* [PATCH V1] accel/amdxdna: Fix race between unmap and free BO
@ 2026-09-17 15:48 Lizhi Hou
2026-09-17 18:23 ` Max Zhen
0 siblings, 1 reply; 2+ messages in thread
From: Lizhi Hou @ 2026-09-17 15:48 UTC (permalink / raw)
To: ogabbay, quic_jhugo, mario.limonciello, karol.wachowski,
dri-devel, max.zhen
Cc: Lizhi Hou, linux-kernel, sonal.santan
A userspace process can concurrently trigger an unmap, which queues
hmm_unreg_work, and free BO. If amdxdna_hmm_unreg_work() executes
list_del(&mapp->node) and is preempted before calling amdxdna_umap_put(),
amdxdna_hmm_unregister_all() can fail to find the mapping in the list
and return without calling cancel_work_sync(). This allows object
destruction to proceed while the worker still holds a pointer to the BO
in the umap structure.
Moving the work item from struct amdxdna_umap to struct amdxdna_gem_obj.
The worker no longer holds a pointer to an individual mapping, so there
is no longer a window where the umap has been unlinked from umap_list but
work is still pending against it. The work item now lives in the BO
itself, so use a single cancel_work_sync() in the free path to drain the
work.
Fixes: 445d20910429 ("accel/amdxdna: Fix unexpected wait when flushing notifier_wq")
Signed-off-by: Lizhi Hou <lizhi.hou@amd.com>
---
drivers/accel/amdxdna/amdxdna_gem.c | 95 +++++++++++++----------------
drivers/accel/amdxdna/amdxdna_gem.h | 3 +-
2 files changed, 42 insertions(+), 56 deletions(-)
diff --git a/drivers/accel/amdxdna/amdxdna_gem.c b/drivers/accel/amdxdna/amdxdna_gem.c
index e861db6f9369..398d0a58b53a 100644
--- a/drivers/accel/amdxdna/amdxdna_gem.c
+++ b/drivers/accel/amdxdna/amdxdna_gem.c
@@ -164,30 +164,6 @@ void amdxdna_gem_heap_free(struct amdxdna_client *client, struct amdxdna_gem_obj
mutex_unlock(&client->mm_lock);
}
-static struct amdxdna_gem_obj *
-amdxdna_gem_create_obj(struct drm_device *dev, size_t size)
-{
- struct amdxdna_gem_obj *abo;
-
- abo = kzalloc_obj(*abo);
- if (!abo)
- return ERR_PTR(-ENOMEM);
-
- abo->pinned = false;
- abo->assigned_hwctx = AMDXDNA_INVALID_CTX_HANDLE;
- mutex_init(&abo->lock);
-
- abo->mem.dma_addr = AMDXDNA_INVALID_ADDR;
- abo->mem.uva = AMDXDNA_INVALID_ADDR;
- abo->mem.size = size;
- abo->open_ref = 0;
- abo->internal = false;
- INIT_LIST_HEAD(&abo->mem.umap_list);
- xa_init_flags(&abo->heap_xa, XA_FLAGS_ALLOC);
-
- return abo;
-}
-
static void
amdxdna_gem_destroy_obj(struct amdxdna_gem_obj *abo)
{
@@ -278,10 +254,8 @@ static bool amdxdna_hmm_invalidate(struct mmu_interval_notifier *mni,
if (range->event == MMU_NOTIFY_UNMAP) {
down_write(&xdna->notifier_lock);
- if (!mapp->unmapped) {
- queue_work(xdna->notifier_wq, &mapp->hmm_unreg_work);
- mapp->unmapped = true;
- }
+ mapp->unmapped = true;
+ queue_work(xdna->notifier_wq, &abo->hmm_unreg_work);
up_write(&xdna->notifier_lock);
}
