Re: Stack Reorganization Patch

["Naveen Sharma, Noida" <naveens at noida dot hcltech dot com>]

Hi,

This is a re-post of my previous patch at 
  http://gcc.gnu.org/ml/gcc-patches/2003-01/msg00019.html

I would request the review of the patch and let me know 
the issues in its acceptance.

I thank you in advance for your kind attention.
Here is the check list filled out for this repost, I hope there are no
problems.
 
[X ] The message is sent to gcc-patches.

[X]  The message has a meaningful subject line.
     
[X] The message contains a description of the change you want to make, and
why.

    I have stated the precise problem, the strategy in my previous postings.
    Here are links:
    Initial Post:
          http://gcc.gnu.org/ml/gcc-patches/2003-01/msg00019.html
    Discussion thread on which this work is based:
          (http://gcc.gnu.org/ml/gcc/2002-05/threads.html#02838).
    Please let me know of any additional information required.
     
[X ] The message says on what platform you tested the patch.

     a) regtested i686 native: 
        No new failures.

     b) tested for sh-elf cross: 
        I have some known problems with nested 
        functions/non-local labels.

    Since I am enabling this patch  only for SH4 (and that too with 
    additional option), it should not break any other targets and I 
    need people's help for other's targets.
        
[X] The patch is attached to the mail, and was generated with `diff -upN' or
    `diff -cpN'.

    The patch was generated by "diff -c2p". It is attached as
          stack_reorg.tar.gz
     It contains two file 
           a) gcc/stack_reorg.c
           b) STACK_REORG.txt
       
     #a is new file, and file #b contains changes to files in current 
     cvs tree.

[X] There is a ChangeLog entry before the patch, but not part of the patch 
    itself.

    ChangeLog is a separate file attached with this mail.

[X ] The ChangeLog entry is properly formatted.

[ ] The patch either fixes an existing test case, or adds a new test case.
    Not applicable. It is a general improvement.

[X] If the patch adds or changes a feature, a documentation change is
included.

    Yes, see invoke.texi changes.

[X] If the patch adds or changes a hook, target macro, RTL type or tree
type,
    a documentation change is included.

   Yep, done in tm.texi.

Please let me know any additional information.

I mentioned 5-6% improvement for some benchmarks (gsm speech compression)
with this patch. Another side effect is about 5% reduction in code size
for sh-elf. Perhaps others could benefit too.

I am aware of certain problems with this patch, notably it has 
problem with TC in PR# 8987. 

Thanks and Best Regards,
    Naveen Sharma.

Attachment: stack_reorg.taz
Description: Binary data

Attachment: ChangeLog
Description: Binary data