This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
Re: mips atomicity.h selection
- From: Loren James Rittle <rittle at latour dot rsch dot comm dot mot dot com>
- To: gcc-patches at gcc dot gnu dot org
- Cc: rsandifo at redhat dot com, libstdc++ at gcc dot gnu dot org
- Date: Mon, 5 May 2003 16:09:14 -0500 (CDT)
- Subject: Re: mips atomicity.h selection
- Organization: Networks and Infrastructure Lab (IL02/2240), Motorola Labs
> At one point Eric committed a patch to fix this:
>
> http://gcc.gnu.org/ml/gcc-patches/2002-07/msg00115.html
>
> but a clean-up seems to have negated its effect:
>
> http://gcc.gnu.org/ml/gcc-patches/2002-11/msg00064.html
> Tested on mips-elf. OK to install?
> * configure.target (mips*): Use the generic atomicity.h by default.
For the record, the libstdc++-v3 maintainers explicitly yield to the
CPU/OS maintainers on this type of issue. However I think (a)
Benjamin ax'd the configuration path without realizing that mips-*-*
would start getting mips/atomicity.h by default; or (b) a later
cleanup affected it thus I will explicitly approve this one for
mainline based on the historical analysis you have done. Either way,
generating illegal instructions for a CPU is not good thus I'd lobby
for the 3.3 branch as well under the same theory as the i386 fix.
Please ask Mark nicely. You might mention that switching to generic
mutex-based atomics is safer than switching from generic...
Regards,
Loren