This is the main merge request for Nouveau 3.18, overview:
- various bits of roy's gt21x clock work
- various bits of kepler memory clock work (don't get too excited, there's at least one more major bit left that's busting higher freqs)
- misc fan control improvements
- kepler hdmi infoframe fixes
- dp audio
- l2 cache + cbc improvements
* 'linux-3.18' of git://anongit.freedesktop.org/git/nouveau/linux-2.6: (68 commits)
drm/gt214-/disp: enable dp audio
drm/gt214-/kms: fix hda eld regression
drm/g94-/disp: calculate some dp audio constants
drm/gt214-/kms: perform hda codec setup on displayport too
drm/gk104-/disp: infoframe registers moved yet again on kepler
drm/nouveau/bios: parse older ramcfg/timing data like we do newer ones
drm/nva3/fb/ram: Per-partition regs
drm/nouveau/fb/ram: Support strided regs
drm/nv50/fb/ram: Store the number of partitions in the designated fields
drm/nv50/kms: Set VBLANK time in modeset script
drm/nouveau/bios: Add rammap support for version 1.0
drm/gf100-/pwr/memx: block host and fifo around reclock
drm/nouveau/pwr/memx: fix command ordering around block/unblock
drm/nouveau/pwr/memx: rename fb off/on to block/unblock
drm/nva3/clk: Pause the GPU before reclocking
drm/nouveau/gpio: rename g92 class to g94
drm/gk104-/fb/ram: move fb enable/disable to same place as nvidia
drm/gk104/fb/ram: twiddle some more bits when reclocking
drm/nouveau/bios: parse another large chunk of random memory config data
drm/gk104-/fb/ram: perform certain steps only when bios data differs
...
Revert parts of f244d8b623 ("ACPIPHP / radeon / nouveau: Fix VGA
switcheroo problem related to hotplug").
A previous commit 5493b31f0b55 ("PCI: Add pci_ignore_hotplug() to ignore
hotplug events for a device") added equivalent functionality implemented in
a different way for both acpiphp and pciehp.
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Acked-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Dave Airlie <airlied@redhat.com>
Acked-by: Rajat Jain <rajatxjain@gmail.com>
NVIDIA appear to have tweaked the algorithm from GF110, this implements
the previous algorithm for them still.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Solves blinking on reclocking memory. The value set is an underestimate, but
with non-reduced vblanking this should give us plenty of time
Signed-off-by: Roy Spliet <rspliet@eclipso.eu>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
More accurate as to the function of the opcodes. Not only is FB disabled,
but the host is prevented from touching the GPU. An upcoming patch for
Kepler will also halt PFIFO (as NVIDIA does).
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
nv92 hardware has only 16 interrupt lines, while nv94 and later
has 32. Accessing 0xe0c{0,4} registers on nv92 can lead to incorrect
PDISP setup. This is a regression introduced with
commit 9d0f5ec9ee0fd5dc5fc1cc2cf559286431e406e3
Author: Ben Skeggs <bskeggs@redhat.com>
Date: Mon May 12 15:22:42 2014 +1000
gpio: split g92 class from nv50
Reported-by: estece on #nouveau
Cc: stable@vger.kernel.org # 3.16+
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
*when* this is done is only a rough approximation of what the binary driver
does.. need to investigate more to see if it matters
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Awful, awful. But, on the GK106 I have, some upcoming patches show
that this is actually necessary after all.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
All the other chipsets should be moved over to this too. It's not needed
yet for the upcoming commits, so left this step as it'll conflict badly
with Roy's GT21x reclocking work.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
NVIDIA binary driver appears to, not sure if it's for a good reason, but
grasping at straws for some GDDR5 reclocking issues here.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Time measured from disabling FB to re-enabling, PPWR_IN reveals status of
heads at the end of script. Helps debug various issues (like flicker).
Signed-off-by: Roy Spliet <rspliet@eclipso.eu>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Needs to be done after wait-for-VBLANK, and NVA3 requires register writes
in between.
Rather than hard-coding register writes, just split out fb_disable and
fb_enable.
v2. Squashed "fb/ramnve0: disable fb before reclocking"
Signed-off-by: Roy Spliet <rspliet@eclipso.eu>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
One of my nv92 has a calibrated internal sensor but it displays 0°C
as the default values use sw calibration values to force the temperature
to 0.
Since we cannot read the temperature from the adt7473 present on this board,
let's re-enable the internal reading!
Signed-off-by: Martin Peres <martin.peres@free.fr>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
We will use this subdev to disable temperature reading on cards that did not
get a sensor calibration in the factory.
v2:
- rename "nouveau_fuse_rd32" to "gxXXX_fuse_rd32" as adviced by Christian Costa
- fold the code a little as adviced by Emil Velikov
Signed-off-by: Martin Peres <martin.peres@free.fr>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
It can help to remove any ambiguity about which options were passed to Nouveau,
especially in case the user had some options set in /etc/modprobe.d/*.conf that
he forgot about, as they won't appear in a dmesg.
Signed-off-by: Pierre Moreau <pierre.morrow@free.fr>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
The problem with the current implementation is that adding a timer improperly
checked which process would time up first by not taking into account how much
time elapsed since their timer got scheduled. Rework the re-scheduling
decision t fix this.
The catch with this fix is that we are limited to scheduling timers of up to
2^31 ticks to avoid any potential overflow. Since we are unlikely to need to
wait for more than a second, this won't be a problem :)
Another possible fix would be to decrement the timeouts of all processes but
it would duplicate a lot of code and dealing with edge cases wasn't pretty
last time I checked.
Signed-off-by: Martin Peres <martin.peres@free.fr>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
For some reason, it is now required to wait a 20 µs after the 0x200 reset of
the engine.
Signed-off-by: Martin Peres <martin.peres@free.fr>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
v2: change the copyright ownership from "Nouveau Community" to myself, as per
Illia's recommendation.
Signed-off-by: Martin Peres <martin.peres@free.fr>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Re-use the therm-exported fan structure with only two minor modifications:
- pwm_freq: u16 -> u32;
- add fan_type (toggle or PWM)
v2:
- Do not memset the table to 0 as it erases the pre-set default values
Signed-off-by: Martin Peres <martin.peres@free.fr>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
The allocation algorithm doesn't expect there to be holes in the mm, which
causes its alignment/cutoff calculations to choke (and go negative) when
encountering the last chunk of a block before a hole.
The least expensive solution is to simply fill in any holes with nodes
that are pre-marked as being allocated.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
This is really the wrong thing to do, but at the time it was our only
option to prevent worse issues.
We no longer cause quite so much anger from LTC, so it's not needed.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Here's the updated topic/core-stuff pull request with the two patches
already merged into drm-fixes dropped.
* tag 'topic/core-stuff-2014-09-15' of git://anongit.freedesktop.org/drm-intel:
drm: Drop modeset locking from crtc init function
drm/i915/hdmi: Enable pipe pixel replication for SD interlaced modes
drm/edid: Reduce horizontal timings for pixel replicated modes
drm: Include task->name and master status in debugfs clients info
drm/gem: Fix kerneldoc typo
drm: use c99 initializers in structures
drm: fix drm_modeset_lock.h kernel-doc notation
In preparation for DT support where panel timings will be described by a
DRM-agnostic video mode, replace the struct drm_mode_modeinfo instance
in the panel platform data with a struct videomode.
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
The "Renesas Corporation" listed in the copyright notice doesn't exist.
Replace it with "Renesas Electronics Corporation" and update the
copyright years.
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
The "Renesas Corporation" listed in the copyright notice doesn't exist.
Replace it with "Renesas Electronics Corporation" and update the
copyright years.
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
At driver init no one can access modeset objects and we're
single-threaded. So locking is just cargo-culting here. Worse, with
the new ww mutexes and ww mutex slowpath debugging the mutex_lock
might actually fail, and we don't have the full-blown ww recovery
dance.
Which then leads to fireworks when we try to unlock the not-locked
crtc lock.
An audit of all the functions called from here shows that none of them
contain locking checks, so there's also no reason to keep the locking
around just for consistency of caller contexts. Besides that I have
the rule (at least in i915) that such places where we take locks just
to simplify locking checks and not for correctness always require a
comment.
This regression was introduced in
commit 51fd371bba
Author: Rob Clark <robdclark@gmail.com>
Date: Tue Nov 19 12:10:12 2013 -0500
drm: convert crtc and connection_mutex to ww_mutex (v5)
v2: Don't drop the lock_init call, spotted by the 0day builder.
