This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [PATCH 20/27] [AARCH64] Make lp64 and ilp32 directories.
- From: Joseph Myers <joseph at codesourcery dot com>
- To: Yury Norov <ynorov at caviumnetworks dot com>
- Cc: <libc-alpha at sourceware dot org>, <linux-kernel at vger dot kernel dot org>, <arnd at arndb dot de>, <catalin dot marinas at arm dot com>, <marcus dot shawcroft at arm dot com>, <philb at gnu dot org>, <davem at davemloft dot net>, <szabolcs dot nagy at arm dot com>, <maxim dot kuvyrkov at linaro dot org>, <pinskia at gmail dot com>, Andrew Pinski <apinski at cavium dot com>
- Date: Tue, 21 Jun 2016 10:44:11 +0000
- Subject: Re: [PATCH 20/27] [AARCH64] Make lp64 and ilp32 directories.
- Authentication-results: sourceware.org; auth=none
- References: <1466485631-3532-1-git-send-email-ynorov at caviumnetworks dot com> <1466485631-3532-22-git-send-email-ynorov at caviumnetworks dot com>
On Tue, 21 Jun 2016, Yury Norov wrote:
> From: Andrew Pinski <apinski@cavium.com>
>
> The patch makes the ilp32 and lp64 have their own directory under aarch64.
> Since ILP32 uses most of the same system calls as LP64 and has a 64bit
> off_t, we need make the functions that end in 64 the same as the ones without.
> We also need not to special case ioctl or use the already provided mmap.c file.
My previous comments at
<https://sourceware.org/ml/libc-alpha/2014-10/msg00640.html> apply. Where
are your ABI test baselines?
> diff --git a/sysdeps/aarch64/ilp32/Implies b/sysdeps/aarch64/ilp32/Implies
> new file mode 100644
> index 0000000..705bc5b
> --- /dev/null
> +++ b/sysdeps/aarch64/ilp32/Implies
> @@ -0,0 +1,6 @@
> +aarch64
> +wordsize-32
> +ieee754/ldbl-128
> +ieee754/dbl-64
> +ieee754/flt-32
> +aarch64/soft-fp
You can use ieee754/dbl-64/wordsize-64 here. See my commit
b75bc69cdfe5247b2156ce249518f1c1df4d797d.
> diff --git a/sysdeps/unix/sysv/linux/aarch64/ilp32/getdents.c b/sysdeps/unix/sysv/linux/aarch64/ilp32/getdents.c
The formatting in this file is thoroughly wrong.
> diff --git a/sysdeps/unix/sysv/linux/aarch64/ilp32/preadv.c b/sysdeps/unix/sysv/linux/aarch64/ilp32/preadv.c
> new file mode 100644
> index 0000000..b5b9994
> --- /dev/null
> +++ b/sysdeps/unix/sysv/linux/aarch64/ilp32/preadv.c
> @@ -0,0 +1 @@
> +/* See preadv.c */
> \ No newline at end of file
All files should have final newlines (but with Adhemerval's preadv
unification, you shouldn't need this; make the unified version more
general if necessary).
> diff --git a/sysdeps/unix/sysv/linux/aarch64/ilp32/shlib-versions b/sysdeps/unix/sysv/linux/aarch64/ilp32/shlib-versions
> new file mode 100644
> index 0000000..17db764
> --- /dev/null
> +++ b/sysdeps/unix/sysv/linux/aarch64/ilp32/shlib-versions
> @@ -0,0 +1,7 @@
> +DEFAULT GLIBC_2.21
No, GLIBC_2.24 if this gets in by the end of June, GLIBC_2.25 otherwise
(since glibc is frozen for release during July). The symbol version is
always the version of the first release with the new port.
--
Joseph S. Myers
joseph@codesourcery.com