This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [PATCH] Remove socket.S implementation
- From: Adhemerval Zanella <adhemerval dot zanella at linaro dot org>
- To: Joseph Myers <joseph at codesourcery dot com>
- Cc: GNU C Library <libc-alpha at sourceware dot org>
- Date: Wed, 20 May 2015 14:01:45 -0300
- Subject: Re: [PATCH] Remove socket.S implementation
- Authentication-results: sourceware.org; auth=none
- References: <555C9805 dot 1010100 at linaro dot org> <alpine dot DEB dot 2 dot 10 dot 1505201613020 dot 23246 at digraph dot polyomino dot org dot uk>
On 20-05-2015 13:18, Joseph Myers wrote:
> On Wed, 20 May 2015, Adhemerval Zanella wrote:
>
>> The patch idea is to simplify the socket function implementation that
>> uses the socketcall to be based on C implemetation instead of a pseudo
>> assembly implementation with arch specific parts. All the affect
>> architectures (i386, microblaze, mips, powerpc, and sparc) have support
>> for 6 argument syscalls.
>
> m68k, s390 and sh are also socketcall architectures. I don't see m68k
> changes in this patch.
Indeed, this is the only missing cross-compiler I forgot to create. I will
remove its socket implementation and check with a cross-build.
>
> (6-argument syscalls seem irrelevant here since socketcall is a 2-argument
> syscall.)
Yes, this comments is out of place here.
>
>> * sysdeps/sysv/linux/mips/mips32/internal_recvmmsg.S: Likewise.
>> * sysdeps/sysv/linux/mips/mips32/internal_sendmmsg.S: Likewise.
>
> Missing unix/ in path in ChangeLog entry.
>
> As I understand it, the patch should also remove
> internal_{accept4,recvmmsg,sendmmsg}.S for ARM (where they exist only to
> ensure the correct EABI attributes on the output for those objects, which
> is irrelevant when nothing will try to build those objects at all).
>
I will remove them as well, thanks for this out.