FAILED: patch "[PATCH] drm/amdkfd: stop restore retries on device unplug" failed to apply to 6.1-stable tree

FAILED: patch "[PATCH] drm/amdkfd: stop restore retries on device unplug" failed to apply to 6.1-stable tree

[Sasha Levin]

The patch below does not apply to the 6.1-stable tree.
If someone wants it applied there, or to any other stable or longterm
tree, then please email the backport, including the original git commit
id to <stable@vger.kernel.org>.

To reproduce the conflict and resubmit, you may use the following commands:

git fetch https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/ linux-6.1.y
git checkout FETCH_HEAD
git cherry-pick -x 3254ed0b15f2fe656c9cb36ccb59b92422440bf8
# <resolve conflicts, build, test, etc.>
git commit -s
git send-email --to '<stable@vger.kernel.org>' --in-reply-to '20261107055554-failed-6.1-3254ed0b15f2@kernel.org' --subject-prefix 'PATCH 6.1.y' 'HEAD^..'

Thanks,
Sasha

------------------ original commit in Linus's tree ------------------

From 3254ed0b15f2fe656c9cb36ccb59b92422440bf8 Mon Sep 17 00:00:00 2001
From: Olivier Kaloudoff <olivier.kaloudoff@gmail.com>
Date: Tue, 13 Oct 2026 20:53:12 +0200
Subject: [PATCH] drm/amdkfd: stop restore retries on device unplug

amdgpu_amdkfd_restore_userptr_worker() and svm_range_restore_work()
reschedule themselves when validation fails with evicted objects
pending. After a surprise removal the restore paths fail with -ENODEV
permanently, so rescheduling just spins the worker uselessly until
cancel_delayed_work_sync() in fini finally stops it - which can take
a while, since process teardown itself is delayed by the dead device.

Treat -ENODEV as terminal in both workers.

Cc: Xiaogang Chen <xiaogang.chen@amd.com>
Signed-off-by: Olivier Kaloudoff <olivier.kaloudoff@gmail.com>
Signed-off-by: Felix Kuehling <felix.kuehling@amd.com>
Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit 266fab83139600e58db7e900e930c722e619b78c)
Cc: stable@vger.kernel.org
---
 .../gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c   | 18 +++++++++++++-----
 drivers/gpu/drm/amd/amdkfd/kfd_svm.c           |  8 +++++---
 2 files changed, 18 insertions(+), 8 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
index ff77e8ae9fe6f..3a4b3424b5b29 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
@@ -2854,6 +2854,7 @@ static void amdgpu_amdkfd_restore_userptr_worker(struct work_struct *work)
 	struct task_struct *usertask;
 	struct mm_struct *mm;
 	uint32_t evicted_bos;
+	int ret;
 
 	mutex_lock(&process_info->notifier_lock);
 	evicted_bos = process_info->evicted_bos;
@@ -2873,14 +2874,16 @@ static void amdgpu_amdkfd_restore_userptr_worker(struct work_struct *work)
 
 	mutex_lock(&process_info->lock);
 
-	if (update_invalid_user_pages(process_info, mm))
+	ret = update_invalid_user_pages(process_info, mm);
+	if (ret)
 		goto unlock_out;
 	/* userptr_inval_list can be empty if all evicted userptr BOs
 	 * have been freed. In that case there is nothing to validate
 	 * and we can just restart the queues.
 	 */
 	if (!list_empty(&process_info->userptr_inval_list)) {
-		if (validate_invalid_user_pages(process_info))
+		ret = validate_invalid_user_pages(process_info);
+		if (ret)
 			goto unlock_out;
 	}
 	/* Final check for concurrent evicton and atomic update. If
@@ -2889,11 +2892,14 @@ static void amdgpu_amdkfd_restore_userptr_worker(struct work_struct *work)
 	 * reference counting inside KFD will handle this case.
 	 */
 	mutex_lock(&process_info->notifier_lock);
-	if (process_info->evicted_bos != evicted_bos)
+	if (process_info->evicted_bos != evicted_bos) {
+		ret = -EAGAIN;
 		goto unlock_notifier_out;
+	}
 
 	if (confirm_valid_user_pages_locked(process_info)) {
 		WARN(1, "User pages unexpectedly invalid");
+		ret = -EINVAL;
 		goto unlock_notifier_out;
 	}
 
@@ -2911,8 +2917,10 @@ static void amdgpu_amdkfd_restore_userptr_worker(struct work_struct *work)
 unlock_out:
 	mutex_unlock(&process_info->lock);
 
-	/* If validation failed, reschedule another attempt */
-	if (evicted_bos) {
+	/* If validation failed, reschedule another attempt, unless the
+	 * device is gone for good
+	 */
+	if (evicted_bos && ret != -ENODEV) {
 		queue_delayed_work(system_freezable_wq,
 			&process_info->restore_userptr_work,
 			msecs_to_jiffies(AMDGPU_USERPTR_RESTORE_DELAY_MS));
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
index 1f5d73f0ebce3..a4ce54b06975b 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
@@ -1963,7 +1963,7 @@ static void svm_range_restore_work(struct work_struct *work)
 	struct mm_struct *mm;
 	int evicted_ranges;
 	int invalid;
-	int r;
+	int r = 0;
 
 	svms = container_of(dwork, struct svm_range_list, restore_work);
 	evicted_ranges = atomic_read(&svms->evicted_ranges);
@@ -2037,8 +2037,10 @@ static void svm_range_restore_work(struct work_struct *work)
 	mmap_write_unlock(mm);
 	mutex_unlock(&process_info->lock);
 