@@ -311,13 +285,13 @@ static void amdxdna_hmm_unregister(struct amdxdna_gem_obj *abo,
if (!compare_range(mapp, vma->vm_mm, vma->vm_start, vma->vm_end))
continue;
- queue_work(xdna->notifier_wq, &mapp->hmm_unreg_work);
mapp->unmapped = true;
+ queue_work(xdna->notifier_wq, &abo->hmm_unreg_work);
}
up_write(&xdna->notifier_lock);
}
-static void amdxdna_hmm_unregister_all(struct amdxdna_gem_obj *abo)
+static void amdxdna_hmm_unreg_umaps(struct amdxdna_gem_obj *abo, bool force)
{
struct amdxdna_dev *xdna = to_xdna_dev(to_gobj(abo)->dev);
struct amdxdna_umap *mapp, *tmp;
@@ -325,16 +299,18 @@ static void amdxdna_hmm_unregister_all(struct amdxdna_gem_obj *abo)
down_write(&xdna->notifier_lock);
list_for_each_entry_safe(mapp, tmp, &abo->mem.umap_list, node) {
+ if (!force && !mapp->unmapped)
+ continue;
+
mapp->unmapped = true;
- mapp->cleanup = true;
list_move(&mapp->node, &dead);
}
+ if (list_empty(&abo->mem.umap_list))
+ abo->mem.uva = AMDXDNA_INVALID_ADDR;
up_write(&xdna->notifier_lock);
- list_for_each_entry_safe(mapp, tmp, &dead, node) {
- cancel_work_sync(&mapp->hmm_unreg_work);
+ list_for_each_entry_safe(mapp, tmp, &dead, node)
amdxdna_umap_put(mapp);
- }
}
static void amdxdna_umap_release(struct kref *ref)
@@ -353,24 +329,10 @@ void amdxdna_umap_put(struct amdxdna_umap *mapp)
static void amdxdna_hmm_unreg_work(struct work_struct *work)
{
- struct amdxdna_umap *mapp = container_of(work, struct amdxdna_umap,
- hmm_unreg_work);
- struct amdxdna_gem_obj *abo = mapp->abo;
- struct amdxdna_dev *xdna;
+ struct amdxdna_gem_obj *abo = container_of(work, struct amdxdna_gem_obj,
+ hmm_unreg_work);
- xdna = to_xdna_dev(to_gobj(mapp->abo)->dev);
- down_write(&xdna->notifier_lock);
- if (mapp->cleanup) {
- up_write(&xdna->notifier_lock);
- return;
- }
-
- list_del(&mapp->node);
- if (list_empty(&abo->mem.umap_list))
- abo->mem.uva = AMDXDNA_INVALID_ADDR;
- up_write(&xdna->notifier_lock);
-
- amdxdna_umap_put(mapp);
+ amdxdna_hmm_unreg_umaps(abo, false);
}
static int amdxdna_hmm_register(struct amdxdna_gem_obj *abo,
@@ -422,8 +384,6 @@ static int amdxdna_hmm_register(struct amdxdna_gem_obj *abo,
mapp->abo = abo;
kref_init(&mapp->refcnt);
- INIT_WORK(&mapp->hmm_unreg_work, amdxdna_hmm_unreg_work);
-
ret = mmu_interval_notifier_insert_locked(&mapp->notifier,
current->mm,
addr,
@@ -449,6 +409,31 @@ static int amdxdna_hmm_register(struct amdxdna_gem_obj *abo,
return ret;
}
+static struct amdxdna_gem_obj *
+amdxdna_gem_create_obj(struct drm_device *dev, size_t size)
+{
+ struct amdxdna_gem_obj *abo;
+
+ abo = kzalloc_obj(*abo);
+ if (!abo)
+ return ERR_PTR(-ENOMEM);
+
+ abo->pinned = false;
+ abo->assigned_hwctx = AMDXDNA_INVALID_CTX_HANDLE;
+ mutex_init(&abo->lock);
+
+ abo->mem.dma_addr = AMDXDNA_INVALID_ADDR;
+ abo->mem.uva = AMDXDNA_INVALID_ADDR;
+ abo->mem.size = size;
+ abo->open_ref = 0;
