|
From: | Paul Eggert |
Subject: | bug#17598: Emacs pretest 24.3.91 - AIX - SIGSEV when customize |
Date: | Fri, 13 Jun 2014 07:03:26 -0700 |
User-agent: | Mozilla/5.0 (X11; Linux x86_64; rv:24.0) Gecko/20100101 Thunderbird/24.5.0 |
Gilles PION wrote:
>Which GCC compiler are you using? I am using 4.8.1, and it works for >me; see attached files.4.8.2, boostratpped on a AIX 6.1 Host
I'm afraid that does not work. If you are using GCC on AIX 7.1, your GCC must be built on AIX 7.1. A GCC bootstrapped on AIX 6.1 will run on AIX 7.1 but will generate incorrect code sometimes. That appears to be what's happening in your case, as the truss output contains lines like this:
18809032: 42074209: 0.2175: faccessat(-3041965, "<<HOMEDIR>>/src/emacs/emacs-gcc-crashtest/emacs-24.3.91/src/lib-src", 0, 4) Err#22 EINVAL
which means Emacs thinks AT_FDCWD equals -3041965, but this is wrong for AIX 7.1: /usr/include/fcntl.h #defines AT_FDCWD to be -2. Apparently your GCC is incorrectly using fixincluded files from AIX 6.1 (which do not define AT_FDCWD) and Emacs is therefore incorrectly falling back on gnulib's definition of AT_FDCWD.
Bad fixincluded files would explain a lot of problems, including the subshell problem you reported earlier, as well as the Xpm problems. Possibly we were too hasty in suppressing Emacs's use of Xpm on AIX?
[Prev in Thread] | Current Thread | [Next in Thread] |