public inbox for igt-dev@lists.freedesktop.org
 help / color / mirror / Atom feed
* [PATCH i-g-t] tests/intel/xe_wedged: Skip subtests using wedged_mode=2 in SR-IOV cases
@ 2026-01-09 15:48 Lukasz Laguna
  2026-01-09 16:17 ` Rodrigo Vivi
                   ` (4 more replies)
  0 siblings, 5 replies; 7+ messages in thread
From: Lukasz Laguna @ 2026-01-09 15:48 UTC (permalink / raw)
  To: igt-dev
  Cc: marcin.bernatowicz, satyanarayana.k.v.p, jakub1.kolakowski,
	lukasz.laguna

In upon-any-hang-no-reset (2) wedged mode, engine resets need to be
disabled, which requires changing the GuC reset policy.

VFs are not permitted to modify this policy, so the driver now rejects
attempts to set wedged_mode=2 on a VF.

On the PF, modifying the reset policy is allowed, but the change is
global and affects all VFs. To prevent unintended cross-VF impact, the
driver permits wedged_mode=2 on the PF only when the driver is built
with the debug config. Otherwise, the operation fails with -EPERM.

Update the affected subtests to skip in such cases to avoid false
failures.

Link: https://lore.kernel.org/all/20260107174741.29163-4-lukasz.laguna@intel.com/
Link: https://lore.kernel.org/all/20260107174741.29163-5-lukasz.laguna@intel.com/
Signed-off-by: Lukasz Laguna <lukasz.laguna@intel.com>
---
 tests/intel/xe_wedged.c | 12 +++++++++++-
 1 file changed, 11 insertions(+), 1 deletion(-)

diff --git a/tests/intel/xe_wedged.c b/tests/intel/xe_wedged.c
index bcd78b49d7..e1ac629ca7 100644
--- a/tests/intel/xe_wedged.c
+++ b/tests/intel/xe_wedged.c
@@ -18,6 +18,7 @@
 #include "igt.h"
 #include "igt_device.h"
 #include "igt_kmod.h"
+#include "igt_sriov_device.h"
 #include "igt_syncobj.h"
 #include "igt_sysfs.h"
 
@@ -247,11 +248,16 @@ int igt_main()
 		int err;
 
 		igt_require(igt_debugfs_exists(fd, "wedged_mode", O_RDWR));
+		igt_require(!intel_is_vf_device(fd));
+
+		igt_debugfs_write(fd, "wedged_mode", "2");
+		igt_skip_on_f(igt_sriov_is_pf(fd) && errno == EPERM,
+			      "On PF, upon-any-hang-no-reset (2) wedged mode is restricted to debug builds only\n");
+
 		ignore_wedged_in_dmesg();
 
 		hang_sync.handle = syncobj_create(fd, 0);
 
-		igt_debugfs_write(fd, "wedged_mode", "2");
 		simple_hang(fd, &hang_sync);
 
 		/*
@@ -282,8 +288,12 @@ int igt_main()
 
 	igt_subtest_f("wedged-mode-toggle") {
 		igt_require(igt_debugfs_exists(fd, "wedged_mode", O_RDWR));
+		igt_require(!intel_is_vf_device(fd));
 
 		igt_debugfs_write(fd, "wedged_mode", "2");
+		igt_skip_on_f(igt_sriov_is_pf(fd) && errno == EPERM,
+			      "On PF, upon-any-hang-no-reset (2) wedged mode is restricted to debug builds only\n");
+
 		igt_assert_eq(simple_ioctl(fd), 0);
 		igt_debugfs_write(fd, "wedged_mode", "1");
 		ignore_wedged_in_dmesg();
-- 
2.40.0


^ permalink raw reply related	[flat|nested] 7+ messages in thread

end of thread, other threads:[~2026-01-09 23:28 UTC | newest]

Thread overview: 7+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-01-09 15:48 [PATCH i-g-t] tests/intel/xe_wedged: Skip subtests using wedged_mode=2 in SR-IOV cases Lukasz Laguna
2026-01-09 16:17 ` Rodrigo Vivi
2026-01-09 19:49   ` Rodrigo Vivi
2026-01-09 18:09 ` ✓ Xe.CI.BAT: success for " Patchwork
2026-01-09 18:28 ` ✓ i915.CI.BAT: " Patchwork
2026-01-09 21:49 ` ✗ i915.CI.Full: failure " Patchwork
2026-01-09 23:28 ` ✗ Xe.CI.Full: " Patchwork

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox