Matthew Brost [Fri, 9 Aug 2024 23:28:30 +0000 (16:28 -0700)]
drm/xe: Fix tile fini sequence
Only set tile->mmio.regs to NULL if not the root tile in tile_fini. The
root tile mmio regs is setup ealier in MMIO init thus it should be set
to NULL in mmio_fini.
The BO cleanup touches the GGTT and therefore requires the HW to be
available, so we need to use devm instead of drmm.
Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/1160 Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com> Cc: Lucas De Marchi <lucas.demarchi@intel.com> Cc: Matthew Auld <matthew.auld@intel.com> Reviewed-by: Matthew Auld <matthew.auld@intel.com> Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20240809231237.1503796-2-daniele.ceraolospurio@intel.com Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Matthew Auld [Mon, 12 Aug 2024 14:13:32 +0000 (15:13 +0100)]
drm/xe: fix engine_class bounds check again
This was fixed in commit b7dce525c4fc ("drm/xe/queue: fix engine_class
bounds check"), but then re-introduced in commit 6f20fc09936e ("drm/xe:
Move and export xe_hw_engine lookup.") which should only be simple code
movement of the existing function.
Fixes: 6f20fc09936e ("drm/xe: Move and export xe_hw_engine lookup.") Signed-off-by: Matthew Auld <matthew.auld@intel.com> Cc: Dominik Grzegorzek <dominik.grzegorzek@intel.com> Cc: Mika Kuoppala <mika.kuoppala@linux.intel.com> Cc: Matthew Brost <matthew.brost@intel.com> Reviewed-by: Jonathan Cavitt <jonathan.cavitt@intel.com> Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20240812141331.729843-2-matthew.auld@intel.com Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Tejas Upadhyay [Wed, 14 Aug 2024 09:56:12 +0000 (15:26 +0530)]
drm/xe: Move enable host l2 VRAM post MCR init
xe_gt_enable_host_l2_vram() is reading the XE2_GAMREQSTRM_CTRL register
that is currently missing the MCR annotation. However, just adding the
annotation doesn't work as this function is called before MCR handling
is initialized in xe_gt_mcr_init().
xe_gt_enable_host_l2_vram() is used to implement WA 16023588340 that
needs to be done as early as possible during initialization in order to
be effective since the MMIO writes impact it. In the failure scenario,
driver would simply not be able to bind successfully.
Moving xe_gt_enable_host_l2_vram() later, after MCR initialization is
done, only incurs a few additional HW accesses, particularly when
loading GuC for hwconfig. Binding/unbinding the driver 100 times in BMG
still works so it should be ok to start handling the WA a little bit
later. This is sufficient to allow adding the MCR annotation to
XE2_GAMREQSTRM_CTRL.
Lucas De Marchi [Tue, 13 Aug 2024 14:19:31 +0000 (07:19 -0700)]
drm/xe: Rename enable_display module param
The different approach used by xe regarding the initialization of
display HW has been proved a great addition for early driver bring up:
core xe can be tested without having all the bits sorted out on the
display side.
On the other hand, the approach exposed by i915-display is to *actively*
disable the display by programming it if needed, i.e. if it was left
enabled by firmware. It also has its use to make sure the HW is actually
disabled and not wasting power.
However having both the way it is in xe doesn't expose a good interface
wrt module params. From modinfo:
Rename enable_display to probe_display to try to convey the message that
the HW is being touched and improve the module param description. To
avoid confusion, the enable_display is renamed everywhere, not only in
the module param. New description for the parameters:
Matt Roper [Mon, 12 Aug 2024 18:10:43 +0000 (11:10 -0700)]
drm/xe: Name and document Wa_14019789679
Early in the development of Xe we identified an issue with SVG state
handling on DG2 and MTL (and later on Xe2 as well). In
commit 72ac304769dd ("drm/xe: Emit SVG state on RCS during driver load
on DG2 and MTL") and commit fb24b858a20d ("drm/xe/xe2: Update SVG state
handling") we implemented our own workaround to prevent SVG state from
leaking from context A to context B in cases where context B never
issues a specific state setting.
The hardware teams have now created official workaround Wa_14019789679
to cover this issue. The workaround description only requires emitting
3DSTATE_MESH_CONTROL, since they believe that's the only SVG instruction
that would potentially remain unset by a context B, but still cause
notable issues if unwanted values were inherited from context A.