-	/* If validation failed, reschedule another attempt */
-	if (evicted_ranges) {
+	/* If validation failed, reschedule another attempt, unless the
+	 * device is gone for good
+	 */
+	if (evicted_ranges && r != -ENODEV) {
 		pr_debug("reschedule to restore svm range\n");
 		queue_delayed_work(system_freezable_wq, &svms->restore_work,
 			msecs_to_jiffies(AMDGPU_SVM_RANGE_RESTORE_DELAY_MS));
-- 
2.53.0

[PATCH 6.1.y] drm/amdkfd: stop restore retries on device unplug

[Sasha Levin]

From: Olivier Kaloudoff <olivier.kaloudoff@gmail.com>

[ Upstream commit 3254ed0b15f2fe656c9cb36ccb59b92422440bf8 ]

amdgpu_amdkfd_restore_userptr_worker() and svm_range_restore_work()
reschedule themselves when validation fails with evicted objects
pending. After a surprise removal the restore paths fail with -ENODEV
permanently, so rescheduling just spins the worker uselessly until
cancel_delayed_work_sync() in fini finally stops it - which can take
a while, since process teardown itself is delayed by the dead device.

Treat -ENODEV as terminal in both workers.

Cc: Xiaogang Chen <xiaogang.chen@amd.com>
Signed-off-by: Olivier Kaloudoff <olivier.kaloudoff@gmail.com>
Signed-off-by: Felix Kuehling <felix.kuehling@amd.com>
Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(cherry picked from commit 266fab83139600e58db7e900e930c722e619b78c)
Cc: stable@vger.kernel.org
[ sashal: Adapted retry handling to existing atomic eviction counters instead of notifier locking. ]
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
 .../gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c    | 17 ++++++++++++-----
 drivers/gpu/drm/amd/amdkfd/kfd_svm.c            |  8 +++++---
 2 files changed, 17 insertions(+), 8 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
index 59ea50b0962e5..4372317b63019 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
@@ -2609,6 +2609,7 @@ static void amdgpu_amdkfd_restore_userptr_worker(struct work_struct *work)
 	struct task_struct *usertask;
 	struct mm_struct *mm;
 	int evicted_bos;
+	int ret;
 
 	evicted_bos = atomic_read(&process_info->evicted_bos);
 	if (!evicted_bos)
@@ -2626,7 +2627,8 @@ static void amdgpu_amdkfd_restore_userptr_worker(struct work_struct *work)
 
 	mutex_lock(&process_info->lock);
 
-	if (update_invalid_user_pages(process_info, mm))
+	ret = update_invalid_user_pages(process_info, mm);
+	if (ret)
 		goto unlock_out;
 	/* userptr_inval_list can be empty if all evicted userptr BOs
 	 * have been freed. In that case there is nothing to validate
@@ -2636,7 +2638,8 @@ static void amdgpu_amdkfd_restore_userptr_worker(struct work_struct *work)
 		if (atomic_read(&process_info->evicted_bos) != evicted_bos)
 			goto unlock_out; /* Concurrent eviction, try again */
 
-		if (validate_invalid_user_pages(process_info))
+		ret = validate_invalid_user_pages(process_info);
+		if (ret)
 			goto unlock_out;
 	}
 	/* Final check for concurrent evicton and atomic update. If
@@ -2645,8 +2648,10 @@ static void amdgpu_amdkfd_restore_userptr_worker(struct work_struct *work)
 	 * reference counting inside KFD will handle this case.
 	 */
 	if (atomic_cmpxchg(&process_info->evicted_bos, evicted_bos, 0) !=
-	    evicted_bos)
+	    evicted_bos) {
+		ret = -EAGAIN;
 		goto unlock_out;
+	}
 	evicted_bos = 0;
 	if (kgd2kfd_resume_mm(mm)) {
 		pr_err("%s: Failed to resume KFD\n", __func__);
@@ -2658,8 +2663,10 @@ static void amdgpu_amdkfd_restore_userptr_worker(struct work_struct *work)
 unlock_out:
 	mutex_unlock(&process_info->lock);
 
-	/* If validation failed, reschedule another attempt */
-	if (evicted_bos) {
+	/* If validation failed, reschedule another attempt, unless the
+	 * device is gone for good
+	 */
+	if (evicted_bos && ret != -ENODEV) {
 		schedule_delayed_work(&process_info->restore_userptr_work,
 			msecs_to_jiffies(AMDGPU_USERPTR_RESTORE_DELAY_MS));
 
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
index e60ab76d3e6f6..13ae155bd5325 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
@@ -1692,7 +1692,7 @@ static void svm_range_restore_work(struct work_struct *work)
 	struct mm_struct *mm;
 	int evicted_ranges;
 	int invalid;
-	int r;
+	int r = 0;
 
 	svms = container_of(dwork, struct svm_range_list, restore_work);
 	evicted_ranges = atomic_read(&svms->evicted_ranges);
@@ -1766,8 +1766,10 @@ static void svm_range_restore_work(struct work_struct *work)
 	mmap_write_unlock(mm);
 	mutex_unlock(&process_info->lock);
 
-	/* If validation failed, reschedule another attempt */
-	if (evicted_ranges) {
+	/* If validation failed, reschedule another attempt, unless the
+	 * device is gone for good
+	 */
+	if (evicted_ranges && r != -ENODEV) {
 		pr_debug("reschedule to restore svm range\n");
 		schedule_delayed_work(&svms->restore_work,
 			msecs_to_jiffies(AMDGPU_SVM_RANGE_RESTORE_DELAY_MS));
-- 
2.53.0