This reverts commit b08425fa77.
Revert this to align with 6.17 because the fixes tag
was wrong on this commit.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit be33e8a239)
If the doorbell is invalid, be sure to set the r to an error
state so the function returns an error.
Reviewed-by: David (Ming Qiang) Wu <David.Wu3@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The minimum firmware versions required for user queue functionality
have been increased to address an issue where the queue privilege
state was lost during queue connect operations.
The problem occurred because the privilege state was being restored
to its initial value at the beginning of the function, overwriting
the state that was properly set during the queue connect case.
This commit updates the minimum version requirements:
- ME firmware from 2390 to 2420
- PFP firmware from 2530 to 2580
- MEC firmware from 2600 to 2650
- MES firmware remains at 120
These updated firmware versions contain the necessary fixes to
properly maintain queue privilege state throughout connect operations.
Fixes: 61ca97e959 ("drm/amdgpu: Add fw minimum version check for usermode queue")
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Jesse Zhang <Jesse.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
We need to cancel any outstanding work at both suspend
and driver teardown. Move the cancel to hw_fini which
gets called in both cases.
Reviewed-by: David (Ming Qiang) Wu <David.Wu3@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Fix a UBSAN shift-out-of-bounds warning in amdgpu_debugfs_jpeg_sched_mask_set
when the shift exponent reaches or exceeds 32 bits. The issue occurred because
a 32-bit integer '1' was being shifted by up to 32 bits, which is undefined
behavior.
Replace '1' with '1ULL' to ensure 64-bit arithmetic, matching the u64 type of
'val' and preventing the shift overflow. This is consistent with the existing
mask calculation that already uses 1ULL.
The error manifested as:
UBSAN: shift-out-of-bounds in drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c:373:17
shift exponent 32 is too large for 32-bit type 'int'
v2: remove debug log
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Suggested-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Jesse Zhang <Jesse.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
It should use vm flags instead of pte flags
to specify bo vm attributes.
This reverts commit 1263ceea2a1327014d9de2858a122f3c27dfa4dd.
Reapply this patch with the proper fixes tag.
Fixes: 6716a823d1 ("drm/amdgpu: rework how PTE flags are generated v3")
Signed-off-by: Jack Xiao <Jack.Xiao@amd.com>
Reviewed-by: Likun Gao <Likun.Gao@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This reverts commit b08425fa77.
Revert this to align with 6.17 because the fixes tag
was wrong on this commit.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Implement ring reset for VPE. Similar to VCN and JPEG,
just powergate the the IP to reset it.
v2: Properly set per queue reset flag
Reviewed-by: Jesse Zhang <jesse.zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
We already call this in the hw_fini() methods for all
VCN instances, so no need to call it again in
amdgpu_vcn_suspend().
Tested-by: David (Ming Qiang) Wu <David.Wu3@amd.com>
Reviewed-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
AMDGPU_HAS_VRAM is redundant with is_app_apu, as both refer to
APUs with no carve-out. Since AMDGPU_HAS_VRAM only occurs once,
remove AMDGPU_HAS_VRAM definition. The tmr allocation can be covered
with AMDGPU_GEM_DOMAIN_GTT | AMDGPU_GEM_DOMAIN_VRAM in both vram and
non vram ASICs.
Signed-off-by: Yifan Zhang <yifan1.zhang@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
By polling, poll ACA bank count to ensure that valid
ACA bank reg info can be obtained
v2: add corresponding delay before send msg to SMU to query mca bank info
(Stanley)
v3: the loop cannot exit. (Thomas)
v4: remove amdgpu_aca_clear_bank_count. (Kevin)
v5: continuously inject ce. If a creation interruption
occurs at this time, bank reg info will be lost. (Thomas)
v5: each cycle is delayed by 100ms. (Tao)
Signed-off-by: Ce Sun <cesun102@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
It's not necessarily to allocate psp firmware private
buffer in different memory domain in sriov and bare
metal environment
Signed-off-by: Hawking Zhang <Hawking.Zhang@amd.com>
Reviewed-by: Le Ma <le.ma@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Current dma-buf vmap semantics require that the mapped buffer remains
in place until the corresponding vunmap has completed.
For GEM-SHMEM, this used to be guaranteed by a pin operation while creating
an S/G table in import. GEM-SHMEN can now import dma-buf objects without
creating the S/G table, so the pin is missing. Leads to page-fault errors,
such as the one shown below.
[ 102.101726] BUG: unable to handle page fault for address: ffffc90127000000
[...]
[ 102.157102] RIP: 0010:udl_compress_hline16+0x219/0x940 [udl]
[...]
