engine->breadcrumbs;
/* The ordering of irq_posted versus applying the barrier
* is crucial. The clearing of the current irq_posted must
* be visible before we perform the barrier operation,
* such that if a subsequent interrupt arrives, irq_posted
* is reasserted and our task rewoken (which causes us to
* do another __i915_request_irq_complete() immediately
* and reapply the barrier). Conversely, if the clear
* occurs after the barrier, then an interrupt that arrived
* whilst we waited on the barrier would not trigger a
* barrier on the next pass, and the read may not see the
* seqno update.
*/
engine->irq_seqno_barrier(engine);
/* If we consume the irq, but we are no longer the bottom-half,
* the real bottom-half may not have serialised their own
* seqno check with the irq-barrier (i.e. may have inspected
* the seqno before we believe it coherent since they see
* irq_posted == false but we are still running).
*/
spin_lock_irq(&b->irq_lock);
if (b->irq_wait && b->irq_wait->tsk != current)
/* Note that if the bottom-half is changed as we
* are sending the wake-up, the new bottom-half will
* be woken by whomever made the change. We only have
* to worry about when we steal the irq-posted for
* ourself.
*/
wake_up_process(b->irq_wait->tsk);
spin_unlock_irq(&b->irq_lock);
if (__i915_gem_request_completed(req, seqno))
return true;
}
drm/i915: Slaughter the thundering i915_wait_request herd One particularly stressful scenario consists of many independent tasks all competing for GPU time and waiting upon the results (e.g. realtime transcoding of many, many streams). One bottleneck in particular is that each client waits on its own results, but every client is woken up after every batchbuffer - hence the thunder of hooves as then every client must do its heavyweight dance to read a coherent seqno to see if it is the lucky one. Ideally, we only want one client to wake up after the interrupt and check its request for completion. Since the requests must retire in order, we can select the first client on the oldest request to be woken. Once that client has completed his wait, we can then wake up the next client and so on. However, all clients then incur latency as every process in the chain may be delayed for scheduling - this may also then cause some priority inversion. To reduce the latency, when a client is added or removed from the list, we scan the tree for completed seqno and wake up all the completed waiters in parallel. Using igt/benchmarks/gem_latency, we can demonstrate this effect. The benchmark measures the number of GPU cycles between completion of a batch and the client waking up from a call to wait-ioctl. With many concurrent waiters, with each on a different request, we observe that the wakeup latency before the patch scales nearly linearly with the number of waiters (before external factors kick in making the scaling much worse). After applying the patch, we can see that only the single waiter for the request is being woken up, providing a constant wakeup latency for every operation. However, the situation is not quite as rosy for many waiters on the same request, though to the best of my knowledge this is much less likely in practice. Here, we can observe that the concurrent waiters incur extra latency from being woken up by the solitary bottom-half, rather than directly by the interrupt. This appears to be scheduler induced (having discounted adverse effects from having a rbtree walk/erase in the wakeup path), each additional wake_up_process() costs approximately 1us on big core. Another effect of performing the secondary wakeups from the first bottom-half is the incurred delay this imposes on high priority threads - rather than immediately returning to userspace and leaving the interrupt handler to wake the others. To offset the delay incurred with additional waiters on a request, we could use a hybrid scheme that did a quick read in the interrupt handler and dequeued all the completed waiters (incurring the overhead in the interrupt handler, not the best plan either as we then incur GPU submission latency) but we would still have to wake up the bottom-half every time to do the heavyweight slow read. Or we could only kick the waiters on the seqno with the same priority as the current task (i.e. in the realtime waiter scenario, only it is woken up immediately by the interrupt and simply queues the next waiter before returning to userspace, minimising its delay at the expense of the chain, and also reducing contention on its scheduler runqueue). This is effective at avoid long pauses in the interrupt handler and at avoiding the extra latency in realtime/high-priority waiters. v2: Convert from a kworker per engine into a dedicated kthread for the bottom-half. v3: Rename request members and tweak