+ abo->internal = false;
+ INIT_LIST_HEAD(&abo->mem.umap_list);
+ xa_init_flags(&abo->heap_xa, XA_FLAGS_ALLOC);
+ INIT_WORK(&abo->hmm_unreg_work, amdxdna_hmm_unreg_work);
+
+ return abo;
+}
+
static void amdxdna_gem_dev_obj_free(struct drm_gem_object *gobj)
{
struct amdxdna_dev *xdna = to_xdna_dev(gobj->dev);
@@ -756,7 +741,9 @@ static void amdxdna_gem_obj_free(struct drm_gem_object *gobj)
struct amdxdna_dev *xdna = to_xdna_dev(gobj->dev);
struct amdxdna_gem_obj *abo = to_xdna_obj(gobj);
- amdxdna_hmm_unregister_all(abo);
+ /* No notifier survives this, so no new work can be queued. */
+ amdxdna_hmm_unreg_umaps(abo, true);
+ cancel_work_sync(&abo->hmm_unreg_work);
if (abo->pinned)
amdxdna_gem_unpin(abo);
diff --git a/drivers/accel/amdxdna/amdxdna_gem.h b/drivers/accel/amdxdna/amdxdna_gem.h
index 5dfefdcf1356..9b4aa21a37c9 100644
--- a/drivers/accel/amdxdna/amdxdna_gem.h
+++ b/drivers/accel/amdxdna/amdxdna_gem.h
@@ -14,13 +14,11 @@
struct amdxdna_umap {
struct mmu_interval_notifier notifier;
struct hmm_range range;
- struct work_struct hmm_unreg_work;
struct amdxdna_gem_obj *abo;
struct list_head node;
struct kref refcnt;
bool invalid;
bool unmapped;
- bool cleanup;
};
struct amdxdna_mem {
@@ -44,6 +42,7 @@ struct amdxdna_gem_obj {
struct mutex lock; /* Protects: pinned, mem.kva, open_ref */
struct amdxdna_mem mem;
int open_ref;
+ struct work_struct hmm_unreg_work;
/* Below members are initialized when needed */
struct drm_mm_node mm_node; /* For AMDXDNA_BO_DEV */
--
2.34.1
^ permalink raw reply [flat|nested] 2+ messages in thread* Re: [PATCH V1] accel/amdxdna: Fix race between unmap and free BO
2026-09-17 15:48 [PATCH V1] accel/amdxdna: Fix race between unmap and free BO Lizhi Hou
@ 2026-09-17 18:23 ` Max Zhen
0 siblings, 0 replies; 2+ messages in thread
From: Max Zhen @ 2026-09-17 18:23 UTC (permalink / raw)
To: Lizhi Hou, ogabbay, quic_jhugo, mario.limonciello,
karol.wachowski, dri-devel
Cc: linux-kernel, sonal.santan
On 9/17/2026 Thu 08:48, Lizhi Hou wrote:
> A userspace process can concurrently trigger an unmap, which queues
> hmm_unreg_work, and free BO. If amdxdna_hmm_unreg_work() executes
> list_del(&mapp->node) and is preempted before calling amdxdna_umap_put(),
> amdxdna_hmm_unregister_all() can fail to find the mapping in the list
> and return without calling cancel_work_sync(). This allows object
> destruction to proceed while the worker still holds a pointer to the BO
> in the umap structure.
>
> Moving the work item from struct amdxdna_umap to struct amdxdna_gem_obj.
> The worker no longer holds a pointer to an individual mapping, so there
> is no longer a window where the umap has been unlinked from umap_list but
> work is still pending against it. The work item now lives in the BO
> itself, so use a single cancel_work_sync() in the free path to drain the
> work.
