If page table PDEs is evicted and restored, after updating PDEs, need
increase vm->tlb_seq, then amdgpu_vm_flush will flush TLB before command
submission.
Signed-off-by: Philip Yang <Philip.Yang@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
For mes manages gds allocation, skip gds switch.
Signed-off-by: Jack Xiao <Jack.Xiao@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
VM might already be freed when amdgpu_vm_tlb_seq_cb() is called.
We see the calltrace below.
Fix it by keeping the last flush fence around and wait for it to signal
BUG kmalloc-4k (Not tainted): Poison overwritten
0xffff9c88630414e8-0xffff9c88630414e8 @offset=5352. First byte 0x6c
instead of 0x6b Allocated in amdgpu_driver_open_kms+0x9d/0x360 [amdgpu]
age=44 cpu=0 pid=2343
__slab_alloc.isra.0+0x4f/0x90
kmem_cache_alloc_trace+0x6b8/0x7a0
amdgpu_driver_open_kms+0x9d/0x360 [amdgpu]
drm_file_alloc+0x222/0x3e0 [drm]
drm_open+0x11d/0x410 [drm]
Freed in amdgpu_driver_postclose_kms+0x3e9/0x550 [amdgpu] age=22 cpu=1
pid=2485
kfree+0x4a2/0x580
amdgpu_driver_postclose_kms+0x3e9/0x550 [amdgpu]
drm_file_free+0x24e/0x3c0 [drm]
drm_close_helper.isra.0+0x90/0xb0 [drm]
drm_release+0x97/0x1a0 [drm]
__fput+0xb6/0x280
____fput+0xe/0x10
task_work_run+0x64/0xb0
Suggested-by: Christian König <christian.koenig@amd.com>
Signed-off-by: xinhui pan <xinhui.pan@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Add an usage for submissions independent of implicit sync but still
interesting for memory management.
v2: cleanup the kerneldoc a bit
v3: separate amdgpu changes from this
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20220407085946.744568-10-christian.koenig@amd.com
This change adds the dma_resv_usage enum and allows us to specify why a
dma_resv object is queried for its containing fences.
Additional to that a dma_resv_usage_rw() helper function is added to aid
retrieving the fences for a read or write userspace submission.
This is then deployed to the different query functions of the dma_resv
object and all of their users. When the write paratermer was previously
true we now use DMA_RESV_USAGE_WRITE and DMA_RESV_USAGE_READ otherwise.
v2: add KERNEL/OTHER in separate patch
v3: some kerneldoc suggestions by Daniel
v4: some more kerneldoc suggestions by Daniel, fix missing cases lost in
the rebase pointed out by Bas.
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20220407085946.744568-2-christian.koenig@amd.com
Audit all the users of dma_resv_add_excl_fence() and make sure they
reserve a shared slot also when only trying to add an exclusive fence.
This is the next step towards handling the exclusive fence like a
shared one.
v2: fix missed case in amdgpu
v3: and two more radeon, rename function
v4: add one more case to TTM, fix i915 after rebase
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20220406075132.3263-2-christian.koenig@amd.com
For VG20 + XGMI bridge, all mappings PTEs cache in TC, this may have
stall invalid PTEs in TC because one cache line has 8 pages. Need always
flush_tlb after updating mapping.
Signed-off-by: Philip Yang <Philip.Yang@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Testing the valid bit is not enough to figure out if we
need to invalidate the TLB or not.
During eviction it is quite likely that we move a BO from VRAM to GTT and
update the page tables immediately to the new GTT address.
Rework the whole function to get all the necessary parameters directly as
value.
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Philip Yang <Philip.Yang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Fix two incorrect kerneldocs for the recent VM code changes.
v2: fix one more typo
Signed-off-by: Christian König <christian.koenig@amd.com>
Reported-by: kernel test robot <lkp@intel.com>
Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Flush TLB needs wait for GPU update fence done. MMU notify callback to
unmap range from GPUs uses unlocked GPU page table update, so add tlb_cb
to unlocked update fence to increase vm->tlb_seq.
Signed-off-by: Philip Yang <Philip.Yang@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Instead of providing the bulk move structure for each LRU update set
this as property of the BO. This should avoid costly bulk move rebuilds
with some games under RADV.
v2: some name polishing, add a few more kerneldoc words.
v3: add some lockdep
v4: fix bugs, handle pin/unpin as well
v5: improve kerneldoc
Signed-off-by: Christian König <christian.koenig@amd.com>
Tested-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20220321132601.2161-5-christian.koenig@amd.com
This way we finally fix the problem that new resource are
not immediately evict-able after allocation.
That has caused numerous problems including OOM on GDS handling
and not being able to use TTM as general resource manager.
v2: stop assuming in ttm_resource_fini that res->bo is still valid.
v3: cleanup kerneldoc, add more lockdep annotation
v4: consistently use res->num_pages
Signed-off-by: Christian König <christian.koenig@amd.com>
Tested-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20220321132601.2161-1-christian.koenig@amd.com
Better to leave the decision when to flush the VM changes in the TLB to
the VM code.
