blk-throttle: Use task_subsys_state() to determine a task's blkio_cgroup

Currentlly we first map the task to cgroup and then cgroup to blkio_cgroup. There is a more direct way to get to blkio_cgroup from task using task_subsys_state(). Use that.

The real reason for the fix is that it also avoids a race in generic cgroup code. During remount/umount rebind_subsystems() is called and it can do following with and rcu protection.

cgrp->subsys[i] = NULL;

That means if somebody got hold of cgroup under rcu and then it tried to do cgroup->subsys[] to get to blkio_cgroup, it would get NULL which is wrong. I was running into this race condition with ltp running on a upstream derived kernel and that lead to crash.

So ideally we should also fix cgroup generic code to wait for rcu grace period before setting pointer to NULL. Li Zefan is not very keen on introducing synchronize_wait() as he thinks it will slow down moun/remount/umount operations.

So for the time being atleast fix the kernel crash by taking a more direct route to blkio_cgroup.

One tester had reported a crash while running LTP on a derived kernel and with this fix crash is no more seen while the test has been running for over 6 days.

Signed-off-by Vivek Goyal <[email protected]>

Currentlly we first map the task to cgroup and then cgroup to blkio_cgroup. There is a more direct way to get to blkio_cgroup from task using task_subsys_state(). Use that.

The real reason for the fix is that it also avoids a race in generic cgroup code. During remount/umount rebind_subsystems() is called and it can do following with and rcu protection.

cgrp->subsys[i] = NULL;

That means if somebody got hold of cgroup under rcu and then it tried to do cgroup->subsys[] to get to blkio_cgroup, it would get NULL which is wrong. I was running into this race condition with ltp running on a upstream derived kernel and that lead to crash.

So ideally we should also fix cgroup generic code to wait for rcu grace period before setting pointer to NULL. Li Zefan is not very keen on introducing synchronize_wait() as he thinks it will slow down moun/remount/umount operations.

So for the time being atleast fix the kernel crash by taking a more direct route to blkio_cgroup.

One tester had reported a crash while running LTP on a derived kernel and with this fix crash is no more seen while the test has been running for over 6 days.

Signed-off-by Vivek Goyal <[email protected]>

---
 block/blk-cgroup.c   |    7 +++++++
 block/blk-cgroup.h   |    3 +++
 block/blk-throttle.c |    9 ++++-----
 block/cfq-iosched.c  |   11 +++++------
 4 files changed, 19 insertions(+), 11 deletions(-)

Index: linux-2.6/block/blk-throttle.c
===================================================================

block/blk-throttle.c +4 -5

@@ -160,9 +160,8 @@ static void throtl_put_tg(struct throtl_
 }
 
 static struct throtl_grp * throtl_find_alloc_tg(struct throtl_data *td,
-			struct cgroup *cgroup)
+			struct blkio_cgroup *blkcg)
 {
-	struct blkio_cgroup *blkcg = cgroup_to_blkio_cgroup(cgroup);
 	struct throtl_grp *tg = NULL;
 	void *key = td;
 	struct backing_dev_info *bdi = &td->queue->backing_dev_info;

@@ -229,12 +228,12 @@ done:
 
 static struct throtl_grp * throtl_get_tg(struct throtl_data *td)
 {
-	struct cgroup *cgroup;
 	struct throtl_grp *tg = NULL;
+	struct blkio_cgroup *blkcg;
 
 	rcu_read_lock();
-	cgroup = task_cgroup(current, blkio_subsys_id);
-	tg = throtl_find_alloc_tg(td, cgroup);
+	blkcg = task_blkio_cgroup(current);
+	tg = throtl_find_alloc_tg(td, blkcg);
 	if (!tg)
 		tg = &td->root_tg;
 	rcu_read_unlock();

Index: linux-2.6/block/blk-cgroup.c
===================================================================

block/blk-cgroup.c +7 -0

@@ -114,6 +114,13 @@ struct blkio_cgroup *cgroup_to_blkio_cgr
 }
 EXPORT_SYMBOL_GPL(cgroup_to_blkio_cgroup);
 
+struct blkio_cgroup *task_blkio_cgroup(struct task_struct *tsk)
+{
+	return container_of(task_subsys_state(tsk, blkio_subsys_id),
+			    struct blkio_cgroup, css);
+}
+EXPORT_SYMBOL_GPL(task_blkio_cgroup);
+
 static inline void
 blkio_update_group_weight(struct blkio_group *blkg, unsigned int weight)
 {

