ocfs2: avoid access invalid address when read o2dlm debug messages

In such a race case, invalid address access may occurs. So we should delete list res->tracking before resA->refs decrease to 0.

diff -puN fs/ocfs2/dlm/dlmmaster.c~ocfs2-avoid-access-invalid-address-when-read-o2dlm-debug-messages fs/ocfs2/dlm/dlmmaster.c

fs/ocfs2/dlm/dlmmaster.c renamed from fs/ocfs2/dlm/dlmmaster.c~ocfs2-avoid-access-invalid-address-when-read-o2dlm-debug-messages +0 -10

@@ -498,16 +498,6 @@ static void dlm_lockres_release(struct k
 	mlog(0, "destroying lockres %.*s\n", res->lockname.len,
 	     res->lockname.name);
 
-	spin_lock(&dlm->track_lock);
-	if (!list_empty(&res->tracking))
-		list_del_init(&res->tracking);
-	else {
-		mlog(ML_ERROR, "Resource %.*s not on the Tracking list\n",
-		     res->lockname.len, res->lockname.name);
-		dlm_print_one_lock_resource(res);
-	}
-	spin_unlock(&dlm->track_lock);
-
 	atomic_dec(&dlm->res_cur_count);
 
 	if (!hlist_unhashed(&res->hash_node) ||

diff -puN fs/ocfs2/dlm/dlmthread.c~ocfs2-avoid-access-invalid-address-when-read-o2dlm-debug-messages fs/ocfs2/dlm/dlmthread.c

fs/ocfs2/dlm/dlmthread.c renamed from fs/ocfs2/dlm/dlmthread.c~ocfs2-avoid-access-invalid-address-when-read-o2dlm-debug-messages +10 -0

@@ -211,6 +211,16 @@ static void dlm_purge_lockres(struct dlm
 
 	__dlm_unhash_lockres(dlm, res);
 
+	spin_lock(&dlm->track_lock);
+	if (!list_empty(&res->tracking))
+		list_del_init(&res->tracking);
+	else {
+		mlog(ML_ERROR, "Resource %.*s not on the Tracking list\n",
+				res->lockname.len, res->lockname.name);
+		__dlm_print_one_lock_resource(res);
+	}
+	spin_unlock(&dlm->track_lock);
+
 	/* lockres is not in the hash now.  drop the flag and wake up
 	 * any processes waiting in dlm_get_lock_resource. */
 	if (!master) {

_

Quoted 45 lines

In such a race case, invalid address access may occurs. So we should delete list res->tracking before resA->refs decrease to 0.

diff -puN fs/ocfs2/dlm/dlmmaster.c~ocfs2-avoid-access-invalid-address-when-read-o2dlm-debug-messages fs/ocfs2/dlm/dlmmaster.c

fs/ocfs2/dlm/dlmmaster.c renamed from fs/ocfs2/dlm/dlmmaster.c~ocfs2-avoid-access-invalid-address-when-read-o2dlm-debug-messages +0 -10

@@ -498,16 +498,6 @@ static void dlm_lockres_release(struct k
 	mlog(0, "destroying lockres %.*s\n", res->lockname.len,
 	     res->lockname.name);
 
-	spin_lock(&dlm->track_lock);
-	if (!list_empty(&res->tracking))
-		list_del_init(&res->tracking);
-	else {
-		mlog(ML_ERROR, "Resource %.*s not on the Tracking list\n",
-		     res->lockname.len, res->lockname.name);
-		dlm_print_one_lock_resource(res);
-	}
-	spin_unlock(&dlm->track_lock);
-
 	atomic_dec(&dlm->res_cur_count);
 
 	if (!hlist_unhashed(&res->hash_node) ||

diff -puN fs/ocfs2/dlm/dlmthread.c~ocfs2-avoid-access-invalid-address-when-read-o2dlm-debug-messages fs/ocfs2/dlm/dlmthread.c

fs/ocfs2/dlm/dlmthread.c renamed from fs/ocfs2/dlm/dlmthread.c~ocfs2-avoid-access-invalid-address-when-read-o2dlm-debug-messages +10 -0

