From: Bengt Richter <bokr@bokr.com>
To: "Ludovic Courtès" <ludo@gnu.org>
Cc: 36747@debbugs.gnu.org
Subject: bug#36747: Official MesCC bootstrap binaries differ from my locally built ones
Date: Thu, 29 Aug 2019 15:28:14 -0700 [thread overview]
Message-ID: <20190829222814.GA2437@PhantoNv4ArchGx.localdomain> (raw)
In-Reply-To: <875zmjrlrc.fsf@gnu.org>
On +2019-08-27 11:38:31 +0200, Ludovic Courtès wrote:
> Hi Mark,
>
> Mark H Weaver <mhw@netris.org> skribis:
>
> > Ludovic Courtès <ludo@gnu.org> writes:
> >
> >> Mark H Weaver <mhw@netris.org> skribis:
>
> [...]
>
> >>> commit 82eaac49ac983f28768d6623d802f41cbd7f779b
> >>> Author: Mark H Weaver <mhw@netris.org>
> >>> Date: Thu Aug 15 16:44:36 2019 -0400
> >>>
> >>> gnu: bash: Unconditionally configure PGRP_PIPE for *-linux systems.
> >>>
> >>> * gnu/packages/patches/bash-linux-pgrp-pipe.patch: New file.
> >>> * gnu/local.mk (dist_patch_DATA): Add it.
> >>> * gnu/packages/bash.scm (bash)[source]: Add the patch.
>
> [...]
>
> >> OK. The Bash change means we’re rebuilding from scratch on
> >> architectures, not just x86. So I’ll probably ungraft my Ghostscript
> >> fix (466ff55c72959ba1499ce3ec69f534b3038eb30b) while we’re at it.
> >
> > Hmm, good point. Perhaps we should postpone the Bash fix until the next
> > core-updates cycle. The problem was quite severe in bash-4.4, which
> > builds incorrectly on linux-5.0 or later, but the issue in bash-5.0 is
> > far less likely to cause problems in practice: it will build differently
> > on linux-2.3 or earlier.
> >
> > What do you think?
>
> Your call: if you think this Bash fix can be delayed without causing
> problems, then please revert it and we’ll apply it on the next cycle.
> That would allow us to merge ‘core-updates’ more quickly for sure!
>
> Thanks,
> Ludo’.
>
>
>
Hi,
(Just in case this might be relevant, sorry if not :)
I noticed that the guix-built version of bash is not position-independent,
and also differences in built-for versions, wondered why:
$ type bash which
bash is /home/bokr/.guix-profile/bin/bash
which is /home/bokr/.guix-profile/bin/which
$ which -a bash
/home/bokr/.guix-profile/bin/bash
/usr/bin/bash
$
$ which -a bash|xargs readlink -f
/gnu/store/qn1ax1fkj16x280m1rv7mcimfmn9l2pf-bash-4.4.23/bin/bash
/usr/bin/bash
$
$ which -a bash|xargs readlink -f|xargs file
/gnu/store/qn1ax1fkj16x280m1rv7mcimfmn9l2pf-bash-4.4.23/bin/bash:
ELF 64-bit LSB executable, x86-64, version 1 (SYSV), dynamically linked,
^^^^^^^^^^^^^^
interpreter /gnu/store/h90vnqw0nwd0hhm1l5dgxsdrigddfmq4-glibc-2.28/lib/ld-linux-x86-64.so.2,
for GNU/Linux 2.6.32, not stripped
^^^^^^
/usr/bin/bash:
ELF 64-bit LSB pie executable, x86-64, version 1 (SYSV), dynamically linked,
^^^^^^^^^^^^^^^^^^
interpreter /lib64/ld-linux-x86-64.so.2, BuildID[sha1]=21a51cb5f7d727370e4d8099d283d7cd20222571,
for GNU/Linux 3.2.0, stripped
^^^^^
If you substitute guile for bash in the above, it shows the same differences.
Could something need position-independence only under special circumstances
and silently give up if not repositionable, causing mystery errors?
Also, I am in the foreign (ArchLinux) host environment twilight zone
of mixed /usr vs ~/.guix-profile name searches and wonder if
there are hidden /usr dependencies baked statically into some /gnu/store executables
that are causing me problems.
exec is a bash built-in, and I suppose that is usually doesn't matter
if a guile script is launched via
#!/usr/bin/bash
exec guile ...
but I have a hunch that sometimes it does matter, because editing scripts
to do hash-bang with
#!/home/bokr/.guix-profile/bin/bash
has made a locale error ("guile: warning: failed to install locale")
go away and the script execute successfully, vs not executing beyond the error.
Also, what about the shell that runs first, when I log in?
Should I set SHELL=/gnu/... ? And/or reconfigure to get login/systemd
to launch /gnu/.../bash ?
I had hoped a binary install could give me clean separation of
guix pull goodies vs pacman -Syu (ArchLinux's package installer/updater)
stuff, but it seems difficult to control the env name environment that
both guix and non-guix use.
