On Sat, Nov 09, 2013 at 05:19:31AM +0000, Al Viro wrote: > On Sat, Nov 09, 2013 at 11:06:36AM +0800, Fengguang Wu wrote: > > Greetings, > > > > I got the below dmesg and the first bad commit is > > > > commit 84550b9356af50c7cbd6b6ce6e8fd06585eebf14 > > Author: Al Viro <viro@xxxxxxxxxxxxxxxxxx> > > Date: Sun Sep 29 22:06:07 2013 -0400 > > Got it... See if the following fixes the problem - it's > shrink_dcache_for_umount() needing to be careful after possible > run of d_genocide(); we are left with dentries that have zero > refcount and are not on any LRU. And yes, it's d_genocide() being > not nice, exactly because it relies upon shrink_dcache_for_umount() > coming right after it to sweep the garbage. So we'd better be > ready to deal with said garbage... > > Diff below fixes that junk here. Folded and re-pushed... Thanks! I can confirm that it fixed the bug. Tested-by: Fengguang Wu <fengguang.wu@xxxxxxxxx> > diff --git a/fs/dcache.c b/fs/dcache.c > index 50d4357..72c99b8 100644 > --- a/fs/dcache.c > +++ b/fs/dcache.c > @@ -1391,7 +1391,8 @@ static enum d_walk_ret umount_collect(void *_data, struct dentry *dentry) > * need to get the global LRU lock and do the > * LRU accounting. > */ > - d_lru_del(dentry); > + if (dentry->d_flags & DCACHE_LRU_LIST) > + d_lru_del(dentry); > d_shrink_add(dentry, &data->dispose); > data->found++; > ret = D_WALK_NORETRY; -- To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
