This is the mail archive of the 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: unab le to find a register to spill i n class âSSE_FIRST_REGâ


Yes, you are correct.

I got the regiter classes mixed up it's been so long since I last looked at the x86 backend in any detail.


----- Original Message ----- From: "Richard Henderson" <>
To: "Graham Stott" <>
Cc: "H.J. Lu" <>; "Jeff Law" <>; "Vladimir Makarov" <>; "Steven Bosscher" <>; "Richard Guenther" <>; <>; <>
Sent: Thursday, June 18, 2009 9:24 PM
Subject: Re: PATCH: PR target/40470: unable to find a register to spill in class âSSE_FIRST_REGâ

Graham Stott wrote:

cant_combine_insn_p looks to be where the combind should get rejected.

the combine if it finds a HARD reg which is likely to be spilled.

Now on the x86 XMM0 is in MMX_REGS class which is NOT a likely spilled class hence it doen't get rejected.

Adding MMX_REGS to the set of likely spilled classes on x86 should do the

%xmm0 is *supposed* to be in SSE_FIRST_REG class. Are you sure you saw MMX_REG?

Now, CLASS_LIKELY_SPILLED_P in i386.h didn't get
updated when the SSE_FIRST_REG class was introduced,
so fixing that could fix the problem described.


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