From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from lithops.sigma-star.at ([195.201.40.130])
 by bombadil.infradead.org with esmtps (Exim 4.90_1 #2 (Red Hat Linux))
 id 1fcVxI-0002Nf-C0
 for linux-mtd@lists.infradead.org; Mon, 09 Jul 2018 13:11:54 +0000
From: Richard Weinberger <richard@nod.at>
To: Adrian Hunter <adrian.hunter@intel.com>
Cc: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>,
 linux-mtd@lists.infradead.org
Subject: Re: UBIFS question: Power-cuts after ubifs_leb_unmap()
Date: Mon, 09 Jul 2018 15:11:39 +0200
Message-ID: <3206844.7NrM75eRyS@blindfold>
In-Reply-To: <6f88e3c1-4f27-cb59-5ab4-b3553d20ce27@intel.com>
References: <1756392.9dNXh6ho7o@blindfold>
 <6f88e3c1-4f27-cb59-5ab4-b3553d20ce27@intel.com>
MIME-Version: 1.0
Content-Transfer-Encoding: 7Bit
Content-Type: text/plain; charset="us-ascii"
List-Id: Linux MTD discussion mailing list <linux-mtd.lists.infradead.org>
List-Unsubscribe: <http://lists.infradead.org/mailman/options/linux-mtd>,
 <mailto:linux-mtd-request@lists.infradead.org?subject=unsubscribe>
List-Archive: <http://lists.infradead.org/pipermail/linux-mtd/>
List-Post: <mailto:linux-mtd@lists.infradead.org>
List-Help: <mailto:linux-mtd-request@lists.infradead.org?subject=help>
List-Subscribe: <http://lists.infradead.org/mailman/listinfo/linux-mtd>,
 <mailto:linux-mtd-request@lists.infradead.org?subject=subscribe>

Am Montag, 9. Juli 2018, 14:21:33 CEST schrieb Adrian Hunter:
> On 09/07/18 13:11, Richard Weinberger wrote:
> > Artem, Adrian,
> > 
> > While playing with a new UBI/UBIFS test framework I managed to hit this error, 
> > with lprops self-checks enabled:
> > 
> > [ 2412.268964] UBIFS error (ubi0:0 pid 708): scan_check_cb: bad accounting of 
> > LEB 11: free 0, dirty 118072 flags 0x1, should be free 126976, dirty 0
> > 
> > LEB 11 is unmapped but LPT still thinks that some data is used,
> > lp->free + lp >dirty < leb-size.
> > Even without lprobs self-checks, the same filesystem will later hit this 
> > assertion in ubifs_garbage_collect_leb():
> > 
> > ubifs_assert(!list_empty(&sleb->nodes));
> > 
> > The assert makes sure that the LEB actually contains nodes.
> > ubifs_garbage_collect_leb() handles the special case lp->free + lp->dirty == 
> > c->leb_size.
> > But not lp->free + lp->dirty < leb-size.
> > 
> > Now I'm not sure where to fix that, maybe you can remember some design 
> > decisions.
> > 1. Shall we massage ubifs_garbage_collect_leb() to deal with this special case 
> > too?
> > 2. Is it already a bug when this case happens?
> > 
> >>From reviewing the code, I think the said situation can arise when we face 
> > power-cut
> > in ubifs_garbage_collect_leb():
> > 
> > if (snod->type == UBIFS_IDX_NODE) {
> > 	...
> > } else {
> > 	...
> > 
> > 	err = ubifs_change_one_lp(c, lnum, c->leb_size, 0, 0, 0, 0);
> > 
> > 	...
> > 
> > 	err = ubifs_leb_unmap(c, lnum);
> > 
> > 	// POWER CUT
> > }
> > 
> > We mark the LEB as free and unmap it.
> > ubifs_change_one_lp() does not immediately write a new LPT, if we lose power 
> > right after ubifs_leb_unmap() it can happen that the LEB already got erased 
> > but the LPT has the old accounting information.
> 
> Doesn't GC copy the nodes into the journal, so after the journal is
> replayed, the old nodes will be dirtied and lprops will be correct again.

Yes, this is the theory. But the assert proves that something is not as we expect it. ;-\

Thanks,
//richard