@@ -211,6 +211,16 @@ static void dlm_purge_lockres(struct dlm
 
 	__dlm_unhash_lockres(dlm, res);
 
+	spin_lock(&dlm->track_lock);
+	if (!list_empty(&res->tracking))
+		list_del_init(&res->tracking);
+	else {
+		mlog(ML_ERROR, "Resource %.*s not on the Tracking list\n",
+				res->lockname.len, res->lockname.name);
+		__dlm_print_one_lock_resource(res);
+	}
+	spin_unlock(&dlm->track_lock);
+
 	/* lockres is not in the hash now.  drop the flag and wake up
 	 * any processes waiting in dlm_get_lock_resource. */
 	if (!master) {

Quoted 1 line

_

How do we know that dlm_purge_lockres() is the last caller of dlm_lockres_put()? Don't we now have a problem where if the last ref is dropped by any other function than dlm_purge_lockres() the lockres is freed while on the tracking list?

	--Mark

--
Mark Fasheh

.

? 2014/12/17 6:26, Mark Fasheh ??:

Quoted 57 lines · Andrew Morton

Quoted 45 lines

In such a race case, invalid address access may occurs. So we should delete list res->tracking before resA->refs decrease to 0.

diff -puN fs/ocfs2/dlm/dlmmaster.c~ocfs2-avoid-access-invalid-address-when-read-o2dlm-debug-messages fs/ocfs2/dlm/dlmmaster.c

fs/ocfs2/dlm/dlmmaster.c renamed from fs/ocfs2/dlm/dlmmaster.c~ocfs2-avoid-access-invalid-address-when-read-o2dlm-debug-messages +0 -10

@@ -498,16 +498,6 @@ static void dlm_lockres_release(struct k
 	mlog(0, "destroying lockres %.*s\n", res->lockname.len,
 	     res->lockname.name);
 
-	spin_lock(&dlm->track_lock);
-	if (!list_empty(&res->tracking))
-		list_del_init(&res->tracking);
-	else {
-		mlog(ML_ERROR, "Resource %.*s not on the Tracking list\n",
-		     res->lockname.len, res->lockname.name);
-		dlm_print_one_lock_resource(res);
-	}
-	spin_unlock(&dlm->track_lock);
-
 	atomic_dec(&dlm->res_cur_count);
 
 	if (!hlist_unhashed(&res->hash_node) ||

diff -puN fs/ocfs2/dlm/dlmthread.c~ocfs2-avoid-access-invalid-address-when-read-o2dlm-debug-messages fs/ocfs2/dlm/dlmthread.c

fs/ocfs2/dlm/dlmthread.c renamed from fs/ocfs2/dlm/dlmthread.c~ocfs2-avoid-access-invalid-address-when-read-o2dlm-debug-messages +10 -0

@@ -211,6 +211,16 @@ static void dlm_purge_lockres(struct dlm
 
 	__dlm_unhash_lockres(dlm, res);
 
+	spin_lock(&dlm->track_lock);
+	if (!list_empty(&res->tracking))
+		list_del_init(&res->tracking);
+	else {
+		mlog(ML_ERROR, "Resource %.*s not on the Tracking list\n",
+				res->lockname.len, res->lockname.name);
+		__dlm_print_one_lock_resource(res);
+	}
+	spin_unlock(&dlm->track_lock);
+
 	/* lockres is not in the hash now.  drop the flag and wake up
 	 * any processes waiting in dlm_get_lock_resource. */
 	if (!master) {

Quoted 1 line

_

How do we know that dlm_purge_lockres() is the last caller of dlm_lockres_put()? Don't we now have a problem where if the last ref is dropped by any other function than dlm_purge_lockres() the lockres is freed while on the tracking list?

	--Mark

--
Mark Fasheh

Quoted 2 lines

.

dlm_purge_lockres is not necessarily the last caller of dlm_lockres_put(), but it means lockres will be purged if dlm_purge_lockres is called. Besides, lockres is also unhashed in dlm_purge_lockres, so lockres can be removed from tracking list. contents of dlm->tracking_list will be consistent with dlm->lockres_hash.

