On Sat, Nov 24, 2001 at 07:50:45AM +0100, Andrea Arcangeli wrote:
> On Sat, Nov 24, 2001 at 01:37:22AM -0500, Alexander Viro wrote:
> >
> >
> > On Sat, 24 Nov 2001, Alexander Viro wrote:
> >
> > > On Sat, 24 Nov 2001, Andrea Arcangeli wrote:
> > >
> > > > you are screwed because you were running a broken filesystem: it is its
> > > > own business to drop the inodes if it fails, all it needs to do is to
> > > > call invalidate_inodes(s) internally before returning from the read_super
> > > > in the failure case.
> > >
> > > Cute. Do you realize that _every_ fs would have to do that?
> >
> > Put it that way:
> > * if ->read_super() decides to fail, it should evict all inodes
> > it had put into icache.
> > * if ->put_super() does any iput(), it should take care to evict
> > that inode from icache.
>
> exactly.
>
> > IOW,
> > * if we do iput() while we are outside of (success of ->read_super(),
> > call of ->put_super()) - we want that inode to be evicted ASAP.
> >
> > Which is precisely what 2.4.15+patch does.
>
> and it's slower and overlay complex compared to the right fix:
>
> --- 2.4.15aa1/fs/ext2/super.c.~1~ Fri Nov 23 08:21:00 2001
> +++ 2.4.15aa1/fs/ext2/super.c Sat Nov 24 07:50:19 2001
> @@ -643,6 +643,7 @@
> printk(KERN_ERR "EXT2-fs: corrupt root inode, run e2fsck\n");
> } else
> printk(KERN_ERR "EXT2-fs: get root inode failed\n");
> + invalidate_inodes(sb);
> goto failed_mount2;
> }
> ext2_setup_super (sb, es, sb->s_flags & MS_RDONLY);
this one looks even better (and it doesn't need to propagate the fix to
the lowlevel):
--- 2.4.15aa1/fs/super.c.~1~ Fri Nov 23 08:21:01 2001
+++ 2.4.15aa1/fs/super.c Sat Nov 24 07:58:37 2001
@@ -470,6 +470,7 @@
return s;
out_fail:
+ invalidate_inodes(s);
s->s_dev = 0;
s->s_bdev = 0;
s->s_type = NULL;
Andrea
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Fri Nov 30 2001 - 21:00:16 EST