Fix PR24908: ICE in dwarf2out for cpu2000 with -funroll-loops

[Geoff Keating]

On 18/11/2005, at 5:04 AM, Alan Modra wrote:

> This PR shows regrename replacing a save of lr in r12 with a save  
> of lr
> in r26.  Perfectly legal.  Shouldn't be a problem even if there was a
> call somewhere in the testcase, so I don't know why the assert  
> excluded
> call-saved regs.

The assert is because it's ambiguous; for instance, you can't tell if  
it's a save of lr in r26, or a restore of r26 from lr.  You might be  
able to reason that it's clearly not ambiguous, by looking at what is  
known about the contents of registers...

>   Bootstrapped and regression tested powerpc64-linux.
> OK to apply?
>
> :ADDPATCH debug:
>
> 	PR debug/24908
> 	* dwarf2out.c (dwarf2out_frame_debug_expr): Don't assert that
> 	call_used_regs can't be used to save reg in another reg.
>
> Index: gcc/dwarf2out.c
> ===================================================================
> --- gcc/dwarf2out.c	(revision 107176)
> +++ gcc/dwarf2out.c	(working copy)
> @@ -1578,11 +1578,10 @@ dwarf2out_frame_debug_expr (rtx expr, co
>  	  else
>  	    {
>  	      /* Saving a register in a register.  */
> -	      gcc_assert (call_used_regs [REGNO (dest)]
> -			  && (!fixed_regs [REGNO (dest)]
> -			      /* For the SPARC and its register window.  */
> -			      || DWARF_FRAME_REGNUM (REGNO (src))
> -				   == DWARF_FRAME_RETURN_COLUMN));
> +	      gcc_assert (!fixed_regs [REGNO (dest)]
> +			  /* For the SPARC and its register window.  */
> +			  || (DWARF_FRAME_REGNUM (REGNO (src))
> +			      == DWARF_FRAME_RETURN_COLUMN));
>  	      queue_reg_save (label, src, dest, 0);
>  	    }
>  	  break;
>
> -- 
> Alan Modra
> IBM OzLabs - Linux Technology Centre

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 2410 bytes
Desc: not available
URL: <http://gcc.gnu.org/pipermail/gcc-patches/attachments/20051118/43b99ba4/attachment.p7s>