This is the mail archive of the
libc-alpha@sources.redhat.com
mailing list for the glibc project.
Re: [PATCH] Add ABI mandated AT_ entries, prepare powerpc/dl-sysdep.c for new kernels
- To: drepper at cygnus dot com (Ulrich Drepper)
- Subject: Re: [PATCH] Add ABI mandated AT_ entries, prepare powerpc/dl-sysdep.c for new kernels
- From: Franz Sirl <Franz dot Sirl-kernel at lauterbach dot com>
- Date: Wed, 11 Apr 2001 12:04:15 +0200
- Cc: Geoff Keating <geoffk at redhat dot com>,libc-alpha at sources dot redhat dot com
- References: <Geoff Keating's message of "Tue, 10 Apr 2001 18:17:14 -0700"><01041100593300.01204@enzo.bigblue.local><200104110117.SAA00911@geoffk.org>
At 05:01 11.04.2001, Ulrich Drepper wrote:
>Geoff Keating <geoffk@geoffk.org> writes:
>
> > This is OK. Do you have the ability to commit it?
>
>I'll check it in. But are you all sure this patch is OK? I have no
>problems with most of it but was unsure about the
>sysdeps/unix/sysv/linux/powerpc/dl-sysdep.c change. Is this code not
>necessary anymore?
Yes, it can go away with newer kernels. Newer kernels will generate a block
of 4 AT_IGNOREPPC words (16bytes) at the start, making sure that the test
"<=0x10" (or <=AT_PHDR in old glibc's) in powerpc/dl-sysdep.c doesn't
trigger anymore. The only thing glibc has to do now is to never use the
AT_IGNOREPPC value for real information.
Besides unifying the glibc code, the kernel code is much cheaper now too,
as the expensive __get_user/__put_user memmove shifting of the AUXV can go
away. The only overhead that remains are the 4 extra __put_user calls to
setup the AT_IGNOREPPC block.
Franz.