[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Patches] MIPS eglibc build problem
- To: Steve Ellcey <sellcey@xxxxxxxx>
- Subject: Re: [Patches] MIPS eglibc build problem
- From: "Joseph S. Myers" <joseph@xxxxxxxxxxxxxxxx>
- Date: Fri, 11 May 2012 21:41:32 +0000 (UTC)
On Fri, 11 May 2012, Steve Ellcey wrote:
> In the older (2.14) eglibc Makefile, there is this code:
> # A callable macro that expands to a shell command. Preprocess file $(1)
> # using ABI option $(2) and see which macros it defines. Print FOO for each
> # macro of the form __NR$(3)_FOO, filtering out ABI-specific __NR macros
> # that have a prefix other than $(3).
> mips_list_syscalls = $(filter-out -m%,$(CC)) -E -x c $(+includes) \
> $(sysincludes) -D_LIBC -dM -mabi=$(2) $(1) | \
> sed -n 's@^\#define __NR$(3)_\([^ ]*\) .*@\1@p' | \
> sed -e '/^[ON]32_/d' -e '/^N64_/d' -e '/^64_/d' | \
> LC_ALL=C sort
> I dont see this anymore and I think it would have filtered out the -mips32
> in my CC command and allowed things to work. Did this code get moved
> somewhere? Or was it removed because it wasn't thought to be needed?
This is now in a fully generic form in libc,
sysdeps/unix/sysv/linux/Makefile, with each architecture providing only
variables not makefile rules for building this file.
There is nothing EGLIBC-specific here so I advise raising things on
libc-ports. Maybe syscall-list-n32-options and syscall-list-n64-options
need to include appropriate -march= options to ensure the -mabi= options
are accepted by overriding any previous options in $(CC)? Or maybe -D and
-U options should be used instead, as on x86_64?
Joseph S. Myers
Patches mailing list