However since we already have a more extensive implementation that emits
the entire SVG state and prevents _any_ SVG state from unintentionally
leaking, we'll stick with our existing implementation just to be safe.
xe_call_for_each_device() has been unused since commit 57ecead343e7
("drm/xe/tests: Convert xe_mocs live tests"). Remove it and the related
dev_to_xe_device_fn() and struct kunit_test_data.
Nirmoy Das [Fri, 9 Aug 2024 22:03:47 +0000 (00:03 +0200)]
drm/xe/migrate: Parameterize ccs and bo data clear in xe_migrate_clear()
Parameterize clearing ccs and bo data in xe_migrate_clear() which higher
layers can utilize. This patch will be used later on when doing bo data
clear for igfx as well.
v2: Replace multiple params with flags in xe_migrate_clear (Matt B)
v3: s/CLEAR_BO_DATA_FLAG_*/XE_MIGRATE_CLEAR_FLAG_* and move to
xe_migrate.h. other nits(Matt B)
Cc: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com> Cc: Matthew Auld <matthew.auld@intel.com> Cc: Matthew Brost <matthew.brost@intel.com> Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com> Signed-off-by: Akshata Jahagirdar <akshata.jahagirdar@intel.com> Reviewed-by: Matthew Auld <matthew.auld@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20240809220347.25330-1-nirmoy.das@intel.com Signed-off-by: Nirmoy Das <nirmoy.das@intel.com>
Nirmoy Das [Sat, 10 Aug 2024 19:15:19 +0000 (21:15 +0200)]
drm/xe: Add stats for tlb invalidation count
Add stats for tlb invalidation count which can be viewed with per GT
stat debugfs file.
Example output:
cat /sys/kernel/debug/dri/0/gt0/stats
tlb_inval_count: 22
v2: fix #include order(Tejas)
Cc: Lucas De Marchi <lucas.demarchi@intel.com> Cc: Matthew Brost <matthew.brost@intel.com> Cc: Michal Wajdeczko <michal.wajdeczko@intel.com> Cc: Sai Gowtham Ch <sai.gowtham.ch@intel.com> Reviewed-by: Tejas Upadhyay <tejas.upadhyay@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20240810191522.18616-2-nirmoy.das@intel.com Signed-off-by: Nirmoy Das <nirmoy.das@intel.com>
Nirmoy Das [Sat, 10 Aug 2024 19:15:18 +0000 (21:15 +0200)]
drm/xe/gt: Add APIs for printing stats over debugfs
Add skeleton APIs for recording and printing various stats over
debugfs. This currently only added counter types stats which is backed
by atomic_t and wrapped with CONFIG_DRM_XE_STATS so this can be disabled
on production system.
v4: Rebase and other minor fixes (Matt)
v3: s/CONFIG_DRM_XE_STATS/CONFIG_DEBUG_FS(Lucas)
v2: add missing docs
Add boundary checks for stats id and other improvements (Michal)
Fix build when CONFIG_DRM_XE_STATS is disabled(Matt)
Cc: Lucas De Marchi <lucas.demarchi@intel.com> Cc: Matthew Brost <matthew.brost@intel.com> Cc: Michal Wajdeczko <michal.wajdeczko@intel.com> Cc: Sai Gowtham Ch <sai.gowtham.ch@intel.com> Reviewed-by: Matthew Brost <matthew.brost@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20240810191522.18616-1-nirmoy.das@intel.com Signed-off-by: Nirmoy Das <nirmoy.das@intel.com>
Lucas De Marchi [Thu, 8 Aug 2024 17:11:21 +0000 (10:11 -0700)]
drm/xe: Allow to compile out debugfs
Use a dummy xe_debugfs_register() if debugfs is not enabled and move all
debugfs-related files under `ifeq ($(CONFIG_DEBUG_FS),y)` in the
Makefile. This is similar to what was done for display in
commit 439987f6f471 ("drm/xe: don't build debugfs files when
CONFIG_DEBUG_FS=n").
This removes the following warning while loading xe with
CONFIG_DEUBG_FS=n:
Matthew Brost [Wed, 24 Jul 2024 15:28:31 +0000 (08:28 -0700)]
drm/xe: Move VM dma-resv lock from xe_exec_queue_create to __xe_exec_queue_init
The critical section which requires the VM dma-resv is the call
xe_lrc_create in __xe_exec_queue_init. Move this lock to
__xe_exec_queue_init holding it just around xe_lrc_create. Not only is
good practice, this also fixes a locking double of the VM dma-resv in
the error paths of __xe_exec_queue_init as xe_lrc_put tries to acquire
this too resulting in a deadlock.
