GNU bug report logs - #43152
28.0.50; when building emacs 28.0.50 ./temacs is stopped with core

Previous Next

Full log

Message #77 received at 43152 <at> debbugs.gnu.org (full text, mbox):

From: Paul Eggert <eggert <at> cs.ucla.edu>
To: Lars Ingebrigtsen <larsi <at> gnus.org>
Cc: Philippe Spiesser <ann.onymous <at> orange.fr>, 43152 <at> debbugs.gnu.org,
 Bruno Haible <bruno <at> clisp.org>, Stefan Kangas <stefankangas <at> gmail.com>,
 Angelo Graziosi <angelo.g0 <at> libero.it>
Subject: Re: bug#43152: 28.0.50; when building emacs 28.0.50 ./temacs is
 stopped with core
Date: Sat, 5 Sep 2020 17:58:22 -0700

On 9/5/20 2:30 PM, Lars Ingebrigtsen wrote:
> I've got git 2.17 on high sierra, so I did a normal git bisect, and it
> reports:
> 
> Bisecting: 0 revisions left to test after this (roughly 0 steps)
> [42ec41251584c480ee3286ff369c18629f52a7d5] Update from Gnulib
> 
> The next commit is:
> 
> commit 2c389455c72250b579f5225b99bc7de0cf435e4a (refs/bisect/good-2c389455c72250b579f5225b99bc7de0cf435e4a)

Thanks, so Bug#43152 is due to verify.h's recent attempt to use __builtin_assume 
on Clang. As I understand it, using __builtin_assume was only for minor 
performance improvements when compiling with older Clang versions, which does 
not appear to be worth the software debugging hassle that we've been enduring on 
GNU Emacs master. So I reverted those changes to verify.h in Gnulib and updated 
GNU Emacs to match. I'll cc this to Bruno Haible, who wrote those 
__builtin_assume changes.

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.