From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <linux-kernel-owner@vger.kernel.org>
Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand
        id S934534AbcIPLYd (ORCPT <rfc822;w@1wt.eu>);
        Fri, 16 Sep 2016 07:24:33 -0400
Received: from down.free-electrons.com ([37.187.137.238]:34747 "EHLO
        mail.free-electrons.com" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org
        with ESMTP id S934388AbcIPLYJ (ORCPT
        <rfc822;linux-kernel@vger.kernel.org>);
        Fri, 16 Sep 2016 07:24:09 -0400
Date: Fri, 16 Sep 2016 13:23:55 +0200
From: Boris Brezillon <boris.brezillon@free-electrons.com>
To: Richard Weinberger <richard@nod.at>
Cc: Artem Bityutskiy <dedekind1@gmail.com>,
        David Woodhouse <dwmw2@infradead.org>,
        Brian Norris <computersforpeace@gmail.com>,
        linux-mtd@lists.infradead.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH v2 11/17] UBI: simplify recover_peb() code
Message-ID: <20160916132355.22c354f8@bbrezillon>
In-Reply-To: <bfcc6841-b5a7-8e40-5a6b-9e64f42390d7@nod.at>
References: <1473087908-27862-1-git-send-email-boris.brezillon@free-electrons.com>
        <1473087908-27862-12-git-send-email-boris.brezillon@free-electrons.com>
        <bfcc6841-b5a7-8e40-5a6b-9e64f42390d7@nod.at>
X-Mailer: Claws Mail 3.13.2 (GTK+ 2.24.30; x86_64-pc-linux-gnu)
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org

On Fri, 16 Sep 2016 12:14:04 +0200
Richard Weinberger <richard@nod.at> wrote:

> Boris,
> 
> On 05.09.2016 17:05, Boris Brezillon wrote:
> > + * This function is called in case of a write failure and moves all good data
> > + * from the potentially bad physical eraseblock to a good physical eraseblock.
> > + * This function also writes the data which was not written due to the failure.
> > + * Returns 0 in case of success, and a negative error code in case of failure.
> > + * This function tries %UBI_IO_RETRIES before giving up.
> > + */
> > +static int recover_peb(struct ubi_device *ubi, int pnum, int vol_id, int lnum,
> > +		       const void *buf, int offset, int len)
> > +{
> > +	int err, idx = vol_id2idx(ubi, vol_id), tries;
> > +	struct ubi_volume *vol = ubi->volumes[idx];
> > +	struct ubi_vid_hdr *vid_hdr;
> > +
> > +	vid_hdr = ubi_zalloc_vid_hdr(ubi, GFP_NOFS);
> > +	if (!vid_hdr)
> > +		return -ENOMEM;
> > +
> > +	for (tries = 0; tries <= UBI_IO_RETRIES; tries++) {
> > +		err = try_recover_peb(vol, pnum, lnum, buf, offset, len,
> > +				      vid_hdr);
> > +		if (!err || err == -ENOSPC)
> > +			break;  
> 
> Why do you handle ENOSPC as fatal error? Since the loop is bound by UBI_IO_RETRIES
> IMHO we can retry also upon ENOSPC.

I was just trying to mimic the existing behavior: if ubi_wl_get_peb()
fails to return a free PEB it returns -ENOSPC, and the current
implementation does not retry in this case.

I also realize that we should not retry if the error happened when
reading from the source PEB.

Maybe we should have an extra 'bool *retry' parameter to let
recover_peb() know whether the operation should be retried or not.
What do you think?