Convert existing drm_exec transactions, like GT pagefault validation,
non-LR exec() IOCTL and the rebind worker to support
exhaustive eviction using the xe_validation_guard().
v2:
- Adapt to signature change in xe_validation_guard() (Matt Brost)
- Avoid gotos from within xe_validation_guard() (Matt Brost)
- Check error return from xe_validation_guard()
v3:
- Rebase on gpu_madvise()
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com> #v1
Link: https://lore.kernel.org/r/20250908101246.65025-6-thomas.hellstrom@linux.intel.com
We want all validation (potential backing store allocation) to be part
of a drm_exec transaction. Therefore add a drm_exec pointer argument
to xe_bo_validate() and ___xe_bo_create_locked(). Upcoming patches
will deal with making all (or nearly all) calls to these functions
part of a drm_exec transaction. In the meantime, define special values
of the drm_exec pointer:
XE_VALIDATION_UNIMPLEMENTED: Implementation of the drm_exec transaction
has not been done yet.
XE_VALIDATION_UNSUPPORTED: Some Middle-layers (dma-buf) doesn't allow
the drm_exec context to be passed down to map_attachment where
validation takes place.
XE_VALIDATION_OPT_OUT: May be used only for kunit tests where exhaustive
eviction isn't crucial and the ROI of converting those is very
small.
For XE_VALIDATION_UNIMPLEMENTED and XE_VALIDATION_OPT_OUT there is also
a lockdep check that a drm_exec transaction can indeed start at the
location where the macro is expanded. This is to encourage
developers to take this into consideration early in the code
development process.
v2:
- Fix xe_vm_set_validation_exec() imbalance. Add an assert that
hopefully catches future instances of this (Matt Brost)
v3:
- Extend to psmi_alloc_object
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com> #v3
Link: https://lore.kernel.org/r/20250908101246.65025-2-thomas.hellstrom@linux.intel.com
When the xe pm_notifier evicts for suspend / hibernate, there might be
racing tasks trying to re-validate again. This can lead to suspend taking
excessive time or get stuck in a live-lock. This behaviour becomes
much worse with the fix that actually makes re-validation bring back
bos to VRAM rather than letting them remain in TT.
Prevent that by having exec and the rebind worker waiting for a completion
that is set to block by the pm_notifier before suspend and is signaled
by the pm_notifier after resume / wakeup.
It's probably still possible to craft malicious applications that block
suspending. More work is pending to fix that.
v3:
- Avoid wait_for_completion() in the kernel worker since it could
potentially cause work item flushes from freezable processes to
wait forever. Instead terminate the rebind workers if needed and
re-launch at resume. (Matt Auld)
v4:
- Fix some bad naming and leftover debug printouts.
- Fix kerneldoc.
- Use drmm_mutex_init() for the xe->rebind_resume_lock (Matt Auld).
- Rework the interface of xe_vm_rebind_resume_worker (Matt Auld).
Link: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/4288
Fixes: c6a4d46ec1 ("drm/xe: evict user memory in PM notifier")
Cc: Matthew Auld <matthew.auld@intel.com>
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: <stable@vger.kernel.org> # v6.16+
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://lore.kernel.org/r/20250904160715.2613-4-thomas.hellstrom@linux.intel.com
Goal here is cut over to gpusvm and remove xe_hmm, relying instead on
common code. The core facilities we need are get_pages(), unmap_pages()
and free_pages() for a given useptr range, plus a vm level notifier
lock, which is now provided by gpusvm.
v2:
- Reuse the same SVM vm struct we use for full SVM, that way we can
use the same lock (Matt B & Himal)
v3:
- Re-use svm_init/fini for userptr.
v4:
- Allow building xe without userptr if we are missing DRM_GPUSVM
config. (Matt B)
- Always make .read_only match xe_vma_read_only() for the ctx. (Dafna)
v5:
- Fix missing conversion with CONFIG_DRM_XE_USERPTR_INVAL_INJECT
v6:
- Convert the new user in xe_vm_madise.
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Dafna Hirschfeld <dafna.hirschfeld@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://lore.kernel.org/r/20250828142430.615826-17-matthew.auld@intel.com
copy_from_user() has more checks and is more safer than
__copy_from_user()
Suggested-by: Kees Cook <kees@kernel.org>
Signed-off-by: Harish Chegondi <harish.chegondi@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
Signed-off-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
Link: https://lore.kernel.org/r/acabf20aa8621c7bc8de09b1bffb8d14b5376484.1746126614.git.harish.chegondi@intel.com
The driver needs to know if a BO is encrypted with PXP to enable the
display decryption at flip time.
