- Mali 4xx GPUs have two kinds of processors GP and PP. GP is for
OpenGL vertex shader processing and PP is for fragment shader
processing. Each processor has its own MMU so prcessors work in
virtual address space.
- There's only one GP but multiple PP (max 4 for mali 400 and 8
for mali 450) in the same mali 4xx GPU. All PPs are grouped
togather to handle a single fragment shader task divided by
FB output tiled pixels. Mali 400 user space driver is
responsible for assign target tiled pixels to each PP, but mali
450 has a HW module called DLBU to dynamically balance each
PP's load.
- User space driver allocate buffer object and map into GPU
virtual address space, upload command stream and draw data with
CPU mmap of the buffer object, then submit task to GP/PP with
a register frame indicating where is the command stream and misc
settings.
- There's no command stream validation/relocation due to each user
process has its own GPU virtual address space. GP/PP's MMU switch
virtual address space before running two tasks from different
user process. Error or evil user space code just get MMU fault
or GP/PP error IRQ, then the HW/SW will be recovered.
- Use GEM+shmem for MM. Currently just alloc and pin memory when
gem object creation. GPU vm map of the buffer is also done in
the alloc stage in kernel space. We may delay the memory
allocation and real GPU vm map to command submission stage in the
furture as improvement.
- Use drm_sched for GPU task schedule. Each OpenGL context should
have a lima context object in the kernel to distinguish tasks
from different user. drm_sched gets task from each lima context
in a fair way.
mesa driver can be found here before upstreamed:
https://gitlab.freedesktop.org/lima/mesa
v8:
- add comments for in_sync
- fix ctx free miss mutex unlock
v7:
- remove lima_fence_ops with default value
- move fence slab create to device probe
- check pad ioctl args to be zero
- add comments for user/kernel interface
v6:
- fix comments by checkpatch.pl
v5:
- export gp/pp version to userspace
- rebase on drm-misc-next
v4:
- use get param interface to get info
- separate context create/free ioctl
- remove unused max sched task param
- update copyright time
- use xarray instead of idr
- stop using drmP.h
v3:
- fix comments from kbuild robot
- restrict supported arch to tested ones
v2:
- fix syscall argument check
- fix job finish fence leak since kernel 5.0
- use drm syncobj to replace native fence
- move buffer object GPU va map into kernel
- reserve syscall argument space for future info
- remove kernel gem modifier
- switch TTM back to GEM+shmem MM
- use time based io poll
- use whole register name
- adopt gem reservation obj integration
- use drm_timeout_abs_to_jiffies
Cc: Eric Anholt <eric@anholt.net>
Cc: Rob Herring <robh@kernel.org>
Cc: Christian König <ckoenig.leichtzumerken@gmail.com>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: Alex Deucher <alexdeucher@gmail.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Rob Clark <robdclark@gmail.com>
Cc: Dave Airlie <airlied@gmail.com>
Signed-off-by: Andreas Baierl <ichgeh@imkreisrum.de>
Signed-off-by: Erico Nunes <nunes.erico@gmail.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Signed-off-by: Marek Vasut <marex@denx.de>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: Simon Shields <simon@lineageos.org>
Signed-off-by: Vasily Khoruzhick <anarsoul@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Rob Herring <robh@kerrnel.org>
Signed-off-by: Eric Anholt <eric@anholt.net>
Link: https://patchwork.freedesktop.org/patch/291200/
This makes sure the vc4_reset doesn't hit an obscure race with the
GET_PARAM ioctl, fixes a decrement outside of the lock, and prevents
future code from making mistakes with the weird return value of
pm_runtime_get_sync().
Signed-off-by: Eric Anholt <eric@anholt.net>
Link: https://patchwork.freedesktop.org/patch/msgid/20190220210343.28157-6-eric@anholt.net
Reviewed-by: Paul Kocialkowski <paul.kocialkowski@bootlin.com>
The shmem->pages[] array has "num_pages" elements so the > should be >=
to prevent reading beyond the end of the array. The shmem->pages[]
array is allocated in drm_gem_shmem_prime_import_sg_table().
Fixes: 2194a63a81 ("drm: Add library for shmem backed GEM objects")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Eric Anholt <eric@anholt.net>
Link: https://patchwork.freedesktop.org/patch/msgid/20190322064125.GA12551@kadam
The debugfs_regset32 is nice to use for reducing boilerplate in
dumping a bunch of regs in debugfs, but we also want to be able to
print to dmesg them at runtime for driver debugging. drm_printer lets
us format debugfs and the printk the same way.
v2: Add some kerneldoc for the function (requested by danvet)
Signed-off-by: Eric Anholt <eric@anholt.net>
Link: https://patchwork.freedesktop.org/patch/msgid/20190220210343.28157-1-eric@anholt.net
Reviewed-by: Paul Kocialkowski <paul.kocialkowski@bootlin.com> (v1)
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Add a debugfs node "register" and entry function dump_register to
dev/pipeline/component to register dump, then user can read
"/sys/kernel/debug/komeda/register" to get the register values via these
chip function.
