From: Matthew Auld <matthew.auld@intel.com>
To: "Matthew Brost" <matthew.brost@intel.com>,
"Thomas Hellström" <thomas.hellstrom@linux.intel.com>,
"Himal Prasad Ghimiray" <himal.prasad.ghimiray@intel.com>,
"Stuart Summers" <stuart.summers@intel.com>
Cc: intel-xe@lists.freedesktop.org, Matthew Auld <matthew.auld@intel.com>
Subject: [PATCH v2 3/6] drm/xe: Requeue the rebind worker instead of retrying in-line on -EAGAIN
Date: Fri, 20 Nov 2026 13:02:18 +0000 [thread overview]
Message-ID: <20261120-userptr-repin-v2-3-65e1f6e582ef@intel.com> (raw)
In-Reply-To: <20261120-userptr-repin-v2-0-65e1f6e582ef@intel.com>
From: Thomas Hellström <thomas.hellstrom@linux.intel.com>
When the preempt rebind worker needs to retry, it currently jumps back
and keeps holding vm->lock in write mode across the retry. Only VFs
with migration support drop the lock and requeue the worker.
Holding the vm lock across retries needlessly blocks other users of the
lock and other work on the ordered workqueue, and makes the retry unfit
for conditions that may last for a while, like repeatedly invalidated
userptrs.
Always drop the vm lock and requeue the worker on -EAGAIN.
The userptr invalidation injection state that forces one retry per
rebind sequence was a local variable that would be reset by a requeue.
Move it to the vm.
v2 (Matt):
- Split out the VF preempt fence leak fix into a separate patch.
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Himal Prasad Ghimiray <himal.prasad.ghimiray@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
---
drivers/gpu/drm/xe/xe_vm.c | 61 +++++++++++++++++++++++++---------------
drivers/gpu/drm/xe/xe_vm_types.h | 8 ++++++
2 files changed, 46 insertions(+), 23 deletions(-)
diff --git a/drivers/gpu/drm/xe/xe_vm.c b/drivers/gpu/drm/xe/xe_vm.c
index 82e716b12b49..ff8d355728fc 100644
--- a/drivers/gpu/drm/xe/xe_vm.c
+++ b/drivers/gpu/drm/xe/xe_vm.c
@@ -542,6 +542,31 @@ void xe_vm_resume_rebind_worker(struct xe_vm *vm)
queue_work(vm->xe->ordered_wq, &vm->preempt.rebind_work);
}
+#if IS_ENABLED(CONFIG_DRM_XE_USERPTR_INVAL_INJECT)
+/*
+ * Force one retry per rebind sequence. The sequence spans requeues of the
+ * worker, so the state must live in the vm.
+ */
+static bool xe_vm_rebind_worker_inject_retry(struct xe_vm *vm)
+{
+ return !vm->preempt.inval_inject_tries++;
+}
+
+static void xe_vm_rebind_worker_reset_inject(struct xe_vm *vm)
+{
+ vm->preempt.inval_inject_tries = 0;
+}
+#else
+static bool xe_vm_rebind_worker_inject_retry(struct xe_vm *vm)
+{
+ return false;
+}
+
+static void xe_vm_rebind_worker_reset_inject(struct xe_vm *vm)
+{
+}
+#endif
+
static void preempt_rebind_work_func(struct work_struct *w)
{
struct xe_vm *vm = container_of(w, struct xe_vm, preempt.rebind_work);
@@ -551,7 +576,6 @@ static void preempt_rebind_work_func(struct work_struct *w)
LIST_HEAD(preempt_fences);
int err = 0;
long wait;
- int __maybe_unused tries = 0;
xe_assert(vm->xe, xe_vm_in_preempt_fence_mode(vm));
trace_xe_vm_rebind_worker_enter(vm);
@@ -564,7 +588,6 @@ static void preempt_rebind_work_func(struct work_struct *w)
return;
}
-retry:
if (!try_wait_for_completion(&vm->xe->pm_block) && vm_suspend_rebind_worker(vm)) {
up_write(&vm->lock);
/* We don't actually block but don't make progress. */
@@ -614,20 +637,14 @@ static void preempt_rebind_work_func(struct work_struct *w)
goto out_unlock;
}
-#define retry_required(__tries, __vm) \
- (IS_ENABLED(CONFIG_DRM_XE_USERPTR_INVAL_INJECT) ? \
- (!(__tries)++ || __xe_vm_userptr_needs_repin(__vm)) : \
- __xe_vm_userptr_needs_repin(__vm))
-
xe_svm_notifier_lock(vm);
- if (retry_required(tries, vm)) {
+ if (xe_vm_rebind_worker_inject_retry(vm) ||
+ __xe_vm_userptr_needs_repin(vm)) {
xe_svm_notifier_unlock(vm);
err = -EAGAIN;
goto out_unlock;
}
-#undef retry_required
-
xe_vm_bump_lru(vm, &exec);
/* Point of no return. */
@@ -642,22 +659,20 @@ static void preempt_rebind_work_func(struct work_struct *w)
trace_xe_vm_rebind_worker_retry(vm);
/*
- * We can't block in workers on a VF which supports migration
- * given this can block the VF post-migration workers from
- * getting scheduled.
+ * Drop all locks and requeue rather than retrying in-line.
+ * This avoids blocking other work items, such as VF
+ * post-migration workers, and lets other users of the vm
+ * lock make progress between attempts.
*/
- if (IS_SRIOV_VF(vm->xe) &&
- xe_sriov_vf_migration_supported(vm->xe)) {
- up_write(&vm->lock);
- free_preempt_fences(&preempt_fences);
- xe_vm_queue_rebind_worker(vm);
- trace_xe_vm_rebind_worker_exit(vm);
- return;
- }
-
- goto retry;
+ up_write(&vm->lock);
+ free_preempt_fences(&preempt_fences);
+ xe_vm_queue_rebind_worker(vm);
+ trace_xe_vm_rebind_worker_exit(vm);
+ return;
}
+ xe_vm_rebind_worker_reset_inject(vm);
+
if (err) {
drm_warn(&vm->xe->drm, "VM worker error: %d\n", err);
xe_vm_kill(vm, true);
diff --git a/drivers/gpu/drm/xe/xe_vm_types.h b/drivers/gpu/drm/xe/xe_vm_types.h
index fb9305124679..756cf2dbee8a 100644
--- a/drivers/gpu/drm/xe/xe_vm_types.h
+++ b/drivers/gpu/drm/xe/xe_vm_types.h
@@ -350,6 +350,14 @@ struct xe_vm {
* kicked on resume.
*/
struct list_head pm_activate_link;
+#if IS_ENABLED(CONFIG_DRM_XE_USERPTR_INVAL_INJECT)
+ /**
+ * @inval_inject_tries: Number of rebind attempts in the
+ * current rebind sequence. Used to force a retry for testing.
+ * Protected by @lock.
+ */
+ unsigned int inval_inject_tries;
+#endif
} preempt;
/** @exec_queues: Manages list of exec queues attached to this VM, protected by lock. */
--
2.56.1
next prev parent reply other threads:[~2026-11-20 13:03 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-11-20 13:02 [PATCH v2 0/6] Non-fatal repin retries and NUMA-local rebind worker Matthew Auld
2026-11-20 13:02 ` [PATCH v2 1/6] drm/xe/vm: restore userptr invalidations on VMA op unwind Matthew Auld
2026-11-20 13:02 ` [PATCH v2 2/6] drm/xe/vf: Free preempt fences on rebind worker retry Matthew Auld
2026-11-20 13:02 ` Matthew Auld [this message]
2026-11-20 13:02 ` [PATCH v2 4/6] drm/xe: Add a userptr repin timeout sysfs attribute Matthew Auld
2026-11-20 13:02 ` [PATCH v2 5/6] drm/xe/userptr: Make the repin retry loop non-fatal Matthew Auld
2026-11-20 13:02 ` [PATCH v2 6/6] drm/xe: run compute rebind worker on device NUMA node Matthew Auld
2026-11-20 13:12 ` ✓ CI.KUnit: success for Non-fatal repin retries and NUMA-local rebind worker (rev2) 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=20261120-userptr-repin-v2-3-65e1f6e582ef@intel.com \
--to=matthew.auld@intel.com \
--cc=himal.prasad.ghimiray@intel.com \
--cc=intel-xe@lists.freedesktop.org \
--cc=matthew.brost@intel.com \
--cc=stuart.summers@intel.com \
--cc=thomas.hellstrom@linux.intel.com \
/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