[Buildroot] [PATCH v4] package/busybox: patch for glibc-2.31 and musl-1.2.0 compatibility

[Yann E. MORIN <yann.morin.1998@free.fr>]

Romain, Carlos, All,

On 2020-03-21 21:36 +0100, Romain Naour spake thusly:
> Le 15/03/2020 ? 12:59, unixmania at gmail.com a ?crit?:
> > From: Evgeniy Didin <Evgeniy.Didin@synopsys.com>
> > Current release of busybox 1_31_1 is failing to build with glibc-2.31
> > due to removal of stime() from glibc. It also fails to build with musl
> > 1.2.0 due to the direct use of __NR_clock_gettime.
> This patch is far too big IMHO ;-)

Well, if that's just backports of upstream patches, that's OKish.

> It would be great if you can split this patch an create a patch series.

There I agree. Of course, if upstream fixed both in a single commit, use
a single patch to backport that commit. Since this can be a bit
ambiguous, here's what I mean:

  [1/2] package/busybox: fix builg with glib 2.31
        Note one of the backpoted patch also fix a musl-related failure
        in addition ot fixing a glibc issue.

  [2/2] package/busybox: fix builg with musl 1.2.0

> > This commit adds six patches: one is a prerequisite for the subsequent
> > patches, three to fix 64-bit time_t issues, one to fix the direct use of
> > __NR_clock_gettime, which is not safe, and one that removes
> > stime() function calls.
> 
> Instead of backporting patch 0003, try to rebase patch 0008 over busybox 1.31.1.
> 
> But for the remaining patches, maybe you can squash then since the patch 0007
> remove the syscall wrappers around clock_gettime that was modified by patch
> 0004, 0005, 0006.

There I highly disagree: we want to backport uptream patches as-is, even
if they borked their fixes and needed three or four patches on top of
each others: when we eventually update, we'll want to be able to quickly
asses whther we can drop patches.

> It would be great to ask upstream for a new release 1.32 or backport those
> patches to 1.31.x.

That'd be ideal, or even if they could backport to 1.31.x.

> Another solution would be using the current busybox master as a last resort.

As a *last* *resort*, I'd be OK to take a bump to a sha1, yes, but that
would need to demonstrate that upstream is not ready by far to make a
release, and that backporting individual commits is too tedious...

Regards,
Yann E. MORIN.

> Best regards,
> Romain
> 
> > 
> > Fixes:
> >   http://autobuild.buildroot.net/results/f45f91aea6deee6699eabdfa618ac44873b8da51/
> > 
> > Signed-off-by: Evgeniy Didin <Evgeniy.Didin@synopsys.com>
> > Signed-off-by: Carlos Santos <unixmania@gmail.com>
> > Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
> > Cc: arc-buildroot at synopsys.com
> > ---
> > Supersedes: https://patchwork.ozlabs.org/patch/1249916/
> > ---
> > Changes v4:
> > - Rabase on the master branch
> > - Add SOB lines to patches
> > ---
> >  ...-overhead-of-single-parameter-bb_err.patch | 7637 +++++++++++++++++
> >  ...-Use-64-prefix-syscall-if-we-have-to.patch |   59 +
> >  ...-Use-64-prefix-syscall-if-we-have-to.patch |   48 +
> >  ...-Use-64-prefix-syscall-if-we-have-to.patch |   48 +
> >  ...rappers-around-clock_gettime-closes-.patch |  132 +
> >  .../0008-Remove-stime-function-calls.patch    |   95 +
> >  6 files changed, 8019 insertions(+)
> >  create mode 100644 package/busybox/0003-libbb-reduce-the-overhead-of-single-parameter-bb_err.patch
> >  create mode 100644 package/busybox/0004-date-Use-64-prefix-syscall-if-we-have-to.patch
> >  create mode 100644 package/busybox/0005-time-Use-64-prefix-syscall-if-we-have-to.patch
> >  create mode 100644 package/busybox/0006-runsv-Use-64-prefix-syscall-if-we-have-to.patch
> >  create mode 100644 package/busybox/0007-Remove-syscall-wrappers-around-clock_gettime-closes-.patch
> >  create mode 100644 package/busybox/0008-Remove-stime-function-calls.patch
> > 
> _______________________________________________
> buildroot mailing list
> buildroot at busybox.net
> http://lists.busybox.net/mailman/listinfo/buildroot

-- 
.-----------------.--------------------.------------------.--------------------.
|  Yann E. MORIN  | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: |
| +33 662 376 056 | Software  Designer | \ / CAMPAIGN     |  ___               |
| +33 561 099 427 `------------.-------:  X  AGAINST      |  \e/  There is no  |
| http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL    |   v   conspiracy.  |
'------------------------------^-------^------------------^--------------------'