lkml.org 
[lkml]   [2022]   [Apr]   [12]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
From
Subject[PATCH 5.17 032/343] drm/amdkfd: svm range restore work deadlock when process exit
Date
From: Philip Yang <Philip.Yang@amd.com>

[ Upstream commit 6225bb3a88d22594aacea2485dc28ca12d596721 ]

kfd_process_notifier_release flush svm_range_restore_work
which calls svm_range_list_lock_and_flush_work to flush deferred_list
work, but if deferred_list work mmput release the last user, it will
call exit_mmap -> notifier_release, it is deadlock with below backtrace.

Move flush svm_range_restore_work to kfd_process_wq_release to avoid
deadlock. Then svm_range_restore_work take task->mm ref to avoid mm is
gone while validating and mapping ranges to GPU.

Workqueue: events svm_range_deferred_list_work [amdgpu]
Call Trace:
wait_for_completion+0x94/0x100
__flush_work+0x12a/0x1e0
__cancel_work_timer+0x10e/0x190
cancel_delayed_work_sync+0x13/0x20
kfd_process_notifier_release+0x98/0x2a0 [amdgpu]
__mmu_notifier_release+0x74/0x1f0
exit_mmap+0x170/0x200
mmput+0x5d/0x130
svm_range_deferred_list_work+0x104/0x230 [amdgpu]
process_one_work+0x220/0x3c0

Signed-off-by: Philip Yang <Philip.Yang@amd.com>
Reported-by: Ruili Ji <ruili.ji@amd.com>
Tested-by: Ruili Ji <ruili.ji@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
---
drivers/gpu/drm/amd/amdkfd/kfd_process.c | 1 -
drivers/gpu/drm/amd/amdkfd/kfd_svm.c | 15 +++++++++------
2 files changed, 9 insertions(+), 7 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process.c b/drivers/gpu/drm/amd/amdkfd/kfd_process.c
index d1145da5348f..74f162887d3b 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_process.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_process.c
@@ -1150,7 +1150,6 @@ static void kfd_process_notifier_release(struct mmu_notifier *mn,

cancel_delayed_work_sync(&p->eviction_work);
cancel_delayed_work_sync(&p->restore_work);
- cancel_delayed_work_sync(&p->svms.restore_work);

mutex_lock(&p->mutex);

diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
index 225affcddbc1..1cf9041c9727 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
@@ -1643,13 +1643,14 @@ static void svm_range_restore_work(struct work_struct *work)

pr_debug("restore svm ranges\n");

- /* kfd_process_notifier_release destroys this worker thread. So during
- * the lifetime of this thread, kfd_process and mm will be valid.
- */
p = container_of(svms, struct kfd_process, svms);
- mm = p->mm;
- if (!mm)
+
+ /* Keep mm reference when svm_range_validate_and_map ranges */
+ mm = get_task_mm(p->lead_thread);
+ if (!mm) {
+ pr_debug("svms 0x%p process mm gone\n", svms);
return;
+ }

svm_range_list_lock_and_flush_work(svms, mm);
mutex_lock(&svms->lock);
@@ -1703,6 +1704,7 @@ static void svm_range_restore_work(struct work_struct *work)
out_reschedule:
mutex_unlock(&svms->lock);
mmap_write_unlock(mm);
+ mmput(mm);

/* If validation failed, reschedule another attempt */
if (evicted_ranges) {
@@ -2840,6 +2842,8 @@ void svm_range_list_fini(struct kfd_process *p)

pr_debug("pasid 0x%x svms 0x%p\n", p->pasid, &p->svms);

+ cancel_delayed_work_sync(&p->svms.restore_work);
+
/* Ensure list work is finished before process is destroyed */
flush_work(&p->svms.deferred_list_work);

@@ -2850,7 +2854,6 @@ void svm_range_list_fini(struct kfd_process *p)
atomic_inc(&p->svms.drain_pagefaults);
svm_range_drain_retry_fault(&p->svms);

-
list_for_each_entry_safe(prange, next, &p->svms.list, list) {
svm_range_unlink(prange);
svm_range_remove_notifier(prange);
--
2.35.1


\
 
 \ /
  Last update: 2022-04-12 10:27    [W:1.004 / U:0.048 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site