[ 102.243250] Call Trace:
[ 102.245695] <TASK>
[ 102.2477V95] ? validate_chain+0x24e/0x5e0
[ 102.251805] ? __lock_acquire+0x568/0xae0
[ 102.255807] udl_render_hline+0x165/0x341 [udl]
[ 102.260338] ? __pfx_udl_render_hline+0x10/0x10 [udl]
[ 102.265379] ? local_clock_noinstr+0xb/0x100
[ 102.269642] ? __lock_release.isra.0+0x16c/0x2e0
[ 102.274246] ? mark_held_locks+0x40/0x70
[ 102.278177] udl_primary_plane_helper_atomic_update+0x43e/0x680 [udl]
[ 102.284606] ? __pfx_udl_primary_plane_helper_atomic_update+0x10/0x10 [udl]
[ 102.291551] ? lockdep_hardirqs_on_prepare.part.0+0x92/0x170
[ 102.297208] ? lockdep_hardirqs_on+0x88/0x130
[ 102.301554] ? _raw_spin_unlock_irq+0x24/0x50
[ 102.305901] ? wait_for_completion_timeout+0x2bb/0x3a0
[ 102.311028] ? drm_atomic_helper_calc_timestamping_constants+0x141/0x200
[ 102.317714] ? drm_atomic_helper_commit_planes+0x3b6/0x1030
[ 102.323279] drm_atomic_helper_commit_planes+0x3b6/0x1030
[ 102.328664] drm_atomic_helper_commit_tail+0x41/0xb0
[ 102.333622] commit_tail+0x204/0x330
[...]
[ 102.529946] ---[ end trace 0000000000000000 ]---
[ 102.651980] RIP: 0010:udl_compress_hline16+0x219/0x940 [udl]
In this stack strace, udl (based on GEM-SHMEM) imported and vmap'ed a
dma-buf from amdgpu. Amdgpu relocated the buffer, thereby invalidating the
mapping.
Provide a custom dma-buf vmap method in amdgpu that pins the object before
mapping it's buffer's pages into kernel address space. Do the opposite in
vunmap.
Note that dma-buf vmap differs from GEM vmap in how it handles relocation.
While dma-buf vmap keeps the buffer in place, GEM vmap requires the caller
to keep the buffer in place. Hence, this fix is in amdgpu's dma-buf code
instead of its GEM code.
A discussion of various approaches to solving the problem is available
at [1].
v3:
- try (GTT | VRAM); drop CPU domain (Christian)
v2:
- only use mapable domains (Christian)
- try pinning to domains in preferred order
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Fixes: 660cd44659 ("drm/shmem-helper: Import dmabuf without mapping its sg_table")
Reported-by: Thomas Zimmermann <tzimmermann@suse.de>
Closes: https://lore.kernel.org/dri-devel/ba1bdfb8-dbf7-4372-bdcb-df7e0511c702@suse.de/
Cc: Shixiong Ou <oushixiong@kylinos.cn>
Cc: Thomas Zimmermann <tzimmermann@suse.de>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: Maxime Ripard <mripard@kernel.org>
Cc: David Airlie <airlied@gmail.com>
Cc: Simona Vetter <simona@ffwll.ch>
Cc: Sumit Semwal <sumit.semwal@linaro.org>
Cc: "Christian König" <christian.koenig@amd.com>
Cc: dri-devel@lists.freedesktop.org
Cc: linux-media@vger.kernel.org
Cc: linaro-mm-sig@lists.linaro.org
Link: https://lore.kernel.org/dri-devel/9792c6c3-a2b8-4b2b-b5ba-fba19b153e21@suse.de/ # [1]
Reviewed-by: Christian König <christian.koenig@amd.com>
Link: https://lore.kernel.org/r/20250821064031.39090-1-tzimmermann@suse.de
The function amdgpu_vcn_sysfs_reset_mask_init already returns 0, which
makes the check of the result unnecessary in the vcn_v4_0_3_sw_init().
Just return the amdgpu_vcn_sysfs_reset_mask_init directly.
Signed-off-by: Rodrigo Siqueira <siqueira@igalia.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
When a poison is consumed on the guest before the guest receives the host's poison creation msg, a corner case may occur to have poison_handler complete processing earlier than it should to cause the guest to hang waiting for the req_bad_pages reply during a VF FLR, resulting in the VM becoming inaccessible in stress tests.
To fix this issue, this patch refactored the mailbox sequence by seperating the bad_page_work into two parts req_bad_pages_work and handle_bad_pages_work.
