linux-stable.git/drivers/gpu/drm/xe, branch v6.8.2

drm/xe: Skip VMAs pin when requesting signal to the last XE_EXEC

2024-03-26T22:17:33+00:00

[ Upstream commit dd8a07f06dfd946e0eea1a3323d52e7c28a6ed80 ]

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: eb9702ad2986 ("drm/xe: Allow num_batch_buffer / num_binds == 0 in IOCTLs")
Cc: Thomas Hellstrom 
Co-developed-by: Matthew Brost 
Signed-off-by: José Roberto de Souza 
Reviewed-by: Matthew Brost 
Signed-off-by: Matthew Brost 
Link: https://patchwork.freedesktop.org/patch/msgid/20240313171318.121066-1-jose.souza@intel.com
(cherry picked from commit 58480c1c912ff8146d067301a0d04cca318b4a66)
Signed-off-by: Lucas De Marchi 
Signed-off-by: Sasha Levin

drm/xe: Invalidate userptr VMA on page pin fault

2024-03-26T22:17:33+00:00

[ Upstream commit 386021394394eccef248dc5eb9c9370240821a8c ]

Rather than return an error to the user or ban the VM when userptr VMA
page pin fails with -EFAULT, invalidate VMA mappings. This supports the
UMD use case of freeing userptr while still having bindings.

Now that non-faulting VMs can invalidate VMAs, drop the usm prefix for
the tile_invalidated member.

v2:
 - Fix build error (CI)
v3:
 - Don't invalidate VMA if in fault mode, rather kill VM (Thomas)
 - Update commit message with tile_invalidated name chagne (Thomas)
 - Wait VM bookkeep slots with VM resv lock (Thomas)
v4:
 - Move list_del_init(&userptr.repin_link) after error check (Thomas)
 - Assert not in fault mode (Matthew)

Fixes: dd08ebf6c352 ("drm/xe: Introduce a new DRM driver for Intel GPUs")
Signed-off-by: Matthew Brost 
Reviewed-by: Thomas Hellström 
Link: https://patchwork.freedesktop.org/patch/msgid/20240312183907.933835-1-matthew.brost@intel.com
(cherry picked from commit 521db22a1d70dbc596a07544a738416025b1b63c)
Signed-off-by: Lucas De Marchi 
Signed-off-by: Sasha Levin

drm/xe: Replace 'grouped target' in Makefile with pattern rule

2024-03-26T22:17:16+00:00

[ Upstream commit e62d2e00780b4a465c77d2229837495fcbc480d3 ]

Since 'grouped target' is used only in 'make' 4.3, it should
be avoided. Replace it with 'multi-target pattern rule' which
has the same behavior.

Fixes: 9616e74b796c ("drm/xe: Add support for OOB workarounds")
Signed-off-by: Dafna Hirschfeld 
Reviewed-by: Lucas De Marchi 
Link: https://patchwork.freedesktop.org/patch/msgid/20240302153927.2602241-1-dhirschfeld@habana.ai
[ reword commit message ]
Signed-off-by: Lucas De Marchi 
(cherry picked from commit 5224ed586ba7f9bba956655a1bfe5b75df7394d4)
Signed-off-by: Lucas De Marchi 
Signed-off-by: Sasha Levin

drm/xe: Fix ref counting leak on page fault

2024-03-26T22:17:16+00:00

[ Upstream commit f7da398935f7ddabf1a098714593e032c875cd74 ]

If a page fault occurs on VM not in fault a ref can be leaked. Fix this.

Fixes: dd08ebf6c352 ("drm/xe: Introduce a new DRM driver for Intel GPUs")
Signed-off-by: Matthew Brost 
Reviewed-by: Mika Kuoppala 
Link: https://patchwork.freedesktop.org/patch/msgid/20240301041036.238471-1-matthew.brost@intel.com
(cherry picked from commit 27b5a3f237fe66dbf2288c2b50973aee8a427e41)
Signed-off-by: Lucas De Marchi 
Signed-off-by: Sasha Levin

drm/xe/tests: Fix printf format specifiers in xe_migrate test

2024-03-26T22:16:28+00:00

[ Upstream commit 689a930b93c5c20294df5da0407df361c5412eac ]

KUNIT_FAIL() is used to fail the xe_migrate test when an error occurs.
However, there's a mismatch in the format specifier: '%li' is used to
log 'err', which is an 'int'.

Use '%i' instead of '%li', and for the case where we're printing an
error pointer, just use '%pe', instead of extracting the error code
manually with PTR_ERR(). (This also results in a nicer output when the
error code is known.)

