This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
RE: [PATCH] Remove atomic_compare_and_exchange_bool_rel.
- From: Lei Xu <lei dot xu at nxp dot com>
- To: Torvald Riegel <triegel at redhat dot com>, GLIBC Devel <libc-alpha at sourceware dot org>
- Date: Wed, 22 Jun 2016 09:35:49 +0000
- Subject: RE: [PATCH] Remove atomic_compare_and_exchange_bool_rel.
- Authentication-results: sourceware.org; auth=none
- Authentication-results: spf=none (sender IP is ) smtp dot mailfrom=lei dot xu at nxp dot com;
- References: <1465911341 dot 19633 dot 101 dot camel at localhost dot localdomain>
- Spamdiagnosticmetadata: NSPM
- Spamdiagnosticoutput: 1:99
Hello, Torvald
Sorry for the late.
Currently we did not use the latest glibc version.
So I port your patch (and other two patches it depends) to the eglibc 2.15 version for verification on PowerPC e6500,
because at first we found the segmentation issue on PowerPC E6500 under eglibc 2.15 , and it seems ok.
Considering the reason resulted in the segmentation fault is that atomic_compare_and_exchange_bool_rel is defined as
atomic_compare_and_exchange_bool_acq, after using your patch, it will call the atomic_compare_and_exchange_val_rel finally
(USE_ATOMIC_COMPILER_BUILTINS was defined to 0 on powerpc), which could fix the issue I meet before.
Regards
Lei
-----Original Message-----
From: Torvald Riegel [mailto:triegel@redhat.com]
Sent: Tuesday, June 14, 2016 9:36 PM
To: GLIBC Devel
Cc: Lei Xu
Subject: [PATCH] Remove atomic_compare_and_exchange_bool_rel.
Removing this operation and the matching (unused) catomic_ operation seemed to be easier than fixing powerpc's definition of it, only for it to be removed anyway in the future. There were just three call sites of it.
Tested on x86_64-linux. Lei Xu, could you test on Power?