Hi Nathan, David,
I had the vacation too.
David Chinner <[email protected]> wrote:
>Hmmm - Idon't think we should iput() before we wake up any pinned waiters.
>When we have a waiter on i_ipin_wait (called from xfs_iflush()), we have
>a thread sleeping with the inode locked.
>
>If we then call iput() and it drops the last reference, we can call back
>into the filesystem and start transactions. Those transactions will need
>to lock the inode. Hence I think the above can deadlock when racing against
>an inode flush.
>
>The code should probably read:
>
> if (dropped last pincount) {
> int need_iput = 0;
> struct inode *inode;
>
> spin_lock(i_flags_lock)
> if (!reclaimable) {
> if (!vp) {
> if (!(i_state & (NEW|CLEAR))) {
> inode = igrab(inode)
> if (inode) {
> need_iput = 1
> mark_inode_dirty_sync(inode)
> }
> }
> }
> }
> spin_unlock(i_flags_lock)
> wake_up(&ip->i_ipin_wait)
> if (need_iput)
> iput(inode);
> }
>
>to avoid this possible deadlock.
OK, I see your point. There is wait_event(in xfs_iunpin_wait) that should
be wake_up'ed(in xfs_iunpin), so deadlock can occur.
I'll update my patch and test it. Please wait for a few moments.
-
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/
[Index of Archives]
[Kernel Newbies]
[Netfilter]
[Bugtraq]
[Photo]
[Stuff]
[Gimp]
[Yosemite News]
[MIPS Linux]
[ARM Linux]
[Linux Security]
[Linux RAID]
[Video 4 Linux]
[Linux for the blind]
[Linux Resources]
