This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
IFUNC and libc-abis
- From: "Joseph S. Myers" <joseph at codesourcery dot com>
- To: <libc-alpha at sourceware dot org>
- Date: Wed, 22 Jan 2014 00:22:24 +0000
- Subject: IFUNC and libc-abis
- Authentication-results: sourceware.org; auth=none
I notice that IFUNC support has been added for S/390, ARM and AArch64
without any corresponding additions to the libc-abis file (or
architecture-specific files, in the ARM and AArch64 cases).
This appears not to have caused any adverse effects - what *should* the
effect be of this entry being missing, and should the fix simply be to add
entries to the main file / add new architecture-specific files (as
applicable - I've added another entry to the libc/ports differences list
on the wiki todo list to find a better way of handling
architecture-specific aspects of this file without duplicating
architecture-independent pieces - c.f.
<https://sourceware.org/ml/libc-alpha/2010-04/msg00021.html>, "An actually
clean scheme would not require any duplication of information across ports
files")?
As far as I can see, what I said in
<https://sourceware.org/ml/libc-ports/2010-03/msg00046.html> about
powerpc64 and sparc64 entries not being needed as the match is against
$(base-machine)-$(config-vendor)-$(config-os) where base-machine has been
canonicalized by configure still applies. (But I wouldn't now suggest
doing anything with $(abi-name), since as far as I can tell that's now an
unused variable, and should be removed along with the associated
shlib-versions entries and Makeconfig / soversions.awk handling.)
--
Joseph S. Myers
joseph@codesourcery.com