Quoted 58 lines

? 2014/12/17 6:26, Mark Fasheh ??:

Quoted 57 lines · Andrew Morton

Quoted 45 lines

In such a race case, invalid address access may occurs. So we should delete list res->tracking before resA->refs decrease to 0.

diff -puN fs/ocfs2/dlm/dlmmaster.c~ocfs2-avoid-access-invalid-address-when-read-o2dlm-debug-messages fs/ocfs2/dlm/dlmmaster.c

fs/ocfs2/dlm/dlmmaster.c renamed from fs/ocfs2/dlm/dlmmaster.c~ocfs2-avoid-access-invalid-address-when-read-o2dlm-debug-messages +0 -10

@@ -498,16 +498,6 @@ static void dlm_lockres_release(struct k
 	mlog(0, "destroying lockres %.*s\n", res->lockname.len,
 	     res->lockname.name);
 
-	spin_lock(&dlm->track_lock);
-	if (!list_empty(&res->tracking))
-		list_del_init(&res->tracking);
-	else {
-		mlog(ML_ERROR, "Resource %.*s not on the Tracking list\n",
-		     res->lockname.len, res->lockname.name);
-		dlm_print_one_lock_resource(res);
-	}
-	spin_unlock(&dlm->track_lock);
-
 	atomic_dec(&dlm->res_cur_count);
 
 	if (!hlist_unhashed(&res->hash_node) ||

diff -puN fs/ocfs2/dlm/dlmthread.c~ocfs2-avoid-access-invalid-address-when-read-o2dlm-debug-messages fs/ocfs2/dlm/dlmthread.c

fs/ocfs2/dlm/dlmthread.c renamed from fs/ocfs2/dlm/dlmthread.c~ocfs2-avoid-access-invalid-address-when-read-o2dlm-debug-messages +10 -0

@@ -211,6 +211,16 @@ static void dlm_purge_lockres(struct dlm
 
 	__dlm_unhash_lockres(dlm, res);
 
+	spin_lock(&dlm->track_lock);
+	if (!list_empty(&res->tracking))
+		list_del_init(&res->tracking);
+	else {
+		mlog(ML_ERROR, "Resource %.*s not on the Tracking list\n",
+				res->lockname.len, res->lockname.name);
+		__dlm_print_one_lock_resource(res);
+	}
+	spin_unlock(&dlm->track_lock);
+
 	/* lockres is not in the hash now.  drop the flag and wake up
 	 * any processes waiting in dlm_get_lock_resource. */
 	if (!master) {

Quoted 1 line

_

How do we know that dlm_purge_lockres() is the last caller of dlm_lockres_put()? Don't we now have a problem where if the last ref is dropped by any other function than dlm_purge_lockres() the lockres is freed while on the tracking list?

	--Mark

--
Mark Fasheh

Quoted 8 lines

Quoted 2 lines

.

dlm_purge_lockres is not necessarily the last caller of dlm_lockres_put(), but it means lockres will be purged if dlm_purge_lockres is called. Besides, lockres is also unhashed in dlm_purge_lockres, so lockres can be removed from tracking list. contents of dlm->tracking_list will be consistent with dlm->lockres_hash.

I'm still confused. This is what I'm worried about:

1) a procss calls dlm_lockres_put(), NOT from dlm_purge_lockres().

2) if the count goes to zero, then that process will call dlm_lockres_release()

3) dlm_lockres_release() will free the lockres without removing it from the tracking list. Thus we will have a corrupted list.

Does that make sense? Am I wrong here?

	--Mark

--
Mark Fasheh

.

? 2014/12/20 4:25, Mark Fasheh ??:

Quoted 68 lines · jiangyiwen

Quoted 58 lines

? 2014/12/17 6:26, Mark Fasheh ??:

Quoted 57 lines · Andrew Morton

Quoted 45 lines

In such a race case, invalid address access may occurs. So we should delete list res->tracking before resA->refs decrease to 0.

diff -puN fs/ocfs2/dlm/dlmmaster.c~ocfs2-avoid-access-invalid-address-when-read-o2dlm-debug-messages fs/ocfs2/dlm/dlmmaster.c

fs/ocfs2/dlm/dlmmaster.c renamed from fs/ocfs2/dlm/dlmmaster.c~ocfs2-avoid-access-invalid-address-when-read-o2dlm-debug-messages +0 -10

