From: Gao Xiang <hsiangkao@redhat.com>
To: Eric Sandeen <sandeen@sandeen.net>
Cc: linux-xfs@vger.kernel.org,
"Darrick J. Wong" <darrick.wong@oracle.com>,
Brian Foster <bfoster@redhat.com>,
Eric Sandeen <sandeen@redhat.com>,
Dave Chinner <david@fromorbit.com>
Subject: Re: [PATCH v6 3/3] mkfs: make use of xfs_validate_stripe_geometry()
Date: Fri, 19 Feb 2021 08:39:42 +0800 [thread overview]
Message-ID: <20210219003942.GA392963@xiangao.remote.csb> (raw)
In-Reply-To: <1f63b1b7-71ec-2a03-1053-58a1abd0088a@sandeen.net>
On Thu, Feb 18, 2021 at 10:38:17AM -0600, Eric Sandeen wrote:
> On 2/17/21 11:24 PM, Gao Xiang wrote:
>
...
> since we have this check already in xfs_validate_stripe_geometry, it seems best to
> keep using it there, and not copy it ... which I think you accomplish below.
>
> >> btw, do we have some range test about these variables? I could rearrange the code
> >> snippet, but I'm not sure if it could introduce some new potential regression as well...
> >>
> >> Thanks,
> >> Gao Xiang
> >
> > Or how about applying the following incremental patch, although the maximum dswidth
> > would be smaller I think, but considering libxfs_validate_stripe_geometry() accepts
> > dswidth in 64-bit bytes as well. I think that would be fine. Does that make sense?
> >
> > I've confirmed "# mkfs/mkfs.xfs -f -d su=4097,sw=1 /dev/loop0" now report:
> > stripe unit (4097) must be a multiple of the sector size (512)
> >
> > and xfs/191-input-validation passes now...
> >
> > diff --git a/mkfs/xfs_mkfs.c b/mkfs/xfs_mkfs.c
> > index f152d5c7..80405790 100644
> > --- a/mkfs/xfs_mkfs.c
> > +++ b/mkfs/xfs_mkfs.c
> > @@ -2361,20 +2361,24 @@ _("both data su and data sw options must be specified\n"));
> > usage();
> > }
>
> Just thinking through this... I think this is the right idea.
>
> > - dsunit = (int)BTOBBT(dsu);
> > - big_dswidth = (long long int)dsunit * dsw;
> > + big_dswidth = (long long int)dsu * dsw;
>
> dsu is in bytes; this would mean big_dswidth is now also in bytes...
> the original goal here, I think, is to not overflow the 32-bit superblock value
> for dswidth.
Yeah, agreed. Thanks for catching this.
>
> > if (big_dswidth > INT_MAX) {
> > fprintf(stderr,
> > _("data stripe width (%lld) is too large of a multiple of the data stripe unit (%d)\n"),
> > big_dswidth, dsunit);
>
> so this used to test big_dswidth in BB (sectors); but now it tests in bytes.
>
> Perhaps this should change to check and report sectors again:
>
> if (BTOBBT(big_dswidth) > INT_MAX) {
> fprintf(stderr,
> _("data stripe width (%lld) is too large of a multiple of the data stripe unit (%d)\n"),
> BTOBBT(big_dswidth), dsunit);
>
> I think the goal is to not overflow the 32-bit on-disk values, which would be
> easy to do with "dsw" specified as a /multiplier/ of "dsu"
>
> So I think that if we keep range checking the value in BB units, it will be
> OK.
>
> > usage();
> > }
> > - dswidth = big_dswidth;
> > - }
> >
> > - if (!libxfs_validate_stripe_geometry(NULL, BBTOB(dsunit), BBTOB(dswidth),
> > - cfg->sectorsize, false))
> > + if (!libxfs_validate_stripe_geometry(NULL, dsu, big_dswidth,
> > + cfg->sectorsize, false))
> > + usage();
> > +
> > + dsunit = BTOBBT(dsu);
> > + dswidth = BTOBBT(big_dswidth);
> > + } else if (!libxfs_validate_stripe_geometry(NULL, BBTOB(dsunit),
> > + BBTOB(dswidth), cfg->sectorsize, false)) {
> > usage();
> > + }
> Otherwise this looks reasonable to me; now it's basically:
>
> 1) If we got geometry in bytes, validate them directly
> 2) If we got geometry in BB, convert to bytes, and validate
> 3) If we got no geometry, validate the device-reported defaults
>
Ok, let me send the next version.
Thanks,
Gao Xiang
> Thanks,
> -Eric
>
> > /* If sunit & swidth were manually specified as 0, same as noalign */
> > if ((cli_opt_set(&dopts, D_SUNIT) || cli_opt_set(&dopts, D_SU)) &&
> >
>
next prev parent reply other threads:[~2021-02-19 0:41 UTC|newest]
Thread overview: 14+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-10-13 4:06 [PATCH v6 0/3] xfsprogs: consolidate stripe validation Gao Xiang
2020-10-13 4:06 ` [PATCH v6 1/3] libxfs: allow i18n to xfs printk Gao Xiang
2020-10-13 4:06 ` [PATCH v6 2/3] xfs: introduce xfs_validate_stripe_geometry() Gao Xiang
2020-10-13 4:06 ` [PATCH v6 3/3] mkfs: make use of xfs_validate_stripe_geometry() Gao Xiang
2020-10-13 13:44 ` Brian Foster
2020-10-14 16:20 ` Darrick J. Wong
2021-02-16 1:04 ` Eric Sandeen
2021-02-18 2:41 ` Gao Xiang
2021-02-18 5:24 ` Gao Xiang
2021-02-18 16:38 ` Eric Sandeen
2021-02-19 0:39 ` Gao Xiang [this message]
2021-02-19 1:37 ` [PATCH v7 " Gao Xiang
2021-02-24 0:10 ` Eric Sandeen
2021-02-24 0:55 ` Gao Xiang
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20210219003942.GA392963@xiangao.remote.csb \
--to=hsiangkao@redhat.com \
--cc=bfoster@redhat.com \
--cc=darrick.wong@oracle.com \
--cc=david@fromorbit.com \
--cc=linux-xfs@vger.kernel.org \
--cc=sandeen@redhat.com \
--cc=sandeen@sandeen.net \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).