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=83341
Cc: Rob Clark <robdclark@gmail.com>
Cc: thellstrom@vmware.com
Cc: maarten.lankhorst@canonical.com
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Enable 2x pixel replication for modes the mode flag DBLCLK to double
horizontal timings and pixel clock across TMDS.
Signed-off-by: Clint Taylor <clinton.a.taylor@intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Pixel replicated modes should be non-2x horizontal timings and pixel
replicated by the HW across the HDMI cable at 2X pixel clock. Current
horizontal resolution of 1440 does not allow pixel duplication to
occur and scaling artifacts occur on the TV. HDMI certification
7-26 currently fails for all pixel replicated modes. This change will
allow HDMI certification with 480i/576i modes once pixel replication
is turned on.
Signed-off-by: Clint Taylor <clinton.a.taylor@intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Showing who is the current master is useful for trying to decypher
errors when trying to acquire master (e.g. a race with X taking over
from plymouth). By including the process name as well as the pid
simplifies the task of grabbing enough information remotely at the point
of error.
v2: Add the command column header and flesh out a couple of comments.
(David Herrmann)
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: David Herrmann <dh.herrmann@gmail.com>
Reviewed-by: David Herrmann <dh.herrmann@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The drm_gem_private_object_init function is called drm_gem_object_init
in its kerneldoc. Fix it.
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
One small change I forgot to make in
commit c4d69da167
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date: Mon Sep 8 14:25:41 2014 +0100
drm/i915: Evict CS TLBs between batches
was to update the copy width for the compact BLT copy instruction.
Reported-by: Thomas Richter <thor@math.tu-berlin.de>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Thomas Richter <thor@math.tu-berlin.de>
Cc: Jani Nikula <jani.nikula@intel.com>
Cc: stable@vger.kernel.org
Tested-by: Thomas Richter <thor@math.tu-berlin.de>
Acked-by: Daniel Vetter <daniel@ffwll.ch>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Move the MMIO mangling to a separate routine and actually
disable the DVO output when using pure analog.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>
It looks like the AST2400 comes up with the DVO enable bit set,
which causes us to incorrectly assume we have a SIL164 regardless
of the value of the scratch registers setup by the BMC firmware.
So let's limit that test to the case where the chip has already
been setup by a BIOS.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>
If the P2A has been used to target other SOC registers before that
call, we're going to hit the wrong place so make sure we set the
base address up properly before using it.
(P2A stands for PCIe to AHB bridge and is the bride that allows
accessing the AST's internal AHB bus using a relocatable 64k
window in the second half of the PCIe MMIO BAR)
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>
We need to do it on machines without a BIOS such as POWER8. Also
for detection to work without triggering PCIe errors, we need
to enable VGA early on, inside ast_detect_chip().
While touching those files, replace a few hard coded register
numbers with the corresponding symbolic constant.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>
If the PIO resources haven't been assigned, then we have no choice
but try to use the MMIO version. This is the case for example on
POWER8 which doesn't support PIO at all.
Chips rev 0x20 or later have MMIO decoding enabled by default.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Use c99 initializers for structures.
Drop 0 initializers in drivers/gpu/drm/sti/sti_vtac.c. A 0x0 initializer
is left in vtac_mode_aux in drivers/gpu/drm/sti/sti_vtac.c to highlight the
relation to vtac_mode_main.
A simplified version of the semantic match that finds the first problem is
as follows: (http://coccinelle.lip6.fr/)
// <smpl>
@decl@
identifier i1,fld;
type T;
field list[n] fs;
@@
struct i1 {
fs
T fld;
...};
@bad@
identifier decl.i1,i2;
expression e;
initializer list[decl.n] is;
@@
struct i1 i2 = { is,
+ .fld = e
- e
,...};
// </smpl>
v2: Drop 0 initializers and add trailing commas at the suggestions of Josh
Triplett.
Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
A few odd cases:
- mgag200 someho had a totally unused drm_dma_handle_t. Remove it.
- i915 still uses the legacy pci dma alloc api, so grows an include.
Everything else fairly standard.
v2: Include "drm_legacy.h" in drm.ko source files for consistency.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
And replace the drm_core_ prefix with drm_legacy_ since really, this
isn't core stuff.
Also drop drm_core_dropmap since it's unused.
v2: Fix up i810.ko fully which somehow slipped through.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
If we push down the ioctl table in drm_ioctl.c all the forward
declarations in drmP.h are not required any more.
v2: Fold in fixup from Fenugguang Wu to declare functions as static.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Drivers really, really have no business even looking at this lock. And
thankfully they don't.
So unexport it and move the declaration to drm_internal.h.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
This way drivers can't grow crazy ideas any more, and it also
helps a bit in reviewing EXPORT_SYMBOLS.
v2: Even more stuff. Unfortunately we can't move drm_vm_open_locked
because exynos does some horrible stuff with it.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
This allows us to drop 2 header declarations from drmP.h. The 3rd one
is also used in drm_ioctl.c, so for that create a new drm_internal.h
header for non-legacy non-kms (since we have internal headers for
those parts already) declarations private to drm.ko.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Also sprinkle the customary legacy_ prefix.
Unfortunately we can't move the other functions since i915 is still
using them. Shame on me for that one :(
v2: Fix patch subject as spotted by David Herrmann.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
And move a few legayc functions to start things over there.
It compiles ...
Inspired by a patch from Dave Airlie, but with a split between drm.ko
private legacy functions and stuff used by drivers.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Also sprinkle the drm_legacy_ prefix where missing.
v2: Drop extern from function declarations and include "drm_legacy.h"
in drm_scatter.c, spotted by David.
Cc: David Herrmann <dh.herrmann@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Also drop the unneeded EXPORT_SYMBOL and sprinkle drm_legacy_ prefixes
where missing.
v2: Drop the confusing _core_ and drop extern, both suggested by
David.
Cc: David Herrmann <dh.herrmann@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
So updated vblank-rework pull request, now with the polish that Mario
requested applied (and reviewed by him). Also with backmerge like you've
requested for easier merging.
The neat thing this finally allows is to immediately disable the vblank
interrupt on the last drm_vblank_put if the hardware has perfectly
accurate vblank counter and timestamp readout support. On i915 that
required piles of small adjustements from Ville since depending upon the
platform and port the vblank happens at different scanout lines.
Of course this is fully opt-in and per-device (we need that since gen2
doesn't have a hw vblank counter).
* tag 'topic/vblank-rework-2014-09-12' of git://anongit.freedesktop.org/drm-intel: (22 commits)
drm: Clarify vblank ts/scanoutpos sampling #defines
drm: Simplify return value of drm_get_last_vbltimestamp
drm: Only update final vblank count when precise ts is available
drm: Really never disable vblank irqs for offdelay==0
drm: Use vblank_disable_and_save in drm_vblank_cleanup()
drm: Remove drm_vblank_cleanup from drm_vblank_init error path.
drm: Store the vblank timestamp when adjusting the counter during disable
drm: Fix confusing debug message in drm_update_vblank_count()
drm/i915: Update scanline_offset only for active crtcs
drm: Kick start vblank interrupts at drm_vblank_on()
drm/i915: Opt out of vblank disable timer on >gen2
drm: Add dev->vblank_disable_immediate flag
drm: Disable vblank interrupt immediately when drm_vblank_offdelay<0
drm: Fix race between drm_vblank_off() and drm_queue_vblank_event()
drm: Fix deadlock between event_lock and vbl_lock/vblank_time_lock
drm: Reduce the amount of dev->vblank[crtc] in the code
drm: Avoid random vblank counter jumps if the hardware counter has been reset
drm: Have the vblank counter account for the time between vblank irq disable and drm_vblank_off()
drm: Move drm_update_vblank_count()
drm: Don't clear vblank timestamps when vblank interrupt is disabled
...
This fixes problems on ppc64 platforms, where we could end up using
a WC mapping for migrating BOs with memcpy, when really we want to
use cached memory.
Tested-by: Ben Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Just move this into a separate header file, and make the
two users use it.
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Type error and cause AST2000 cannot be detected correctly
Signed-off-by: Y.C. Chen <yc_chen@aspeedtech.com>
Reviewed-by: Egbert Eich <eich@suse.de>
Cc: stable@vger.kernel.org
Signed-off-by: Dave Airlie <airlied@redhat.com>
Some config settings like 3rd TX chips will not get correctly
if the extended reg is protected
Signed-off-by: Y.C. Chen <yc_chen@aspeedtech.com>
Reviewed-by: Egbert Eich <eich@suse.de>
Cc: stable@vger.kernel.org
Signed-off-by: Dave Airlie <airlied@redhat.com>
This allows us to specify if we want to sync to
the shared fences of a reservation object or not.
Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This patch adds a new flag to the ttm_validate_buffer list to
add the fence as shared to the reservation object.
Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Dave asked me to do the backmerge before sending him the revised pull
request, so here we go. Nothing fancy in the conflicts, just a few
things changed right next to each another.
Conflicts:
drivers/gpu/drm/drm_irq.c
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
I've read INVBL as "invalid backlight" and got mightly confused.
The #defines are already fairly long and we can afford to extend
them a bit more without resulting in ugly code all over.
I'm not sure how useful the complicated bitmask return value of these
functions really are since no one checks them. But for now let's keep
things as is.
Cc: Mario Kleiner <mario.kleiner.de@gmail.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Mario Kleiner <mario.kleiner.de@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Imo u32 hints at a register value, but in reality all callers only
care whether the sampled timestamp is precise or not. So give them
just a bool.
Also move the declaration out of drmP.h, it's only used in drm_irq.c.
v2: Also drop the EXPORT_SYMBOL, spotted by Mario.
Cc: Mario Kleiner <mario.kleiner.de@gmail.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Mario Kleiner <mario.kleiner.de@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Drivers without a hardware vblank counter simply can't account for the
vblanks that happened while the vblank interrupt was off. To check
this grab a vblank timestamp and if the result is dubious follow the
normal save-and-disable logic.
Drivers should prevent this by setting vblank_disable_allowed = false,
but since running vblank interrupts constantly is not good for power
consumption most drivers lie. Testing for precise vblank timestamps is
the next best thing we can check for.
Suggested-by: Mario Kleiner <mario.kleiner.de@gmail.com>
Cc: Mario Kleiner <mario.kleiner.de@gmail.com>
Cc: Matt Roper <matthew.d.roper@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Mario Kleiner <mario.kleiner.de@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
With the new support for immediate vblank disabling we always disabled
the vblank interrupt right away, irrespective of the vblank offdelay
setting.
But being able to let vblanks run forever is fairly useful for
debugging, so restore that behaviour.
Suggested-by: Mario Kleiner <mario.kleiner.de@gmail.com>
Cc: Mario Kleiner <mario.kleiner.de@gmail.com>
Cc: Matt Roper <matthew.d.roper@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Mario Kleiner <mario.kleiner.de@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
1) add LVDS support for mdp4 (tested with auo B101XTN01.0 panel)
2) add B101XTN01.0 panel
3) bit of gpu refactoring to prepare for addition of addition gpu
generations beyond just a3xx
* 'msm-next' of git://people.freedesktop.org/~robclark/linux:
drm/msm/adreno: push dump/show stuff to base class
drm/msm/adreno: bit of init refactoring
drm/msm/adreno: move decision about what gpu to to load
drm/msm/adreno: split adreno device out into it's own file
drm/panel/simple: add optronics B101XTN01.0 (v3)
drm/msm/mdp4: add LVDS panel support
drm/msm/mdp4: fix blend setup with multiple crtcs
drm/msm: update generated headers
If VRAM carveout is used, due to no IOMMU, we should have a default
value for msm.vram so that we don't simply crash.
Reported-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Signed-off-by: Rob Clark <robdclark@gmail.com>
There is currently a nested function in Russel King's tree
for the msm HDMI driver.
The last nested function was removed from the Linux kernel
when the Thinkpad driver was fixed.
I believe nested functions are not desired upstream, and it
also breaks compilation with clang so here is a patch to
change the nested function into static function. The patch
works with both clang and gcc.
Signed-off-by: Mark Charlebois <charlebm@gmail.com>
Signed-off-by: Rob Clark <robdclark@gmail.com>
more fixes for 3.17, almost all Cc: stable material.
* tag 'drm-intel-fixes-2014-09-10' of git://anongit.freedesktop.org/drm-intel:
drm/i915: Wait for vblank before enabling the TV encoder
drm/i915: Evict CS TLBs between batches
drm/i915: Fix irq enable tracking in driver load
drm/i915: Fix EIO/wedged handling in gem fault handler
drm/i915: Prevent recursive deadlock on releasing a busy userptr
Powering off a hot-pluggable device, e.g., with pci_set_power_state(D3cold),
normally generates a hot-remove event that unbinds the driver.
Some drivers expect to remain bound to a device even while they power it
off and back on again. This can be dangerous, because if the device is
removed or replaced while it is powered off, the driver doesn't know that
anything changed. But some drivers accept that risk.
Add pci_ignore_hotplug() for use by drivers that know their device cannot
be removed. Using pci_ignore_hotplug() tells the PCI core that hot-plug
events for the device should be ignored.
The radeon and nouveau drivers use this to switch between a low-power,
integrated GPU and a higher-power, higher-performance discrete GPU. They
power off the unused GPU, but they want to remain bound to it.
This is a reimplementation of f244d8b623 ("ACPIPHP / radeon / nouveau:
Fix VGA switcheroo problem related to hotplug") but extends it to work with
both acpiphp and pciehp.
This fixes a problem where systems with dual GPUs using the radeon drivers
become unusable, freezing every few seconds (see bugzillas below). The
resume of the radeon device may also fail, e.g.,
This fixes problems on dual GPU systems where the radeon driver becomes
unusable because of problems while suspending the device, as in bug 79701:
[drm] radeon: finishing device.
radeon 0000:01:00.0: Userspace still has active objects !
radeon 0000:01:00.0: ffff8800cb4ec288 ffff8800cb4ec000 16384 4294967297 force free
...
WARNING: CPU: 0 PID: 67 at /home/apw/COD/linux/drivers/gpu/drm/radeon/radeon_gart.c:234 radeon_gart_unbind+0xd2/0xe0 [radeon]()
trying to unbind memory from uninitialized GART !
or while resuming it, as in bug 77261:
radeon 0000:01:00.0: ring 0 stalled for more than 10158msec
radeon 0000:01:00.0: GPU lockup ...
radeon 0000:01:00.0: GPU pci config reset
pciehp 0000:00:01.0:pcie04: Card not present on Slot(1-1)
radeon 0000:01:00.0: GPU reset succeeded, trying to resume
*ERROR* radeon: dpm resume failed
radeon 0000:01:00.0: Wait for MC idle timedout !
Link: https://bugzilla.kernel.org/show_bug.cgi?id=77261
Link: https://bugzilla.kernel.org/show_bug.cgi?id=79701
Reported-by: Shawn Starr <shawn.starr@rogers.com>
Reported-by: Jose P. <lbdkmjdf@sharklasers.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Acked-by: Rajat Jain <rajatxjain@gmail.com>
Acked-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Acked-by: Dave Airlie <airlied@redhat.com>
CC: stable@vger.kernel.org # v3.15+
Allows pinning of buffers in the non-CPU visible portion of
vram.
v2: incorporate Michel's comments.
v3: rebase on Michel's patch
v4: rebase on Michel's v2 patch
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
This sets the lpfn member to 0 instead of the full domain size. TTM uses
the full domain size when lpfn is 0.
Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This flag is a hint that userspace expects the BO to be accessed by the
CPU. We can use that hint to prevent such BOs from ever being stored in
the CPU inaccessible part of VRAM.
Signed-off-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Add ptr to list of interesting registers to 'struct adreno_gpu' and use
that to move most of the debugfs show and register dump bits down into
adreno_gpu. This will avoid duplication as support for additional
adreno generations is added.
Signed-off-by: Rob Clark <robdclark@gmail.com>
Push a few bits down into adreno_gpu so they won't have to be duplicated
as support for additional adreno generations is added.
Signed-off-by: Rob Clark <robdclark@gmail.com>
LVDS panel, make/model described as:
AU Optronics Corporation - B101XTN01.0 (H/W:0A)
See:
http://www.encore-electronic.com/media/B101XTN01.0.pdf
Tested with panel attached to an Inforce IFC6410 board.
Signed-off-by: Rob Clark <robdclark@gmail.com>
LVDS panel support uses the LCDC (parallel) encoder. Unlike with HDMI,
there is not a separate LVDS block, so no need to split things into a
bridge+connector. Nor is there is anything re-used with mdp5.
Note that there can be some regulators shared between HDMI and LVDS (in
particular, on apq8064, ext_3v3p), so we should not use the _exclusive()
variants of devm_regulator_get().
The drm_panel framework is used for panel-specific driver.
Signed-off-by: Rob Clark <robdclark@gmail.com>
In particular, blend_setup() should not overwrite the other crtc's mixer
settings. Also, the encoder needs to be able to specify the mixer-id
explicitly, since both LVDS and DTV use 'INTF_LVDC_DTV', so we cannot
guess the mixer-id from the interface.