@@ -498,16 +498,6 @@ static void dlm_lockres_release(struct k
 	mlog(0, "destroying lockres %.*s\n", res->lockname.len,
 	     res->lockname.name);
 
-	spin_lock(&dlm->track_lock);
-	if (!list_empty(&res->tracking))
-		list_del_init(&res->tracking);
-	else {
-		mlog(ML_ERROR, "Resource %.*s not on the Tracking list\n",
-		     res->lockname.len, res->lockname.name);
-		dlm_print_one_lock_resource(res);
-	}
-	spin_unlock(&dlm->track_lock);
-
 	atomic_dec(&dlm->res_cur_count);
 
 	if (!hlist_unhashed(&res->hash_node) ||

diff -puN fs/ocfs2/dlm/dlmthread.c~ocfs2-avoid-access-invalid-address-when-read-o2dlm-debug-messages fs/ocfs2/dlm/dlmthread.c

fs/ocfs2/dlm/dlmthread.c renamed from fs/ocfs2/dlm/dlmthread.c~ocfs2-avoid-access-invalid-address-when-read-o2dlm-debug-messages +10 -0

@@ -211,6 +211,16 @@ static void dlm_purge_lockres(struct dlm
 
 	__dlm_unhash_lockres(dlm, res);
 
+	spin_lock(&dlm->track_lock);
+	if (!list_empty(&res->tracking))
+		list_del_init(&res->tracking);
+	else {
+		mlog(ML_ERROR, "Resource %.*s not on the Tracking list\n",
+				res->lockname.len, res->lockname.name);
+		__dlm_print_one_lock_resource(res);
+	}
+	spin_unlock(&dlm->track_lock);
+
 	/* lockres is not in the hash now.  drop the flag and wake up
 	 * any processes waiting in dlm_get_lock_resource. */
 	if (!master) {

Quoted 1 line

_

How do we know that dlm_purge_lockres() is the last caller of dlm_lockres_put()? Don't we now have a problem where if the last ref is dropped by any other function than dlm_purge_lockres() the lockres is freed while on the tracking list?

	--Mark

--
Mark Fasheh

Quoted 8 lines

Quoted 2 lines

.

dlm_purge_lockres is not necessarily the last caller of dlm_lockres_put(), but it means lockres will be purged if dlm_purge_lockres is called. Besides, lockres is also unhashed in dlm_purge_lockres, so lockres can be removed from tracking list. contents of dlm->tracking_list will be consistent with dlm->lockres_hash.

Quoted 6 lines

I'm still confused. This is what I'm worried about:

1) a procss calls dlm_lockres_put(), NOT from dlm_purge_lockres().

2) if the count goes to zero, then that process will call dlm_lockres_release()

But at this time, it has already called dlm_purge_lockres(). The reasons are as follows:
1) lockres is created in dlm_init_lockres(), it call kref_init(), count is 1;
2) Only when lockres is unused, it will call dlm_lockres_put() twice by dlm_run_purge_list(). So dlm_purge_lockres() has been called if the count goes to zero.

Quoted 3 lines

3) dlm_lockres_release() will free the lockres without removing it from the tracking list. Thus we will have a corrupted list.

Without this scene. dlm_purge_lockres() is called before dlm_lockres_release().

Quoted 8 lines

Does that make sense? Am I wrong here?

	--Mark

--
Mark Fasheh

.

_______________________________________________
Ocfs2-devel mailing list
Ocfs2-devel at oss.oracle.com
https://oss.oracle.com/mailman/listinfo/ocfs2-devel

Hi jiangyiwen,

Quoted 80 lines · jiangyiwen

? 2014/12/20 4:25, Mark Fasheh ??:

Quoted 68 lines · jiangyiwen

Quoted 58 lines

? 2014/12/17 6:26, Mark Fasheh ??:

Quoted 57 lines · Andrew Morton

Quoted 45 lines

In such a race case, invalid address access may occurs. So we should delete list res->tracking before resA->refs decrease to 0.

diff -puN fs/ocfs2/dlm/dlmmaster.c~ocfs2-avoid-access-invalid-address-when-read-o2dlm-debug-messages fs/ocfs2/dlm/dlmmaster.c

fs/ocfs2/dlm/dlmmaster.c renamed from fs/ocfs2/dlm/dlmmaster.c~ocfs2-avoid-access-invalid-address-when-read-o2dlm-debug-messages +0 -10