Index: linux-2.6/block/blk-cgroup.h
===================================================================

block/blk-cgroup.h +3 -0

@@ -291,6 +291,7 @@ static inline void blkiocg_set_start_emp
 #if defined(CONFIG_BLK_CGROUP) || defined(CONFIG_BLK_CGROUP_MODULE)
 extern struct blkio_cgroup blkio_root_cgroup;
 extern struct blkio_cgroup *cgroup_to_blkio_cgroup(struct cgroup *cgroup);
+extern struct blkio_cgroup *task_blkio_cgroup(struct task_struct *tsk);
 extern void blkiocg_add_blkio_group(struct blkio_cgroup *blkcg,
 	struct blkio_group *blkg, void *key, dev_t dev,
 	enum blkio_policy_id plid);

@@ -314,6 +315,8 @@ void blkiocg_update_io_remove_stats(stru
 struct cgroup;
 static inline struct blkio_cgroup *
 cgroup_to_blkio_cgroup(struct cgroup *cgroup) { return NULL; }
+static inline struct blkio_cgroup *
+task_blkio_cgroup(struct task_struct *tsk) { return NULL; }
 
 static inline void blkiocg_add_blkio_group(struct blkio_cgroup *blkcg,
 		struct blkio_group *blkg, void *key, dev_t dev,

Index: linux-2.6/block/cfq-iosched.c
===================================================================

block/cfq-iosched.c +5 -6

@@ -1014,10 +1014,9 @@ void cfq_update_blkio_group_weight(void 
 	cfqg->needs_update = true;
 }
 
-static struct cfq_group *
-cfq_find_alloc_cfqg(struct cfq_data *cfqd, struct cgroup *cgroup, int create)
+static struct cfq_group * cfq_find_alloc_cfqg(struct cfq_data *cfqd,
+		struct blkio_cgroup *blkcg, int create)
 {
-	struct blkio_cgroup *blkcg = cgroup_to_blkio_cgroup(cgroup);
 	struct cfq_group *cfqg = NULL;
 	void *key = cfqd;
 	int i, j;

@@ -1079,12 +1078,12 @@ done:
  */
 static struct cfq_group *cfq_get_cfqg(struct cfq_data *cfqd, int create)
 {
-	struct cgroup *cgroup;
+	struct blkio_cgroup *blkcg;
 	struct cfq_group *cfqg = NULL;
 
 	rcu_read_lock();
-	cgroup = task_cgroup(current, blkio_subsys_id);
-	cfqg = cfq_find_alloc_cfqg(cfqd, cgroup, create);
+	blkcg = task_blkio_cgroup(current);
+	cfqg = cfq_find_alloc_cfqg(cfqd, blkcg, create);
 	if (!cfqg && create)
 		cfqg = &cfqd->root_group;
 	rcu_read_unlock();

Currentlly we first map the task to cgroup and then cgroup to blkio_cgroup. There is a more direct way to get to blkio_cgroup from task using task_subsys_state(). Use that.

The real reason for the fix is that it also avoids a race in generic cgroup code. During remount/umount rebind_subsystems() is called and it can do following with and rcu protection.

cgrp->subsys[i] = NULL;

That means if somebody got hold of cgroup under rcu and then it tried to do cgroup->subsys[] to get to blkio_cgroup, it would get NULL which is wrong. I was running into this race condition with ltp running on a upstream derived kernel and that lead to crash.

So ideally we should also fix cgroup generic code to wait for rcu grace period before setting pointer to NULL. Li Zefan is not very keen on introducing synchronize_wait() as he thinks it will slow down moun/remount/umount operations.

So for the time being atleast fix the kernel crash by taking a more direct route to blkio_cgroup.

One tester had reported a crash while running LTP on a derived kernel and with this fix crash is no more seen while the test has been running for over 6 days.

Signed-off-by Vivek Goyal <[email protected]>

---
 block/blk-cgroup.c   |    7 +++++++
 block/blk-cgroup.h   |    3 +++
 block/blk-throttle.c |    9 ++++-----
 block/cfq-iosched.c  |   11 +++++------
 4 files changed, 19 insertions(+), 11 deletions(-)

Index: linux-2.6/block/blk-throttle.c
===================================================================

block/blk-throttle.c +4 -5

@@ -160,9 +160,8 @@ static void throtl_put_tg(struct throtl_
 }
 
 static struct throtl_grp * throtl_find_alloc_tg(struct throtl_data *td,
-			struct cgroup *cgroup)
+			struct blkio_cgroup *blkcg)
 {
-	struct blkio_cgroup *blkcg = cgroup_to_blkio_cgroup(cgroup);
 	struct throtl_grp *tg = NULL;
 	void *key = td;
 	struct backing_dev_info *bdi = &td->queue->backing_dev_info;