Signed-off-by: Rob Clark <robdclark@gmail.com>
Same as the other legacy APIs, most of this is internal, so prefix it with
drm_legacy_* and move into drm_legacy.h.
Signed-off-by: David Herrmann <dh.herrmann@gmail.com>
Reviewed-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
This merges all the remains of drm_usb into its only user, udl. We can
then drop all the drm_usb stuff, including dev->usbdev.
Signed-off-by: David Herrmann <dh.herrmann@gmail.com>
Reviewed-by: Thierry Reding <treding@nvidia.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Dave Airlie <airlied@redhat.com>
..we will not miss you..
Signed-off-by: David Herrmann <dh.herrmann@gmail.com>
Reviewed-by: Thierry Reding <treding@nvidia.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Dave Airlie <airlied@redhat.com>
One step closer to dropping all the drm_bus_* code:
Add a driver->set_busid() callback and make all drivers use the generic
helpers. Nouveau is the only driver that uses two different bus-types with
the same drm_driver. This is totally broken if both buses are available on
the same machine (unlikely, but lets be safe). Therefore, we create two
different drivers for each platform during module_init() and set the
set_busid() callback respectively.
Signed-off-by: David Herrmann <dh.herrmann@gmail.com>
Reviewed-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
This field is unused and there is really no reason to optimize
unique-allocations. Drop it.
Signed-off-by: David Herrmann <dh.herrmann@gmail.com>
Reviewed-by: Thierry Reding <treding@nvidia.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Lets use kasprintf() to avoid pre-allocating the buffer. This is really
nothing to optimize for speed and the input is trusted, so kasprintf() is
just fine.
Signed-off-by: David Herrmann <dh.herrmann@gmail.com>
Reviewed-by: Thierry Reding <treding@nvidia.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Dave Airlie <airlied@redhat.com>
The sigdata structure is only used to group two fields in drm_device.
Inline it and make it an unnamed object.
Signed-off-by: David Herrmann <dh.herrmann@gmail.com>
Reviewed-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Calling vblank_disable_fn() will cause that function to no-op
if !dev->vblank_disable_allowed for some kms drivers, e.g.,
on nouveau-kms. This can cause the gpu vblank irq's to not get
disabled before freeing the dev->vblank array, so if a
vblank irq fires and calls into drm_handle_vblank() after
drm_vblank_cleanup() completes, it will cause use-after-free
access to dev->vblank array.
Call vblank_disable_and_save unconditionally, so vblank irqs
are guaranteed to be off, before we delete the data structures
on which they operate.
Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
[danvet: Fix subsystem name in patch subject.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
DRM_DEBUG_CODE is currently always set, so distributions enable it. The
only reason to keep support in code is if developers wanted to disable
debug support. Sounds unlikely.
All the DRM_DEBUG() printks are still guarded by a drm_debug read. So if
its cacheline is read once, they're discarded pretty fast.. There should
hardly be any performance penalty, it's even guarded by unlikely().
Signed-off-by: David Herrmann <dh.herrmann@gmail.com>
Reviewed-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
The drm_memory.h header is only used to define PAGE_AGP, which is only
used in drm_memory.c. Fold the header into drm_memory.c and drop it.
Signed-off-by: David Herrmann <dh.herrmann@gmail.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Dave Airlie <airlied@redhat.com>
pte_wrprotect() is only used by drm_vm.c, so move the include there. Also
include it unconditionally, all architectures provide this header!
Furthermore, replace asm/current.h with sched.h, which includes
asm/current.h unconditionally. This way we get the same effect and avoid
direct asm/ includes. Furthermore, drop the weird __alpha__ protection.
It's safe to include sched.h everywhere (and the wait.h comment doesn't
apply, anyway).
Signed-off-by: David Herrmann <dh.herrmann@gmail.com>
Reviewed-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Move drm_agp_head to drm_agpsupport.h and drm_agp_mem into drm_legacy.h.
Unfortunately, drivers still heavily access drm_agp_head so we cannot
move it to drm_legacy.h. However, at least it's no longer visible in
drmP.h now (it's directly included from it, though).
Signed-off-by: David Herrmann <dh.herrmann@gmail.com>
Reviewed-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
In drm_release(), we currently call drm_remove_magic() if the drm_file
has a drm-magic attached. Therefore, once drm_master_release() is called,
the magic-list _must_ be empty.
By dropping the no-op cleanup, we can move "struct drm_magic_entry" to
drm_auth.c and avoid exposing it to all of DRM.
Signed-off-by: David Herrmann <dh.herrmann@gmail.com>
Reviewed-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Make all the drm_vma_entry handling local to drm_vm.c and hide it from
global headers. This requires to extract the inlined legacy drm_vma_entry
cleanup into a small helper and also move a weirdly placed drm_vma_info
helper into drm_vm.c.
Signed-off-by: David Herrmann <dh.herrmann@gmail.com>
Reviewed-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Move internal declarations to drm_legacy.h and add drm_legacy_*() prefix
to all legacy functions.
[airlied: add a bit of an explaination to drm_legacy.h]
Signed-off-by: David Herrmann <dh.herrmann@gmail.com>
Reviewed-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Radeon UMS is the last user of drm_buffer. Move it out of sight so radeon
can drop it together with UMS.
Signed-off-by: David Herrmann <dh.herrmann@gmail.com>
Reviewed-by: Thierry Reding <treding@nvidia.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
On systems with special thermal configurations make sure we make
note of the thermal setup. This is required for proper firmware
configuration on these systems.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
There is no need to use hex_dump_to_buffer() since we have a kernel helper to
dump up to 64 bytes just via printk(). In our case the actual size is 15 bytes.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Semaphore values have 64 bits, not 32. This fixes a very subtle bug
that disables synchronization when the upper 32bits wasn't zero.
Signed-off-by: Christian König <christian.koenig@amd.com>
Cc: stable@vger.kernel.org
Reviewed-By: Grigori Goronzy <greg@chown.ath.cx>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The vblank waits in intel_tv_detect_type() are timing out for some
reason. This is a regression caused removing seemingly useless vblank
waits from the modeset seqeuence in:
commit 56ef52cad5
Author: Ville Syrjälä <ville.syrjala@linux.intel.com>
Date: Thu May 8 19:23:15 2014 +0300
drm/i915: Kill vblank waits after pipe enable on gmch platforms
So it turns out they weren't all entirely useless. Apparently the pipe
has to go through one full frame before we enable the TV port. Add a
vblank wait to intel_enable_tv() to make sure that happens.
Another approach was attempted by placing the vblank wait just after
enabling the port. The theory behind that attempt was that we need to
let the port stay enabled for one full frame before disabling it again
during load detection. But that didn't work, and we definitely must
have the vblank wait before enabling the port.
Cc: stable@vger.kernel.org
Cc: Alan Bartlett <ajb@elrepo.org>
Tested-by: Alan Bartlett <ajb@elrepo.org>
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=79311
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Daniel Vetter <daniel@ffwll.ch>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Running igt, I was encountering the invalid TLB bug on my 845g, despite
that it was using the CS workaround. Examining the w/a buffer in the
error state, showed that the copy from the user batch into the
workaround itself was suffering from the invalid TLB bug (the first
cacheline was broken with the first two words reversed). Time to try a
fresh approach. This extends the workaround to write into each page of
our scratch buffer in order to overflow the TLB and evict the invalid
entries. This could be refined to only do so after we update the GTT,
but for simplicity, we do it before each batch.
I suspect this supersedes our current workaround, but for safety keep
doing both.
v2: The magic number shall be 2.
This doesn't conclusively prove that it is the mythical TLB bug we've
been trying to workaround for so long, that it requires touching a number
of pages to prevent the corruption indicates to me that it is TLB
related, but the corruption (the reversed cacheline) is more subtle than
a TLB bug, where we would expect it to read the wrong page entirely.
Oh well, it prevents a reliable hang for me and so probably for others
as well.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Cc: stable@vger.kernel.org
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
A bunch of warnings fire on some ->irq_postinstall hooks since those
can enable interrupts (e.g. rps interrupts). And then our ordering
self-checks fire and complain.
To fix that set the tracking boolen before enabling the irqs with
drm_irq_install. Quoting the discussion with Jesse why that's safe:
On Tue, Aug 26, 2014 at 11:18 PM, Jesse Barnes <jbarnes@virtuousgeek.org> wrote:
> Yes, it might work, but if you look through the history, we set this
> field carefully; first to true in the irq_init code, then to false only
> after the irq_install completes. So I think your fragility arguments
> apply to this change too.