Old sequence:
1.Stop data exchange work
2.Guest sends MB_REQ_RAS_BAD_PAGES to host and keep polling for IDH_RAS_BAD_PAGES_READY
3.If the IDH_RAS_BAD_PAGES_READY arrives within timeout limit, re-init the data exchange region for updated bad page info
else timeout with error message
New sequence:
req_bad_pages_work:
1.Stop data exhange work
2.Guest sends MB_REQ_RAS_BAD_PAGES to host
Once Guest receives IDH_RAS_BAD_PAGES_READY event
handle_bad_pages_work:
3.re-init the data exchange region for updated bad page info
Signed-off-by: Chenglei Xie <Chenglei.Xie@amd.com>
Reviewed-by: Shravan Kumar Gande <Shravankumar.Gande@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This change reorganizes VCN reset capability detection by:
1. Moving reset mask configuration from sw_init to new late_init phase
2. Adding vcn_v4_0_3_late_init() to properly check for per-queue reset support
3. Only setting soft full reset mask as fallback when per-queue reset isn't supported
4. Removing TODO comment now that queue reset support is implemented
V2: Removed unrelated changes. Keep amdgpu_get_soft_full_reset_mask in place
and remove TODO comment. (Alex)
v3: set the flags at one place (all in late_init) (Lijo)
Suggested-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Ruili Ji <ruiliji2@amd.com>
Signed-off-by: Jesse Zhang <Jesse.Zhang@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
If a amdgpu_bo_va is fpriv->prt_va, the bo of this one is always NULL.
So, such kind of amdgpu_bo_va should be updated separately before
amdgpu_vm_handle_moved.
Signed-off-by: Heng Zhou <Heng.Zhou@amd.com>
Reviewed-by: Kasiviswanathan, Harish <Harish.Kasiviswanathan@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This reverts commit 515986100d.
The dma_buf field in struct drm_gem_object is not stable over the
object instance's lifetime. The field becomes NULL when user space
releases the final GEM handle on the buffer object. This resulted
in a NULL-pointer deref.
Workarounds in commit 5307dce878 ("drm/gem: Acquire references on
GEM handles for framebuffers") and commit f6bfc9afc7 ("drm/framebuffer:
Acquire internal references on GEM handles") only solved the problem
partially. They especially don't work for buffer objects without a DRM
framebuffer associated.
Hence, this revert to going back to using .import_attach->dmabuf.
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Simona Vetter <simona.vetter@ffwll.ch>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Link: https://lore.kernel.org/r/20250715082635.34974-1-tzimmermann@suse.de
During process kill, drm_sched_entity_flush() will kill the vm
entities. The following job submissions of this process will fail, and
the resources of these jobs have not been released, nor have the fences
been signalled, causing tasks to hang and timeout.
Fix by check entity status in amdgpu_vm_ready() and avoid submit jobs to
stopped entity.
v2: add amdgpu_vm_ready() check before amdgpu_vm_clear_freed() in
function amdgpu_cs_vm_handling().
Fixes: 1f02f2044b ("drm/amdgpu: Avoid extra evict-restore process.")
Signed-off-by: Liu01 Tong <Tong.Liu01@amd.com>
Signed-off-by: Lin.Cao <lincao12@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit f101c13a87)
It should use vm flags instead of pte flags
to specify bo vm attributes.
Fixes: 7946340fa3 ("drm/amdgpu: Move csa related code to separate file")
Signed-off-by: Jack Xiao <Jack.Xiao@amd.com>
Reviewed-by: Likun Gao <Likun.Gao@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit b08425fa77)
The vram block allocation flag must be cleared
before making vram reservation, otherwise reserving
addresses within the currently freed memory range
will always fail.
Fixes: c9cad937c0 ("drm/amdgpu: add drm buddy support to amdgpu")
Signed-off-by: YiPeng Chai <YiPeng.Chai@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit d38eaf27de)
The fw reserved GFX command is only supported starting from PSP fw
version 0x3a0e14 and 0x3b0e0d. Older versions do not support this command.
Add a version guard to ensure the command is only used when the running
PSP fw meets the minimum version requirement.
This ensures backward compatibility and safe operation across fw
revisions.
Fixes: a3b7f9c306 ("drm/amdgpu: reclaim psp fw reservation memory region")
Signed-off-by: Frank Min <Frank.Min@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit 065e23170a)
During process kill, drm_sched_entity_flush() will kill the vm
entities. The following job submissions of this process will fail, and
the resources of these jobs have not been released, nor have the fences
been signalled, causing tasks to hang and timeout.
Fix by check entity status in amdgpu_vm_ready() and avoid submit jobs to
stopped entity.
v2: add amdgpu_vm_ready() check before amdgpu_vm_clear_freed() in
function amdgpu_cs_vm_handling().
Signed-off-by: Liu01 Tong <Tong.Liu01@amd.com>
Signed-off-by: Lin.Cao <lincao12@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
It should use vm flags instead of pte flags
to specify bo vm attributes.