Fixes: dd08ebf6c352 ("drm/xe: Introduce a new DRM driver for Intel GPUs")
Signed-off-by: David Gow 
Tested-by: Guenter Roeck 
Reviewed-by: Lucas De Marchi 
Acked-by: Thomas Hellström 
Signed-off-by: Shuah Khan 
Signed-off-by: Sasha Levin

drm/xe: Return immediately on tile_init failure

2024-03-07T08:13:38+00:00

There's no reason to proceed with applying workaround and initing
sysfs if we are going to abort the probe upon failure.

Fixes: e5a845fd8fa4 ("drm/xe: Add sysfs entry for tile")
Cc: Lucas De Marchi 
Cc: Matt Roper 
Cc: Matthew Auld 
Reviewed-by: Matt Roper 
Link: https://patchwork.freedesktop.org/patch/msgid/20240306203110.146387-1-rodrigo.vivi@intel.com
Signed-off-by: Rodrigo Vivi 
(cherry picked from commit af7b93d1d7eeeef674681ddea875be6a29857a5d)
Signed-off-by: Thomas Hellström

drm/xe/xe_trace: Add move_lacks_source detail to xe_bo_move trace

2024-02-29T11:32:15+00:00

Add move_lacks_source detail to xe_bo_move trace to make it readable
that is to check if it is migrate clear or migrate copy.

Cc: Thomas Hellström 
Signed-off-by: Priyanka Dandamudi 
Reviewed-by: Thomas Hellström 
Fixes: a09946a9a903 ("drm/xe/xe_bo_move: Enhance xe_bo_move trace")
Signed-off-by: Thomas Hellström 
Link: https://patchwork.freedesktop.org/patch/msgid/20240221101950.1019312-1-priyanka.dandamudi@intel.com
(cherry picked from commit 8034f6b070cc3716e81b1846f8a4ca5339c3f29b)
Signed-off-by: Thomas Hellström

drm/xe: Deny unbinds if uapi ufence pending

2024-02-29T09:39:03+00:00

If user fence was provided for MAP in vm_bind_ioctl
and it has still not been signalled, deny UNMAP of said
vma with EBUSY as long as unsignalled fence exists.

This guarantees that MAP vs UNMAP sequences won't
escape under the radar if we ever want to track the
client's state wrt to completed and accessible MAPs.
By means of intercepting the ufence release signalling.

v2: find ufence with num_fences > 1 (Matt)
v3: careful on clearing vma ufence (Matt)

Link: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/1159
Cc: Thomas Hellström 
Cc: Matthew Brost 
Cc: Joonas Lahtinen 
Signed-off-by: Mika Kuoppala 
Reviewed-by: Matthew Brost 
Signed-off-by: Thomas Hellström 
Link: https://patchwork.freedesktop.org/patch/msgid/20240215181152.450082-3-mika.kuoppala@linux.intel.com
(cherry picked from commit 158900ade92cce5ab85a06d618eb51e6c7ffb28a)
Signed-off-by: Thomas Hellström

drm/xe: Expose user fence from xe_sync_entry

2024-02-29T09:39:02+00:00

By allowing getting reference to user fence, we can
control the lifetime outside of sync entries.

This is needed to allow vma to track the associated
user fence that was provided with bind ioctl.

v2: xe_user_fence can be kept opaque (Jani, Matt)
v3: indent fix (Matt)

Cc: Thomas Hellström 
Cc: Matthew Brost 
Cc: Jani Nikula 
Signed-off-by: Mika Kuoppala 
Reviewed-by: Matthew Brost 
Signed-off-by: Thomas Hellström 
Link: https://patchwork.freedesktop.org/patch/msgid/20240215181152.450082-2-mika.kuoppala@linux.intel.com
(cherry picked from commit 977e5b82e0901480bc201342d39f855fc0a2ef47)
Signed-off-by: Thomas Hellström

drm/xe: Use pointers in trace events

2024-02-29T09:39:02+00:00

Commit a0df2cc858c3 ("drm/xe/xe_bo_move: Enhance xe_bo_move trace")
inadvertently reverted commit 8d038f49c1f3 ("drm/xe: Fix cast on trace
variable"), breaking the build on 32bits.

As noted by Ville, there's no point in converting the pointers to u64
and add casts everywhere. In fact, it's better to just use %p and let
the address be hashed. Convert all the cases in xe_trace.h to use
pointers.

Cc: Ville Syrjälä 
Cc: Matt Roper 
Cc: Priyanka Dandamudi 
Cc: Oak Zeng 
Cc: Thomas Hellström 
Signed-off-by: Lucas De Marchi 
Reviewed-by: Thomas Hellström 
Link: https://patchwork.freedesktop.org/patch/msgid/20240222144125.2862546-1-lucas.demarchi@intel.com
(cherry picked from commit 7a975748d4dc0a524c99a390c6f74b7097ef8cf7)
Signed-off-by: Thomas Hellström