Well we've done it in 4 commits or so, but currently we have:
- Set irqs_disabled to true early in driver load to make sure checks
that. That's done in irq_init, which is totally not the function that
enables interrupts, only the function that initializes all the vtables
and similar things. We actually have a fairly sane naming scheme
nowadays (not fully consistent ofc): _init is sw setup,
_enable/_hw_init is the actual hw setup. That is done in
95f25beddb
- Set irqs_disabled to false right after the irqs are actually
enabled. This is done in ed2e6df189
So my change should only move the flag change over the ->preinstall
and ->postinstall hooks. I've done a little audit and didn't spot
anything amiss. Furthermore the runtime pm setup already clears
irqs_disabled _before_ calling these two hooks.
This regression has been introduced in
commit ed2e6df189
Author: Jesse Barnes <jbarnes@virtuousgeek.org>
Date: Fri Jun 20 09:39:36 2014 -0700
drm/i915: clear pm._irqs_disabled field after installing IRQs
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Cc: Oliver Hartkopp <socketcan@hartkopp.net>
Tested-by: Oliver Hartkopp <socketcan@hartkopp.net>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Tested-by: Chris Wilson <chris@chris-wilson.co.uk> # gm45, ilk
Reviewed-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
In
commit 1f83fee08d
Author: Daniel Vetter <daniel.vetter@ffwll.ch>
Date: Thu Nov 15 17:17:22 2012 +0100
drm/i915: clear up wedged transitions
I've accidentally inverted the EIO/wedged handling in the fault
handler: We want to return the EIO as a SIGBUS only if it's not
because of the gpu having died, to prevent userspace from unduly
dying.
In my defence the comment right above is completely misleading, so fix
both.
v2: Drop the WARN_ON, it's not actually a bug to e.g. receive an -EIO
when swap-in fails.
v3: Don't remove too much ... oops.
Reported-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: stable@vger.kernel.org
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
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>
If we successfully confuse the hardware, and cause it to drop a queued
pageflip, we wait for 60s and issue a warning before continuing on with
the modeset. However, this leaves the pending pageflip still stuck
indefinitely. Pretend to userspace that it does complete, and let us
start afresh following the modeset.
v2: Rebase after refactor
v3: Rebase, rebase.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
References: https://bugs.freedesktop.org/show_bug.cgi?id=82612
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Long ago, back in the racy haydays of 915gm interrupt handling, page
flips would occasionally go astray and leave the hardware stuck, and the
display not updating. This annoyed people who relied on their systems
being able to display continuously updating information 24/7, and so
some code to detect when the driver missed the page flip completion
signal was added. Until recently, it was presumed that the interrupt
handling was now flawless, but once again Simon Farnsworth has found a
system whose display will stall. Reinstate the pageflip stall detection,
which works by checking to see if the hardware has been updated to the
new framebuffer address following each vblank. If the hardware is
scanning out from the new framebuffer, but we still think the flip is
pending, then we kick our driver into submision.
This is a continuation of the effort started with
commit 4e5359cd05
Author: Simon Farnsworth <simon.farnsworth@onelan.co.uk>
Date: Wed Sep 1 17:47:52 2010 +0100
drm/i915: Avoid pageflipping freeze when we miss the flip prepare interrupt
This now includes a belt-and-braces approach to make sure the driver
(or the hardware) doesn't miss an interrupt and cause us to stop
updating the display should the unthinkable happen and the pageflip fail - i.e.
that the user is able to continue submitting flips.
v2: Cleanup, refactor, and rename
v3: Only start counting vblanks after the flip command has been seen by
the hardware.
v4: Record the seqno after we touch the ring, or else there may be no
seqno allocated yet.
v5: Rebase on mmio-flip.
v6: Rebase, rebase.
Reported-by: Simon Farnsworth <simon@farnz.org.uk>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=75502
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com> [v4]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Fixes not being able to init fence subsystem when multiple boards are
present.
Reported-by: Ilia Mirkin <imirkin@alum.mit.edu>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Tired of copy/pasting things around.
v2: Rebase on top of the for_each_pipe() change adding dev_priv as first
argument.
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The new version of the macro does a few things better:
- protect the arguments,
- only evaluate the arguments once,
- check that the arguments are of the same type,
Change LC_FREQ_2K to be a unsigned 64bit constant and removed the '()'
from the caller as a result.
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Jani wanted some comments to explain why we call certain vdd on/off
functions in certain places.
v2: Make the comments more thorough (Imre)
Reviewed-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
We need to turn the DP port off after the pipe, otherwise the pipe won't
turn off properly on certain pch platforms at least (happens on my ILK for
example). This also matches the BSpec modeset sequence better. We still
don't match the spec exactly though (eg. audio disable should happen
much earlier), but at last this eliminates the nasty
wait_for_pipe_off() timeouts.
We already did the port disable after the pipe for VLV/CHV and for CPU
eDP.
For g4x leave the port disable where it is since that matches the
modeset sequence in the documentation and I don't have a suitable
machine to test if the other order would work.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Bspec says we should enable the DP port before enabling panel power,
and that the port must be enabled with training pattern 1. Do so.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
On VLV/CHV the panel power sequencer may need to be "kicked" a bit to
lock onto the new port, and that needs to happen before any aux
transfers are attempted if we want the aux transfers to actaully
succeed. So turn on panel power (part of the "kick") before aux
transfers (DPMS_ON + link training).
This also matches the documented modeset sequence better for pch
platforms. The documentation doesn't explicitly state anything about the
DPMS or link training DPCD writes, but the panel power on step is
always listed before link training is mentioned.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Imre Deak <imre.deak@intel.com>
References: https://bugs.freedesktop.org/show_bug.cgi?id=70117
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Try to make sure we find the power sequencer that the BIOS used
by first looking for one which has the panel power enabled, then
fall back to one with VDD force bit enabled, and finally look at
just the port select bits. This should make us pick the correct
power sequencer when the BIOS has already enabled the panel.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Imre Deak <imre.deak@intel.com>
[danvet: Shorten the vlv_intial_pps_pipe to make lines fit into 80
chars.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The power sequencer loses its state when the disp2d power well is down.
Clear the dev_priv->pps_pipe tracking so that the power sequencer state
gets reinitialized the next time it's needed.
v2: Fix the pps_mutex vs. power_domain mutex deadlock by taking power
domain reference first
v3: Rename from edp_pps_(un)lock() to just pps_(un)lock() for the future,
update due to backlight code changes
Reviewed-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
VLV/CHV have a per-pipe panel power sequencer which locks onto the
port once used. We need to keep track wich power sequencers are
locked to which ports.
v2: remove spurious whitespace change, rebase due to backlight changes (Imre)
Reviewed-by: Antti Koskipaa <antti.koskipaa@linux.intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
[danvet: Break some really long lines to appease checkpatch a bit.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Introduce a new mutex (pps_mutex) to protect the power sequencer
state. For now this state includes want_panel_vdd as well as the
power sequencer registers.
We need a single mutex (as opposed to per port) because later on we
will need to deal with VLV/CHV which have multiple power sequencer
which can be reassigned to different ports.
v2: Add the locking to intel_dp_encoder_suspend too (Imre)
v3: Take care intel_edp_backlight_power() and
_intel_edp_backlight_on/off(), deal with reboot notifier
vlv_power_sequencer_pipe() call (Imre)
Reviewed-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Ville found an old w/a documented for g4x that suggested that we need to
reset the HEAD after writing START. This is a useful fixup for some of
the g4x ring initialisation woes, but as usual, not all.
v2: Do the rewrite unconditionally anyway
References: https://bugs.freedesktop.org/show_bug.cgi?id=76554
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
When unbinding, there is a possibility that we drop the active reference
on the object, thereby freeing it. If that happens, we may destroy the
vm link as well as the object and vma. So iterate carefully.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
here's a couple of display regression fixes for 3.17.
* tag 'drm-intel-fixes-2014-09-03' of git://anongit.freedesktop.org/drm-intel:
drm/i915: Fix lock dropping in intel_tv_detect()
drm/i915: handle G45/GM45 pulse detection connected state.
This crash was already here before the conversion, but qxl never leaked
hard enough to hit this.
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@canonical.com>
The locking of release_lock was stupid; t should have been be called with
fence_lock_irq if it was legitimately used. Unfortunately it never protected
anything except the fence implementation correctly.
