[PATCH] Re: Buggy peephole prevents bootstrap (i386)

Jan Hubicka jh@suse.cz
Mon Oct 16 07:13:00 GMT 2000


> This did the trick.  Jan may know how to go about modifying a later
> branch or set<cc> instead of just disabling the optimization, but it seems
> like too much work to me.
I am having almost completed patch for this at home. I will send it soon.
Thanks for fixing this bit.

Honza
> 
> bootstrapped on i686-linux
> 
> gcc/ChangeLog
> 	* config/i386/i386.md (cmp to add peep2): Enable only for CCGCmode.
> 
> Alan Modra
> -- 
> Linuxcare.  Support for the Revolution.
> 
> --- gcc/config/i386/i386.md~	Fri Oct 13 16:37:19 2000
> +++ gcc/config/i386/i386.md	Sat Oct 14 14:10:24 2000
> @@ -11447,7 +11447,7 @@
>    [(set (reg 17)
>  	(compare (match_operand:SI 0 "register_operand" "")
>  		 (const_int 128)))]
> -  "ix86_match_ccmode (insn, CCmode)
> +  "ix86_match_ccmode (insn, CCGCmode)
>     && find_regno_note (insn, REG_DEAD, true_regnum (operands[0]))"
>    [(parallel [(set (reg:CC 17)
>  		   (compare:CC (neg:SI (const_int -128))
> @@ -11459,7 +11459,7 @@
>    [(set (reg 17)
>  	(compare (match_operand:HI 0 "register_operand" "")
>  		 (const_int 128)))]
> -  "ix86_match_ccmode (insn, CCmode)
> +  "ix86_match_ccmode (insn, CCGCmode)
>     && find_regno_note (insn, REG_DEAD, true_regnum (operands[0]))"
>    [(parallel [(set (reg:CC 17)
>  		   (compare:CC (neg:HI (const_int -128))
> 
> 


More information about the Gcc-bugs mailing list