Set SAHF flag correctly in ia32

Set SAHF flag correctly for ia32

sahf flag will not be set for ia32 on some old platform because some old processors does not support CPUID's extended features.
This also avoids redundant cpuid check in ia32 for sahf.

BUG=
R=danno@chromium.org

Committed: https://code.google.com/p/v8/source/detail?r=21766

[Weiliang, 2014-05-28 06:14:06]

On 2014/05/23 09:18:03, Weiliang wrote:

Friendly ping. 

It is not a specific fix for x87 port, but for overall ia32.

[danno, 2014-05-28 08:17:47]

lgtm, although IIRC we have already removed any explicit checks for sahf support
in the ia32 backend when generating code, so  I don't think this change has any
effect. However, it does seem to make sense for consistency.

Please fix the indentation nit and I will land this for you.

https://codereview.chromium.org/298823008/diff/1/src/cpu.cc
File src/cpu.cc (right):

https://codereview.chromium.org/298823008/diff/1/src/cpu.cc#newcode304
src/cpu.cc:304: // SAHF is always available in compat/legacy mode,
nit: indentation is off

[Weiliang, 2014-05-28 08:30:45]

On 2014/05/28 08:17:47, danno wrote:
> lgtm, although IIRC we have already removed any explicit checks for sahf
support
> in the ia32 backend when generating code, so  I don't think this change has
any
> effect. However, it does seem to make sense for consistency.
> 
> Please fix the indentation nit and I will land this for you.
> 
> https://codereview.chromium.org/298823008/diff/1/src/cpu.cc
> File src/cpu.cc (right):
> 
> https://codereview.chromium.org/298823008/diff/1/src/cpu.cc#newcode304
> src/cpu.cc:304: // SAHF is always available in compat/legacy mode,
> nit: indentation is off

Fix. I just find that I can remove the check from x87 port. I will submit
another CL to remove it. 

Thanks a lot

[danno, 2014-06-02 08:19:28]

lgtm

[Sven Panne, 2014-06-11 09:44:25]

Committed patchset #2 manually as r21766 (presubmit successful).

[Yang, 2014-06-11 10:49:28]

On 2014/06/11 09:44:25, Sven Panne wrote:
> Committed patchset #2 manually as r21766 (presubmit successful).

huh? I thought this CL has been dismissed after Weiliang's last comment.

[Sven Panne, 2014-06-11 11:29:12]

On 2014/06/11 10:49:28, Yang wrote:
> On 2014/06/11 09:44:25, Sven Panne wrote:
> > Committed patchset #2 manually as r21766 (presubmit successful).
> 
> huh? I thought this CL has been dismissed after Weiliang's last comment.

Hmmm, reading it again, it seems that this CL was really not necessary. Shall we
revert it? Danno? Wailiang? I've got no strong opinion about that...

[danno, 2014-06-11 12:31:09]

I'm fine reverting it