[PATCH/RFA] SH: Rename ashlhi3 insn

[Joern RENNECKE]

Hans-Peter Nilsson wrote:

>On Mon, 21 Mar 2005, Alexandre Oliva wrote:
>  
>
>>Also, it would probably be a good idea to try to avoid attempting to
>>emit the insn that would violate the predicates in the first place.
>>    
>>
>
>IMHO that's where the real bug is.  A non-matching predicate
>should work like a FAIL in the expander, if GCC can't make the
>predicate match (usually by copying to a register).
>
>brgds, H-P
>
Agreed.  But it's a middle-end bug that is trivial to work around in the
target port.  I'm still waiting for review on dozens of patches for more 
serious
problems.