Michal Wajdeczko [Tue, 6 Aug 2024 18:05:16 +0000 (20:05 +0200)]
drm/xe/pf: Fix VF config validation on multi-GT platforms
When validating VF config on the media GT, we may wrongly report
that VF is already partially configured on it, as we consider GGTT
and LMEM provisioning done on the primary GT (since both GGTT and
LMEM are tile-level resources, not a GT-level).
This will cause skipping a VF auto-provisioning on the media-GT and
in result will block a VF from successfully initialize that GT.
Fix that by considering GGTT and LMEM configurations only when
checking if a VF provisioning is complete, and omit GGTT and LMEM
when reporting empty/partial provisioning.
This WA is applied while initializing the media GT, but it a primary
GT WA (because it modifies a register on the primary GT), so the XE_WA
macro is returning false even when the WA should be applied.
Fix this by using the primary GT in the macro.
Note that this WA only applies to PXP and we don't yet support that in
Xe, so there are no negative effects to this bug, which is why we didn't
see any errors in testing.
v2: use the primary GT in the macro instead of marking the WA as
platform-wide (Lucas, Matt).
Enable workarounds for HW bug where render engine reset fails. Given
that we're bumping the minimum required GuC version to 70.29, we're
guaranteed to always have support for this KLV in the GuC.
v2: Enable KLV correctly for either workaround (Lucas)
v4: Add check for minimum supported GuC firmware version. Enable w/a for
hw version 20.01 too. (Daniele)
v5 (Daniele): remove now unneeded fw type and version checks (JohnH)
Julia Filipchuk [Fri, 2 Aug 2024 22:21:27 +0000 (15:21 -0700)]
drm/xe/guc: Bump minimum required GuC version to v70.29.2
The VF API version for this release is 1.13.4.
Bumping the minimum required GuC version just before force-probe
removal allows us to set a baseline for what API features are expected
to be available. I.e., at this point there is no need for any version
checking in the code before using a feature. Of course, if/when the
API is extended in future GuC releases, those new features will need
API version checks in the code.
Bump the recommended GuC versions to match.
Also add numerical comparison helpers to simplify the version number
checks.
v2: Reword commit message and make comparison helpers GuC specific -
review feedback from Daniele, done by JohnH
Matthew Brost [Thu, 1 Aug 2024 15:41:18 +0000 (08:41 -0700)]
drm/xe: Faster devcoredump
The current algorithm to read out devcoredump is O(N*N) where N is the
size of coredump due to usage of the drm_coredump_printer in
xe_devcoredump_read. Switch to a O(N) algorithm which prints the
devcoredump into a readable format in snapshot work and update
xe_devcoredump_read to memcpy from the readable format directly.
v2:
- Fix double free on devcoredump removal (Testing)
- Set read_data_size after snap work flush
- Adjust remaining in iterator upon realloc (Testing)
- Set read_data upon realloc (Testing)
v3:
- Kernel doc
v4:
- Two pass algorithm to determine size (Maarten)
v5:
- Use scope for reading variables (Johnathan)
Reported-by: Paulo Zanoni <paulo.r.zanoni@intel.com> Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2408 Cc: Rodrigo Vivi <rodrigo.vivi@intel.com> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Signed-off-by: Matthew Brost <matthew.brost@intel.com> Reviewed-by: Jonathan Cavitt <jonathan.cavitt@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20240801154118.2547543-4-matthew.brost@intel.com
Matthew Brost [Thu, 1 Aug 2024 15:41:17 +0000 (08:41 -0700)]
drm/printer: Allow NULL data in devcoredump printer
We want to determine the size of the devcoredump before writing it out.
To that end, we will run the devcoredump printer with NULL data to get
the size, alloc data based on the generated offset, then run the
devcorecump again with a valid data pointer to print. This necessitates
not writing data to the data pointer on the initial pass, when it is
NULL.
Matthew Brost [Thu, 1 Aug 2024 15:41:16 +0000 (08:41 -0700)]
drm/xe: Take ref to VM in delayed snapshot
Kernel BO's don't take a ref to the VM, we need the VM for the
delayed snapshot, so take a ref to the VM in delayed snapshot.
v2:
- Check for lrc_bo before taking a VM ref (CI)
- Check lrc_bo->vm before taking / dropping a VM ref (CI)
- Drop VM in xe_lrc_snapshot_free
v5:
- Fix commit message wording (Johnathan)
Karthik Poosa [Thu, 1 Aug 2024 11:24:24 +0000 (16:54 +0530)]
drm/xe/hwmon: Fix PL1 disable flow in xe_hwmon_power_max_write
In xe_hwmon_power_max_write, for PL1 disable supported case, instead of
returning after PL1 disable, PL1 enable path was also being run.
