(02/12/10 12:14), Andrew Morton wrote: > On Thu, 11 Feb 2010 21:44:17 -0500 Valdis.Kletnieks@xxxxxx wrote: > >> On Wed, 10 Feb 2010 16:17:41 PST, akpm@xxxxxxxxxxxxxxxxxxxx said: >>> The mm-of-the-moment snapshot 2010-02-10-16-17 has been uploaded to >>> >>> http://userweb.kernel.org/~akpm/mmotm/ >> >> Hit this at shutdown. ext4 filesystems. Brought it up to single-user >> mode, then 'shutdown -h now'. >> >> [ 58.311786] ------------[ cut here ]------------ >> [ 58.312261] kernel BUG at fs/dcache.c:677! ... > Yeah, Sachin reported that against linux-next too (I think - need to > check the exact line numbers): > http://permalink.gmane.org/gmane.linux.file-systems/38148 > > It seems to have been met with shrugs thus far. According to the Sachin's report: > While running hackbench against today's next-20100208 (9056d31..) > on a s390 box, came across following BUG. ... > Did not observe this with next-20100205. Also there doesn't seem to be > any patches applied to this code after the last release. So not > sure what could have caused this. Corresponding code is : So the bug seems introduced between next-20100205 and next-20100208. Isn't this change missing the equivalent of "dput(old)"? In patch-v2.6.33-rc7-next-20100208: > diff --git a/fs/namei.c b/fs/namei.c > index 94a5e60..a056dad 100644 > --- a/fs/namei.c > +++ b/fs/namei.c > @@ -689,33 +689,17 @@ static __always_inline void follow_dotdot(struct nameidata *nd) ... > - spin_lock(&dcache_lock); > if (nd->path.dentry != nd->path.mnt->mnt_root) { > - nd->path.dentry = dget(nd->path.dentry->d_parent); > - spin_unlock(&dcache_lock); > - dput(old); > + /* rare case of legitimate dget_parent()... */ > + nd->path.dentry = dget_parent(nd->path.dentry); > break; > } -- Jun'ichi Nomura, NEC Corporation -- To unsubscribe from this list: send the line "unsubscribe linux-next" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
