[sdcc-devel] [ sdcc-Bugs-1464657 ] peephole chews the jumptables

[SourceForge.net <no...@so...>]

Bugs item #1464657, was opened at 2006-04-05 05:38
Message generated for change (Comment added) made by frief
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=100599&aid=1464657&group_id=599

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: msc51(8051) target
Group: fixed
Status: Closed
Resolution: Fixed
Priority: 5
Submitted By: Stas Sergeev (stsp)
Assigned to: Maarten Brock (maartenbrock)
Summary: peephole chews the jumptables

Initial Comment:
Hi.

There seems to be no peephole category here, so filling
as an assembler bug.

The bug is in fact trivial - peephole replaces jumps
to ret with ret, but in jumptables this yields a bad
code. I beleive I already reported that problem a
few years ago, but it is still not addressed.

Attached is the test-case. It is supposed to return 5,
but instead returns 12, after jumping to some random
code.

----------------------------------------------------------------------

>Comment By: Frieder Ferlemann (frief)
Date: 2006-04-08 10:22

Message:
Logged In: YES 
user_id=589052

Unfortuately Sourceforge's CVS service is currently seeing
major problems. See:
http://sourceforge.net/docs/A04/
for the current status. Hopefully fixed soon.

you might want to grap the latest source-snapshot
(sdcc-src-20060407.tar.gz) directly from:
http://sdcc.sourceforge.net/snap.php 

(Extract this to a different directory to avoid spoiling
your local cvs repository)

----------------------------------------------------------------------

Comment By: Stas Sergeev (stsp)
Date: 2006-04-08 09:47

Message:
Logged In: YES 
user_id=501371

Thanks, but is it in CVS? I do cvs update and do not
get your fix.

----------------------------------------------------------------------

Comment By: Maarten Brock (maartenbrock)
Date: 2006-04-06 17:25

Message:
Logged In: YES 
user_id=888171

Fixed in SDCC 2.5.5 #1237.

----------------------------------------------------------------------

Comment By: Maarten Brock (maartenbrock)
Date: 2006-04-05 20:58

Message:
Logged In: YES 
user_id=888171

Before I add a new category I first want to ask the other 
developers about it.

Thanks for looking this up. Indeed it's not exactly the 
same but similar. The solution gave me a good hint at how 
to solve this one.

----------------------------------------------------------------------

Comment By: Stas Sergeev (stsp)
Date: 2006-04-05 18:28

Message:
Logged In: YES 
user_id=501371

> I would call it target related, not assembler.
I wanted to fill it as a peephole bug, but there is no
such category. Maybe you can fix this by adding one? :)

> Also if it was reported before I doubt it was a bug
> report.
Just for the record, I digged it, here it is (#814558):
http://sourceforge.net/tracker/index.php?func=detail&aid=814558&group_id=599&atid=100599
It is actually not exactly the same, but I wonder if
they have the similar roots.

----------------------------------------------------------------------

Comment By: Maarten Brock (maartenbrock)
Date: 2006-04-05 17:54

Message:
Logged In: YES 
user_id=888171

I would call it target related, not assembler. I assume 
you're referring to the mcs51 target. Also if it was 
reported before I doubt it was a bug report.

Nonetheless thanks for the report (and test-case!) because 
it definitely is a bug.

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=100599&aid=1464657&group_id=599