Re: [PATCH 1/1] e2fsprogs: 1.42.9 -> 1.43-WIP

[Patrick Ohly <patrick.ohly@intel.com>]

On Fri, 2016-01-15 at 11:15 +0000, Richard Purdie wrote:
> On Thu, 2016-01-14 at 18:05 -0800, Robert Yang wrote:
> > Upgrade to 1.43-WIP to make "mke2fs -d" support xattr, so that the
> > layer
> > which requires xattr such as meta-selinux can populate images easily.
> > 
> > * Remove the following patches since they are alredy in the source.
> >   0001-e2fsprogs-fix-cross-compilation-problem.patch
> >   0001-libext2fs-fix-potential-buffer-overflow-in-closefs.patch
> >   0001-mke2fs-add-the-ability-to-copy-files-from-a-given-di.patch
> >   0002-misc-create_inode.c-copy-files-recursively.patch
> >   0003-misc-create_inode.c-create-special-file.patch
> >   0004-misc-create_inode.c-create-symlink.patch
> >   0005-misc-create_inode.c-copy-regular-file.patch
> >   0006-misc-create_inode.c-create-directory.patch
> >   0007-misc-create_inode.c-set-owner-mode-time-for-the-inod.patch
> >   0008-mke2fs.c-add-an-option-d-root-directory.patch
> >   0009-misc-create_inode.c-handle-hardlinks.patch
> >   0010-debugfs-use-the-functions-in-misc-create_inode.c.patch
> >   0011-mke2fs.8.in-update-the-manual-for-the-d-option.patch
> >   0012-Fix-musl-build-failures.patch
> >   CVE-2015-0247.patch
> >   copy-in-create-hardlinks-with-the-correct-directory-.patch
> >   fix-icache.patch
> >   misc-mke2fs.c-return-error-when-failed-to-populate-fs.patch
> > 
> > * Remove cache_inode.patch since it is not needed any more
> > 
> > * Updated mkdir.patch and ptest.patch
> > 
> > * Add --enable-libblkid to EXTRA_OECONF since libblkid is not created
> > by
> >   default.
> > 
> > * Time of core-image-sato-sdk do_rootfs:
> >   - Before upgrade
> >     real    3m18.508s
> >     user    7m42.088s
> >     sys     1m1.984s
> > 
> >   - After upgrade
> >     real    3m21.552s
> >     user    7m38.496s
> >     sys     1m0.644s
> > 
> >    The are nearly the same
> > 
> > * The "fsck -fn" shows the image is OK, and also can boot.
> > 
> > [YOCTO #8622]
> > 
> > Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
> [...]
> 
> > -SRC_URI[md5sum] = "3f8e41e63b432ba114b33f58674563f7"
> > -SRC_URI[sha256sum] =
> > "2f92ac06e92fa00f2ada3ee67dad012d74d685537527ad1241d82f2d041f2802"
> > +SRCREV = "0f26747167cc9d82df849b0aad387bf824f04544"
> > +PV = "1.43-WIP+git${SRCPV}"
> 
> What happens when 1.43 is released? 1.43 < 1.43-WIP so we'd have to
> bump PE.
> 
> Can we just call this 1.42+1.43-git${SRCPV}?

However, that is not a more recent version than the one that was in
OE-core before:

$ dpkg --compare-versions 1.42+1.43 gt 1.42.9 && echo greater || echo less
less

As a result, the version upgrade (which is in OE-core master now) became
a downgrade as far as distros with stable package feeds are concerned,
didn't it?

The version for OE-core could have been: 1.42.9+1.43-git${SRCPV}

However, I've had a "1.42.9+git${SRCPV}" in meta-intel-iot-security for
a while now, and 1.42.9+1.43-git${SRCPV} is considered older than that
because of the embedded 1.43. While I understand that external layers
should not be something that OE-core needs to be concerned about too
much, some consistency still helps.

I believe the "1.42.9+git${SRCPV}" string goes back to Ross, so I'd
assume that it is not too unusual. Can we perhaps use "1.42.9+git
${SRCPV}" also in OE-core? Then if I'm not mistaken, the magic behind
${SRCPV} will ensure that the final version number will be higher.

-- 
Best Regards, Patrick Ohly

The content of this message is my personal opinion only and although
I am an employee of Intel, the statements I make here in no way
represent Intel's position on the issue, nor am I authorized to speak
on behalf of Intel on this matter.