From: Xin Wang <x.wang@intel.com>
To: igt-dev@lists.freedesktop.org
Cc: Xin Wang <x.wang@intel.com>, Daniel Charles <daniel.charles@intel.com>
Subject: [PATCH] tests/intel: only check multi-LRC support for parallel submission paths
Date: Wed, 8 Apr 2026 00:14:16 -0700 [thread overview]
Message-ID: <20260408071416.808087-1-x.wang@intel.com> (raw)
The previous commit "tests/intel: skip or adjust tests for non-multi-LRC
engine classes" unconditionally guarded balancer test functions with
xe_engine_class_supports_multi_lrc(), causing virtual (non-parallel)
subtests to be skipped on engine classes that have multiple placements
but do not support multi-LRC.
Only parallel submission requires multi-LRC support; virtual balancer
tests should still run whenever num_placements >= 2 regardless of
multi-LRC capability.
Fix this by making the multi-LRC check conditional on the PARALLEL flag
in test_exec(), test_cm(), test_balancer(), and utilization_multi().
Remove the check entirely from test_all_active() which is a pure virtual
test. In xe_exec_threads, move the filter into the caller threads() so
that VIRTUAL threads are still spawned while PARALLEL threads are skipped
for non-multi-LRC classes.
Fixes: 09d91fcf2961 ("tests/intel: skip or adjust tests for non-multi-LRC engine classes")
Cc: Daniel Charles <daniel.charles@intel.com>
Signed-off-by: Xin Wang <x.wang@intel.com>
---
tests/intel/xe_drm_fdinfo.c | 3 ++-
tests/intel/xe_exec_balancer.c | 8 +++++---
tests/intel/xe_exec_reset.c | 3 ++-
tests/intel/xe_exec_threads.c | 7 ++++++-
4 files changed, 15 insertions(+), 6 deletions(-)
diff --git a/tests/intel/xe_drm_fdinfo.c b/tests/intel/xe_drm_fdinfo.c
index 3c113ed5d..8449a4f17 100644
--- a/tests/intel/xe_drm_fdinfo.c
+++ b/tests/intel/xe_drm_fdinfo.c
@@ -673,7 +673,8 @@ utilization_multi(int fd, int gt, int class, unsigned int flags)
igt_assert(virtual ^ parallel);
num_placements = xe_gt_fill_engines_by_class(fd, gt, class, eci);
- if (num_placements < 2 || !xe_engine_class_supports_multi_lrc(fd, class))
+ if (num_placements < 2 ||
+ (parallel && !xe_engine_class_supports_multi_lrc(fd, class)))
return;
igt_debug("Target class: %s\n", engine_map[class]);
diff --git a/tests/intel/xe_exec_balancer.c b/tests/intel/xe_exec_balancer.c
index 3c6ec45f1..4aaef7666 100644
--- a/tests/intel/xe_exec_balancer.c
+++ b/tests/intel/xe_exec_balancer.c
@@ -57,7 +57,7 @@ static bool test_all_active(int fd, int gt, int class)
int i, num_placements;
num_placements = xe_gt_fill_engines_by_class(fd, gt, class, eci);
- if (num_placements < 2 || !xe_engine_class_supports_multi_lrc(fd, class))
+ if (num_placements < 2)
return false;
vm = xe_vm_create(fd, 0, 0);
@@ -187,7 +187,8 @@ test_exec(int fd, int gt, int class, int n_exec_queues, int n_execs,
igt_assert_lte(n_exec_queues, MAX_N_EXEC_QUEUES);
num_placements = xe_gt_fill_engines_by_class(fd, gt, class, eci);
- if (num_placements < 2 || !xe_engine_class_supports_multi_lrc(fd, class))
+ if (num_placements < 2 ||
+ ((flags & PARALLEL) && !xe_engine_class_supports_multi_lrc(fd, class)))
return false;
vm = xe_vm_create(fd, 0, 0);
@@ -402,7 +403,8 @@ test_cm(int fd, int gt, int class, int n_exec_queues, int n_execs,
igt_assert_lte(n_exec_queues, MAX_N_EXEC_QUEUES);
num_placements = xe_gt_fill_engines_by_class(fd, gt, class, eci);
- if (num_placements < 2 || !xe_engine_class_supports_multi_lrc(fd, class))
+ if (num_placements < 2 ||
+ ((flags & PARALLEL) && !xe_engine_class_supports_multi_lrc(fd, class)))
return false;
vm = xe_vm_create(fd, DRM_XE_VM_CREATE_FLAG_LR_MODE, 0);
diff --git a/tests/intel/xe_exec_reset.c b/tests/intel/xe_exec_reset.c
index 95191139d..66580ea44 100644
--- a/tests/intel/xe_exec_reset.c
+++ b/tests/intel/xe_exec_reset.c
@@ -184,7 +184,8 @@ test_balancer(int fd, int gt, int class, int n_exec_queues, int n_execs,
fd = drm_open_driver(DRIVER_XE);
num_placements = xe_gt_fill_engines_by_class(fd, gt, class, eci);
- if (num_placements < 2 || !xe_engine_class_supports_multi_lrc(fd, class))
+ if (num_placements < 2 ||
+ ((flags & PARALLEL) && !xe_engine_class_supports_multi_lrc(fd, class)))
return;
vm = xe_vm_create(fd, 0, 0);
diff --git a/tests/intel/xe_exec_threads.c b/tests/intel/xe_exec_threads.c
index 7b8100c5b..40a6c0c6d 100644
--- a/tests/intel/xe_exec_threads.c
+++ b/tests/intel/xe_exec_threads.c
@@ -85,7 +85,6 @@ test_balancer(int fd, int gt, uint32_t vm, uint64_t addr, uint64_t userptr,
num_placements = xe_gt_fill_engines_by_class(fd, gt, class, eci);
igt_assert_lt(1, num_placements);
- igt_assert(xe_engine_class_supports_multi_lrc(fd, class));
bo_size = sizeof(*data) * n_execs;
bo_size = xe_bb_size(fd, bo_size);
@@ -1211,6 +1210,12 @@ static void threads(int fd, int flags)
continue;
while (*data_flags >= 0) {
+ if ((*data_flags & PARALLEL) &&
+ !xe_engine_class_supports_multi_lrc(fd, class)) {
+ data_flags++;
+ continue;
+ }
+
threads_data[i].mutex = &mutex;
threads_data[i].cond = &cond;
if (flags & SHARED_VM)
--
2.43.0
next reply other threads:[~2026-04-08 7:14 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-04-08 7:14 Xin Wang [this message]
2026-04-08 17:27 ` tests/intel: only check multi-LRC support for parallel submission paths Daniel Charles
2026-04-09 13:41 ` Kamil Konieczny
2026-04-09 8:14 ` ✓ Xe.CI.BAT: success for " Patchwork
2026-04-09 8:31 ` ✓ i915.CI.BAT: " Patchwork
2026-04-09 9:25 ` ✓ Xe.CI.FULL: " Patchwork
2026-04-09 20:43 ` ✗ i915.CI.Full: failure " Patchwork
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20260408071416.808087-1-x.wang@intel.com \
--to=x.wang@intel.com \
--cc=daniel.charles@intel.com \
--cc=igt-dev@lists.freedesktop.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox