target/9594: [3.3/3.4 regression] [sh4-elf] Assembler complai ns pcrel too far.

[Naveen Sharma, Noida]

Hi,
 
> Synopsis: [3.3/3.4 regression] [sh4-elf] Assembler complains 
> pcrel too far.
> 
> Responsible-Changed-From-To: unassigned->amylaar
> Responsible-Changed-By: amylaar
> Responsible-Changed-When: Mon Apr 28 13:07:29 2003
> Responsible-Changed-Why:
>     Because I fixed this problem.
> State-Changed-From-To: open->feedback
> State-Changed-By: amylaar
> State-Changed-When: Mon Apr 28 13:07:29 2003
> State-Changed-Why:
>     Fixed on 15th April in mainline and 3.3 branch.
> 
> http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&
> database=gcc&pr=9594

Thanks, 
The patch fixes the problem. So this PR may be closed

On a related note, I notice that gcc is known to generate 
out of bound branches for SH.

http://gcc.gnu.org/ml/gcc-patches/2002-01/msg00818.html
http://gcc.gnu.org/ml/gcc-bugs/2002-06/msg00896.html

>From the discussion thread following the latter, it seems 
there needs to be some design changes to solve these sort 
of problems. In this message
 http://gcc.gnu.org/ml/gcc-patches/2002-11/msg00192.html

You suggest to "marry" the local constant pool generation 
with an early branch shortening pass.

I would like to analyse this more, and implement it, if this can 
avoid problems of this nature. What do you think ?

Best Regards,
  Naveen Sharma.