With commit 89773b8559
("drm/amdkfd: Let VRAM allocations go to GTT domain on small APUs")
big and small APU "VRAM" handling in KFD was unified. Since AMD_IS_APU
is set for both big and small APUs, we can simplify the checks in
the code.
v2: clean up a few more places (Lang)
Acked-by: Felix Kuehling <felix.kuehling@amd.com>
Reviewed-by: Lang Yu <Lang.Yu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This reverts commit 28ebbb4981.
Revert this commit as apparently the LLVM code to take advantage of
this never landed.
Reviewed-by: Feifei Xu <Feifei.Xu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: Feifei Xu <feifei.xu@amd.com>
The pointer parent may be NULLed by the function amdgpu_vm_pt_parent.
To make the code more robust, check the pointer parent.
Signed-off-by: Jesse Zhang <Jesse.Zhang@amd.com>
Suggested-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Convert a variable sized array from [1] to [].
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Use current speed/width on devices which don't support
dynamic PCIe switching.
Fixes: 466a7d1153 ("drm/amd: Use the first non-dGPU PCI device for BW limits")
Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/3289
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
MST colorspace property support was disabled due to a series of warnings
that came up when the device was plugged in since the properties weren't
made at device creation. Create the properties in advance instead.
Suggested-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Fixes: 69a9596102 ("drm/amd/display: Temporary Disable MST DP Colorspace Property").
Reported-and-tested-by: Tyler Schneider <tyler.schneider@amd.com>
Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/3353
Reviewed-by: Harry Wentland <harry.wentland@amd.com>
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
hbm filed takes bit 13 and bit 14 in boot status.
Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Tao Zhou <tao.zhou1@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
nouveau:
- use tile_mode and pte_kind for VM_BIND bo allocations
-----BEGIN PGP SIGNATURE-----
iQEzBAABCgAdFiEEchf7rIzpz2NEoWjlaA3BHVMLeiMFAmZFtP4ACgkQaA3BHVML
eiO0KAf9HshJjz+QEi+S0s6Z4p1pQqZgq1sgIhAzNl7LpINYeSHtlFK4erGgjWzu
lGat8s1kh/7n4NDvsxf7+F0SSd7S8Pqx5DqaFrDDeW1L3YdRQ0X59IVvqFbPZCGH
3GAFgg3gKWJBOEaHCYdMTDXj5mVM2OGxwCjw6x1hek/CUyuKgMTFSHRAEpQI7W2j
GYU5Gite/gO+/GlNr+6rPcOGPghdDACZqMRT/r8H0PP1SZoyI0lzJzZt0R/PrIDw
EClWH01l+M+OM7p6VzJ/RmvSP9lI7z8w3xvCkeQgRow0J6+jkSxC0JsRa6GXq9u7
g04YSb5fNpRLDzIA9KZKJyuO1Cuhkg==
=kvVW
-----END PGP SIGNATURE-----
Merge tag 'drm-misc-fixes-2024-05-16' of https://gitlab.freedesktop.org/drm/misc/kernel into drm-next
Short summary of fixes pull:
nouveau:
- use tile_mode and pte_kind for VM_BIND bo allocations
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
From: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20240516072658.GA8395@linux.fritz.box
[Why]
The vram width value is 0.
Because the integratedsysteminfo table in VBIOS has updated to 2.3.
[How]
Driver needs a new intergrated info v2.3 table too.
Then the vram width value will be correct.
Signed-off-by: Li Ma <li.ma@amd.com>
Reviewed-by: Yifan Zhang <yifan1.zhang@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Small APUs(i.e., consumer, embedded products) usually have a small
carveout device memory which can't satisfy most compute workloads
memory allocation requirements.
We can't even run a Basic MNIST Example with a default 512MB carveout.
https://github.com/pytorch/examples/tree/main/mnist. Error Log:
"torch.cuda.OutOfMemoryError: HIP out of memory. Tried to allocate
84.00 MiB. GPU 0 has a total capacity of 512.00 MiB of which 0 bytes
is free. Of the allocated memory 103.83 MiB is allocated by PyTorch,
and 22.17 MiB is reserved by PyTorch but unallocated"
Though we can change BIOS settings to enlarge carveout size,
which is inflexible and may bring complaint. On the other hand,
the memory resource can't be effectively used between host and device.
The solution is MI300A approach, i.e., let VRAM allocations go to GTT.