>
> Fixes: 445d20910429 ("accel/amdxdna: Fix unexpected wait when flushing notifier_wq")
> Signed-off-by: Lizhi Hou <lizhi.hou@amd.com>
Reviewed-by: Max Zhen <max.zhen@amd.com>
> ---
> drivers/accel/amdxdna/amdxdna_gem.c | 95 +++++++++++++----------------
> drivers/accel/amdxdna/amdxdna_gem.h | 3 +-
> 2 files changed, 42 insertions(+), 56 deletions(-)
>
> diff --git a/drivers/accel/amdxdna/amdxdna_gem.c b/drivers/accel/amdxdna/amdxdna_gem.c
> index e861db6f9369..398d0a58b53a 100644
> --- a/drivers/accel/amdxdna/amdxdna_gem.c
> +++ b/drivers/accel/amdxdna/amdxdna_gem.c
> @@ -164,30 +164,6 @@ void amdxdna_gem_heap_free(struct amdxdna_client *client, struct amdxdna_gem_obj
> mutex_unlock(&client->mm_lock);
> }
>
> -static struct amdxdna_gem_obj *
> -amdxdna_gem_create_obj(struct drm_device *dev, size_t size)
> -{
> - struct amdxdna_gem_obj *abo;
> -
> - abo = kzalloc_obj(*abo);
> - if (!abo)
> - return ERR_PTR(-ENOMEM);
> -
> - abo->pinned = false;
> - abo->assigned_hwctx = AMDXDNA_INVALID_CTX_HANDLE;
> - mutex_init(&abo->lock);
> -
> - abo->mem.dma_addr = AMDXDNA_INVALID_ADDR;
> - abo->mem.uva = AMDXDNA_INVALID_ADDR;
> - abo->mem.size = size;
> - abo->open_ref = 0;
> - abo->internal = false;
> - INIT_LIST_HEAD(&abo->mem.umap_list);
> - xa_init_flags(&abo->heap_xa, XA_FLAGS_ALLOC);
> -
> - return abo;
> -}
> -
> static void
> amdxdna_gem_destroy_obj(struct amdxdna_gem_obj *abo)
> {
> @@ -278,10 +254,8 @@ static bool amdxdna_hmm_invalidate(struct mmu_interval_notifier *mni,
>
> if (range->event == MMU_NOTIFY_UNMAP) {
> down_write(&xdna->notifier_lock);
> - if (!mapp->unmapped) {
> - queue_work(xdna->notifier_wq, &mapp->hmm_unreg_work);
> - mapp->unmapped = true;
> - }
> + mapp->unmapped = true;
> + queue_work(xdna->notifier_wq, &abo->hmm_unreg_work);
> up_write(&xdna->notifier_lock);
> }
>
> @@ -311,13 +285,13 @@ static void amdxdna_hmm_unregister(struct amdxdna_gem_obj *abo,
> if (!compare_range(mapp, vma->vm_mm, vma->vm_start, vma->vm_end))
> continue;
>
> - queue_work(xdna->notifier_wq, &mapp->hmm_unreg_work);
> mapp->unmapped = true;
> + queue_work(xdna->notifier_wq, &abo->hmm_unreg_work);
> }
> up_write(&xdna->notifier_lock);
> }
>
> -static void amdxdna_hmm_unregister_all(struct amdxdna_gem_obj *abo)
> +static void amdxdna_hmm_unreg_umaps(struct amdxdna_gem_obj *abo, bool force)
> {
> struct amdxdna_dev *xdna = to_xdna_dev(to_gobj(abo)->dev);
> struct amdxdna_umap *mapp, *tmp;
> @@ -325,16 +299,18 @@ static void amdxdna_hmm_unregister_all(struct amdxdna_gem_obj *abo)
>
> down_write(&xdna->notifier_lock);
> list_for_each_entry_safe(mapp, tmp, &abo->mem.umap_list, node) {
> + if (!force && !mapp->unmapped)
> + continue;
> +
> mapp->unmapped = true;
> - mapp->cleanup = true;
> list_move(&mapp->node, &dead);
> }
> + if (list_empty(&abo->mem.umap_list))
> + abo->mem.uva = AMDXDNA_INVALID_ADDR;
> up_write(&xdna->notifier_lock);
>
> - list_for_each_entry_safe(mapp, tmp, &dead, node) {
> - cancel_work_sync(&mapp->hmm_unreg_work);
> + list_for_each_entry_safe(mapp, tmp, &dead, node)
> amdxdna_umap_put(mapp);
> - }
> }
>
> static void amdxdna_umap_release(struct kref *ref)
> @@ -353,24 +329,10 @@ void amdxdna_umap_put(struct amdxdna_umap *mapp)
>
> static void amdxdna_hmm_unreg_work(struct work_struct *work)
> {
> - struct amdxdna_umap *mapp = container_of(work, struct amdxdna_umap,
> - hmm_unreg_work);