Signed-off-by: James Qian Wang (Arm Technology China) <james.qian.wang@arm.com>
[Added the d71_layer_dump() function that was in a previous commit]
Signed-off-by: Liviu Dudau <liviu.dudau@arm.com>
1. Added irq_handler/irq_enable/irq_disable to komeda_dev_func, then the
Komeda-CORE can control the HW irq via these chip function.
2. Install irq and register irq_handler to system by DRM, so once the IRQ
coming, the handling sequence is:
komeda_kms_irq_handler(int irq, void *data)
/* step 1. call into the CHIP to recognize event */
mdev->funcs->irq_handler(mdev, &evts);
/* step 2. notify the crtc to handle the events */
for (i = 0; i < kms->n_crtcs; i++)
komeda_crtc_handle_event(&kms->crtcs[i], &evts);
v2:
- Move get IRQ number into this change.
- Enable irq before drm_dev_register.
Signed-off-by: James Qian Wang (Arm Technology China) <james.qian.wang@arm.com>
Signed-off-by: Liviu Dudau <liviu.dudau@arm.com>
komeda_accemble_pipelines is for:
1. Verifing the component->supported_inputs according to the
pipeline->avail_components.
2. Generating component->supported_outputs.
v2: Lower the debug message of komeda_component_dump to DRM_DEBUG.
Signed-off-by: James Qian Wang (Arm Technology China) <james.qian.wang@arm.com>
Signed-off-by: Liviu Dudau <liviu.dudau@arm.com>
Add and initialize improc and timing_ctrlr according to D71 capablitites
v2: Rebase.
Signed-off-by: James Qian Wang (Arm Technology China) <james.qian.wang@arm.com>
Signed-off-by: Liviu Dudau <liviu.dudau@arm.com>
1. Add detailed layer/layer_state definitions
2. Add d71_layer_init to report layer features and capabilities according
to D71 layer block.
3. Add d71_layer_updat/disable
v2: Rebase.
Signed-off-by: James Qian Wang (Arm Technology China) <james.qian.wang@arm.com>
[removed d71_layer_dump() from this commit]
Signed-off-by: Liviu Dudau <liviu.dudau@arm.com>
The drm_gem_shmem_create() returns error pointers and v3d_bo_create() is
also supposed to return error pointers.
Fixes: 40609d4820 ("drm/v3d: Use the new shmem helpers to reduce driver boilerplate.")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Eric Anholt <eric@anholt.net>
Link: https://patchwork.freedesktop.org/patch/msgid/20190321062731.GC21489@kadam
The HW only executes a load once the tile coordinates packet happens,
and only tracks one at a time, so by emitting our two MSAA loads back
to back we would end up with an undefined color or Z buffer.
Fixes dEQP-EGL.functional.render.multi_context.gles2.rgb888_window
Signed-off-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Paul Kocialkowski <paul.kocialkowski@bootlin.com>
Cc: Maxime Ripard <maxime.ripard@bootlin.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190206232550.12012-1-eric@anholt.net
v2: individually allocate chain array, since chain node is free independently.
v3: all existing points must be already signaled before cpu perform signal operation,
so add check condition for that.
v4: remove v3 change and add checking to prevent out-of-order
v5: unify binary and timeline
Signed-off-by: Chunming Zhou <david1.zhou@amd.com>
Signed-off-by: Christian König <christian.koenig@amd.com>
Cc: Tobias Hector <Tobias.Hector@amd.com>
Cc: Jason Ekstrand <jason@jlekstrand.net>
Cc: Dave Airlie <airlied@redhat.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Link: https://patchwork.freedesktop.org/patch/295792/?series=58813&rev=1
we need to import/export timeline point.
v2: unify to one transfer ioctl
Signed-off-by: Chunming Zhou <david1.zhou@amd.com>
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Link: https://patchwork.freedesktop.org/patch/295790/?series=58813&rev=1
Implement finding the right timeline point in drm_syncobj_find_fence.
v2: return -EINVAL when the point is not submitted yet.
v3: fix reference counting bug, add flags handling as well
v4: add timeout for find fence
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Link: https://patchwork.freedesktop.org/patch/295786/?series=58813&rev=1
user mode can query timeline payload.
v2: check return value of copy_to_user
v3: handle querying entry by entry
v4: rebase on new chain container, simplify interface
v5: query last signaled timeline point, not last point.
