Skip to contentSkip to conversation

fsnotify: fix a crash due to invalid virtual address

PATCH Ashish Sangwan 1 participant 1 message Jun 24, 2015 VFS

lore
Share & tools
lore
#0 lore raw
Quoted 66 lines

So I was looking more into the issue today and fsnotify_destroy_marks() is fine in the end since it holds a reference to each mark (so it cannot be freed) and it also uses a special free_list list_head within mark for its temporary list so nobody removes any mark from that list.

Quoted 61 lines
fs/notify/mark.c +20 -12
index d90deaa..d83ec7d 100644
@@ -124,14 +124,6 @@ void fsnotify_destroy_mark_locked(struct fsnotify_mark *mark,
 
        spin_lock(&mark->lock);
 
-       /* something else already called this function on this mark */
-       if (!(mark->flags & FSNOTIFY_MARK_FLAG_ALIVE)) {
-               spin_unlock(&mark->lock);
-               return;
-       }
-
-       mark->flags &= ~FSNOTIFY_MARK_FLAG_ALIVE;
-
        if (mark->flags & FSNOTIFY_MARK_FLAG_INODE) {
                inode = mark->i.inode;
                fsnotify_destroy_inode_mark(mark);
@@ -188,7 +180,10 @@ void fsnotify_destroy_mark(struct fsnotify_mark *mark,
                           struct fsnotify_group *group)
 {
        mutex_lock_nested(&group->mark_mutex, SINGLE_DEPTH_NESTING);
-       fsnotify_destroy_mark_locked(mark, group);
+       if (mark->flags & FSNOTIFY_MARK_FLAG_ALIVE) {
+               mark->flags &= ~FSNOTIFY_MARK_FLAG_ALIVE;
+               fsnotify_destroy_mark_locked(mark, group);
+       }
        mutex_unlock(&group->mark_mutex);
 }
 
@@ -293,14 +288,27 @@ void fsnotify_clear_marks_by_group_flags(struct fsnotify_group *group,
                                         unsigned int flags)
 {
        struct fsnotify_mark *lmark, *mark;
-
+       LIST_HEAD(free_list);
+
        mutex_lock_nested(&group->mark_mutex, SINGLE_DEPTH_NESTING);
-       list_for_each_entry_safe(mark, lmark, &group->marks_list, g_list) {
+       /* Pass 1 : clear the alive flag and move to free list */
+       list_for_each_entry_safe(mark, lmark, &group->marks_list, g_list)
                if (mark->flags & flags) {
+                       /*
+                        * If the mark is present on group's mark list
+                        * it has to be alive.
+                        */
+                       WARN_ON(!(mark->flags & FSNOTIFY_MARK_FLAG_ALIVE));
+                       list_del_init(&mark->g_list);
+                       list_add(&mark->g_list, &free_list);
+                       mark->flags &= ~FSNOTIFY_MARK_FLAG_ALIVE;
+               }
+
+       /* Pass 2: remove mark */
+       list_for_each_entry_safe(mark, lmark, &free_list, g_list) {
                        fsnotify_get_mark(mark);
                        fsnotify_destroy_mark_locked(mark, group);
                        fsnotify_put_mark(mark);
-               }
        }
        mutex_unlock(&group->mark_mutex);
Quoted 8 lines

fs/notify/mark.c void fsnotify_clear_marks_by_group_flags(struct fsnotify_group *group, · line 314

Quoted 1 line
fs/notify/mark.c +0 -0
 }

So this should fix the problem as well. However the lifetime rules are so convoluted that I'm thinking how to make things simpler (especially to unify how clearing via inode/mount and via group works). When investigating the code I also found out there are possible issues with mark->inode getting cleared while we are delivering some event. So I'm looking into fixing all the problems.

Thanks for your help Jan. Looking forward to your patch(es).

Thanks,
Ashish

Quoted 8 lines

Honza

Signature · 2 lines
Jan Kara <[email protected]>
SUSE Labs, CR