Furthermore, we want to keep track of the status of the encryption and
reject any operation that involves a BO that is encrypted using an old
key. There are two points in time where such checks can kick in:
1 - at VM bind time, all operations except for unmapping will be
rejected if the key used to encrypt the BO is no longer valid. This
check is opt-in via a new VM_BIND flag, to avoid a scenario where a
malicious app purposely shares an invalid BO with a non-PXP aware
app (such as a compositor). If the VM_BIND was failed, the
compositor would be unable to display anything at all. Allowing the
bind to go through means that output still works, it just displays
garbage data within the bounds of the illegal BO.
2 - at job submission time, if the queue is marked as using PXP, all
objects bound to the VM will be checked and the submission will be
rejected if any of them was encrypted with a key that is no longer
valid.
Note that there is no risk of leaking the encrypted data if a user does
not opt-in to those checks; the only consequence is that the user will
not realize that the encryption key is changed and that the data is no
longer valid.
v2: Better commnnts and descriptions (John), rebase
v3: Properly return the result of key_assign up the stack, do not use
xe_bo in display headers (Jani)
v4: improve key_instance variable documentation (John)
Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: John Harrison <John.C.Harrison@Intel.com>
Cc: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: John Harrison <John.C.Harrison@Intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20250129174140.948829-11-daniele.ceraolospurio@intel.com
In couple of places the wrong error handling goto was used to release
locks. Fix these to ensure all locks dropped on exec IOCTL errors.
Cc: Francois Dugast <francois.dugast@intel.com>
Fixes: d16ef1a18e ("drm/xe/exec: Switch hw engine group execution mode upon job submission")
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Francois Dugast <francois.dugast@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20241106224944.30130-1-matthew.brost@intel.com
Upon failure all locks need to be dropped before returning to the user.
Fixes: 58480c1c91 ("drm/xe: Skip VMAs pin when requesting signal to the last XE_EXEC")
Cc: <stable@vger.kernel.org>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Tejas Upadhyay <tejas.upadhyay@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20241105043524.4062774-3-matthew.brost@intel.com
In a couple of places after an exec queue is looked up the exec IOCTL
returns on input errors without dropping the exec queue ref. Fix this
ensuring the exec queue ref is dropped on input error.
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: Tejas Upadhyay <tejas.upadhyay@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20241105043524.4062774-2-matthew.brost@intel.com
Fix external BO's dma-resv usage in exec IOCTL using bookkeep slots
rather than write slots. This leaves syncing to user space rather than
the KMD blindly enforcing write semantics on every external BO.
Fixes: dd08ebf6c3 ("drm/xe: Introduce a new DRM driver for Intel GPUs")
Cc: José Roberto de Souza <jose.souza@intel.com>
Cc: Kenneth Graunke <kenneth.w.graunke@intel.com>
Cc: Paulo Zanoni <paulo.r.zanoni@intel.com>
Reported-by: Simona Vetter <simona.vetter@ffwll.ch>
Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2673
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20240911152622.903058-1-matthew.brost@intel.com
If the job about to be submitted is a dma-fence job, update the current
execution mode of the hw engine group. This triggers an immediate suspend
of the exec queues running faulting long-running jobs.
If the job about to be submitted is a long-running job, kick a new worker
used to resume the exec queues running faulting long-running jobs once
the dma-fence jobs have completed.
v2: Kick the resume worker from exec IOCTL, switch to unordered workqueue,
destroy it after use (Matt Brost)
v3: Do not resume if no exec queue was suspended (Matt Brost)
v4: Squash commits (Matt Brost)
v5: Do not kick the worker when xe_vm_in_preempt_fence_mode (Matt Brost)
Signed-off-by: Francois Dugast <francois.dugast@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240809155156.1955925-10-francois.dugast@intel.com
Increment num_syncs after xe_sync_entry_parse() is successful to ensure
the xe_sync_entry_cleanup() logic under "err_syncs" label works correctly.
v2: Use the same pattern as that in xe_vm.c (Matt Brost)
Fixes: dd08ebf6c3 ("drm/xe: Introduce a new DRM driver for Intel GPUs")
Signed-off-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240711211203.3728180-1-ashutosh.dixit@intel.com
Move it out of the xe_migrate compilation unit so it can be re-used in
other places.