v6: add unorder point check
Signed-off-by: Chunming Zhou <david1.zhou@amd.com>
Signed-off-by: Christian König <christian.koenig@amd.com>
Cc: Tobias Hector <Tobias.Hector@amd.com>
Cc: Jason Ekstrand <jason@jlekstrand.net>
Cc: Dave Airlie <airlied@redhat.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Link: https://patchwork.freedesktop.org/patch/295784/?series=58813&rev=1
points array is one-to-one match with syncobjs array.
v2:
add seperate ioctl for timeline point wait, otherwise break uapi.
v3:
userspace can specify two kinds waits::
a. Wait for time point to be completed.
b. and wait for time point to become available
v4:
rebase
v5:
add comment for xxx_WAIT_AVAILABLE
v6: rebase and rework on new container
v7: drop _WAIT_COMPLETED, it is the default anyway
v8: correctly handle garbage collected fences
Signed-off-by: Chunming Zhou <david1.zhou@amd.com>
Signed-off-by: Christian König <christian.koenig@amd.com>
Cc: Tobias Hector <Tobias.Hector@amd.com>
Cc: Jason Ekstrand <jason@jlekstrand.net>
Cc: Dave Airlie <airlied@redhat.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Link: https://patchwork.freedesktop.org/patch/295782/?series=58813&rev=1
Use the dma_fence_chain object to create a timeline of fence objects
instead of just replacing the existing fence.
v2: rebase and cleanup
v3: fix garbage collection parameters
v4: add unorder point check, print a warn calltrace
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Link: https://patchwork.freedesktop.org/patch/295780/?series=58813&rev=1
Lockless container implementation similar to a dma_fence_array, but with
only two elements per node and automatic garbage collection.
v2: properly document dma_fence_chain_for_each, add dma_fence_chain_find_seqno,
drop prev reference during garbage collection if it's not a chain fence.
v3: use head and iterator for dma_fence_chain_for_each
v4: fix reference count in dma_fence_chain_enable_signaling
v5: fix iteration when walking each chain node
v6: add __rcu for member 'prev' of struct chain node
v7: fix rcu warnings from kernel robot
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Link: https://patchwork.freedesktop.org/patch/295778/?series=58813&rev=1
We can't just map any randome page we get when memory encryption is
active.
Signed-off-by: Christian König <christian.koenig@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.kernel.org/patch/10850833/
Sometimes it is desirabled to use a separate i2c controller for ddc
access. This adds support for the ddc-i2c-bus property of the
hdmi-connector node, using the specified controller if provided.
Signed-off-by: Mans Rullgard <mans@mansr.com>
Signed-off-by: Maxime Ripard <maxime.ripard@bootlin.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190328130249.19356-1-mans@mansr.com
The RK3066 HDMI TX serves as interface between a LCD Controller and
a HDMI bus. A HDMI TX consists of one HDMI transmitter controller and
one HDMI transmitter PHY. The interface has three (3) 8-bit data channels
which can be configured for a number of bus widths (8/10/12/16/20/24-bit)
and different video formats (RGB, YCbCr).
Features:
HDMI version 1.4a, HDCP revision 1.4 and
DVI version 1.0 compliant transmitter.
Supports DTV resolutions from 480i to 1080i/p HD.
Master I2C interface for a DDC connection.
HDMI TX supports multiple power save modes.
The HDMI TX input can switch between LCDC0 and LCDC1.
(Sound support is not included in this patch)
Signed-off-by: Zheng Yang <zhengyang@rock-chips.com>
Signed-off-by: Johan Jonker <jbx6244@gmail.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20190330095639.14626-2-jbx6244@gmail.com
Splits out the dither register bits and introduces
the same config enumerations as in the rockchip kernel tree.
Tested to fix the banding on my ASUS C201.
Signed-off-by: Urja Rannikko <urjaman@gmail.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20190318154412.26994-1-urjaman@gmail.com
UAPI Changes:
- Make mmap code more asynchronous. Avoid full SET_DOMAIN on GTT mmap pagefault,
and flushes pages on acquisition instead. Moves some of the work from mmap fault
time to execbuf time to avoid lock contention during mmap access.
Has neutral to positive impact on perf as the flushing moves to execbuf time
in real world workloads on the current known userspaces due to recycling of BOs.
If there exist an unknown non-recycling userspace, they should explicitly do the
SET_DOMAIN and not rely on kernel doing implicit SET_DOMAIN because swapout/in
might have happenedt.
- Restore the accidentally removed behaviour of returning object size on GEM_CREATE
From 2011: ff72145bad ("drm: dumb scanout create/mmap for intel/radeon (v3)")
- Includes a some neutered patches to prepare to complete the earlier Mesa
recovery feature uAPI. Looking to enable this in the next PR.
