Re: [PATCH] Add support for sparc VIS3 fp<-->int moves.

2011-10-24 Thread David Miller
From: Richard Henderson Date: Mon, 24 Oct 2011 14:05:28 -0700 > You shouldn't need to split these anymore. See the enabled attribute, as > used on several other targets so far. See the patch I posted 2 hours after this one.

Re: [PATCH] Add support for sparc VIS3 fp<-->int moves.

2011-10-24 Thread Richard Henderson
On 10/23/2011 08:53 PM, David Miller wrote: > -(define_insn "*movsi_insn" > +(define_insn "*movsi_insn_novis3" >[(set (match_operand:SI 0 "nonimmediate_operand" "=r,r,r,m,!f,!f,!m,d,d") > (match_operand:SI 1 "input_operand" "rI,K,m,rJ,f,m,f,J,P"))] > - "(register_operand (operands[0],

[PATCH] Add support for sparc VIS3 fp<-->int moves.

2011-10-23 Thread David Miller
The non-trivial aspects (and what took the most time for me) of these changes are: 1) Getting the register move costs and class preferencing right such that the VIS3 moves do get effectively used for incoming float/vector argument passing on 32-bit, yet IRA and reload don't go nuts alloc