Fixed it by returning after disable.
v2: Correct typo and grammar in commit message. (Jonathan)
Enable feature to allow memory reads to take a priority memory path.
This will reduce latency on the read path, but may introduce read after
write (RAW) hazards as read and writes will no longer be ordered.
To avoid RAW hazards, SW can use the MI_MEM_FENCE command or any other
MI command that generates non posted memory writes. This will ensure
data is coherent in memory prior to execution of commands which read
data from memory. RCS,BCS and CCS support this feature.
No pattern identified in KMD that could lead to a hazard.
v2: Modify commit message, enable priority mem read feature for media,
modify version range, modify bspec detail (Matt Roper)
v3: Rebase, fix cramped line-wrapping (jcavitt)
v4: Rebase
v5: Media does not support Priority Mem Read. Modify commit
to reflect the same.
Matthew Brost [Sat, 27 Jul 2024 01:22:16 +0000 (18:22 -0700)]
drm/xe: Use dma_fence_chain_free in chain fence unused as a sync
A chain fence is uninitialized if not installed in a drm sync obj. Thus
if xe_sync_entry_cleanup is called and sync->chain_fence is non-NULL the
proper cleanup is dma_fence_chain_free rather than a dma-fence put.
Reported-by: Paulo Zanoni <paulo.r.zanoni@intel.com> Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2411 Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/2261 Fixes: dd08ebf6c352 ("drm/xe: Introduce a new DRM driver for Intel GPUs") Signed-off-by: Matthew Brost <matthew.brost@intel.com> Reviewed-by: Matthew Auld <matthew.auld@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20240727012216.2118276-1-matthew.brost@intel.com
In function ‘decode_oa_format.isra.26’,
inlined from ‘xe_oa_set_prop_oa_format’ at drivers/gpu/drm/xe/xe_oa.c:1664:6:
././include/linux/compiler_types.h:510:38: error: call to ‘__compiletime_assert_1336’ declared with attribute error: FIELD_GET: mask is not constant
[...]
./include/linux/bitfield.h:155:3: note: in expansion of macro ‘__BF_FIELD_CHECK’
__BF_FIELD_CHECK(_mask, _reg, 0U, "FIELD_GET: "); \
^
drivers/gpu/drm/xe/xe_oa.c:1573:18: note: in expansion of macro ‘FIELD_GET’
u32 bc_report = FIELD_GET(DRM_XE_OA_FORMAT_MASK_BC_REPORT, fmt);
^
Lucas De Marchi [Sat, 27 Jul 2024 01:59:07 +0000 (18:59 -0700)]
drm/xe: Migrate OOB WAs to OR rules
Now that rtp has OR rules, it's not needed to extend it to process OOB
WAs. Previously if an entry had no name, it was considered as "a set of
rules OR'ed with the last named entry".
Instead of generating new entries, add OR rules. The syntax for
xe_wa_oob.rules remains the same, with xe_gen_wa_oob generating the
slightly different table. Object sizes delta are negligible, but having
just one logic makes it easier to maintain:
add/remove: 0/0 grow/shrink: 1/2 up/down: 160/-269 (-109)
Function old new delta
__compound_literal 6104 6264 +160
xe_wa_dump 1839 1810 -29
oob_was 816 576 -240
Total: Before=17257, After=17148, chg -0.63%
Lucas De Marchi [Sat, 27 Jul 2024 01:59:06 +0000 (18:59 -0700)]
drm/xe/rtp: Expand max rules/actions per entry again
Like commit 512660cd1f1a ("drm/xe/rtp: Expand max rules/actions per
entry") did, expand the maximum number of actions/rules. That commit was
too conservative, just incrementing 2. Other than the ugliness of these
macros and additional preprocessor steps when they are used, there are
no downsides on increasing the maximum: the tables in which they are
used use a sentinel to mark the last element.
With rtp processing now supporting OR rules, it's possible to migrate
the extension made for OOB WAs that "entries with name are OR'ed in
previous entry". For that the maximum number of rules needs to be
increased.
Just double it. Hopefully 12 is sufficient for longer than 6 was.
Lucas De Marchi [Sat, 27 Jul 2024 01:59:05 +0000 (18:59 -0700)]
drm/xe/rtp: Simplify marking active workarounds
Stop doing the calculation both in rtp_mark_active() and in its caller.
