[PATCH/RFC] SH: libstdc++-v3 testsuite failures

[Joern Rennecke]

> Then the next_active_insn for the above call_insn returns NULL
> and causes the segfault.  So the first check for the variable
> next in my patch might be useless.  The patch below gets rid of
> all 4 failures of this type.  How this looks like?
> 
> --- ORIG/gcc/gcc/config/sh/sh.c	2004-08-08 10:19:34.000000000 +0900
> +++ LOCAL/gcc/gcc/config/sh/sh.c	2004-08-18 09:28:04.000000000 +0900
> @@ -4599,7 +4599,8 @@ split_branches (rtx first)
>  	    next = next_active_insn (insn);
>  
>  	    if ((GET_CODE (next) == JUMP_INSN
> -		 || GET_CODE (next = next_active_insn (next)) == JUMP_INSN)
> +		 || ((next = next_active_insn (next))
> +		     && GET_CODE (next) == JUMP_INSN))
>  		&& GET_CODE (PATTERN (next)) == SET
>  		&& recog_memoized (next) == CODE_FOR_jump_compact
>  		&& ((INSN_ADDRESSES

OK.