fanotify: Avoid softlockups when reading many events
Subsystems (3)
Jan Kara
maintainerPATCH 1/1 fanotify: Avoid softlockups when reading many events
When user provides large buffer for events and there are lots of events available, we can try to copy them all to userspace without scheduling which can softlockup the kernel (furthermore exacerbated by the contention on notification_lock). Add a scheduling point after copying each event.
Note that usually the real underlying problem is the cost of fanotify event merging and the resulting contention on notification_lock but this is a cheap way to somewhat reduce the problem until we can properly address that.
--- fs/notify/fanotify/fanotify_user.c | 5 +++++ 1 file changed, 5 insertions(+)
This is a quick mending we can do immediately and is probably a good idea nevertheless... I'll queue it up if Amir agrees.
fs/notify/fanotify/fanotify_user.c +5 -0
@@ -412,6 +412,11 @@ static ssize_t fanotify_read(struct file *file, char __user *buf,add_wait_queue (&group->notification_waitq, &wait);while (1 ) { +/* +* User can supply arbitrarily large buffer. Avoid softlockups +* in case there are lots of available events. +*/ +cond_resched (); event =get_one_event (group, count);if (IS_ERR (event)) { ret =PTR_ERR (event);
Signature · 1 line
2.16.4
Quoted 12 lines · Jan Kara
When user provides large buffer for events and there are lots of events available, we can try to copy them all to userspace without scheduling which can softlockup the kernel (furthermore exacerbated by the contention on
notification_lock). Add a scheduling point after copying each event.Note that usually the real underlying problem is the cost of fanotify event merging and the resulting contention on
notification_lockbut this is a cheap way to somewhat reduce the problem until we can properly address that.
Quoted 9 lines
Reported-by Francesco Ruggeri <[email protected]>Signed-off-by Jan Kara <[email protected]>--- fs/notify/fanotify/fanotify_user.c | 5 +++++ 1 file changed, 5 insertions(+)This is a quick mending we can do immediately and is probably a good idea nevertheless... I'll queue it up if Amir agrees.
Sure. fine by me.
Maybe add a lore link to the issue report.
Quoted 19 lines
fs/notify/fanotify/fanotify_user.c +5
-1@@ -412,6 +412,11 @@ static ssize_t fanotify_read(struct file *file, char __user *buf,add_wait_queue (&group->notification_waitq, &wait);while (1 ) { +/* +* User can supply arbitrarily large buffer. Avoid softlockups +* in case there are lots of available events. +*/ +cond_resched (); event =get_one_event (group, count);if (IS_ERR (event)) { ret =PTR_ERR (event); --2.16.4
Jan Kara
maintainerQuoted 19 lines · Amir Goldstein
Quoted 18 lines · Jan Kara
When user provides large buffer for events and there are lots of events available, we can try to copy them all to userspace without scheduling which can softlockup the kernel (furthermore exacerbated by the contention on
notification_lock). Add a scheduling point after copying each event.Note that usually the real underlying problem is the cost of fanotify event merging and the resulting contention on
notification_lockbut this is a cheap way to somewhat reduce the problem until we can properly address that.Reported-by Francesco Ruggeri <[email protected]>Signed-off-by Jan Kara <[email protected]>--- fs/notify/fanotify/fanotify_user.c | 5 +++++ 1 file changed, 5 insertions(+)
Quoted 8 lines
Quoted 3 lines
This is a quick mending we can do immediately and is probably a good idea nevertheless... I'll queue it up if Amir agrees.
Sure. fine by me.
Maybe add a lore link to the issue report.Reviewed-by Amir Goldstein <[email protected]>
Thanks. I've added the link and pushed out the patch to my tree.
Honza
Quoted 20 lines
Quoted 20 lines
fs/notify/fanotify/fanotify_user.c +5
-1@@ -412,6 +412,11 @@ static ssize_t fanotify_read(struct file *file, char __user *buf,add_wait_queue (&group->notification_waitq, &wait);while (1 ) { +/* +* User can supply arbitrarily large buffer. Avoid softlockups +* in case there are lots of available events. +*/ +cond_resched (); event =get_one_event (group, count);if (IS_ERR (event)) { ret =PTR_ERR (event); --2.16.4
Signature · 2 lines
Jan Kara <[email protected]> SUSE Labs, CR