Then device and host can flexibly and effectively share memory resource.
v2: Report local_mem_size_private as 0. (Felix)
Signed-off-by: Lang Yu <Lang.Yu@amd.com>
Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The connector is created by either this ADV7511 bridge driver or
any DRM device driver/previous bridge driver, so this ADV7511
bridge driver should not let the next bridge driver create connector.
If the next bridge is a HDMI connector, the next bridge driver
would fail to attach bridge from display_connector_attach() without
the DRM_BRIDGE_ATTACH_NO_CONNECTOR flag.
Add that flag to drm_bridge_attach() function call in
adv7511_bridge_attach() to fix the issue.
This fixes the issue where the HDMI connector bridge fails to attach
to the previous ADV7535 bridge on i.MX8MP EVK platform:
[ 2.216442] [drm:drm_bridge_attach] *ERROR* failed to attach bridge /hdmi-connector to encoder None-37: -22
[ 2.220675] mmc1: SDHCI controller on 30b50000.mmc [30b50000.mmc] using ADMA
[ 2.226262] [drm:drm_bridge_attach] *ERROR* failed to attach bridge /soc@0/bus@30800000/i2c@30a30000/hdmi@3d to encoder None-37: -22
[ 2.245204] [drm:drm_bridge_attach] *ERROR* failed to attach bridge /soc@0/bus@32c00000/dsi@32e60000 to encoder None-37: -22
[ 2.256445] imx-lcdif 32e80000.display-controller: error -EINVAL: Failed to attach bridge for endpoint0
[ 2.265850] imx-lcdif 32e80000.display-controller: error -EINVAL: Cannot connect bridge
[ 2.274009] imx-lcdif 32e80000.display-controller: probe with driver imx-lcdif failed with error -22
Fixes: 14b3cdbd0e ("drm/bridge: adv7511: make it honour next bridge in DT")
Signed-off-by: Liu Ying <victor.liu@nxp.com>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Reviewed-by: Robert Foss <rfoss@kernel.org>
Acked-by: Sui Jingfeng <suijingfeng@bosc.ac.cn>
Reviewed-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240513080243.3952292-1-victor.liu@nxp.com
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Allocate cleared blocks in the bias range when the DRM
buddy's clear avail is zero. This will validate the bias
range allocation in scenarios like system boot when no
cleared blocks are available and exercise the fallback
path too. The resulting blocks should always be dirty.
v1:(Matthew)
- move the size to the variable declaration section.
- move the mm.clear_avail init to allocator init.
Signed-off-by: Arunpravin Paneer Selvam <Arunpravin.PaneerSelvam@amd.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240514145636.16253-2-Arunpravin.PaneerSelvam@amd.com
Problem statement: During the system boot time, an application request
for the bulk volume of cleared range bias memory when the clear_avail
is zero, we dont fallback into normal allocation method as we had an
unnecessary clear_avail check which prevents the fallback method leads
to fb allocation failure following system goes into unresponsive state.
Solution: Remove the unnecessary clear_avail check in the range bias
allocation function.
v2: add a kunit for this corner case (Daniel Vetter)
Signed-off-by: Arunpravin Paneer Selvam <Arunpravin.PaneerSelvam@amd.com>
Fixes: 96950929eb ("drm/buddy: Implement tracking clear page feature")
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240514145636.16253-1-Arunpravin.PaneerSelvam@amd.com
We need to undo what was done in panthor_sched_pre_reset() even if the
reset failed. We just flag all previously running groups as terminated
when that happens to unblock things.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Steven Price <steven.price@arm.com>
Reviewed-by: Liviu Dudau <liviu.dudau@arm.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240502183813.1612017-5-boris.brezillon@collabora.com
Make sure the user is aware that drm_panthor_tiler_heap_destroy::handle
must be a handle previously returned by
DRM_IOCTL_PANTHOR_TILER_HEAP_CREATE.
v4:
- Add Steve's R-b
v3:
- New patch
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Steven Price <steven.price@arm.com>
Reviewed-by: Liviu Dudau <liviu.dudau@arm.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240502165158.1458959-6-boris.brezillon@collabora.com
The heap ID is used to index the heap context pool, and allocating
in the [1:MAX_HEAPS_PER_POOL] leads to an off-by-one. This was
originally to avoid returning a zero heap handle, but given the handle
is formed with (vm_id << 16) | heap_id, with vm_id > 0, we already can't
end up with a valid heap handle that's zero.
