bug#49516: closed ([core-updates] glibc-2.31 patches fail to apply)

[GNU bug Tracking System]

Your message dated Wed, 21 Jul 2021 16:04:40 -0700
with message-id <87zguf6zg7.fsf@gmail.com>
and subject line Re: bug#49516: [core-updates] glibc-2.31 patches fail to apply
has caused the debbugs.gnu.org bug report #49516,
regarding [core-updates] glibc-2.31 patches fail to apply
to be marked as done.

(If you believe you have received this mail in error, please contact
help-debbugs@gnu.org.)


-- 
49516: http://debbugs.gnu.org/cgi/bugreport.cgi?bug=49516
GNU Bug Tracking System
Contact help-debbugs@gnu.org with problems

--- Begin Message --- Subject: [core-updates] glibc-2.31 patches fail to apply Date: Sat, 10 Jul 2021 17:29:19 -0700 User-agent: Gnus/5.13 (Gnus v5.13) Emacs/27.1 (gnu/linux)

Hi,

On core-updates, the glibc-2.31 patch no longer apply cleanly.  This
causes glibc-2.31 to fail to build.  This causes downstream problems;
for example, building a Guix System from bare-bones.tmpl fails because
%default-locale-libcs still contains glibc-2.31.

Example:

--8<---------------cut here---------------start------------->8---
source is at 'glibc-2.31'
applying 
'/gnu/store/5icxvnlbac76mm49g6sq7vzznjxmn2s6-glibc-ldd-powerpc.patch'...
applying '/gnu/store/v1h2i4i5xmrs9d4c44w5wshv5zyszb8k-glibc-ldd-x86_64.patch'...
applying '/gnu/store/bpds9cz27ghqf64y8xz4vs35p7275d6n-glibc-dl-cache.patch'...
applying 
'/gnu/store/lgrlsr3qnxxvic3y472qwybv5wbyabm6-glibc-hidden-visibility-ldconfig.patch'...
applying 
'/gnu/store/mvq0q2f211bxb4syfxvng9kgdxzkr5f3-glibc-versioned-locpath.patch'...
applying 
'/gnu/store/sz5nmndsway8bq7283ihdgvmm3xb14l8-glibc-allow-kernel-2.6.32.patch'...
applying 
'/gnu/store/vh29xqy3daavjpi0ikpmqzfczzpbscix-glibc-reinstate-prlimit64-fallback.patch'...
applying 
'/gnu/store/rnqkir22908x6z3i1mk4phyvskz15qc4-glibc-supported-locales.patch'...
applying 
'/gnu/store/svva3cym2n04d2x3bpi4rs6qpnw0m162-glibc-hurd-clock_t_centiseconds.patch'...
applying 
'/gnu/store/45ra3k89b3kisarlgvmijy507k6m12ll-glibc-hurd-clock_gettime_monotonic.patch'...
Backtrace:
           5 (primitive-load "/gnu/store/bablv90wm5xkd0vinal0gsifm0l…")
In ice-9/eval.scm:
    619:8  4 (_ #(#(#<directory (guile-user) 7ffff6f53c80> "gli…") #))
In ice-9/boot-9.scm:
    142:2  3 (dynamic-wind #<procedure 7ffff6f16b00 at ice-9/eval.s…> …)
In ice-9/eval.scm:
    619:8  2 (_ #(#(#<directory (guile-user) 7ffff6f53c80>)))
In srfi/srfi-1.scm:
    634:9  1 (for-each #<procedure apply-patch (a)> _)
In guix/build/utils.scm:
    721:6  0 (invoke "/gnu/store/i5md0v46jp7ahap0vbj4hwyh6lxsny3g-p…" …)

guix/build/utils.scm:721:6: In procedure invoke:
ERROR:
  1. &invoke-error:
      program: 
"/gnu/store/i5md0v46jp7ahap0vbj4hwyh6lxsny3g-patch-2.7.6/bin/patch"
      arguments: ("--force" "--no-backup-if-mismatch" "-p1" "--input" 
"/gnu/store/45ra3k89b3kisarlgvmijy507k6m12ll-glibc-hurd-clock_gettime_monotonic.patch")
      exit-status: 1
      term-signal: #f
      stop-signal: #f
builder for `/gnu/store/jlpanpgns01sv3jr63c13fn133pj7ik5-glibc-2.31.tar.xz.drv' 
failed with exit code 1
--8<---------------cut here---------------end--------------->8---

Judging by the git log for gnu/system/locale.scm, it seems that we
sometimes keep old versions of glibc in %default-locale-libs, in order
to make it easier for people to upgrade across glibc version changes.
Therefore, we probably can't just drop glibc-2.31.

The commit which introduced the failure is
87961fc965b96ac0c7a5909ac2faab2d023b5339.  The problem does not occur on
the prior commit.  For glibc-2.31, commit
87961fc965b96ac0c7a5909ac2faab2d023b5339 effectively removed the
glibc-hurd-signal-sa-siginfo.patch and modified
libc-hurd-clock_gettime_monotonic.patch.  We should probably retain
2.31-specific versions of these patches so that users can continue to
build glibc-2.31 until we decide to remove it entirely.

I have attached a patch which accomplishes this.

-- 
Chris

0001-gnu-glibc-2.31-Restore-patches.patch
Description: Text Data

signature.asc
Description: PGP signature

--- End Message ---

--- Begin Message --- Subject: Re: bug#49516: [core-updates] glibc-2.31 patches fail to apply Date: Wed, 21 Jul 2021 16:04:40 -0700 User-agent: Gnus/5.13 (Gnus v5.13) Emacs/27.1 (gnu/linux)

Ludovic Courtès <ludo@gnu.org> writes:

> Hi!
>
> Chris Marusich <cmmarusich@gmail.com> skribis:
>
>> As an aside, when do we remove old versions of glibc?
>
> Good question.  I’d say it’s enough to keep 3 versions in total.
> Currently the main (only?) use case for these is when computing locale
> data via the ‘locale-libcs’ field of operating system definitions.
>
> Thoughts?

3 seems fine to me.  I suppose if they are particularly important for
some reason, an old version could be moved to the Guix-Past channel, but
I guess in most cases we would just remove them and move forward.

>> From ef169adea6f9ca971e22845b839511b015cbc76c Mon Sep 17 00:00:00 2001
>> From: Chris Marusich <cmmarusich@gmail.com>
>> Date: Sat, 10 Jul 2021 16:49:49 -0700
>> Subject: [PATCH] gnu: glibc-2.31: Restore patches.
>>
>> Commit 87961fc965b96ac0c7a5909ac2faab2d023b5339 inadvertently modified the
>> patch set for glibc-2.31.  This change restores the original patch set.
>>
>> Fixes: <https://bugs.gnu.org/49516>.
>>
>> * gnu/packages/base.scm (glibc-2.31) [source]: Use the same patches as glibc,
>> but replace glibc-hurd-clock_gettime_monotonic.patch with
>> glibc-2.31-hurd-clock_gettime_monotonic.patch, and add
>> glibc-hurd-signal-sa-siginfo.patch.
>> * gnu/packages/patches/glibc-2.31-hurd-clock_gettime_monotonic.patch: Add it.
>> * gnu/packages/patches/glibc-hurd-signal-sa-siginfo.patch: Add it.
>> * gnu/local.mk (dist_patch_DATA): Adjust accordingly.
>
> LGTM, thanks!
>
> Ludo’.

Thank you for taking a look!  I've committed this in
93a5e89008af440655527d03d62d4726683a89ac.  I'll close this report now.


-- 
Chris

signature.asc
Description: PGP signature

--- End Message ---