Fixes: 7946340fa3 ("drm/amdgpu: Move csa related code to separate file")
Signed-off-by: Jack Xiao <Jack.Xiao@amd.com>
Reviewed-by: Likun Gao <Likun.Gao@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The vram block allocation flag must be cleared
before making vram reservation, otherwise reserving
addresses within the currently freed memory range
will always fail.
Fixes: c9cad937c0 ("drm/amdgpu: add drm buddy support to amdgpu")
Signed-off-by: YiPeng Chai <YiPeng.Chai@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The fw reserved GFX command is only supported starting from PSP fw
version 0x3a0e14 and 0x3b0e0d. Older versions do not support this command.
Add a version guard to ensure the command is only used when the running
PSP fw meets the minimum version requirement.
This ensures backward compatibility and safe operation across fw
revisions.
Fixes: a3b7f9c306 ("drm/amdgpu: reclaim psp fw reservation memory region")
Signed-off-by: Frank Min <Frank.Min@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
During a DPC error kernel waits for the link to be active before
notifying downstream devices. On certain platforms with Broadcom switch
in synthetiic mode, switch responds with values even though the link is
not fully ready. The config space restoration done by pcie port driver
for SWUS/DS of dGPU is thus not effective as the switch is still doing
internal enumeration.
As a workaround, save state of SWUS/DS device in driver. Add additional
check to see if link is active and restore the values during DPC error
callbacks.
Signed-off-by: Lijo Lazar <lijo.lazar@amd.com>
Reviewed-by: Yang Wang <kevinyang.wang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Pull drm fixes from Dave Airlie:
"This is the fixes that built up in the merge window, mostly amdgpu and
xe with one i915 display fix, seems like things are pretty good for
rc1.
i915:
- DP LPFS fixes
xe:
- SRIOV: PF fixes and removal of need of module param
- Fix driver unbind around Devcoredump
- Mark xe driver as BROKEN if kernel page size is not 4kB
amdgpu:
- GC 9.5.0 fixes
- SMU fix
- DCE 6 DC fixes
- mmhub client ID fixes
- VRR fix
- Backlight fix
- UserQ fix
- Legacy reset fix
- Misc fixes
amdkfd:
- CRIU fix
- Debugfs fix"
* tag 'drm-next-2025-08-08' of https://gitlab.freedesktop.org/drm/kernel: (28 commits)
drm/amdgpu: add missing vram lost check for LEGACY RESET
drm/amdgpu/discovery: fix fw based ip discovery
drm/amdkfd: Destroy KFD debugfs after destroy KFD wq
amdgpu/amdgpu_discovery: increase timeout limit for IFWI init
drm/amdgpu: Update SDMA firmware version check for user queue support
drm/amdgpu: Add NULL check for asic_funcs
drm/amd/display: Revert "drm/amd/display: Fix AMDGPU_MAX_BL_LEVEL value"
drm/amd/display: fix a Null pointer dereference vulnerability
drm/amd/display: Add primary plane to commits for correct VRR handling
drm/amdgpu: update mmhub 3.3 client id mappings
drm/amdgpu: update mmhub 3.0.1 client id mappings
drm/amdgpu: Retain job->vm in amdgpu_job_prepare_job
drm/amd/display: Fix DCE 6.0 and 6.4 PLL programming.
drm/amd/display: Don't overwrite dce60_clk_mgr
drm/amdkfd: Fix checkpoint-restore on multi-xcc
drm/amd: Restore cached manual clock settings during resume
drm/amd: Restore cached power limit during resume
drm/amdgpu: Update external revid for GC v9.5.0
drm/amdgpu: Update supported modes for GC v9.5.0
Mark xe driver as BROKEN if kernel page size is not 4kB
...
We only need the fw based discovery table for sysfs. No
need to parse it. Additionally parsing some of the board
specific tables may result in incorrect data on some boards.
just load the binary and don't parse it on those boards.
Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4441
Fixes: 80a0e82829 ("drm/amdgpu/discovery: optionally use fw based ip discovery")
Reviewed-by: Mario Limonciello (AMD) <superm1@kernel.org>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit 62eedd150f)
Cc: stable@vger.kernel.org
Assign unique id to compute partition. This is the unique id of the
first XCD instance belonging to the partition.
Signed-off-by: Lijo Lazar <lijo.lazar@amd.com>
Reviewed-by: Asad Kamal <asad.kamal@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Legacy resets reset the memory controllers so VRAM contents
may be unreliable after reset.
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
We only need the fw based discovery table for sysfs. No
need to parse it. Additionally parsing some of the board
specific tables may result in incorrect data on some boards.
just load the binary and don't parse it on those boards.
Closes: https://gitlab.freedesktop.org/drm/amd/-/issues/4441
Fixes: 80a0e82829 ("drm/amdgpu/discovery: optionally use fw based ip discovery")
Reviewed-by: Mario Limonciello (AMD) <superm1@kernel.org>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
