Re: Patch (rs6000): Prefer CTR to LR in indirect calls

[Dale Johannesen <dalej at apple dot com>]

On Jun 4, 2005, at 9:05 PM, David Edelsohn wrote:
> 	I just realized that the patch is more intrusive than it needed to
> be.  The `*' modifier modifies the next character and does not need to 
> be
> a separate alternative.  In other words, one should be able to achieve 
> the
> same effect with
>
> 	[(call (mem:SI (match_operand:SI 0 "register_operand" "c*l,c*l"))
>
> Would you please revert the patterns back to fewer alternatives using 
> the
> solution above, after bootstrapping and testing?

The following was bootstrapped and tested on Darwin.  I've also changed 
the
AIX patterns as you suggested, but I don't imagine this part got tested,
and I'm not in a position to do so.  Do you want those bits?

Attachment: diffs3.txt
Description: Text document