The caller easily knows the number of entries to mark, so just pass it
forward. That also simplifies rtp_mark_active() since now it doesn't
have a special case when handling 1 entry.
Lucas De Marchi [Sat, 27 Jul 2024 01:59:04 +0000 (18:59 -0700)]
drm/xe/kunit: Test rtp with no actions
The OOB WAs use xe_rtp_process(), without passing an sr to save result
of the actions since there are none. They are also executed in a gt-only
context, making it harder to share the implementation. Thus, introduce a
new set of tests to check these RTP entries. The only check that can be
done is if the entry was marked as active.
Before commit fd6797ec50c5 ("drm/xe/rtp: Fix off-by-one when processing
rules") several of these tests were failing: the processing of OR'ed
entries would make the subsequent entry to be inadvertently enabled.
Lucas De Marchi [Sat, 27 Jul 2024 01:59:01 +0000 (18:59 -0700)]
drm/xe/kunit: Rename count to count_sr_entries
The RTP tests check both the result of processing the RTP entries and
the outcome saved as SR entries. Rename "count" to be explicit about
what's being counted.
Matt Roper [Fri, 26 Jul 2024 17:17:58 +0000 (10:17 -0700)]
drm/xe/migrate: Future-proof compressed PAT check
Although all current Xe2 platforms support FlatCCS, we probably
shouldn't assume that will be universally true forever. In the past
we've had platforms like PVC that didn't support compression, and the
same could show up again at some point in the future. Future-proof the
migration code by adding an explicit check for FlatCCS support to the
condition that decides whether to use a compressed PAT index for
migration.
While we're at it, we can drop the IS_DGFX check since it's redundant
with the src_is_vram check (only dGPUs have VRAM).
Lucas De Marchi [Fri, 26 Jul 2024 06:43:35 +0000 (23:43 -0700)]
drm/xe/rtp: Fix off-by-one when processing rules
Gustavo noticed an odd "+ 2" in rtp_mark_active() while processing
rtp rules and pointed that it should be "+ 1". In fact, while processing
entries without actions (OOB workarounds), if the WA is activated and
has OR rules, it will also inadvertently activate the very next
workaround.
Test in a LNL B0 platform by moving 18024947630 on top of 16020292621,
makes the latter become active:
drm/xe/mmio: Use single logic for waiting functions
The implementations for xe_mmio_wait32() and xe_mmio_wait32_not() are
almost identical. Let us avoid duplication of logic by having them
calling a common __xe_mmio_wait32() function.
drm/xe: Remove stale declaration of xe_mmio_probe_vram()
The declaration of xe_mmio_probe_vram() became useless since
commit 638d1c79cbf1 ("drm/xe: Promote VRAM initialization function to
own file"). Remove it.
All new binaries are GSC-enabled (and even if they weren't the driver
can detect the type of HuC binary), so the new lnl HuC filename doesn't
use the _gsc postfix to avoid confusion with the GSC binary.
Matthew Brost [Tue, 23 Jul 2024 19:07:14 +0000 (12:07 -0700)]
drm/xe: Remove fence check from send_tlb_invalidation
'fence' argument in send_tlb_invalidation cannot be NULL, remove
non-NULL check from send_tlb_invalidation.
Reported-by: kernel test robot <lkp@intel.com> Reported-by: Dan Carpenter <dan.carpenter@linaro.org> Closes: https://lore.kernel.org/r/202407231049.esig0Fkb-lkp@intel.com/ Fixes: 61ac035361ae ("drm/xe: Drop xe_gt_tlb_invalidation_wait") Signed-off-by: Matthew Brost <matthew.brost@intel.com> Reviewed-by: Nirmoy Das <nirmoy.das@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20240723190714.1744653-1-matthew.brost@intel.com Signed-off-by: Nirmoy Das <nirmoy.das@intel.com>
Matthew Brost [Tue, 23 Jul 2024 15:10:45 +0000 (08:10 -0700)]
drm/xe: Store process name and pid in xe file
An xe file can outlive the associated process as the GPU cleanup is just
triggered upon file close (process kill) and completes sometime later.
