From: Brian Welty <brian.welty@intel.com>
To: igt-dev@lists.freedesktop.org
Subject: [igt-dev] [PATCH i-g-t 1/2] lib/xe_ioctl: Add _xe_wait_ufence()
Date: Fri, 1 Dec 2023 13:00:12 -0800 [thread overview]
Message-ID: <20231201210013.16147-2-brian.welty@intel.com> (raw)
In-Reply-To: <20231201210013.16147-1-brian.welty@intel.com>
xe_wait_ufence is expected to fail if GPU hangs or is reset. We need
a version of calling XE_WAIT_USER_FENCE without the assertion that it
won't expire (timeout), for use with IGTs that intentionally cause
hangs or reset.
Signed-off-by: Brian Welty <brian.welty@intel.com>
---
lib/xe/xe_ioctl.c | 22 ++++++++++++++++------
lib/xe/xe_ioctl.h | 3 +++
2 files changed, 19 insertions(+), 6 deletions(-)
diff --git a/lib/xe/xe_ioctl.c b/lib/xe/xe_ioctl.c
index a5fe7dd792..b9828fc49f 100644
--- a/lib/xe/xe_ioctl.c
+++ b/lib/xe/xe_ioctl.c
@@ -446,9 +446,9 @@ void xe_exec_wait(int fd, uint32_t exec_queue, uint64_t addr)
syncobj_destroy(fd, sync.handle);
}
-int64_t xe_wait_ufence(int fd, uint64_t *addr, uint64_t value,
- struct drm_xe_engine_class_instance *eci,
- int64_t timeout)
+int _xe_wait_ufence(int fd, uint64_t *addr, uint64_t value,
+ struct drm_xe_engine_class_instance *eci,
+ int64_t *timeout)
{
struct drm_xe_wait_user_fence wait = {
.addr = to_user_pointer(addr),
@@ -456,14 +456,24 @@ int64_t xe_wait_ufence(int fd, uint64_t *addr, uint64_t value,
.flags = !eci ? DRM_XE_UFENCE_WAIT_FLAG_SOFT_OP : 0,
.value = value,
.mask = DRM_XE_UFENCE_WAIT_MASK_U64,
- .timeout = timeout,
+ .timeout = *timeout,
.num_engines = eci ? 1 :0,
.instances = eci ? to_user_pointer(eci) : 0,
};
- igt_assert_eq(igt_ioctl(fd, DRM_IOCTL_XE_WAIT_USER_FENCE, &wait), 0);
+ if (igt_ioctl(fd, DRM_IOCTL_XE_WAIT_USER_FENCE, &wait))
+ return -errno;
- return wait.timeout;
+ *timeout = wait.timeout;
+ return 0;
+}
+
+int64_t xe_wait_ufence(int fd, uint64_t *addr, uint64_t value,
+ struct drm_xe_engine_class_instance *eci,
+ int64_t timeout)
+{
+ igt_assert_eq(_xe_wait_ufence(fd, addr, value, eci, &timeout), 0);
+ return timeout;
}
/**
diff --git a/lib/xe/xe_ioctl.h b/lib/xe/xe_ioctl.h
index 6d91d3e8e6..7476a62c81 100644
--- a/lib/xe/xe_ioctl.h
+++ b/lib/xe/xe_ioctl.h
@@ -88,6 +88,9 @@ void xe_exec(int fd, struct drm_xe_exec *exec);
void xe_exec_sync(int fd, uint32_t exec_queue, uint64_t addr,
struct drm_xe_sync *sync, uint32_t num_syncs);
void xe_exec_wait(int fd, uint32_t exec_queue, uint64_t addr);
+int _xe_wait_ufence(int fd, uint64_t *addr, uint64_t value,
+ struct drm_xe_engine_class_instance *eci,
+ int64_t *timeout);
int64_t xe_wait_ufence(int fd, uint64_t *addr, uint64_t value,
struct drm_xe_engine_class_instance *eci,
int64_t timeout);
--
2.38.0
next prev parent reply other threads:[~2023-12-01 21:00 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-12-01 21:00 [igt-dev] [PATCH i-g-t 0/2] Fix xe_exec_reset@cm-gt-reset Brian Welty
2023-12-01 21:00 ` Brian Welty [this message]
2023-12-04 13:09 ` [igt-dev] [PATCH i-g-t 1/2] lib/xe_ioctl: Add _xe_wait_ufence() Kamil Konieczny
2023-12-01 21:00 ` [igt-dev] [PATCH i-g-t 2/2] tests/xe/exec_reset: Fix cm-gt-reset Brian Welty
2023-12-04 3:30 ` Mishra, Pallavi
2023-12-01 22:11 ` [igt-dev] ✓ Fi.CI.BAT: success for Fix xe_exec_reset@cm-gt-reset Patchwork
2023-12-01 22:28 ` [igt-dev] ✓ CI.xeBAT: " Patchwork
2023-12-03 4:03 ` [igt-dev] ✗ Fi.CI.IGT: 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=20231201210013.16147-2-brian.welty@intel.com \
--to=brian.welty@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