Cc: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Francois Dugast <francois.dugast@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240614094433.775866-1-francois.dugast@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Clean up laying violation of setting q->flags EXEC_QUEUE_FLAG_BANNED bit
in GuC backend. Move banned to GuC owned bit and report banned status to
upper layers via reset_status vfunc. This is a slight change in behavior
as reset_status returns true if wedged or killed bits set too, but in
all of these cases submission to queue is no longer allowed.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Jonathan Cavitt <jonathan.cavitt@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240604184700.1946918-1-matthew.brost@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Rebinding might allocate page-table bos, causing evictions.
To support blocking locking during these evictions,
perform the rebinding in the drm_exec locking loop.
Also Reserve fence slots where actually needed rather than trying to
predict how many fence slots will be needed over a complete
wound-wait transaction.
v2:
- Remove a leftover call to xe_vm_rebind() (Matt Brost)
- Add a helper function xe_vm_validate_rebind() (Matt Brost)
v3:
- Add comments and squash with previous patch (Matt Brost)
Fixes: 24f947d58f ("drm/xe: Use DRM GPUVM helpers for external- and evicted objects")
Fixes: 29f424eb87 ("drm/xe/exec: move fence reservation")
Cc: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240327091136.3271-5-thomas.hellstrom@linux.intel.com
Instead of handling the vm's rebind fence separately,
which is error prone if they are not strictly ordered,
attach rebind fences as kernel fences to the vm's resv.
Fixes: dd08ebf6c3 ("drm/xe: Introduce a new DRM driver for Intel GPUs")
Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: <stable@vger.kernel.org> # v6.8+
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240327091136.3271-3-thomas.hellstrom@linux.intel.com
Refactor xe_sync_entry_signal so it doesn't have to
modify xe_sched_job struct instead create a new helper function
to set user fence values for a job.
v2: Move the sync type check to xe_sched_job_init_user_fence(Lucas)
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Cc: Matthew Auld <matthew.auld@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: Michal Wajdeczko <michal.wajdeczko@intel.com>
Signed-off-by: Nirmoy Das <nirmoy.das@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240321161142.4954-1-nirmoy.das@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Doing a XE_EXEC with num_batch_buffer == 0 makes signals passed as
argument to be signaled when the last real XE_EXEC is completed.
But to do that it was first pinning all VMAs in drm_gpuvm_exec_lock(),
this patch remove this pinning as it is not required.
This change also help Mesa implementing memory over-commiting recovery
as it needs to unbind not needed VMAs when the whole VM can't fit
in GPU memory but it can only do the unbiding when the last XE_EXEC
is completed.
So with this change Mesa can get the signal it want without getting
out-of-memory errors.
Fixes: eb9702ad29 ("drm/xe: Allow num_batch_buffer / num_binds == 0 in IOCTLs")
Cc: Thomas Hellstrom <thomas.hellstrom@linux.intel.com>
Co-developed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240313171318.121066-1-jose.souza@intel.com
At label err_unlock_list the condition write_label will never be true.
Remove the deadcode line for write_label true.
Reported by static analyzer.
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: Tejas Upadhyay <tejas.upadhyay@intel.com>
Cc: Bommu Krishnaiah <krishnaiah.bommu@intel.com>
Reviewed-by: Tejas Upadhyay <tejas.upadhyay@intel.com>
Signed-off-by: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240313150545.2830408-3-himal.prasad.ghimiray@intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The way exec ufences are coded only 1 ufence per IOCTL will be signaled.
It is possible to fix this but for current use cases 1 ufence per IOCTL
is sufficient. Enforce a limit of 1 ufence per IOCTL (both exec and bind
to be uniform).
v2:
- Add fixes tag (Thomas)
Fixes: dd08ebf6c3 ("drm/xe: Introduce a new DRM driver for Intel GPUs")
Cc: Mika Kahola <mika.kahola@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Brian Welty <brian.welty@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240124234413.1640825-1-matthew.brost@intel.com
The intent is to return -EWOULDBLOCK to the user if a long running exec
queue is full during the exec IOCTL. -EWOULDBLOCK aliases to -EAGAIN
which results in the exec IOCTL doing a retry loop. Fix this by ensuring
the retry loop is broken when returning -EWOULDBLOCK.
Fixes: 8ae8a2e8dd ("drm/xe: Long running job update")
Reported-by: Sai Gowtham Ch <sai.gowtham.ch@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Brian Welty <brian.welty@intel.com>
(cherry picked from commit 97d0047cbb)
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
The intent is to return -EWOULDBLOCK to the user if a long running exec
queue is full during the exec IOCTL. -EWOULDBLOCK aliases to -EAGAIN
which results in the exec IOCTL doing a retry loop. Fix this by ensuring
the retry loop is broken when returning -EWOULDBLOCK.
Fixes: 8ae8a2e8dd ("drm/xe: Long running job update")
Reported-by: Sai Gowtham Ch <sai.gowtham.ch@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Brian Welty <brian.welty@intel.com>
Looks possible to switch from CPU binding to GPU binding mid exec, and
if that happens for the same dma-resv we might use two fence slots, once
for the dummy fence, and another for the actual GPU bind.
References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/698
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
We currently assume that we can upfront know exactly how many fence
slots we will need at the start of the exec, however the TTM bo_validate
can itself consume numerous fence slots, and due to how the
dma_resv_reserve_fences() works it only ensures that at least that many
fence slots are available. With this it is quite possible that TTM
steals some of the fence slots and then when it comes time to do the vma
binding and final exec stage we are lacking enough fence slots, leading
to some nasty BUG_ON(). A simple fix is to reserve our own fences later,
after the validate stage.
References: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/698
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Tested-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Xe, is a new driver for Intel GPUs that supports both integrated and
discrete platforms. The experimental support starts with Tiger Lake.
i915 will continue be the main production driver for the platforms
up to Meteor Lake and Alchemist. Then the goal is to make this Intel
Xe driver the primary driver for Lunar Lake and newer platforms.
It uses most, if not all, of the key drm concepts, in special: TTM,
drm-scheduler, drm-exec, drm-gpuvm/gpuva and others.
-----BEGIN PGP SIGNATURE-----
iQEzBAABCAAdFiEEbSBwaO7dZQkcLOKj+mJfZA7rE8oFAmWErlkACgkQ+mJfZA7r
E8ptAQf/UILUy63usqMBQx5gGTJ5e6oyGy0r97JdBBWNxhcu3/4uN1SR6V2LSV7p
mnZt/LrHZ/24s73e5D7sh909GCYm/MMWH7v0KypJq5Z74BHb9IePP7+Q4NTfXfqS
5AotYxDEwPBQZpfYLA7y17XrB01yz/gF8wHytBFTPiPOlve7BDAYw4j8o+ocztxy
sLLwPi7A6RJJlpXaDa33DpH5MLpTVYUOuxaIElqu949AJj/Me+LHd26VfoGWt9YT
AWzAm+LLoBhyE5lk9swWxvphGvNoOuKf8soTMEVXr4qxsR6d4LShkrljSYSbtRhl
2QZMgvSL4pGkB3Yvfxb0hidWuVMNfA==
=bubx
-----END PGP SIGNATURE-----
Merge tag 'drm-xe-next-2023-12-21-pr1-1' of https://gitlab.freedesktop.org/drm/xe/kernel into drm-next
Introduce a new DRM driver for Intel GPUs
Xe, is a new driver for Intel GPUs that supports both integrated and
discrete platforms. The experimental support starts with Tiger Lake.
i915 will continue be the main production driver for the platforms
up to Meteor Lake and Alchemist. Then the goal is to make this Intel
Xe driver the primary driver for Lunar Lake and newer platforms.
It uses most, if not all, of the key drm concepts, in special: TTM,
drm-scheduler, drm-exec, drm-gpuvm/gpuva and others.
Signed-off-by: Dave Airlie <airlied@redhat.com>
[airlied: add an extra X86 check, fix a typo, fix drm_exec_init interface
change].
From: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/ZYSwLgXZUZ57qGPQ@intel.com
The idea being out-syncs can signal indicating all previous operations
on the bind queue are complete. An example use case of this would be
support for implementing vkQueueWaitIdle easily.
All in-syncs are waited on before signaling out-syncs. This is
implemented by forming a composite software fence of in-syncs and
installing this fence in the out-syncs and exec queue last fence slot.
The last fence must be added as a dependency for jobs on user exec
queues as it is possible for the last fence to be a composite software
fence (unordered, ioctl with zero bb or binds) rather than hardware
fence (ordered, previous job on queue).
Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Use a flags field instead of severval bools for sync parse as it is
easier to read and less bug prone.
v2: Pull in header change from subsequent patch
Suggested-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Adapt to the DRM_GPUVM helpers moving removing a lot of complicated
driver-specific code.
For now this uses fine-grained locking for the evict list and external
object list, which may incur a slight performance penalty in some
situations.
v2:
- Don't lock all bos and validate on LR exec submissions (Matthew Brost)
- Add some kerneldoc
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Acked-by: Matthew Brost <matthew.brost@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231212100144.6833-2-thomas.hellstrom@linux.intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The name "compute_mode" can be confusing since compute uses either this
mode or fault_mode to achieve the long-running semantics, and compute_mode
can, moving forward, enable fault_mode under the hood to work around
hardware limitations.
Also the name no_dma_fence_mode really refers to what we elsewhere call
long-running mode and the mode contrary to what its name suggests allows
dma-fences as in-fences.
So in an attempt to be more consistent, rename
no_dma_fence_mode -> lr_mode
compute_mode -> preempt_fence_mode
And adjust flags so that
preempt_fence_mode sets XE_VM_FLAG_LR_MODE
fault_mode sets XE_VM_FLAG_LR_MODE | XE_VM_FLAG_FAULT_MODE
v2:
- Fix a typo in the commit message (Oak Zeng)
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Oak Zeng <oak.zeng@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231127123349.23698-1-thomas.hellstrom@linux.intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Async worker is gone. All jobs and memory allocations done in IOCTL to
align with dma fencing rules.
Async vs. sync now means when do bind operations complete relative to
the IOCTL. Async completes when out-syncs signal while sync completes
when the IOCTL returns. In-syncs and out-syncs are only allowed in async
mode.
If memory allocations fail in the job creation step the VM is killed.
This is temporary, eventually a proper unwind will be done and VM will
be usable.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
The XE_WARN_ON macro maps to WARN_ON which is not justified
in many cases where only a simple debug check is needed.
Replace the use of the XE_WARN_ON macro with the new xe_assert
macros which relies on drm_*. This takes a struct drm_device
argument, which is one of the main changes in this commit. The
other main change is that the condition is reversed, as with
XE_WARN_ON a message is displayed if the condition is true,
whereas with xe_assert it is if the condition is false.
v2:
- Rebase
- Keep WARN splats in xe_wopcm.c (Matt Roper)
v3:
- Rebase
Signed-off-by: Francois Dugast <francois.dugast@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Both execs and the preempt rebind worker can issue rebinds. Rebinds
require a fence, per tile, inserted into dma-resv slots of the VM and
BO (if external). The fence reservation accouting did not take into
account the number of fences required for rebinds, fix this.
v2: Rebase
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reported-by: Christopher Snowhill <kode54@gmail.com>
Link: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/518
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Replace the calls to ttm_eu_reserve_buffers() by using the drm_exec
helper instead. Also make sure the locking loop covers any calls to
xe_bo_validate() / ttm_bo_validate() so that these function calls may
easily benefit from being called from within an unsealed locking
transaction and may thus perform blocking dma_resv locks in the future.
For the unlock we remove an assert that the vm->rebind_list is empty
when locks are released. Since if the error path is hit with a partly
locked list, that assert may no longer hold true we chose to remove it.
v3:
- Don't accept duplicate bo locks in the rebind worker.
v5:
- Loop over drm_exec objects in reverse when unlocking.
v6:
- We can't keep the WW ticket when retrying validation on OOM. Fix.
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230908091716.36984-5-thomas.hellstrom@linux.intel.com
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Engine was inappropriately used to refer to execution queues and it
also created some confusion with hardware engines. Where it applies
the exec_queue variable name is changed to q and comments are also
updated.
Link: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/162
Signed-off-by: Francois Dugast <francois.dugast@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
This is a preparation commit for a larger renaming of engine to exec queue.
