On Mon, Oct 07, 2013 at 07:46:34PM -0500, Eric Sandeen wrote:
> On 10/7/13 7:37 PM, Dave Chinner wrote:
> > On Sun, Oct 06, 2013 at 03:07:20PM -0500, Eric Sandeen wrote:
> >> This commit:
> >>
> >> f3edb66a mkfs.xfs: validate options for CRCs up front.
> >>
> >> seems to have inadvertently switched to 32bit project IDs by default:
> >>
> >> - int projid32bit;
> >> + int projid16bit;
> >> ...
> >> - projid32bit = 0;
> >> + projid16bit = 0;
> >>
> >> Switch back to 16 bit project ID by default, at least until we
> >> intentionally make that change.
> >>
> >> Signed-off-by: Eric Sandeen <sandeen@xxxxxxxxxx>
> >> ---
> >>
> >> diff --git a/mkfs/xfs_mkfs.c b/mkfs/xfs_mkfs.c
> >> index eafbed3..ebab09d 100644
> >> --- a/mkfs/xfs_mkfs.c
> >> +++ b/mkfs/xfs_mkfs.c
> >> @@ -965,7 +965,7 @@ main(
> >> textdomain(PACKAGE);
> >>
> >> attrversion = 2;
> >> - projid16bit = 0;
> >> + projid16bit = 1;
> >> blflag = bsflag = slflag = ssflag = lslflag = lssflag = 0;
> >> blocklog = blocksize = 0;
> >> sectorlog = lsectorlog = XFS_MIN_SECTORSIZE_LOG;
> >
> > And so:
> >
> > /* 32 bit project quota always on */
> > /* attr2 always on */
> > if (projid16bit == 1) {
> > fprintf(stderr,
> > _("32 bit Project IDs always enabled on CRC enabled filesytems\n"));
> > usage();
> > }
> >
> > That will prevent mkfs from making a CRC enabled filesystem.
>
> Well, that's quite fixable.
>
> > And, realistically, what's the problem with enabling 32 bit project
> > IDs by default for all new filesystems in 3.2.0? I'm of the opinion
> > we should just leave the code as it is because all kernels since
> > 2.6.37 support 32 bit projid and any kernel picking up
> > xfsprogs-3.2.0 when it is released will have a kernel that supports
> > it....
>
> Doesn't really bother me -
>
> But please send a man page patch. ;)
>
> And then we need to fix up xfstests... at least one fails as a result of
> this IIRC, I'll have to go look again.
See the link in my reply to the xfstests patch....
>
> I don't care if the defaults change but I'd prefer it to not
> have happened accidentally via
>
> "mkfs.xfs: validate options for CRCs up front" ;)
>
> It wasn't documented & didn't seem intentional, so it needs
> fixing one way or the other.
It was intentional (see above), just not documented ;)
Cheers,
Dave.
--
Dave Chinner
david@xxxxxxxxxxxxx
_______________________________________________
xfs mailing list
xfs@xxxxxxxxxxx
http://oss.sgi.com/mailman/listinfo/xfs