v4:
- s/XA_FLAGS_ALLOC1/XA_FLAGS_ALLOC/
v3:
- Allocate in the [0:MAX_HEAPS_PER_POOL-1] range
v2:
- New patch
Fixes: 9cca48fa4f ("drm/panthor: Add the heap logical block")
Reported-by: Eric Smith <eric.smith@collabora.com>
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Tested-by: Eric Smith <eric.smith@collabora.com>
Reviewed-by: Steven Price <steven.price@arm.com>
Reviewed-by: Liviu Dudau <liviu.dudau@arm.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240502165158.1458959-5-boris.brezillon@collabora.com
The field used to store the chunk size if 12 bits wide, and the encoding
is chunk_size = chunk_header.chunk_size << 12, which gives us a
theoretical [4k:8M] range. This range is further limited by
implementation constraints, and all known implementations seem to
impose a [128k:8M] range, so do the same here.
We also relax the power-of-two constraint, which doesn't seem to
exist on v10. This will allow userspace to fine-tune initial/max
tiler memory on memory-constrained devices.
v4:
- Actually fix the range in the kerneldoc
v3:
- Add R-bs
- Fix valid range in the kerneldoc
v2:
- Turn the power-of-two constraint into a page-aligned constraint to allow
fine-tune of the initial/max heap memory size
- Fix the panthor_heap_create() kerneldoc
Fixes: 9cca48fa4f ("drm/panthor: Add the heap logical block")
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Liviu Dudau <liviu.dudau@arm.com>
Reviewed-by: Steven Price <steven.price@arm.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240502165158.1458959-4-boris.brezillon@collabora.com
It doesn't make sense to have a maximum number of chunks smaller than
the initial number of chunks attached to the context.
Fix the uAPI header to reflect the new constraint, and mention the
undocumented "initial_chunk_count > 0" constraint while at it.
v3:
- Add R-b
v2:
- Fix the check
Fixes: 9cca48fa4f ("drm/panthor: Add the heap logical block")
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Liviu Dudau <liviu.dudau@arm.com>
Reviewed-by: Steven Price <steven.price@arm.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240502165158.1458959-3-boris.brezillon@collabora.com
If the kernel couldn't allocate memory because we reached the maximum
number of chunks but no render passes are in flight
(panthor_heap_grow() returning -ENOMEM), we should defer the OOM
handling to the FW by returning a NULL chunk. The FW will then call
the tiler OOM exception handler, which is supposed to implement
incremental rendering (execute an intermediate fragment job to flush
the pending primitives, release the tiler memory that was used to
store those primitives, and start over from where it stopped).
Instead of checking for both ENOMEM and EBUSY, make panthor_heap_grow()
return ENOMEM no matter the reason of this allocation failure, the FW
doesn't care anyway.
v3:
- Add R-bs
v2:
- Make panthor_heap_grow() return -ENOMEM for all kind of allocation
failures
- Document the panthor_heap_grow() semantics
Fixes: de85488138 ("drm/panthor: Add the scheduler logical block")
Signed-off-by: Antonino Maniscalco <antonino.maniscalco@collabora.com>
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Liviu Dudau <liviu.dudau@arm.com>
Reviewed-by: Steven Price <steven.price@arm.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240502165158.1458959-2-boris.brezillon@collabora.com
- Use ordered WQ for G2H handler. (Matthew Brost)
- Use flexible-array rather than zero-sized (Lucas De Marchi)
-----BEGIN PGP SIGNATURE-----
iHUEABYKAB0WIQRskUM7w1oG5rx2IZO4FpNVCsYGvwUCZjz6vwAKCRC4FpNVCsYG
vybvAP4tAOoYsRW35Vb6T7CYMvYVdrp7SywBsWbiFk+KtR6q9AEAhwTEG1Z1sH3e
9yIkcpfTo0O1ARxhN/EFsJQ+OtO4OQE=
=kdLI
-----END PGP SIGNATURE-----
Merge tag 'drm-xe-next-fixes-2024-05-09-1' of https://gitlab.freedesktop.org/drm/xe/kernel into drm-next
Driver Changes:
- Use ordered WQ for G2H handler. (Matthew Brost)
- Use flexible-array rather than zero-sized (Lucas De Marchi)
Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Thomas Hellstrom <thomas.hellstrom@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/Zjz7SzCvfA3vQRxu@fedora
Updates for v6.10
Core:
- Switched to generating register header files during build process
instead of shipping pre-generated headers
- Merged DPU and MDP4 format databases.
