On Sun, 2005-07-03 at 11:34 +0100, Daniel Drake wrote:
> I reverted the patch you sent earlier
> (inotify_unmount_inodes-list-iteration-fix.diff) and applied the one you
> attached here (inotify_unmount_inodes-list-iteration-fix2.diff).
>
> The good news is that the hang is gone. The bad news is that you cured the
> hang by introducing an oops :(
)-: I have addressed the only things I can think off that could cause
the oops and below is the resulting patch. Could you please test it?
Thanks a lot!
Anton
--
Anton Altaparmakov <aia21 at cam.ac.uk> (replace at with @)
Unix Support, Computing Service, University of Cambridge, CB2 3QH, UK
Linux NTFS maintainer / IRC: #ntfs on irc.freenode.net
WWW: http://linux-ntfs.sf.net/ & http://www-stu.christs.cam.ac.uk/~aia21/
--- linux-2.6.13-rc1-mm1-vanilla/fs/inotify.c 2005-07-01 14:51:09.000000000
+0100
+++ linux-2.6.13-rc1-mm1/fs/inotify.c 2005-07-04 15:18:14.000000000 +0100
@@ -560,22 +560,45 @@ EXPORT_SYMBOL_GPL(inotify_get_cookie);
*/
void inotify_unmount_inodes(struct list_head *list)
{
- struct inode *inode, *next_i;
+ struct inode *inode, *next_i, *need_iput = NULL;
list_for_each_entry_safe(inode, next_i, list, i_sb_list) {
+ struct inode *need_iput_tmp;
struct inotify_watch *watch, *next_w;
struct list_head *watches;
/*
+ * If i_count is zero, the inode cannot have any watches and
+ * doing an __iget/iput with MS_ACTIVE clear would actually
+ * evict all inodes from icache which is unnecessarily violent.
+ */
+ if (!atomic_read(&inode->i_count))
+ continue;
+ /*
* We cannot __iget() an inode in state I_CLEAR or I_FREEING,
* which is fine becayse by that point the inode cannot have
* any associated watches.
*/
- if (inode->i_state & (I_CLEAR | I_FREEING))
+ if (inode->i_state & (I_CLEAR | I_FREEING | I_WILL_FREE))
continue;
+ need_iput_tmp = need_iput;
+ need_iput = NULL;
+
/* In case the remove_watch() drops a reference */
- __iget(inode);
+ if (inode != need_iput_tmp)
+ __iget(inode);
+ else
+ need_iput_tmp = NULL;
+
+ /* In case the dropping of a reference would nuke next_i. */
+ if ((&next_i->i_sb_list != list) &&
+ atomic_read(&next_i->i_count) &&
+ !(next_i->i_state & (I_CLEAR | I_FREEING |
+ I_WILL_FREE))) {
+ __iget(next_i);
+ need_iput = next_i;
+ }
/*
* We can safely drop inode_lock here because the per-sb list
@@ -584,6 +607,9 @@ void inotify_unmount_inodes(struct list_
*/
spin_unlock(&inode_lock);
+ if (need_iput_tmp)
+ iput(need_iput_tmp);
+
/* for each watch, send IN_UNMOUNT and then remove it */
down(&inode->inotify_sem);
watches = &inode->inotify_watches;
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