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Philip Yang<Philip.Yang@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Instead of tracking the VM updates through the dependencies just use a
sequence counter for page table updates which indicates the need to
flush the TLB.
This reduces the need to flush the TLB drastically.
v2: squash in NULL check fix (Christian)
Signed-off-by: Christian König <christian.koenig@amd.com>
Acked-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Separate the VM page table backend operations from the state machine since
the amdgpu_vm.c file is becoming to complex.
The allocating, freeing and updating page tables and page directories can
easily be moved into a separate file.
While at it cleanup everything checkpatch.pl reported and rename the
functions a bit to make more clear that they belong together.
Signed-off-by: Christian König <christian.koenig@amd.com>
Acked-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Move the page tables to the idle list after updating the PDEs.
We have gone back and forth with that a couple of times because of problems
with the inter PD dependencies, but it should work now that we have the
state handling cleanly separated.
Signed-off-by: Christian König <christian.koenig@amd.com>
Acked-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Regression has been reported that suspend/resume may hang with
the previous vm ready check commit.
So bring back the evicted list check as a temp fix.
Bug: https://gitlab.freedesktop.org/drm/amd/-/issues/1922
Fixes: c1a66c3bc4 ("drm/amdgpu: check vm ready by amdgpu_vm->evicting flag")
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Qiang Yu <qiang.yu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
-----BEGIN PGP SIGNATURE-----
iQFSBAABCAA8FiEEq68RxlopcLEwq+PEeb4+QwBBGIYFAmIb/PEeHHRvcnZhbGRz
QGxpbnV4LWZvdW5kYXRpb24ub3JnAAoJEHm+PkMAQRiGugMH/R8icwG5gOjAuxBr
fuz9032ECVFS36Cy3ps9Eqgf5EdS4G6yz3joM4aNtp4B7e5FzI9lzBaHS8OAguNL
y7puFtBr9CywsnniJumZzciB9pEHmF/yyEKfMlRZA3JsRfLDacFstETp+duJnXoA
+s49IWsy1ot5zoherhDXcFLqDoFhLVU4hYwE1xpLpW/cllqmSnW8SDZMtWW1Ui/B
Of7zqINR4zBMmRjP4ymGq/ZrPWlFyWLdtOo0xxVoAQkeMgm33kfaaeGzfyK25FqR
JDGUZ7lkKvwz3PYh2hqJ7dc5K+vhJ18I+F1UmOiL6QAAUF/k8jq7i3Qaf6MKqh2Z
2v+A5k0=
=Hiar
-----END PGP SIGNATURE-----
Backmerge tag 'v5.17-rc6' into drm-next
This backmerges v5.17-rc6 so I can merge some amdgpu and some tegra changes on top.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Workstation application ANSA/META v21.1.4 get this error dmesg when
running CI test suite provided by ANSA/META:
[drm:amdgpu_gem_va_ioctl [amdgpu]] *ERROR* Couldn't update BO_VA (-16)
This is caused by:
1. create a 256MB buffer in invisible VRAM
2. CPU map the buffer and access it causes vm_fault and try to move
it to visible VRAM
3. force visible VRAM space and traverse all VRAM bos to check if
evicting this bo is valuable
4. when checking a VM bo (in invisible VRAM), amdgpu_vm_evictable()
will set amdgpu_vm->evicting, but latter due to not in visible
VRAM, won't really evict it so not add it to amdgpu_vm->evicted
5. before next CS to clear the amdgpu_vm->evicting, user VM ops
ioctl will pass amdgpu_vm_ready() (check amdgpu_vm->evicted)
but fail in amdgpu_vm_bo_update_mapping() (check
amdgpu_vm->evicting) and get this error log
This error won't affect functionality as next CS will finish the
waiting VM ops. But we'd better clear the error log by checking
the amdgpu_vm->evicting flag in amdgpu_vm_ready() to stop calling
amdgpu_vm_bo_update_mapping() later.
Another reason is amdgpu_vm->evicted list holds all BOs (both
user buffer and page table), but only page table BOs' eviction
prevent VM ops. amdgpu_vm->evicting flag is set only for page
table BOs, so we should use evicting flag instead of evicted list
in amdgpu_vm_ready().
The side effect of this change is: previously blocked VM op (user
buffer in "evicted" list but no page table in it) gets done
immediately.
v2: update commit comments.
Acked-by: Paul Menzel <pmenzel@molgen.mpg.de>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Qiang Yu <qiang.yu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Since newly added BOs don't have any mappings it's ok to add them
without holding the VM lock. Only when we add per VM BOs the lock is
mandatory.
Signed-off-by: Christian König <christian.koenig@amd.com>
Reported-by: Bhardwaj, Rajneesh <Rajneesh.Bhardwaj@amd.com>
Acked-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Some people complained about the name and this matches much
more Linux naming conventions for object functions.
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Whenever a bo_va structure is added or removed the VM and eventually
added BO should be locked.
Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
trace_amdgpu_vm_update_ptes trace unable to log when nptes too large
Signed-off-by: Somalapuram Amaranath <Amaranath.Somalapuram@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
adev_to_drm is used everywhere, so improve recent changes
when accessing drm_device pointer from amdgpu_device.
Signed-off-by: Guchun Chen <guchun.chen@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
We hit soft hang while doing memory pressure test on one numa system.
After a qucik look, this is because kfd invalid/valid userptr memory
frequently with process_info lock hold.
Looks like update page table mapping use too much cpu time.
perf top says below,
75.81% [kernel] [k] __srcu_read_unlock
6.19% [amdgpu] [k] amdgpu_gmc_set_pte_pde
3.56% [kernel] [k] __srcu_read_lock
2.20% [amdgpu] [k] amdgpu_vm_cpu_update
2.20% [kernel] [k] __sg_page_iter_dma_next
2.15% [drm] [k] drm_dev_enter
1.70% [drm] [k] drm_prime_sg_to_dma_addr_array
1.18% [kernel] [k] __sg_alloc_table_from_pages
1.09% [drm] [k] drm_dev_exit
So move drm_dev_enter/exit outside gmc code, instead let caller do it.
They are gart_unbind, gart_map, vm_clear_bo, vm_update_pdes and
gmc_init_pdb0. vm_bo_update_mapping already calls it.
Signed-off-by: xinhui pan <xinhui.pan@amd.com>
Reviewed-and-tested-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
amdgpu_bo_get_preferred_pin_domain is used for page tables
creation, which is not involved with page pinning. And it is used in
more cases than display scanout, modify its documentation as well.
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>
Check range access permission to restore GPU retry fault, if GPU retry
fault on address which belongs to VMA, and VMA has no read or write
permission requested by GPU, failed to restore the address. The vm fault
event will pass back to user space.
Signed-off-by: Philip Yang <Philip.Yang@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Why: Previously hw fence is alloced separately with job.
It caused historical lifetime issues and corner cases.
The ideal situation is to take fence to manage both job
and fence's lifetime, and simplify the design of gpu-scheduler.
How:
We propose to embed hw_fence into amdgpu_job.
1. We cover the normal job submission by this method.
2. For ib_test, and submit without a parent job keep the
legacy way to create a hw fence separately.
v2:
use AMDGPU_FENCE_FLAG_EMBED_IN_JOB_BIT to show that the fence is
embedded in a job.
v3:
remove redundant variable ring in amdgpu_job
v4:
add tdr sequence support for this feature. Add a job_run_counter to
indicate whether this job is a resubmit job.
v5
add missing handling in amdgpu_fence_enable_signaling
Signed-off-by: Jingwen Chen <Jingwen.Chen2@amd.com>
Signed-off-by: Jack Zhang <Jack.Zhang7@hotmail.com>
Reviewed-by: Andrey Grodzovsky <andrey.grodzovsky@amd.com>
Reviewed by: Monk Liu <monk.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This reverts commit 4e7b93ca52.
Revert reason: The issue has been resolved.
Signed-off-by: Eric Huang <jinhuieric.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This reverts commit 024d8811c9.
Revert reason: The issue has been resolved.
Signed-off-by: Eric Huang <jinhuieric.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This reverts commit 4e7b93ca52.
Revert reason: The issue has been resolved.
Signed-off-by: Eric Huang <jinhuieric.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This reverts commit 024d8811c9.
Revert reason: The issue has been resolved.
Signed-off-by: Eric Huang <jinhuieric.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This reverts commit 075e8080c1.
Reason for revert: the related commit is reverted.
Signed-off-by: Eric Huang <jinhuieric.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This reverts commit 7a68d188d1.
Reason for revert: the related commit is reverted.
Signed-off-by: Eric Huang <jinhuieric.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This reverts commit 075e8080c1.
Reason for revert: the related commit is reverted.
Signed-off-by: Eric Huang <jinhuieric.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This reverts commit 7a68d188d1.
Reason for revert: the related commit is reverted.
Signed-off-by: Eric Huang <jinhuieric.huang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Use new helper function amdgpu_vm_set_pasid() to
assign vm pasid value. This also ensures that we don't free
a pasid from vm code as pasids are allocated somewhere else.
Signed-off-by: Nirmoy Das <nirmoy.das@amd.com>
Acked-by: Felix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Replace idr with xarray as we actually need hash functionality.
Cleanup code related to vm pasid by adding helper function.
Signed-off-by: Nirmoy Das <nirmoy.das@amd.com>
Acked-by: Felix Kuehling <Felix.Kuehling@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Move shadow_list to struct amdgpu_bo_vm as shadow BOs
are part of PT/PD BOs.
Signed-off-by: Nirmoy Das <nirmoy.das@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Page table entries are now in embedded in VM BO, so
we do not need struct amdgpu_vm_pt. This patch replaces
struct amdgpu_vm_pt with struct amdgpu_vm_bo_base.
Signed-off-by: Nirmoy Das <nirmoy.das@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