If the file close triggers error conditions (GPU hangs) the process
cannot be safely referenced to retrieve the name and pid for debug
information. Store the process name and pid directly in the xe file to
be safe.
v2:
- Access file->pid via rcu_access_pointer (Matthew Auld)
Fixes: b10d0c5e9df7 ("drm/xe: Add process name to devcoredump") Fixes: f6ca930d974e ("drm/xe: Add process name and PID to job timedout message") Signed-off-by: Matthew Brost <matthew.brost@intel.com> Acked-by: Rodrigo Vivi <rodrigo.vivi@intel.com> Reviewed-by: Matthew Auld <matthew.auld@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20240723151045.1725417-1-matthew.brost@intel.com
Matthew Brost [Tue, 23 Jul 2024 01:17:02 +0000 (18:17 -0700)]
drm/xe: Return -ENOBUFS if a kmalloc fails which is tied to an array of binds
The size of an array of binds is directly tied to several kmalloc in the
KMD, thus making these kmalloc more likely to fail. Return -ENOBUFS in
the case of these failures.
The expected UMD behavior upon returning -ENOBUFS is to split an array
of binds into a series of single binds.
Matthew Brost [Tue, 23 Jul 2024 01:02:30 +0000 (18:02 -0700)]
drm/xe: Fix xe_pt_abort_unbind
When restoring the children PT entries on a bind failure the incorrect
loop index has used resulting in PT entries being leaked. This is shown
by running xe_vm.bind-array-conflict-error-inject on a VRAM device going
into a suspend state after the test completes.
Lucas De Marchi [Fri, 19 Jul 2024 19:15:34 +0000 (12:15 -0700)]
drm/xe: Fix warning on unreachable statement
eu_type_to_str() relies on -Wswitch to warn (and -Werror) to make sure
it handles all enum values. However it's perfectly legal to pass an int
to that function so in the end that function may happen to return
nothing. There's too much implicit knowledge about the initialization
of eu_type for a compiler to notice eu_type is never assigned to
anything other than those values.
Trying to reproduce this issue, none of gcc-9, gcc-10 and gcc-13
triggered for me, but this was reported in a different system with
gcc-10:
drivers/gpu/drm/xe/xe.o: warning: objtool: xe_gt_topology_dump() falls through to next function xe_gt_topology_init()
Also it was reported these warnings when building with clang:
drivers/gpu/drm/xe/xe.o: warning: objtool: xe_gt_topology_dump+0x77: sibling call from callable instruction with modified stack frame
drivers/gpu/drm/xe/xe.o: warning: objtool: xe_gt_topology_dump() falls through to next function xe_dss_mask_group_ffs()
drivers/gpu/drm/xe/xe.o: warning: objtool: xe_gt_topology_dump+0x77: can't find jump dest instruction at .text.xe_gt_topology_dump+0xc0
Since that value is not really possible in real world, just take the
simple approach and return NULL.
Fixes: 7108b4a589cd ("drm/xe/uapi: Expose SIMD16 EU mask in topology query") Reviewed-by: Nathan Chancellor <nathan@kernel.org> Tested-by: Nathan Chancellor <nathan@kernel.org> Reviewed-by: Michal Wajdeczko <michal.wajdeczko@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20240719191534.3845469-1-lucas.demarchi@intel.com Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Michal Wajdeczko [Mon, 15 Jul 2024 18:05:38 +0000 (20:05 +0200)]
drm/xe: Add NEEDS_2M BO flag
In addition of NEEDS_64K BO flag, add similar one to force 2 MiB
alignment of the buffer objects. Explicitly use this flag during
VF LMEM provisioning as LMTT uses 2 MiB pages and one day we may
drop requirement of allocating pinned objects as contiguous.
Michal Wajdeczko [Sat, 20 Jul 2024 14:25:23 +0000 (16:25 +0200)]
drm/xe/tests: Add helpers for use in live tests
Instead of iterating over available Xe devices within a testcase,
without being able to distinguish potential failures from different
devices on system with many Xe devices, introduce helpers that will
allow to treat each Xe device as a parameter for the testcase like:
Michal Wajdeczko [Sat, 20 Jul 2024 14:25:22 +0000 (16:25 +0200)]
drm/xe: Introduce const cast helper
Typically we want to preserve pointer constness when converting
from one xe pointer to another, but in some rare cases, like kunit
parameter conversions, we might want to discard this constness.
Add a helper that we will use to clearly indicate our intention.
Matthew Brost [Fri, 19 Jul 2024 17:29:05 +0000 (10:29 -0700)]
drm/xe: Build PM into GuC CT layer
Take PM ref when any G2H are outstanding, drop when none are
outstanding.