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@canonical.com>
Really just for completeness - old init function ends up making the plane
exactly the same way due to the way the enums are set up.
Signed-off-by: Derek Foreman <derek.foreman@collabora.co.uk>
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Check for !HAS_PCH_SPLIT() instead of 'gen < 5' in the PCH transcoder
enable functions.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
A few open coded HAS_GMCH_DISPLAY() remain in the underrun reporting
code. Convert them over.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
IS_GEN8() is a bad check in the forcewake code due to bdw vs. chv
differences. Use IS_BROADWELL() instead.
The only actual bug here is that we currently call
__gen7_gt_force_wake_mt_reset() on chv. On the other places we
have checked for chv before using IS_GEN8(), but change them
to use IS_BROADWELL() anyway to reduce the chance of accidents in the
future.
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>
CHV doesn't have FBC, so don't go calling gen8_fbc_sw_flush() on it.
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
[danvet: Add a FIXME comment while at it that we should rework this a
lot more.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
We use the variable name latency_ns in both the local lowlevel wm
calculation routines and at the global level. Rename the global value to
reduce shadow warnings and future confusion.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
We have CHV code that already makes the test obsolete. Besides, when
num_wa_regs is 0 (platforms not gathering that W/A data), we expose
something sensible already.
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Reviewed-by: Arun Siluvery <arun.siluvery@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
If we happen to emit more than I915_MAX_WA_REGS workarounds, we will
currently discard them, not even emit the LRI. Not really what we want,
so warn loudly.
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Reviewed-by: Arun Siluvery <arun.siluvery@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
When entering intel_ring_emit_wa() with num_wa_regs equal to
I915_MAX_WA_REGS, we end up indexing the intel_wa_regs array beyond its
allocation.
Fix the check then.
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Reviewed-by: Arun Siluvery <arun.siluvery@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Those debugfs files are prefixed by i915, the name of the kernel module,
presumably to make the difference with files exposed by core DRM.
Also, add a ',' at the end of the last entry. This is to ease the
conflict resolution when rebasing internal patches that add a member at
the end of the array. Without it, wiggle can't do its job as we need to
modify an existing line (appending the ',').
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Reviewed-by: Arun Siluvery <arun.siluvery@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Currently, CHV is using the same functions as HSW/BDW instead of the
same functions as VLV. This looks wrong, especially since, for
example, valleyview_modeset_global_resouces even has an IS_CHERRYVIEW
check.
This patch has the potential to fix display audio and the CHV CDCLK.
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
As we may query the edid multiple times following a detect, record the
EDID found during output discovery and reuse it. This is a separate
issue from caching the output EDID across detection cycles.
v2: Also hookup the force() callback for audio detection when the user
forces the connection status.
v3: Ville spots a typo, s/==/!=/
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
As we may query the edid multiple times following a detect, record the
EDID found during output discovery and reuse it. This is a separate
issue from caching the output EDID across detection cycles.
v2: Implement connector->force() callback so that edid is associated
with the connector for user overrides as well (Ville)
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Both gmch and pch detection routines used the exact same routine for
eDP, so de-duplicate.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: : Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Rename the defines to have levels instead of values for vswing and
pre-emph levels as the values may differ in other scenarios like low vswing of
eDP1.4 where the values are different.
Done using following cocci patch for each define:
@@
@@
# define DP_TRAIN_VOLTAGE_SWING_400 (0 << 0)
+ # define DP_TRAIN_VOLTAGE_SWING_LEVEL_0 (0 << 0)
...
Signed-off-by: Sonika Jindal <sonika.jindal@intel.com>
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Rename the defines to have levels instead of values for vswing and
pre-emph levels as the values may differ in other scenarios like low vswing of
eDP1.4 where the values are different.
Done using following cocci patch for each define:
@@
@@
# define DP_TRAIN_VOLTAGE_SWING_400 (0 << 0)
+ # define DP_TRAIN_VOLTAGE_SWING_LEVEL_0 (0 << 0)
...
Signed-off-by: Sonika Jindal <sonika.jindal@intel.com>
Acked-by: Dave Airlie <airlied@gmail.com>
Acked-by: Thierry Reding <thierry.reding@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Rename the defines to have levels instead of values for vswing and
pre-emph levels as the values may differ in other scenarios like low vswing of
eDP1.4 where the values are different.
Done using following cocci patch for each define:
@@
@@
# define DP_TRAIN_VOLTAGE_SWING_1200 (3 << 0)
+ # define DP_TRAIN_VOLTAGE_SWING_LEVEL_3 (0 << 0)
...
Signed-off-by: Sonika Jindal <sonika.jindal@intel.com>
Acked-by: Alex Deucher <alexdeucher@gmail.com>
Acked-by: Dave Airlie <airlied@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Rename the defines to have levels instead of values for vswing and
pre-emph levels as the values may differ in other scenarios like low vswing of
eDP1.4 where the values are different.
Done using following cocci patch for each define:
@@
@@
# define DP_TRAIN_VOLTAGE_SWING_400 (0 << 0)
+ # define DP_TRAIN_VOLTAGE_SWING_LEVEL_0 (0 << 0)
...
Signed-off-by: Sonika Jindal <sonika.jindal@intel.com>
Acked-by: Dave Airlie <airlied@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Rename the defines to have levels instead of values for vswing and
pre-emph levels as the values may differ in other scenarios like low vswing of
eDP1.4 where the values are different.
Done using following cocci patch for each define:
@@
@@
# define DP_TRAIN_VOLTAGE_SWING_400 (0 << 0)
+ # define DP_TRAIN_VOLTAGE_SWING_LEVEL_0 (0 << 0)
...
Signed-off-by: Sonika Jindal <sonika.jindal@intel.com>
Acked-by: Dave Airlie <airlied@gmail.com>
Acked-by: Jingoo Han <jg1.han@samsung.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
No point in calling intel_plane_restore() in .set_property() if the
value didn't change.
More importantly this papers over a bug where the current primary plane
code forgets to update the user coordinates we store under intel_plane
unless the primary plane .update_plane() hook is actually called. This
means we have 0 in the coordinates straight after boot and any call
to intel_restore_plane() (such as from restore_fbdev_mode()) will
actually turn off the primary plane. This mess needs to be fixed properly
but that's a bigger task and the first step there is killing off
intel_pipe_set_base() and just calling the primary plane
.update_plane() hook. For the immediate problem of black screen after
boot this small patch is enough to hide it.
The problem originates from these two commits:
commit 3a5f87c286
Author: Thomas Wood <thomas.wood@intel.com>
Date: Wed Aug 20 14:45:00 2014 +0100
drm: fix plane rotation when restoring fbdev configuration
commit d91a2cb8e5104233c02bbde539bd4ee455ec12ac
Author: Sonika Jindal <sonika.jindal@intel.com>
Date: Fri Aug 22 14:06:04 2014 +0530
drm/i915: Add 180 degree primary plane rotation support
Cc: Thomas Wood <thomas.wood@intel.com>
Cc: Sonika Jindal <sonika.jindal@intel.com>
Tested-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Tested-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
There is no need to use hex_dump_to_buffer() since we have a kernel helper to
dump up to 64 bytes just via printk(). In our case the actual size is 15 bytes.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
[danvet: Add cast since %*ph expects and int for the size parameter.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
WaGsvRC0ResidenncyMethod is for vlv, it doesn't deal with chv
appropriately (eg. doesn't limit rps values to even numbers).
Fix a typo in the w/a name while at it.
Cc: Deepak S <deepak.s@linux.intel.com>
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>
My Fujistsu-Siemens Lifebook S6010 doesn't like to resume from
S3 unless VGACNTR has been restore to the original value. The BIOS
value in this case was 0x0124008E. Setting the "VGA disable" bit
doesn't interfere with the S3 resume fortunately.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Tested-by: Thomas Richter <richter@rus.uni-stuttgart.de>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
830M has problems when some of the pipes are disabled. Namely if a
plane, DVO port etc. is currently assigned to a disabled pipe, it
can't moved to the other pipe until the current pipe is also enabled.
To keep things simple just leave both pipes running all the time.
Ideally I think should turn the pipes off if neither is active, and
when either becomes active we enable both. But that would reuquire
proper atomic modeset support, and probably a bit of extra care in
the order things get enabled.
v2: Reorder wrt. double wide handling changes
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Tested-by: Thomas Richter <richter@rus.uni-stuttgart.de>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
830 really does want the pipe A quirk. The planes and ports don't
react to any register writes unless the pipe currently attached
to them is running, so it's impossible to move them to the other
pipe unless both pipes are running.