Signed-off-by: Francois Dugast <francois.dugast@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Combine the userptr, rebind, and destroy links into a union as
the lists these links belong to are mutually exclusive.
v2: Adjust which lists are combined (Thomas H)
v3: Add kernel doc why this is safe (Thomas H), remove related change
of list_del_init -> list_del (Rodrigo)
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reduce the number of warnings reported by checkpatch.pl from 118 to 48 by
addressing those warnings types:
LEADING_SPACE
LINE_SPACING
BRACES
TRAILING_SEMICOLON
CONSTANT_COMPARISON
BLOCK_COMMENT_STYLE
RETURN_VOID
ONE_SEMICOLON
SUSPECT_CODE_INDENT
LINE_CONTINUATIONS
UNNECESSARY_ELSE
UNSPECIFIED_INT
UNNECESSARY_INT
MISORDERED_TYPE
Signed-off-by: Francois Dugast <francois.dugast@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Rather than open coding VM binds and VMA tracking, use the GPUVA
library. GPUVA provides a common infrastructure for VM binds to use mmap
/ munmap semantics and support for VK sparse bindings.
The concepts are:
1) xe_vm inherits from drm_gpuva_manager
2) xe_vma inherits from drm_gpuva
3) xe_vma_op inherits from drm_gpuva_op
4) VM bind operations (MAP, UNMAP, PREFETCH, UNMAP_ALL) call into the
GPUVA code to generate an VMA operations list which is parsed, committed,
and executed.
v2 (CI): Add break after default in case statement.
v3: Rebase
v4: Fix some error handling
v5: Use unlocked version VMA in error paths
v6: Rebase, address some review feedback mainly Thomas H
v7: Fix compile error in xe_vma_op_unwind, address checkpatch
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
This will help with the GPUVA port as the internals of struct xe_vma
will change.
v2: Update comment around helpers
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.kernel.org>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
We cannot recover a VM if a rebind worker hits an error, ban the VM if
happens to ensure we do not attempt to place this VM on the hardware
again.
A follow up will inform the user if this happens.
v2: Return -ECANCELED in exec VM closed or banned, check for closed or
banned within VM lock.
v3: Fix lockdep splat by looking engine outside of vm->lock
v4: Fix error path when engine lookup fails
v5: Add debug message in rebind worker on error, update comments wrt
locking, add xe_vm_close helper
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Use the TTM LRU bulk move for BOs tied to a VM. Update the bulk moves
LRU position on every exec.
v2: Bulk move for compute VMs, use WARN rather than BUG
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
For long running (LR) jobs with the DRM scheduler we must return NULL in
run_job which results in signaling the job's finished fence immediately.
This prevents LR jobs from creating infinite dma-fences.
Signaling job's finished fence immediately breaks flow controlling ring
with the DRM scheduler. To work around this, the ring is flow controlled
and written in the exec IOCTL. Signaling job's finished fence
immediately also breaks the TDR which is used in reset / cleanup entity
paths so write a new path for LR entities.
v2: Better commit, white space, remove rmb(), better comment next to
emit_job()
v3 (Thomas): Change LR reference counting, fix working in commit
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Add uAPI and implementation for NULL bindings. A NULL binding is defined
as writes dropped and read zero. A single bit in the uAPI has been added
which results in a single bit in the PTEs being set.
NULL bindings are intendedd to be used to implement VK sparse bindings,
in particular residencyNonResidentStrict property.
v2: Fix BUG_ON shown in VK testing, fix check patch warning, fix
xe_pt_scan_64K, update __gen8_pte_encode to understand NULL bindings,
remove else if vma_addr
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Suggested-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Padding and reserved fields are declared such that they must be
zeroed, so verify that they're all zero in the respective ioctl
functions.
Derived from original patch by mlankhorst.
v2:
Removed extensions checks where there were none originally. (José)
Moved extraneous parentheses to the correct places. (Lucas)
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: Christopher Snowhill <kode54@gmail.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
With multiple active VMs, under memory pressure, it is possible that
ttm_bo_validate() run into -EDEADLK in ttm_mem_evict_wait_busy() and
return -ENOMEM.
Until ttm properly handles locking in such scenarios, best thing the
driver can do is unwind the lock and retry.
Update xe_exec_begin to retry validating BOs with a timeout upon
-ENOMEM.
Reviewed-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Signed-off-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
If all compute engines of a vm in compute mode are idle,
defer a rebind to the next exec to avoid the VM unnecessarily trying
to make memory resident and compete with other VMs for available
memory space.
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Thomas Hellström