diff options
author | Oleg Nesterov <oleg@redhat.com> | 2013-07-26 11:12:56 -0400 |
---|---|---|
committer | Steven Rostedt <rostedt@goodmis.org> | 2013-07-31 12:16:31 -0400 |
commit | 776164c1faac4966ab14418bb0922e1820da1d19 (patch) | |
tree | c59ca51e2b86468841c2f44aa68073f27b7ca31b /fs/debugfs | |
parent | 8c4f3c3fa9681dc549cd35419b259496082fef8b (diff) |
debugfs: debugfs_remove_recursive() must not rely on list_empty(d_subdirs)
debugfs_remove_recursive() is wrong,
1. it wrongly assumes that !list_empty(d_subdirs) means that this
dir should be removed.
This is not that bad by itself, but:
2. if d_subdirs does not becomes empty after __debugfs_remove()
it gives up and silently fails, it doesn't even try to remove
other entries.
However ->d_subdirs can be non-empty because it still has the
already deleted !debugfs_positive() entries.
3. simple_release_fs() is called even if __debugfs_remove() fails.
Suppose we have
dir1/
dir2/
file2
file1
and someone opens dir1/dir2/file2.
Now, debugfs_remove_recursive(dir1/dir2) succeeds, and dir1/dir2 goes
away.
But debugfs_remove_recursive(dir1) silently fails and doesn't remove
this directory. Because it tries to delete (the already deleted)
dir1/dir2/file2 again and then fails due to "Avoid infinite loop"
logic.
Test-case:
#!/bin/sh
cd /sys/kernel/debug/tracing
echo 'p:probe/sigprocmask sigprocmask' >> kprobe_events
sleep 1000 < events/probe/sigprocmask/id &
echo -n >| kprobe_events
[ -d events/probe ] && echo "ERR!! failed to rm probe"
And after that it is not possible to create another probe entry.
With this patch debugfs_remove_recursive() skips !debugfs_positive()
files although this is not strictly needed. The most important change
is that it does not try to make ->d_subdirs empty, it simply scans
the whole list(s) recursively and removes as much as possible.
Link: http://lkml.kernel.org/r/20130726151256.GC19472@redhat.com
Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Diffstat (limited to 'fs/debugfs')
-rw-r--r-- | fs/debugfs/inode.c | 69 |
1 files changed, 22 insertions, 47 deletions
diff --git a/fs/debugfs/inode.c b/fs/debugfs/inode.c index 4888cb3fdef7..c7c83ff0f752 100644 --- a/fs/debugfs/inode.c +++ b/fs/debugfs/inode.c | |||
@@ -533,8 +533,7 @@ EXPORT_SYMBOL_GPL(debugfs_remove); | |||
533 | */ | 533 | */ |
534 | void debugfs_remove_recursive(struct dentry *dentry) | 534 | void debugfs_remove_recursive(struct dentry *dentry) |
535 | { | 535 | { |
536 | struct dentry *child; | 536 | struct dentry *child, *next, *parent; |
537 | struct dentry *parent; | ||
538 | 537 | ||
539 | if (IS_ERR_OR_NULL(dentry)) | 538 | if (IS_ERR_OR_NULL(dentry)) |
540 | return; | 539 | return; |
@@ -544,61 +543,37 @@ void debugfs_remove_recursive(struct dentry *dentry) | |||
544 | return; | 543 | return; |
545 | 544 | ||
546 | parent = dentry; | 545 | parent = dentry; |
546 | down: | ||
547 | mutex_lock(&parent->d_inode->i_mutex); | 547 | mutex_lock(&parent->d_inode->i_mutex); |
548 | list_for_each_entry_safe(child, next, &parent->d_subdirs, d_u.d_child) { | ||
549 | if (!debugfs_positive(child)) | ||
550 | continue; | ||
548 | 551 | ||
549 | while (1) { | 552 | /* perhaps simple_empty(child) makes more sense */ |
550 | /* | ||
551 | * When all dentries under "parent" has been removed, | ||
552 | * walk up the tree until we reach our starting point. | ||
553 | */ | ||
554 | if (list_empty(&parent->d_subdirs)) { | ||
555 | mutex_unlock(&parent->d_inode->i_mutex); | ||
556 | if (parent == dentry) | ||
557 | break; | ||
558 | parent = parent->d_parent; | ||
559 | mutex_lock(&parent->d_inode->i_mutex); | ||
560 | } | ||
561 | child = list_entry(parent->d_subdirs.next, struct dentry, | ||
562 | d_u.d_child); | ||
563 | next_sibling: | ||
564 | |||
565 | /* | ||
566 | * If "child" isn't empty, walk down the tree and | ||
567 | * remove all its descendants first. | ||
568 | */ | ||
569 | if (!list_empty(&child->d_subdirs)) { | 553 | if (!list_empty(&child->d_subdirs)) { |
570 | mutex_unlock(&parent->d_inode->i_mutex); | 554 | mutex_unlock(&parent->d_inode->i_mutex); |
571 | parent = child; | 555 | parent = child; |
572 | mutex_lock(&parent->d_inode->i_mutex); | 556 | goto down; |
573 | continue; | ||
574 | } | 557 | } |
575 | __debugfs_remove(child, parent); | 558 | up: |
576 | if (parent->d_subdirs.next == &child->d_u.d_child) { | 559 | if (!__debugfs_remove(child, parent)) |
577 | /* | 560 | simple_release_fs(&debugfs_mount, &debugfs_mount_count); |
578 | * Try the next sibling. | ||
579 | */ | ||
580 | if (child->d_u.d_child.next != &parent->d_subdirs) { | ||
581 | child = list_entry(child->d_u.d_child.next, | ||
582 | struct dentry, | ||
583 | d_u.d_child); | ||
584 | goto next_sibling; | ||
585 | } | ||
586 | |||
587 | /* | ||
588 | * Avoid infinite loop if we fail to remove | ||
589 | * one dentry. | ||
590 | */ | ||
591 | mutex_unlock(&parent->d_inode->i_mutex); | ||
592 | break; | ||
593 | } | ||
594 | simple_release_fs(&debugfs_mount, &debugfs_mount_count); | ||
595 | } | 561 | } |
596 | 562 | ||
597 | parent = dentry->d_parent; | 563 | mutex_unlock(&parent->d_inode->i_mutex); |
564 | child = parent; | ||
565 | parent = parent->d_parent; | ||
598 | mutex_lock(&parent->d_inode->i_mutex); | 566 | mutex_lock(&parent->d_inode->i_mutex); |
599 | __debugfs_remove(dentry, parent); | 567 | |
568 | if (child != dentry) { | ||
569 | next = list_entry(child->d_u.d_child.next, struct dentry, | ||
570 | d_u.d_child); | ||
571 | goto up; | ||
572 | } | ||
573 | |||
574 | if (!__debugfs_remove(child, parent)) | ||
575 | simple_release_fs(&debugfs_mount, &debugfs_mount_count); | ||
600 | mutex_unlock(&parent->d_inode->i_mutex); | 576 | mutex_unlock(&parent->d_inode->i_mutex); |
601 | simple_release_fs(&debugfs_mount, &debugfs_mount_count); | ||
602 | } | 577 | } |
603 | EXPORT_SYMBOL_GPL(debugfs_remove_recursive); | 578 | EXPORT_SYMBOL_GPL(debugfs_remove_recursive); |
604 | 579 | ||