Also it's documented that the DPLL must be enabled on both pipes
whenever it's needed.
This reverts commit ac6696d3236bd61503f89a1a99680fd7894d5d53.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Tested-by: Thomas Richter <richter@rus.uni-stuttgart.de>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The vbt on my Fujitsu-Siemens Lifebook S6010 provides two 800x600 modes,
60Hz and 56Hz. The magic register values we have correspond to the 60Hz
mode, and as I don't know how one would trick the VGA BIOS to set up
the 56Hz mode we can't get the magic values for the orther mode. So
when checking whether a mode is valid also check the pixel clock so that
we filter out the 56Hz variant.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Thomas Richter <richter@rus.uni-stuttgart.de>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
In my earlier rewrite I missed a few important registers. Thomas Richter
noticed that they're needed to make his machine resume correctly.
Looks like IEGD does a one time init of these three registers. We don't
have a good one time init place in the ns2501 driver, so let's just
stick them into the .mode_set() hook and see if that helps things along.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Thomas Richter <richter@rus.uni-stuttgart.de>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Try to use the same programming sequence as used by the IEGD driver.
Also shovel the magic register values into a big static const array.
The register values are actually the based on what the BIOS programs
on the Fujitsu-Siemens Lifebook S6010. IEGD seemed to have hardcoded
register values (which also enabled the scaler for 1024x768 mode).
However those didn't actually work so well on the S6010. Possibly the
pipe timings that got used didn't match the ns2501 configuration.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Thomas Richter <richter@rus.uni-stuttgart.de>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Thomas Richter <richter@rus.uni-stuttgart.de>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Calling the mode_set hook on DPMS changes doesn't seem to be necessary
for ns2501. Just drop it.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Tested-by: Thomas Richter <richter@rus.uni-stuttgart.de>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
To more closely match the IEGD ns2501 driver behaviour, call the
mode_set hook while the DVO port is still disabled, then enable the DVO
port, and finally call the dpms hook.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Tested-by: Thomas Richter <richter@rus.uni-stuttgart.de>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
On Fujitsu-Siememens S6010 the ns2501 chip is hooked up to DVOB instead
of DVOC.
FIXME: Maybe need to dig out the correct DVO port from VBT
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Tested-by: Thomas Richter <richter@rus.uni-stuttgart.de>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Disable double wide even if the pipe quirk compels us to leave the
pipe running. Double wide has certain implications for the plane
assignments so best keep it off.
Also helps resuming from S3 on the Fujitsu-Siemens Lifebook S6010
when double wide was enabled prior to suspend.
We do leave the pixel clock ticking at the original rate which would
require double wide to be enabled. But since the planes are all disabled
I'm hoping that the overly fast clock won't cause any problems. Seems
to be fine so far.
v2: Disable double wide also when turning the pipe off
v3: Reorder wrt. force pipe B quirk
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Tested-by: Thomas Richter <richter@rus.uni-stuttgart.de>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Just pass the intel_crtc around instead of dev_priv+pipe.
Also make intel_wait_for_pipe_off() static since it's only used in
intel_display.c.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Tested-by: Thomas Richter <richter@rus.uni-stuttgart.de>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
gen2/3 platforms have a boatload of rings we're not using. On my 830
the BIOS/hw can leave some of those "active" after resume which will
prevent c3 entry. The ring is apparently considered active whenever
head != tail even if the ring is disabled.
Disable and clear all such unused ringbuffers on init/resume.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
My 830 is unhappy with trickle feed enabled. The symptom is that
the image on the screen shifts a bit to right occasionally.
The BIOS initially disables trickle feed, but it gets reset during
suspend, so we need to re-disable it ourselves. Juse disable it
always.
Also disable it for all other gen2/3 platforms since we disable it
for all more recent platforms as well (until HSW that is). At least
my 855 doesn't seem to mind us doing this. I don't have gen3
hardware to test that.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The max watermark value for gen2 planes B and C is 0x1f, instead of
the 0x3f that plane A uses.
Also check against the max even if the pipe is disabled since the
FIFO size exceeds the plane B and C max watermark value.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Tested-by: Thomas Richter <richter@rus.uni-stuttgart.de>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
This gets us out of our init code and out to userspace quite a bit
faster, but does open us up to some bugs given the state of our init
time locking.
v2: switch to async_schedule (Chris)
check with lockdep, seems happy (Jesse)
move hotplug enable flag set to fbdev_initial_config (Jesse)
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
[danvet: Rebase on top of the dev_priv->enable_hotplug_processing
removal.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Follow the BDW example and apply the workarounds touching registers
which are saved in the context image through LRIs in the new
ring->init_context() hook.
This makes Mesa much happier and eg. glxgears doesn't hang after
the first frame.
Cc: Arun Siluvery <arun.siluvery@linux.intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
[danvet: Add missing wa table initialization to avoid a functional
conflict with Arun's wa table debugfs support.]
Reviewed-by: "Barbalho, Rafael" <rafael.barbalho@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
At this point of the code the obj var is already NULL, so we don't
need to set it again to NULL.
Signed-off-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
CHV wants even rps opcodes so print a warning of the
min/max/rpe/rp1 values are odd, and warn if an odd value
slips through to valleyview_set_rps() and truncate it to
an even value.
Also add a comment to chv_freq_opcode() to make sure no one
changes the code without considering this requirement.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Deepak S <deepak.s@linux.intel.com>
[danvet: Help git along in applying the patch, somehow it silently
ended up in the vlv init_gt_powersave function.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The batchbuffer that sets the render context state is submitted
in a different way, and from different places.
We needed to make both the render state preparation and free functions
outside accesible, and namespace accordingly. This mess is so that all
LR, LRC and Execlists functionality can go together in intel_lrc.c: we
can fix all of this later on, once the interfaces are clear.
v2: Create a separate ctx->rcs_initialized for the Execlists case, as
suggested by Chris Wilson.
Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
v3: Setup ring status page in lr_context_deferred_create when the
default context is being created. This means that the render state
init for the default context is no longer a special case. Execute
deferred creation of the default context at the end of
logical_ring_init to allow the render state commands to be submitted.
Fix style errors reported by checkpatch. Rebased.
Signed-off-by: Thomas Daniel <thomas.daniel@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
BDW supports GT C0 residency reporting in constant time unit. Driver
calculates GT utilization based on C0 residency and adjusts RP
frequency up/down accordingly. For offscreen workload specificly,
set frequency to RP0.
Offscreen task is not restricted by frame rate, it can be
executed as soon as possible. Transcoding and serilized workload
between CPU and GPU both need high GT performance, RP0 is a good
option in this case. RC6 will kick in to compensate power
consumption when GT is not active.
v2: Rebase on recent drm-intel-nightly
v3: Add flip timerout monitor, when no flip is deteced within
100ms, set frequency to RP0.
Signed-off-by: Daisy Sun <daisy.sun@intel.com>
[torourke: rebased on latest and resolved conflict]
Signed-off-by: Tom O'Rourke <Tom.O'Rourke@intel.com>
Acked-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
init_clock_gating() is too late to read out the mem_freq. We already
want to print out the GPU MHz numbers before it's called. Move the
mem_freq setup to init_gt_powersave().
v2: Also kill the CHV_CZ_CLOCK_FREQ_MODE_* defines
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>
Use the pixel_size we got from drm_format_plane_cpp() instead of
fb->bits_per_pixel/8 when computing the primary plane page/linear
offsets. Avoids a few divs and makes the code more future proof
against funky pixel formats where bits_per_pixel isn't well defined.
This is what we already did in the sprite code.
Note that the relevant sprite patch was
commit ca320ac456
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date: Wed Dec 19 12:14:22 2012 +0000
drm/i915: Use pixel size for computing linear offsets into a sprite
This change was required on sprites because they support yuv formats
which have fb->bits_per_pixel undefined.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
[danvet: Add Chris' software archeology as a note to the commit
message.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
During driver init we may not have a valid framebuffer for the primary
plane even though the plane is enabled due to failed BIOS fb takeover.
This means we have to avoid dereferencing the fb in
.update_primary_plane() when disabling the plane.