@@ -229,12 +228,12 @@ done:
 
 static struct throtl_grp * throtl_get_tg(struct throtl_data *td)
 {
-	struct cgroup *cgroup;
 	struct throtl_grp *tg = NULL;
+	struct blkio_cgroup *blkcg;
 
 	rcu_read_lock();
-	cgroup = task_cgroup(current, blkio_subsys_id);
-	tg = throtl_find_alloc_tg(td, cgroup);
+	blkcg = task_blkio_cgroup(current);
+	tg = throtl_find_alloc_tg(td, blkcg);
 	if (!tg)
 		tg = &td->root_tg;
 	rcu_read_unlock();

Index: linux-2.6/block/blk-cgroup.c
===================================================================

block/blk-cgroup.c +7 -0

@@ -114,6 +114,13 @@ struct blkio_cgroup *cgroup_to_blkio_cgr
 }
 EXPORT_SYMBOL_GPL(cgroup_to_blkio_cgroup);
 
+struct blkio_cgroup *task_blkio_cgroup(struct task_struct *tsk)
+{
+	return container_of(task_subsys_state(tsk, blkio_subsys_id),
+			    struct blkio_cgroup, css);
+}
+EXPORT_SYMBOL_GPL(task_blkio_cgroup);
+
 static inline void
 blkio_update_group_weight(struct blkio_group *blkg, unsigned int weight)
 {

Index: linux-2.6/block/blk-cgroup.h
===================================================================

block/blk-cgroup.h +3 -0

@@ -291,6 +291,7 @@ static inline void blkiocg_set_start_emp
 #if defined(CONFIG_BLK_CGROUP) || defined(CONFIG_BLK_CGROUP_MODULE)
 extern struct blkio_cgroup blkio_root_cgroup;
 extern struct blkio_cgroup *cgroup_to_blkio_cgroup(struct cgroup *cgroup);
+extern struct blkio_cgroup *task_blkio_cgroup(struct task_struct *tsk);
 extern void blkiocg_add_blkio_group(struct blkio_cgroup *blkcg,
 	struct blkio_group *blkg, void *key, dev_t dev,
 	enum blkio_policy_id plid);

@@ -314,6 +315,8 @@ void blkiocg_update_io_remove_stats(stru
 struct cgroup;
 static inline struct blkio_cgroup *
 cgroup_to_blkio_cgroup(struct cgroup *cgroup) { return NULL; }
+static inline struct blkio_cgroup *
+task_blkio_cgroup(struct task_struct *tsk) { return NULL; }
 
 static inline void blkiocg_add_blkio_group(struct blkio_cgroup *blkcg,
 		struct blkio_group *blkg, void *key, dev_t dev,

Index: linux-2.6/block/cfq-iosched.c
===================================================================

block/cfq-iosched.c +5 -6

@@ -1014,10 +1014,9 @@ void cfq_update_blkio_group_weight(void 
 	cfqg->needs_update = true;
 }
 
-static struct cfq_group *
-cfq_find_alloc_cfqg(struct cfq_data *cfqd, struct cgroup *cgroup, int create)
+static struct cfq_group * cfq_find_alloc_cfqg(struct cfq_data *cfqd,
+		struct blkio_cgroup *blkcg, int create)
 {
-	struct blkio_cgroup *blkcg = cgroup_to_blkio_cgroup(cgroup);
 	struct cfq_group *cfqg = NULL;
 	void *key = cfqd;
 	int i, j;

@@ -1079,12 +1078,12 @@ done:
  */
 static struct cfq_group *cfq_get_cfqg(struct cfq_data *cfqd, int create)
 {
-	struct cgroup *cgroup;
+	struct blkio_cgroup *blkcg;
 	struct cfq_group *cfqg = NULL;
 
 	rcu_read_lock();
-	cgroup = task_cgroup(current, blkio_subsys_id);
-	cfqg = cfq_find_alloc_cfqg(cfqd, cgroup, create);
+	blkcg = task_blkio_cgroup(current);
+	cfqg = cfq_find_alloc_cfqg(cfqd, blkcg, create);
 	if (!cfqg && create)
 		cfqg = &cfqd->root_group;
 	rcu_read_unlock();

Quoted 4 lines · Vivek Goyal

Currentlly we first map the task to cgroup and then cgroup to blkio_cgroup. There is a more direct way to get to blkio_cgroup from task using task_subsys_state(). Use that.

Hi Jens,

Can you please apply this patch. I was hoping to get this fixed in 2.6.39. If it is too late, then we can queue it up for 2.6.40.