UBIFS file-system corruption (missing inode) after power-cut on 4.14.96

Richard Weinberger richard at nod.at
Fri Apr 5 06:23:03 PDT 2019


Am Freitag, 5. April 2019, 15:19:31 CEST schrieb Lars Persson:
> On Thu, Apr 4, 2019 at 11:53 PM Richard Weinberger
> <richard at sigma-star.at> wrote:
> > Can you please run your tests again with this patch applied?
> > It needs further optimization, but we are on the right track.
> >
> > diff --git a/fs/ubifs/orphan.c b/fs/ubifs/orphan.c
> > index 2f1618f300fb..c15e484f73a4 100644
> > --- a/fs/ubifs/orphan.c
> > +++ b/fs/ubifs/orphan.c
> > @@ -691,8 +691,21 @@ static int do_kill_orphans(struct ubifs_info *c, struct ubifs_scan_leb *sleb,
> >                 n = (le32_to_cpu(orph->ch.len) - UBIFS_ORPH_NODE_SZ) >> 3;
> >                 for (i = 0; i < n; i++) {
> >                         union ubifs_key key1, key2;
> > +                       struct ubifs_ino_node *ino;
> > +
> > +                       ino = kmalloc(UBIFS_MAX_INO_NODE_SZ, GFP_NOFS);
> > +                       if (!ino)
> > +                               return -ENOMEM;
> >
> >                         inum = le64_to_cpu(orph->inos[i]);
> > +                       ino_key_init(c, &key1, inum);
> > +                       err = ubifs_tnc_lookup(c, &key1, ino);
> > +                       if (!err && ino->nlink) {
> > +                               kfree(ino);
> > +                               continue;
> > +                       }
> > +                       kfree(ino);
> > +
> >                         dbg_rcvry("deleting orphaned inode %lu",
> >                                   (unsigned long)inum);
> >
> 
> Thanks Richard. The power-cut test is OK now with this change. I run
> it with chk_fs=1 and it passes every time.

Cool, please let it run for some more time, just to be sure.
I will also do more tests on my side and optimize the patch.

Thanks,
//richard





More information about the linux-mtd mailing list