Skip to contentSkip to conversation

drm/amdgpu: keep the userq manager alive as long as its queues

PATCH New Zhu Lingshan 1 participant 1 message Aug 19

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 drm/amdgpu: keep the userq manager alive as long as its queues

The userq manager is embedded in fpriv, so any code paths trigger kfree(fpriv) would destroy the userq_mgr, even some of its queues are still alive, resulting in userq->userq_mgr use-after-free issues.

This commit fixes this problem by introduce a new counter refs representing for the number of its queues, and only free the userq_manager when refs == 0

---
 drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c | 28 +++++++++++++++++++++++
 drivers/gpu/drm/amd/amdgpu/amdgpu_userq.h |  9 ++++++++
 2 files changed, 37 insertions(+)
drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c +28 -0
index 24adad7be251..b7d1b973df6b 100644
@@ -27,6 +27,7 @@
 #include <linux/pm_runtime.h>
 #include <linux/overflow.h>
 #include <drm/drm_drv.h>
+#include <linux/wait_bit.h>
 
 #include "amdgpu.h"
 #include "amdgpu_reset.h"
@@ -533,6 +534,17 @@ amdgpu_userq_get_doorbell_index(struct amdgpu_userq_mgr *uq_mgr,
 	return r;
 }
 
+static void amdgpu_userq_mgr_inc_refs(struct amdgpu_userq_mgr *uq_mgr)
+{
+	atomic_inc(&uq_mgr->refs);
+}
+
+static void amdgpu_userq_mgr_dec_refs(struct amdgpu_userq_mgr *uq_mgr)
+{
+	if (atomic_dec_and_test(&uq_mgr->refs))
+		wake_up_var(&uq_mgr->refs);
+}
+
 static int
 amdgpu_userq_destroy(struct amdgpu_userq_mgr *uq_mgr, struct amdgpu_usermode_queue *queue)
 {
@@ -594,6 +606,8 @@ static void amdgpu_userq_kref_destroy(struct kref *kref)
 	r = amdgpu_userq_destroy(uq_mgr, queue);
 	if (r)
 		drm_file_err(uq_mgr->file, "Failed to destroy usermode queue %d\n", r);
+
+	amdgpu_userq_mgr_dec_refs(uq_mgr);
 }
 
 struct amdgpu_usermode_queue *amdgpu_userq_get(struct amdgpu_userq_mgr *uq_mgr, u32 qid)
@@ -677,6 +691,7 @@ amdgpu_userq_create(struct drm_file *filp, union drm_amdgpu_userq *args)
 	queue->xcp_id = (fpriv->xcp_id != AMDGPU_XCP_NO_PARTITION) ?
 				fpriv->xcp_id : 0;
 	queue->userq_mgr = uq_mgr;
+	amdgpu_userq_mgr_inc_refs(uq_mgr);
 	INIT_DELAYED_WORK(&queue->hang_detect_work,
 			  amdgpu_userq_hang_detect_work);
 
@@ -789,6 +804,7 @@ amdgpu_userq_create(struct drm_file *filp, union drm_amdgpu_userq *args)
 free_queue:
 	trace_amdgpu_userq_create_end(queue, r);
 	kfree(queue);
+	amdgpu_userq_mgr_dec_refs(uq_mgr);
 err_pm_runtime:
 	pm_runtime_put_autosuspend(adev_to_drm(adev)->dev);
 	return r;
@@ -1277,6 +1293,7 @@ int amdgpu_userq_mgr_init(struct amdgpu_userq_mgr *userq_mgr, struct drm_file *f
 {
 	mutex_init(&userq_mgr->userq_mutex);
 	xa_init_flags(&userq_mgr->userq_xa, XA_FLAGS_ALLOC);
+	atomic_set(&userq_mgr->refs, 0);
 	userq_mgr->adev = adev;
 	userq_mgr->file = file_priv;
 	userq_mgr->proc_ctx_allocated = false;
@@ -1326,6 +1343,17 @@ void amdgpu_userq_mgr_fini(struct amdgpu_userq_mgr *userq_mgr)
 		amdgpu_userq_put(queue);
 	}
 
+	/*
+	 * userq_mgr is embedded in struct amdgpu_fpriv and get
+	 * freed when kfree(fpriv). This wait_evet() blocks
+	 * amdgpu_userq_mgr_fini() which is drm_release path,
+	 * so keep userq_mgr alive while any queues holding it.
+	 *
+	 * This prevents queue->userq_mgr use-after-free issues.
+	 */
+	wait_var_event(&userq_mgr->refs,
+		       !atomic_read_acquire(&userq_mgr->refs));
+
 	xa_destroy(&userq_mgr->userq_xa);
 
 	/*
drivers/gpu/drm/amd/amdgpu/amdgpu_userq.h +9 -0
index 6412a7f7b6ef..e65426410cba 100644
@@ -126,6 +126,15 @@ struct amdgpu_userq_mgr {
 	 */
 	struct xarray			userq_xa;
 	struct mutex			userq_mutex;
+
+	/**
+	 * @refs:
+	 *
+	 * Each queue increases this counter when join this manager,
+	 * and decreases it when leave this manager.
+	 */
+	atomic_t			refs;
+
 	struct amdgpu_device		*adev;
 	struct delayed_work		resume_work;
 	struct drm_file			*file;
Signature · 1 line
2.53.0