comments. v4: Use a per-engine spinlock in the breadcrumbs bottom-half. v5: Fix race in locklessly checking waiter status and kicking the task on adding a new waiter. v6: Fix deciding when to force the timer to hide missing interrupts. v7: Move the bottom-half from the kthread to the first client process. v8: Reword a few comments v9: Break the busy loop when the interrupt is unmasked or has fired. v10: Comments, unnecessary churn, better debugging from Tvrtko v11: Wake all completed waiters on removing the current bottom-half to reduce the latency of waking up a herd of clients all waiting on the same request. v12: Rearrange missed-interrupt fault injection so that it works with igt/drv_missed_irq_hang v13: Rename intel_breadcrumb and friends to intel_wait in preparation for signal handling. v14: RCU commentary, assert_spin_locked v15: Hide BUG_ON behind the compiler; report on gem_latency findings. v16: Sort seqno-groups by priority so that first-waiter has the highest task priority (and so avoid priority inversion). v17: Add waiters to post-mortem GPU hang state. v18: Return early for a completed wait after acquiring the spinlock. Avoids adding ourselves to the tree if the is already complete, and skips the awkward question of why we don't do completion wakeups for waits earlier than or equal to ourselves. v19: Prepare for init_breadcrumbs to fail. Later patches may want to allocate during init, so be prepared to propagate back the error code. Testcase: igt/gem_concurrent_blit Testcase: igt/benchmarks/gem_latency Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: "Rogozhkin, Dmitry V" <dmitry.v.rogozhkin@intel.com> Cc: "Gong, Zhipeng" <zhipeng.gong@intel.com> Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> Cc: Dave Gordon <david.s.gordon@intel.com> Cc: "Goel, Akash" <akash.goel@intel.com> Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> #v18 Link: http://patchwork.freedesktop.org/patch/msgid/1467390209-3576-6-git-send-email-chris@chris-wilson.co.uk
2016-07-01 10:23:15 -06:00
return false;
}
drm/i915: Use SSE4.1 movntdqa to accelerate reads from WC memory This patch provides the infrastructure for performing a 16-byte aligned read from WC memory using non-temporal instructions introduced with sse4.1. Using movntdqa we can bypass the CPU caches and read directly from memory and ignoring the page attributes set on the CPU PTE i.e. negating the impact of an otherwise UC access. Copying using movntdqa from WC is almost as fast as reading from WB memory, modulo the possibility of both hitting the CPU cache or leaving the data in the CPU cache for the next consumer. (The CPU cache itself my be flushed for the region of the movntdqa and on later access the movntdqa reads from a separate internal buffer for the cacheline.) The write back to the memory is however cached. This will be used in later patches to accelerate accessing WC memory. v2: Report whether the accelerated copy is successful/possible. v3: Function alignment override was only necessary when using the function target("sse4.1") - which is not necessary for emitting movntdqa from __asm__. v4: Improve notes on CPU cache behaviour vs non-temporal stores. v5: Fix byte offsets for unrolled moves. v6: Find all remaining typos of "movntqda", use kernel_fpu_begin. Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Akash Goel <akash.goel@intel.com> Cc: Damien Lespiau <damien.lespiau@intel.com> Cc: Mika Kuoppala <mika.kuoppala@intel.com> Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1471001999-17787-2-git-send-email-chris@chris-wilson.co.uk
2016-08-12 05:39:59 -06:00
void i915_memcpy_init_early(struct drm_i915_private *dev_priv);
bool i915_memcpy_from_wc(void *dst, const void *src, unsigned long len);
/* The movntdqa instructions used for memcpy-from-wc require 16-byte alignment,
* as well as SSE4.1 support. i915_memcpy_from_wc() will report if it cannot
* perform the operation. To check beforehand, pass in the parameters to
* to i915_can_memcpy_from_wc() - since we only care about the low 4 bits,
* you only need to pass in the minor offsets, page-aligned pointers are
* always valid.
*
* For just checking for SSE4.1, in the foreknowledge that the future use
* will be correctly aligned, just use i915_has_memcpy_from_wc().
*/
#define i915_can_memcpy_from_wc(dst, src, len) \
i915_memcpy_from_wc((void *)((unsigned long)(dst) | (unsigned long)(src) | (len)), NULL, 0)
#define i915_has_memcpy_from_wc() \
i915_memcpy_from_wc(NULL, NULL, 0)
/* i915_mm.c */
int remap_io_mapping(struct vm_area_struct *vma,
unsigned long addr, unsigned long pfn, unsigned long size,
struct io_mapping *iomap);
static inline bool
intel_engine_can_store_dword(struct intel_engine_cs *engine)
{
return __intel_engine_can_store_dword(INTEL_GEN(engine->i915),
engine->class);
}
#endif