To safely ensure we have PM ref when in the GuC CT layer, a PM ref needs
to be held when scheduler messages are pending too.
v2:
- Add outer PM protections to xe_file_close (CI)
v3:
- Only take PM ref 0->1 and drop on 1->0 (Matthew Auld)
v4:
- Add assert to G2H increment function
v5:
- Rebase
v6:
- Declare xe as local variable in xe_file_close (CI)
Fixes: dd08ebf6c352 ("drm/xe: Introduce a new DRM driver for Intel GPUs") Cc: Matthew Auld <matthew.auld@intel.com> Cc: Rodrigo Vivi <rodrigo.vivi@intel.com> Cc: Nirmoy Das <nirmoy.das@intel.com> Signed-off-by: Matthew Brost <matthew.brost@intel.com> Reviewed-by: Matthew Auld <matthew.auld@intel.com> Reviewed-by: Nirmoy Das <nirmoy.das@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20240719172905.1527927-5-matthew.brost@intel.com
Matthew Brost [Fri, 19 Jul 2024 17:29:03 +0000 (10:29 -0700)]
drm/xe: Drop xe_gt_tlb_invalidation_wait
Having two methods to wait on GT TLB invalidations is not ideal. Remove
xe_gt_tlb_invalidation_wait and only use GT TLB invalidation fences.
In addition to two methods being less than ideal, once GT TLB
invalidations are coalesced the seqno cannot be assigned during
xe_gt_tlb_invalidation_ggtt/range. Thus xe_gt_tlb_invalidation_wait
would not have a seqno to wait one. A fence however can be armed and
later signaled.
v3:
- Add explaination about coalescing to commit message
v4:
- Don't put dma fence if defined on stack (CI)
v5:
- Initialize ret to zero (CI)
v6:
- Use invalidation_fence_signal helper in tlb timeout (Matthew Auld)
Michal Wajdeczko [Thu, 18 Jul 2024 20:31:55 +0000 (22:31 +0200)]
drm/xe/vf: Fix register value lookup
We should use the number of actual entries stored in the runtime
register buffer, not the maximum number of entries that this buffer
can hold, otherwise bsearch() may fail and we may miss the data and
wrongly report unexpected access to some registers.
Fixes: 4edadc41a3a4 ("drm/xe/vf: Use register values obtained from the PF") Signed-off-by: Michal Wajdeczko <michal.wajdeczko@intel.com> Cc: Piotr Piórkowski <piotr.piorkowski@intel.com> Cc: Matt Roper <matthew.d.roper@intel.com> Reviewed-by: Matt Roper <matthew.d.roper@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20240718203155.486-1-michal.wajdeczko@intel.com
drm/xe: Fix use after free when client stats are captured
xe_file_close triggers an asynchronous queue cleanup and then frees up
the xef object. Since queue cleanup flushes all pending jobs and the KMD
stores client usage stats into the xef object after jobs are flushed, we
see a use-after-free for the xef object. Resolve this by taking a
reference to xef from xe_exec_queue.
While at it, revert an earlier change that contained a partial work
around for this issue.
v2:
- Take a ref to xef even for the VM bind queue (Matt)
- Squash patches relevant to that fix and work around (Lucas)
v3: Fix typo (Lucas)
Fixes: ce62827bc294 ("drm/xe: Do not access xe file when updating exec queue run_ticks") Fixes: 6109f24f87d7 ("drm/xe: Add helper to accumulate exec queue runtime") Closes: https://gitlab.freedesktop.org/drm/xe/kernel/issues/1908 Signed-off-by: Umesh Nerlige Ramappa <umesh.nerlige.ramappa@intel.com> Reviewed-by: Matthew Brost <matthew.brost@intel.com> Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20240718210548.3580382-5-umesh.nerlige.ramappa@intel.com Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Lucas De Marchi [Wed, 10 Jul 2024 22:02:27 +0000 (15:02 -0700)]
drm/xe/uapi: Expose SIMD16 EU mask in topology query
PVC, Xe2 and later platforms have 16-wide EUs. We were implicitly
reporting for PVC the number of 16-wide EUs without giving userspace any
hint that they were different than for other platforms. Xe2 and later
also have 16-wide, but in those cases the reported number would
correspond to the 8-wide count.