The introduction of the primary plane rotation in
commit d91a2cb8e5104233c02bbde539bd4ee455ec12ac
Author: Sonika Jindal <sonika.jindal@intel.com>
Date: Fri Aug 22 14:06:04 2014 +0530
drm/i915: Add 180 degree primary plane rotation support
caused a regression by trying to look up the pixel format before we can
be sure there's a valid fb available. This isn't entirely unsurprising
since the rotation patches originally predate the change to the primary
plane code that calls .update_primary_plane() also when disabling the
plane:
commit fdd508a641
Author: Ville Syrjälä <ville.syrjala@linux.intel.com>
Date: Fri Aug 8 21:51:11 2014 +0300
drm/i915: Call .update_primary_plane in intel_{enable,
disable}_primary_hw_plane()
v2: Warn but don't blow up when trying to enable a plane w/o an fb (Chris)
Cc: Sonika Jindal <sonika.jindal@intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The workarounds that are applied are exported to a debugfs file;
this is used to verify their state after the test case (reset or
suspend/resume etc). This patch is only required to support i-g-t.
Signed-off-by: Arun Siluvery <arun.siluvery@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
For BDW workarounds are currently initialized in init_clock_gating() but
they are lost during reset, suspend/resume etc; this patch moves the WAs
that are part of register state context to render ring init fn otherwise
default context ends up with incorrect values as they don't get initialized
until init_clock_gating fn.
v2: Add workarounds to golden render state
This method has its own issues, first of all this is different for
each gen and it is generated using a tool so adding new workaround
and mainitaining them across gens is not a straightforward process.
v3: Use LRIs to emit these workarounds (Ville)
Instead of modifying the golden render state the same LRIs are
emitted from within the driver.
v4: Use abstract name when exporting gen specific routines (Chris)
For: VIZ-4092
Signed-off-by: Arun Siluvery <arun.siluvery@linux.intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
According to spec FBC on BDW and HSW are identical without any gaps.
So let's copy the nuke and let FBC really start compressing stuff.
Without this patch we can verify with false color that nothing is being
compressed. With the nuke in place and false color it is possible
to see false color debugs.
Unfortunatelly on some rings like BCS on BDW we have to avoid Bits 22:18 on
LRIs due to a high risk of hung. So, when using Blt ring for frontbuffer rend
cache would never been cleaned and FBC would stop compressing buffer.
One alternative is to cache clean on software frontbuffer tracking.
v2: Fix rebase conflict.
v3: Do not clean cache on BCS ring. Instead use sw frontbuffer tracking.
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Try to avoid confusion with ARRAY_SIZE()/2 and hdmi_level*2.
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
[danvet: Resolve silent patch conflict (didn't even fail to build)
with with Sonika's preceding patch to use the
hsw_ddi_translations_fdi table to driver the fdi link training
iteration loop. Also drop the double-write loop Damien spotted.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Renaming the HSW-specific macros for ddi buffer translation slot to denote the
slot and not the vswing/pre-emph values as they are platform-dependent.
This patch is based on top of the patch series for renaming the DP training
vswing/pre-emph defines:
http://lists.freedesktop.org/archives/intel-gfx/2014-August/050407.html
v2: Creating single macro with argument for slot number (Damien)
v3: Adding macro for num of translation entries (Damien)
Signed-off-by: Sonika Jindal <sonika.jindal@intel.com>
Reviewed-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Previously, it was possible for the GPU memory accesses to be swizzled
to try to optimize the fetches for tiled buffers. This swizzling was on
top of what the memory controller in the uncore already does.
With broadwell, we drop that GPU side swizzling, and the corresponding
initialization in 3 units (GAM, GT, DE). All those bits are reserved, as
specs put it:
Before Gen8, there was a historical configuration control field to
swizzle address bit[6] for in X/Y tiling modes. This was set in three
different places: TILECTL[1:0], ARB_MODE[5:4], and
DISP_ARB_CTL[14:13]"
For Gen8 the swizzle fields are all reserved, and the CPU's memory
controller performs all address swizzling modifications.
This also means that user space doesn't have to manually swizzle when
accessing tiled buffers from the CPU, and so we always return
I915_BIT_6_SWIZZLE_NONE from i915_gem_detect_bit_6_swizzle(), which
short-circuits the initialization of the registers mentionned above in
i915_gem_init_swizzling().
v2: Refine the explanation a bit more (Daniel)
v3: Make it BDW+ specific (Steve)
Cc: Steve Aarnio <steve.j.aarnio@linux.intel.com>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
[danvet: Keep the actual code to set the tiling bits for now, in case
some bios escaped to the wild that uses this - we'd need it for
fastboot.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Instead of going through hoops, just put the driver author directly as
DRM_AUTHOR() argument. This will also make it consistent when we add
Intel to the list.
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Fix assert_panel_unlocked for vlv/chv, and improve it a bit for
non-LVDS. Also don't pretend it works for DDI. There's still work to do
to get this right for eDP on PCH platforms, but this is a start.
v2: WARN_ON(HAS_DDI)
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The patch introduces fixes for the debugfs attributes emitted by
the i915 driver for GEN8. Currently, it is not emitting the correct
attributes which include the status of RC6 states.
Change-Id: Ib2068a0cac9a5wq3f228e547fa1a097ad369d242df
Signed-off-by: Vedang Patel <vedang.patel@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Rather than describing an object as either "snooped or LLC", we can do
better as we should know what machine we are running on!
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
On BDW we're seeing a problem that after we runtime resume, the
outputs connected to DDI C are not detected: they don't appear in the
SDEISR register and GMBUS transactions don't work. They stop working
at the moment we call intel_opregion_notify_adapter() during runtime
suspend, but they don't go back to work when we call the same function
during runtime resume. They only work after we do a modeset and call
intel_opregion_notify_encoder(), but this point is already too late.
While debugging, I tried to pass PCI_D3hot which is the value that
matches the spec, and it seems to have solved the problem. I couldn't
find any explanation of why this solves the problem, but there's also
no documented explanation - besides our code and git log - of why
Haswell should use PCI_D1, so keep this for now in order to keep BDW
runtime PM working.
Also add a comment to point the fact that there's no spec documenting
all the weirdness involved here.
Cc: kristen.c.accardi@intel.com
Testcase: igt/pm_rpm/drm-resources-equal
Testcase: igt/pm_rpm/i2c
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>
Because CHV uses cherryview_init_clock_gating instead of
gen8_init_clock_gating.
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>
Because BDW has WPT, which is equivalent to LPT. This is just like the
CPT/PPT case.
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>
Use the correct mask for the unlock bits. In theory this could have lead
to incorrect asserts but this is unlikely in practise.
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
These two functions make no sense in an Logical Ring Context & Execlists
world.
v2: We got rid of lrc_enabled and centralized everything in the sanitized
i915.enable_execlists instead.
Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
v3: Rebased. Corrected a typo in comment for i915_switch_context and
added a comment that it should not be called in execlist mode. Added
WARN_ON if i915_switch_context is called in execlist mode. Moved check
for execlist mode out of i915_switch_context and into callers. Added
comment in context_reset explaining why nothing is done in execlist
mode.
Signed-off-by: Thomas Daniel <thomas.daniel@intel.com>
[danvet: Simplify the patch subject so I can understand it.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Programing GT IER interrupts was fumbled while enabling Interrupts for
gen8
We forgot to program PM IER interrupt in gen8_gt_irq_postinstall based
on the new re-worked interrupt routines.
v2: Kill the loop and init GT interrupts individually (Ville)
Signed-off-by: Deepak S <deepak.s@linux.intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
[danvet: Adjust commit message as per discussion with Deepak.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
A pending commit removes synchronous mode from switch_mm. This breaks
execlists because switch_mm will always try to write to the legacy ring
buffer.
Return immediately from i915_ppgtt_init_gw in execlists mode.
No longer check for execlists mode in gen8_ppgtt_enable() because this
will no longer be called in execlists mode.
Signed-off-by: Thomas Daniel <thomas.daniel@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Be sure to always flush a stuck pageflip even if we couldn't possibly
expect one to be there.
References: https://bugs.freedesktop.org/show_bug.cgi?id=82612
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Unfortunately, the gem_obj/vma relationship is not symmetrical; a gem_obj
can look up for the same vma more than once (where the ppgtt refcount is
incremented), but will free the vma only once (i915_gem_free_object).
This difference in refcount get/put means that the ppgtt is not removed
after the context and vma are destroyed, because sometimes the refcount
will never go back to zero.
v2: Just move the ppgtt refcount into vma_create.
OTC-Jira: VIZ-3719
Signed-off-by: Michel Thierry <michel.thierry@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Less pointless indentation is always nice. There will be a bit more
code in this function once the power sequencer locking is fixed.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
If we force vdd off warn if someone is still using it. With this
change the delayed vdd off work needs to check want_panel_vdd
itself to make sure it doesn't try to turn vdd off when someone
is using it.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Looks nicer.
Not functional change.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
[danvet: Add "No functional change" as requested by Jani.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>