| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
Add helper which returns valid from hw perspective offset
ready to be inserted to gpu pipeline.
Signed-off-by: Zbigniew Kempczyński <zbigniew.kempczynski@intel.com>
Signed-off-by: Priyanka Dandamudi <priyanka.dandamudi@intel.com>
Reviewed-by: Andrzej Hajda <andrzej.hajda@intel.com>
|
|
In Xe2 and beyond fast-copy between objects with different pat indices
is quite common operation. Add fast-copy helper which requires only
src/dst bo handle, region, pat index and size of operation.
It is however limited to size being multiple of 4KiB and lesser or equal
16MiB but this limitation may removed if there will be user of this.
Cc: Matthew Auld <matthew.auld@intel.com>
Cc: Nakshtra Goyal <nakshtra.goyal@intel.com>
Tested-by: Nakshtra Goyal <nakshtra.goyal@intel.com>
Reviewed-by: Nakshtra Goyal <nakshtra.goyal@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://lore.kernel.org/r/20250930143130.186193-5-zbigniew.kempczynski@intel.com
Signed-off-by: Zbigniew Kempczyński <zbigniew.kempczynski@intel.com>
|
|
There are 2 things happening here: one is converting time in nanoseconds
to ticks by using the refclock, and another asserting a spin duration is
not so close to the maximum duration since there needs to be room for
context switch.
Move the time conversion to xe_util.c and adjust it to maintain
namespace and have better names. Places that do the time conversion to
pass to xe_spin then use the xe_spin_nsec_to_ticks() wrapper to
calculate the ticks.
Reviewed-by: Jonathan Cavitt <jonathan.cavitt@intel.com>
Link: https://lore.kernel.org/r/20250107202615.1306306-4-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
|
|
Some functions are clearly gt-related, so move them to xe_gt.
Reviewed-by: Jonathan Cavitt <jonathan.cavitt@intel.com>
Link: https://lore.kernel.org/r/20250107202615.1306306-2-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
|
|
Add 2 new functions, xe_gt_count_engines_by_class and
xe_gt_fill_engines_by_class, that can be used as helpers to setup
parallel/virtual exec queue. All the tests were basically looping
and filtering engines that match gt and class.
Place these new functions in xe_util.c as currently we lack a better
placement. xe_query.c and xe_gt.c could probably be other candidates,
but they all mix all kind of namespaces too.
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://lore.kernel.org/r/20240814175013.3679997-2-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
|
|
For some cases we are going to need to pass the pat_index for the
vm_bind op. Add a helper for this, such that we can allocate an address
and give the mapping some pat_index.
v2 (Zbigniew)
- Plumb pat_index down into intel_allocator_record and protect against
potential changes.
- Add pat_index to bind_debug().
v3 (Zbigniew)
- A few more improvements
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Zbigniew Kempczyński <zbigniew.kempczynski@intel.com>
Cc: José Roberto de Souza <jose.souza@intel.com>
Cc: Pallavi Mishra <pallavi.mishra@intel.com>
Reviewed-by: Zbigniew Kempczyński <zbigniew.kempczynski@intel.com>
|
|
Align with commit ("drm/xe/uapi: Add missing DRM_ prefix in uAPI constants")
Signed-off-by: Francois Dugast <francois.dugast@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
|
|
When GT is in RC6 actual frequency reported will be 0. So verify actual
on the basic of GT state.
v2:
- Rebased
- Move function xe_is_gt_in_c6, to check rc6 state, to igt_pm library
- Assert if freq reported is not 0 when idle (Ashutosh)
v3:
- Fix review comments (Ashutosh)
v4:
- Move igt_require to igt_main under idle test cases (Ashutosh)
- For idle case wait for C6 instead of usleep (Ashutosh)
v5:
- Remove igt_require from test_freq_range
v6:
- Move xe_is_gt_in_c6 to xe/xe_util.c (Anshuman)
Fixes: acaaca0bf317 ("tests/xe: Add Xe IGT tests")
Signed-off-by: Badal Nilawar <badal.nilawar@intel.com>
Reviewed-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
Signed-off-by: Anshuman Gupta <anshuman.gupta@intel.com>
|
|
Before calling exec we need to prepare vm to contain valid entries.
Bind/unbind in xe expects single bind_op or vector of bind_ops what
makes preparation of it a little bit inconvinient. Add function
which iterates over list of xe_object (auxiliary structure which
describes bind information for object) and performs the bind/unbind
in one step. It also supports passing syncobj in/out to work in
pipelined executions.
Signed-off-by: Zbigniew Kempczyński <zbigniew.kempczynski@intel.com>
Reviewed-by: Karolina Stolarek <karolina.stolarek@intel.com>
|
|
For tests which are working on more than one region using name suffix
like "vram01-system" etc. is common thing. Instead handcrafting this
naming add xe_memregion_dynamic_subtest_name() function which is
similar to memregion_dynamic_subtest_name() for i915.
Signed-off-by: Zbigniew Kempczyński <zbigniew.kempczynski@intel.com>
Reviewed-by: Karolina Stolarek <karolina.stolarek@intel.com>
|