> - struct amdxdna_gem_obj *abo = mapp->abo;
> - struct amdxdna_dev *xdna;
> + struct amdxdna_gem_obj *abo = container_of(work, struct amdxdna_gem_obj,
> + hmm_unreg_work);
>
> - xdna = to_xdna_dev(to_gobj(mapp->abo)->dev);
> - down_write(&xdna->notifier_lock);
> - if (mapp->cleanup) {
> - up_write(&xdna->notifier_lock);
> - return;
> - }
> -
> - list_del(&mapp->node);
> - if (list_empty(&abo->mem.umap_list))
> - abo->mem.uva = AMDXDNA_INVALID_ADDR;
> - up_write(&xdna->notifier_lock);
> -
> - amdxdna_umap_put(mapp);
> + amdxdna_hmm_unreg_umaps(abo, false);
> }
>
> static int amdxdna_hmm_register(struct amdxdna_gem_obj *abo,
> @@ -422,8 +384,6 @@ static int amdxdna_hmm_register(struct amdxdna_gem_obj *abo,
> mapp->abo = abo;
> kref_init(&mapp->refcnt);
>
> - INIT_WORK(&mapp->hmm_unreg_work, amdxdna_hmm_unreg_work);
> -
> ret = mmu_interval_notifier_insert_locked(&mapp->notifier,
> current->mm,
> addr,
> @@ -449,6 +409,31 @@ static int amdxdna_hmm_register(struct amdxdna_gem_obj *abo,
> return ret;
> }
>
> +static struct amdxdna_gem_obj *
> +amdxdna_gem_create_obj(struct drm_device *dev, size_t size)
> +{
> + struct amdxdna_gem_obj *abo;
> +
> + abo = kzalloc_obj(*abo);
> + if (!abo)
> + return ERR_PTR(-ENOMEM);
> +
> + abo->pinned = false;
> + abo->assigned_hwctx = AMDXDNA_INVALID_CTX_HANDLE;
> + mutex_init(&abo->lock);
> +
> + abo->mem.dma_addr = AMDXDNA_INVALID_ADDR;
> + abo->mem.uva = AMDXDNA_INVALID_ADDR;
> + abo->mem.size = size;
> + abo->open_ref = 0;
> + abo->internal = false;
> + INIT_LIST_HEAD(&abo->mem.umap_list);
> + xa_init_flags(&abo->heap_xa, XA_FLAGS_ALLOC);
> + INIT_WORK(&abo->hmm_unreg_work, amdxdna_hmm_unreg_work);
> +
> + return abo;
> +}
> +
> static void amdxdna_gem_dev_obj_free(struct drm_gem_object *gobj)
> {
> struct amdxdna_dev *xdna = to_xdna_dev(gobj->dev);
> @@ -756,7 +741,9 @@ static void amdxdna_gem_obj_free(struct drm_gem_object *gobj)
> struct amdxdna_dev *xdna = to_xdna_dev(gobj->dev);
> struct amdxdna_gem_obj *abo = to_xdna_obj(gobj);
>
> - amdxdna_hmm_unregister_all(abo);
> + /* No notifier survives this, so no new work can be queued. */
> + amdxdna_hmm_unreg_umaps(abo, true);
> + cancel_work_sync(&abo->hmm_unreg_work);
>
> if (abo->pinned)
> amdxdna_gem_unpin(abo);
> diff --git a/drivers/accel/amdxdna/amdxdna_gem.h b/drivers/accel/amdxdna/amdxdna_gem.h
> index 5dfefdcf1356..9b4aa21a37c9 100644
> --- a/drivers/accel/amdxdna/amdxdna_gem.h
> +++ b/drivers/accel/amdxdna/amdxdna_gem.h
> @@ -14,13 +14,11 @@
> struct amdxdna_umap {
> struct mmu_interval_notifier notifier;
> struct hmm_range range;
> - struct work_struct hmm_unreg_work;
> struct amdxdna_gem_obj *abo;
> struct list_head node;
> struct kref refcnt;
> bool invalid;
> bool unmapped;
> - bool cleanup;
> };
>
> struct amdxdna_mem {
> @@ -44,6 +42,7 @@ struct amdxdna_gem_obj {
> struct mutex lock; /* Protects: pinned, mem.kva, open_ref */
> struct amdxdna_mem mem;
> int open_ref;
> + struct work_struct hmm_unreg_work;
>
> /* Below members are initialized when needed */
> struct drm_mm_node mm_node; /* For AMDXDNA_BO_DEV */
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2026-09-17 18:23 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2026-09-17 15:48 [PATCH V1] accel/amdxdna: Fix race between unmap and free BO Lizhi Hou
2026-09-17 18:23 ` Max Zhen
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®