To avoid confusion and make sure the right number is used by userspace
depending on the platform, add a new item to the topology query and drop
the one that is not available. The new mask reported for both PVC and
Xe2 should now match the numbers reported via hwconfig.
v2: Use a different topo item with EU type in its name to report the
new mask instead of adding the type itself as the item (Matt Roper)
Reviewed-by: Matt Roper <matthew.d.roper@intel.com> Acked-by: José Roberto de Souza <jose.souza@intel.com> Acked-by: Mateusz Jablonski <mateusz.jablonski@intel.com> Acked-by: Wenbin Lu <wenbin.lu@intel.com> Acked-by: Effie Yu <effie.yu@intel.com> Acked-by: Rodrigo Vivi <rodrigo.vivi@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20240710220446.2169797-1-lucas.demarchi@intel.com Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Uma Shankar [Wed, 17 Jul 2024 08:22:52 +0000 (13:52 +0530)]
drm/xe/fbdev: Limit the usage of stolen for LNL+
As per recommendation in the workarounds:
WA_22019338487
There is an issue with accessing Stolen memory pages due a
hardware limitation. Limit the usage of stolen memory for
fbdev for LNL+. Don't use BIOS FB from stolen on LNL+ and
assign the same from system memory.
v2: Corrected the WA Number, limited WA to LNL and
Adopted XE_WA framework as suggested by Lucas and Matt.
v3: Introduced the waxxx_display to implement display side
of WA changes on Lunarlake. Used xe_root_mmio_gt and
avoid the for loop (Suggested by Lucas)
In xe2+ dgfx, we don't need to handle the copying of ccs
metadata during migration. This test validates the ccs data post
clear and copy during evict/restore operation. Thus, we can skip
this test on xe2+ dgfx.
drm/xe/migrate: Add kunit to test migration functionality for BMG
This part of kunit verifies that
- main data is decompressed and ccs data is clear post bo eviction.
- main data is raw copied and ccs data is clear post bo restore.
drm/xe/xe_migrate: Handle migration logic for xe2+ dgfx
During eviction (vram->sysmem), we use compressed -> uncompressed mapping.
During restore (sysmem->vram), we need to use mapping from
uncompressed -> uncompressed.
Handle logic for selecting the compressed identity map for eviction,
and selecting uncompressed map for restore operations.
v2: Move check of xe_migrate_ccs_emit() before calling
xe_migrate_ccs_copy(). (Nirmoy)
drm/xe/xe2: Introduce identity map for compressed pat for vram
Xe2+ has unified compression (exactly one compression mode/format),
where compression is now controlled via PAT at PTE level.
This simplifies KMD operations, as it can now decompress freely
without concern for the buffer's original compression format—unlike DG2,
which had multiple compression formats and thus required copying the
raw CCS state during VRAM eviction. In addition mixed VRAM and system
memory buffers were not supported with compression enabled.
On Xe2 dGPU compression is still only supported with VRAM, however we
can now support compression with VRAM and system memory buffers,
with GPU access being seamless underneath. So long as when doing
VRAM -> system memory the KMD uses compressed -> uncompressed,
to decompress it. This also allows CPU access to such buffers,
assuming that userspace first decompress the corresponding
pages being accessed.
If the pages are already in system memory then KMD would have already
decompressed them. When restoring such buffers with sysmem -> VRAM
the KMD can't easily know which pages were originally compressed,
so we always use uncompressed -> uncompressed here.
With this it also means we can drop all the raw CCS handling on such
platforms (including needing to allocate extra CCS storage).
In order to support this we now need to have two different identity
mappings for compressed and uncompressed VRAM.
In this patch, we set up the additional identity map for the VRAM with
compressed pat_index. We then select the appropriate mapping during
migration/clear. During eviction (vram->sysmem), we use the mapping
from compressed -> uncompressed. During restore (sysmem->vram), we need
the mapping from uncompressed -> uncompressed.
Therefore, we need to have two different mappings for compressed and
uncompressed vram. We set up an additional identity map for the vram
with compressed pat_index.
We then select the appropriate mapping during migration/clear.
v2: Formatting nits, Updated code to match recent changes in
xe_migrate_prepare_vm(). (Matt)
v3: Move identity map loop to a helper function. (Matt Brost)
v4: Split helper function in different patch, and
add asserts and nits. (Matt Brost)
v5: Convert the 2 bool arguments of pte_update_size to flags
argument (Matt Brost)
drm/xe/migrate: Add kunit to test clear functionality
This test verifies if the main and ccs data are cleared during bo creation.
The motivation to use Kunit instead of IGT is that, although we can verify
whether the data is zero following bo creation,
we cannot confirm whether the zero value after bo creation is the result of
our clear function or simply because the initial data present was zero.
v2: Updated the mutex_lock and unlock logic,
Changed out_unlock to out_put. (Matt)
projects / users / jedix / linux-maple.git / log