Driver Changes:
- Add Elkhartlake (Gen11) support code and PCI IDs
- Add missing Amberlake PCI ID 0x87CA (Ville)
- Fix to Bugzilla #109780: Pick the first mode from EDID as the fixed mode when there is no preferred mode (Ville)
- Fix GCC 4.8 build by using __is_constexpr() (Chris, Randy, Uma)
- Add "Broadcast RGB", "force_audio" and "max_bpc" properties to DP MST (Ville)
- Remove 8bpc limitation from DP MST (Ville)
- Fix changing between limited and full range RGB output in DP fastsets (Ville)
- Reject unsupported HDR formats (Maarten)
- Handle YUV subpixel support better (Maarten)
- Various plane watermarks fixes and cleaning of the code (Ville)
- Icelake port sync master select fix (Manasi)
- Icelake VEBOX disable bitmask fix (Jose)
- Close a race where userspace could see incompletely initialized GEM context (Chris)
- Avoid C3 on i945gm to keep vblank interrupts steady (Ville)
- Avoid recalculating PLL HW readout each time (Lucas)
- A ton of patches to modularize uncore code (Daniel)
- Instead of storing media fuse value, immediately derive engine masks (Daniele)
- Reduce struct_mutex usage (Chris)
- Iterate over child devices to initialize ddi_port_info (Jani)
- Fixes to return correct error values when bailing out of functions (Dan)
- Use bitmap_zalloc() (Andy)
- Reorder and clarify Gen3/4 code (Ville)
- Refactor out common code in display mode handling (Ville)
- GuC code fixes (Sujaritha, Michal)
- Selftest improvements (Chris)
Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190328151515.GA9606@jlahtine-desk.ger.corp.intel.com
This patch moves the virtio_gpu_cmd_create_resource() call (which
notifies the host about the new resource created) into the
virtio_gpu_object_create() function. That way we can call
virtio_gpu_cmd_create_resource() before ttm_bo_init(), so the host
already knows about the object when ttm initializes the object and calls
our driver callbacks.
Specifically the object is already created when the
virtio_gpu_ttm_tt_bind() callback invokes virtio_gpu_object_attach(),
so the extra virtio_gpu_object_attach() calls done after
virtio_gpu_object_create() are not needed any more.
The fence support for the create ioctl becomes a bit more tricky though.
The code moved into virtio_gpu_object_create() too. We first submit the
(fenced) virtio_gpu_cmd_create_resource() command, then initialize the
ttm object, and finally attach just created object to the fence for the
command in case it didn't finish yet.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Acked-by: Noralf Trønnes <noralf@tronnes.org>
Link: http://patchwork.freedesktop.org/patch/msgid/20190318113332.10900-6-kraxel@redhat.com
Add format, width and height fields to the virtio_gpu_object_params
struct. With that in place we can use the parameter struct for
virtio_gpu_cmd_create_resource() calls too.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Acked-by: Noralf Trønnes <noralf@tronnes.org>
Link: http://patchwork.freedesktop.org/patch/msgid/20190318113332.10900-4-kraxel@redhat.com
Create virtio_gpu_object_params, use that to pass object parameters to
virtio_gpu_object_create. This is just the first step, followup patches
will add more parameters to the struct. The plan is to use the struct
for all object parameters.
Drop unused "kernel" parameter for virtio_gpu_alloc_object(), it is
unused and always false.
Also drop "pinned" parameter. virtio-gpu doesn't shuffle around
objects, so effecively they all are pinned anyway. Hardcode
TTM_PL_FLAG_NO_EVICT so ttm knows. Doesn't change much for the moment
as virtio-gpu supports TTM_PL_FLAG_TT only so there is no opportunity to
move around objects. That'll probably change in the future though.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Acked-by: Noralf Trønnes <noralf@tronnes.org>
Link: http://patchwork.freedesktop.org/patch/msgid/20190318113332.10900-3-kraxel@redhat.com
Drop the dummy ttm backend implementation, add a real one for
TTM_PL_FLAG_TT objects. The bin/unbind callbacks will call
virtio_gpu_object_{attach,detach}, to update the object state
on the host side, instead of invoking those calls from the
move_notify() callback.
With that in place the move and move_notify callbacks are not
needed any more, so drop them.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Acked-by: Noralf Trønnes <noralf@tronnes.org>
Link: http://patchwork.freedesktop.org/patch/msgid/20190318113332.10900-2-kraxel@redhat.com
This file prints which features the virtio-gpu device has.
Also add "virtio-gpu-" prefix to the existing fence file,
to make clear this is a driver-specific debugfs file.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Acked-by: Noralf Trønnes <noralf@tronnes.org>
Link: http://patchwork.freedesktop.org/patch/msgid/20190320083612.32067-1-kraxel@redhat.com