This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: PATCH: PR target/40470: unable to find a registe r to spill in class ‘SSE_FIRST_REG’


On Thu, Jun 18, 2009 at 1:28 PM, Richard Henderson<rth@redhat.com> wrote:
> H.J. Lu wrote:
>>
>> When I add SSE_FIRST_REG/xmm0 to CLASS_LIKELY_SPILLED_P,
>> all testcases compile. However, xmm0 is used for FP return for
>> x86-64, I don't know the performance impact of this change.
>> xmm0 is a fixed register only for a handful SSE4 insns, which
>> can only be generated from intrinsics. Should we really go down
>> this route?
>
> Yes, that is the correct fix. ?As for being a return register,
> it should be no worse than for eax, which is also likely-spilled.
>
>

Here is a patch. OK for 4.3, 4.4 and trunk?

Thanks.


-- 
H.J.
---
gcc/

2009-06-17  H.J. Lu  <hongjiu.lu@intel.com>

	PR target/40470
	* config/i386/i386.h (CLASS_LIKELY_SPILLED_P): Add SSE_FIRST_REG.

gcc/testsuite/

2009-06-17  H.J. Lu  <hongjiu.lu@intel.com>

	PR target/40470
	* gcc.dg/torture/pr40470-1.c: New.
	* gcc.dg/torture/pr40470-2.c: Likewise.
	* gcc.dg/torture/pr40470-3.c: Likewise.
	* gcc.dg/torture/pr40470-4.c: Likewise.

Attachment: gcc-pr40470-3.patch
Description: Text document


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]