[3.3 PATCH] Fix PR optimization/11841

[Gabriel Dos Reis]

Eric Botcazou <ebotcazou@libertysurf.fr> writes:

| > I was waiting for the outcome of your conversation with Richard G.
| > It looks like that this patch won't resolve optimization/13653.  But
| > this is already a good step forward.  So, yes, this patch is OK for
| > gcc-3_3-branch.
| 
| Do you want me to apply it to the branch now?

If you believe it is the final form you want it, yes, please do.

| As for PR opt/13653, I came to the conclusion that the problem is different.  
| Volker identified the patch that had created PR opt/11841 and reverting it 
| doesn't cure PR opt/13653.  Moreover I skimmed through the RTL generated by 
| GCSE and didn't find any problematic REG_EQUAL note.
| 
| I spent a couple of hours 2 weeks ago on it, but the lack of reduced testcase 

Your efforts to address this issue are appreciated.  The size of the
testcase is definitely a big issue.

| is definitely a blocker.  So either the bug shows up only with big code and 
| there is little chance it will be fixed, or the bottleneck is the reduction 
| process and we lack an automated tool in this area.

:-(

-- Gaby