My setup, fwiw:
guix describe:
Generation 11 Aug 25 2019 20:58:43 (current)
guix a707484
repository URL: https://git.savannah.gnu.org/git/guix.git
branch: master
commit: a707484d64e7e46f8cb8401c660fbb6eb77ab9c6
uname -a:
Linux PhantoNv4ArchGx 5.2.9-arch1-1-ARCH #1 SMP PREEMPT Fri Aug 16 11:29:43 UTC 2019 x86_64 GNU/Linux
Well, sorry if these observations are explained in a FAQ.
I note that there is no -pie in the guix version of this:
[15:14 ~/bs]$ which -a gcc|while read line;do { echo -e "$line: ";$line -v --version|& cat -n|egrep -on -e ' -pie ' ; } ;done
/home/bokr/.guix-profile/bin/gcc:
/usr/bin/gcc:
32: -pie
34: -pie
[15:14 ~/bs]$
HTH in some way, PMJI if not.
Regards,
Bengt Richter
next prev parent reply other threads:[~2019-08-29 22:30 UTC|newest]
Thread overview: 58+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-07-20 22:43 bug#36747: Official MesCC bootstrap binaries differ from my locally built ones Mark H Weaver
2019-07-21 13:34 ` Jan Nieuwenhuizen
2019-07-22 0:56 ` Mark H Weaver
2019-07-22 6:18 ` Jan Nieuwenhuizen
2019-07-22 6:26 ` Jan Nieuwenhuizen
2019-07-22 8:26 ` Jan Nieuwenhuizen
2019-07-22 8:31 ` Mark H Weaver
2019-07-22 17:41 ` Jan Nieuwenhuizen
2019-07-23 5:42 ` Mark H Weaver
2019-07-23 6:28 ` Jan Nieuwenhuizen
2019-08-12 0:21 ` Mark H Weaver
2019-08-12 4:11 ` Mark H Weaver
2019-07-23 10:03 ` Ludovic Courtès
2019-08-12 7:08 ` Mark H Weaver
2019-08-12 9:01 ` Jan Nieuwenhuizen
2019-08-13 6:42 ` Mark H Weaver
2019-08-13 10:17 ` Jan Nieuwenhuizen
2019-08-14 15:03 ` Marius Bakke
2019-08-14 17:29 ` Marius Bakke
2019-08-14 18:35 ` Mark H Weaver
2019-08-14 18:43 ` Mark H Weaver
2019-08-14 19:56 ` Marius Bakke
2019-08-14 20:43 ` Mark H Weaver
2019-08-15 19:44 ` Mark H Weaver
2019-08-15 21:19 ` Marius Bakke
2019-08-15 23:16 ` Mark H Weaver
2019-08-15 20:56 ` Mark H Weaver
2019-08-16 7:42 ` Mark H Weaver
2019-08-17 16:49 ` Mark H Weaver
2019-08-16 10:49 ` Ludovic Courtès
2019-08-16 16:59 ` Mark H Weaver
2019-08-17 21:38 ` Ludovic Courtès
2019-08-18 1:17 ` Mark H Weaver
2019-08-18 9:26 ` Ludovic Courtès
2019-08-20 18:40 ` Mark H Weaver
2019-08-21 20:15 ` Mark H Weaver
2019-08-21 21:38 ` Ludovic Courtès
2019-08-21 22:57 ` Mark H Weaver
2019-08-22 10:09 ` Ludovic Courtès
2019-08-24 13:31 ` Ludovic Courtès
2019-08-24 20:34 ` Mark H Weaver
2019-08-26 8:25 ` Ludovic Courtès
2019-08-26 18:36 ` Mark H Weaver
2019-08-27 9:38 ` Ludovic Courtès
2019-08-29 22:28 ` Bengt Richter [this message]
2019-08-27 3:58 ` Mark H Weaver
2019-08-27 9:40 ` Ludovic Courtès
2019-08-27 14:27 ` Mark H Weaver
2019-08-27 16:04 ` Ludovic Courtès
2019-08-27 16:46 ` Mark H Weaver
2019-08-28 0:55 ` Mark H Weaver
2019-08-28 22:12 ` Ludovic Courtès
2019-08-29 5:46 ` Ricardo Wurmus
2019-08-29 6:32 ` Ricardo Wurmus
2019-08-29 19:28 ` Mark H Weaver
2019-08-29 23:23 ` Ludovic Courtès
2019-08-30 19:52 ` Mark H Weaver
2019-08-31 12:44 ` Ludovic Courtès
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
List information: https://guix.gnu.org/
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20190829222814.GA2437@PhantoNv4ArchGx.localdomain \
--to=bokr@bokr.com \
--cc=36747@debbugs.gnu.org \
--cc=ludo@gnu.org \
/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.
Code repositories for project(s) associated with this public inbox
https://git.savannah.gnu.org/cgit/guix.git
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 read-only IMAP folder(s) and NNTP newsgroup(s).