DP:
- Stop using compat string to distinguish DP and eDP cases
- Added support for X Elite platform (X1E80100)
- Reworked DP aux/audio support
- Added SM6350 DP to the bindings (no driver changes, using SM8350
as a fallback compat)
GPU:
- a7xx perfcntr reg fixes
- MAINTAINERS updates
- a750 devcoredump support
Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Rob Clark <robdclark@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/CAF6AEGtpw6dNR9JBikFTQ=TCpt-9FeFW+SGjXWv+Jv3emm0Pbg@mail.gmail.com
System work queues are shared, use a dedicated work queue for G2H
processing to avoid G2H processing getting block behind system tasks.
Fixes: dd08ebf6c3 ("drm/xe: Introduce a new DRM driver for Intel GPUs")
Cc: <stable@vger.kernel.org>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Francois Dugast <francois.dugast@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240506034758.3697397-1-matthew.brost@intel.com
(cherry picked from commit 50aec9665e)
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
We don't need to run the validation of the XML files if we are just
compiling the kernel. Skip the validation unless the user enables
corresponding Kconfig option. This removes a warning from gen_header.py
about lxml being not installed.
Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Closes: https://lore.kernel.org/all/20240409120108.2303d0bd@canb.auug.org.au/
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Reviewed-by: Abhinav Kumar <quic_abhinavk@quicinc.com>
Patchwork: https://patchwork.freedesktop.org/patch/592558/
Signed-off-by: Rob Clark <robdclark@chromium.org>
These tables were made non-const in commit 3cba4a2cdf ("drm/msm/a6xx:
Update ROQ size in coredump") in order to avoid powering up the GPU when
reading back a devcoredump. Instead let's just stash the count that is
potentially read from hw in struct a6xx_gpu_state_obj, and make the
tables const again.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Patchwork: https://patchwork.freedesktop.org/patch/592699/
Import from kgsl commit 809ee24fe560.
Signed-off-by: Connor Abbott <cwabbott0@gmail.com>
Patchwork: https://patchwork.freedesktop.org/patch/592516/
Signed-off-by: Rob Clark <robdclark@chromium.org>
Add myself as a reviewer for Adreno driver changes.
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Patchwork: https://patchwork.freedesktop.org/patch/590705/
Signed-off-by: Rob Clark <robdclark@chromium.org>
The msm driver is.. gigantic and covers display hardware (incl. things
concerning (e)DP, DSI, HDMI), as well as the entire lineup of Adreno
GPUs (with hw bringup, memory mappings, userspace interaction etc.).
Because of that, people listed as M:/R: receive patches concerning
drivers for any part of the display block OR the GPU. Separate the
latter, as it's both a functionally separate block and is of
interest to different folks.
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Patchwork: https://patchwork.freedesktop.org/patch/590704/
Signed-off-by: Rob Clark <robdclark@chromium.org>
Calling a6xx_destroy() before adreno_gpu_init() leads to a null pointer
dereference on:
msm_gpu_cleanup() : platform_set_drvdata(gpu->pdev, NULL);
as gpu->pdev is only assigned in:
a6xx_gpu_init()
|_ adreno_gpu_init
|_ msm_gpu_init()
Instead of relying on handwavy null checks down the cleanup chain,
explicitly de-allocate the LLC data and free a6xx_gpu instead.
Fixes: 76efc2453d ("drm/msm/gpu: Fix crash during system suspend after unbind")
Signed-off-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Patchwork: https://patchwork.freedesktop.org/patch/588919/
Signed-off-by: Rob Clark <robdclark@chromium.org>
a7xx_submit() should use the a7xx variant of the RBBM_PERFCTR_CP register
for retrieving the CP cycles value before and after the submitted command
stream execution.
Signed-off-by: Zan Dobersek <zdobersek@igalia.com>
Fixes: af66706acc ("drm/msm/a6xx: Add skeleton A7xx support")
Patchwork: https://patchwork.freedesktop.org/patch/588445/
Signed-off-by: Rob Clark <robdclark@chromium.org>
In addition to the CP_PERFCTR_CP_SEL register range, allow writes to the
CP_BV_PERFCTR_CP_SEL registers in the 0x8e0-0x8e6 range for profiling
purposes of tools like fdperf and perfetto.
Signed-off-by: Zan Dobersek <zdobersek@igalia.com>
Patchwork: https://patchwork.freedesktop.org/patch/580548/
[fixup a730_protect size]
Signed-off-by: Rob Clark <robdclark@chromium.org>