@@ -498,16 +498,6 @@ static void dlm_lockres_release(struct k
 	mlog(0, "destroying lockres %.*s\n", res->lockname.len,
 	     res->lockname.name);
 
-	spin_lock(&dlm->track_lock);
-	if (!list_empty(&res->tracking))
-		list_del_init(&res->tracking);
-	else {
-		mlog(ML_ERROR, "Resource %.*s not on the Tracking list\n",
-		     res->lockname.len, res->lockname.name);
-		dlm_print_one_lock_resource(res);
-	}
-	spin_unlock(&dlm->track_lock);
-
 	atomic_dec(&dlm->res_cur_count);
 
 	if (!hlist_unhashed(&res->hash_node) ||

diff -puN fs/ocfs2/dlm/dlmthread.c~ocfs2-avoid-access-invalid-address-when-read-o2dlm-debug-messages fs/ocfs2/dlm/dlmthread.c

fs/ocfs2/dlm/dlmthread.c renamed from fs/ocfs2/dlm/dlmthread.c~ocfs2-avoid-access-invalid-address-when-read-o2dlm-debug-messages +10 -0

@@ -211,6 +211,16 @@ static void dlm_purge_lockres(struct dlm
 
 	__dlm_unhash_lockres(dlm, res);
 
+	spin_lock(&dlm->track_lock);
+	if (!list_empty(&res->tracking))
+		list_del_init(&res->tracking);
+	else {
+		mlog(ML_ERROR, "Resource %.*s not on the Tracking list\n",
+				res->lockname.len, res->lockname.name);
+		__dlm_print_one_lock_resource(res);
+	}
+	spin_unlock(&dlm->track_lock);
+
 	/* lockres is not in the hash now.  drop the flag and wake up
 	 * any processes waiting in dlm_get_lock_resource. */
 	if (!master) {

Quoted 1 line

_

How do we know that dlm_purge_lockres() is the last caller of dlm_lockres_put()? Don't we now have a problem where if the last ref is dropped by any other function than dlm_purge_lockres() the lockres is freed while on the tracking list?

	--Mark

--
Mark Fasheh

Quoted 8 lines

Quoted 2 lines

.

dlm_purge_lockres is not necessarily the last caller of dlm_lockres_put(), but it means lockres will be purged if dlm_purge_lockres is called. Besides, lockres is also unhashed in dlm_purge_lockres, so lockres can be removed from tracking list. contents of dlm->tracking_list will be consistent with dlm->lockres_hash.

Quoted 6 lines

I'm still confused. This is what I'm worried about:

1) a procss calls dlm_lockres_put(), NOT from dlm_purge_lockres().

2) if the count goes to zero, then that process will call dlm_lockres_release()

But at this time, it has already called dlm_purge_lockres(). The reasons are as follows:
1) lockres is created in dlm_init_lockres(), it call kref_init(), count is 1;
2) Only when lockres is unused, it will call dlm_lockres_put() twice by dlm_run_purge_list(). So dlm_purge_lockres() has been called if the count goes to zero.

Quoted 4 lines

Quoted 3 lines

3) dlm_lockres_release() will free the lockres without removing it from the tracking list. Thus we will have a corrupted list.

Without this scene. dlm_purge_lockres() is called before dlm_lockres_release().

Once lock resource is inserted into hash list, it should call dlm_purge_lockres()->__dlm_unhash_lockres() to put the last ref. However, if lock resource is initialized and inserted into tracking list but not inserted into hash list, it can happen that dlm_lockres_release() will free the lockres without removing it from the tracking list.
So we should remove lockres from tracking list if we call dlm_lockres_put() after lockres is created but not inserted into hash list yet.

Thanks,
Xuejiufei

Quoted 11 lines

Quoted 8 lines

Does that make sense? Am I wrong here?

	--Mark

--
Mark Fasheh

.

Quoted 5 lines

_______________________________________________
Ocfs2-devel mailing list
Ocfs2-devel at oss.oracle.com
https://oss.oracle.com/mailman/listinfo/ocfs2-devel

.