Re: optimization/10030: gcc 3.2(.1) produces broken code with -O2, but not -O

Re: optimization/10030: gcc 3.2(.1) produces broken code with -O2, but not -O

[bangerth]

Synopsis: gcc 3.2(.1) produces broken code with -O2, but not -O

State-Changed-From-To: open->feedback
State-Changed-By: bangerth
State-Changed-When: Tue Mar 11 20:11:56 2003
State-Changed-Why:
    We need a smaller testcase that a whole package. Can you try 
    to identify which function is being miscompiled? 
    
    Please understand that we are, in general, not able to
    debug whole packages to find out what goes wrong. For such
    cases, we really need the help of those who submit bug
    reports. Please see
      http://gcc.gnu.org/bugs.html
    
    Thanks
      Wolfgang

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=10030

Re: optimization/10030: gcc 3.2(.1) produces broken code with -O2, but not -O

[ljrittle]

Synopsis: gcc 3.2(.1) produces broken code with -O2, but not -O

Responsible-Changed-From-To: unassigned->ljrittle
Responsible-Changed-By: ljrittle
Responsible-Changed-When: Wed Mar 26 10:22:56 2003
Responsible-Changed-Why:
    Mine.  (I lowered the severity: Anything so nebulous with a posted workaround is, by default, not critical!)
State-Changed-From-To: feedback->closed
State-Changed-By: ljrittle
State-Changed-When: Wed Mar 26 10:22:56 2003
State-Changed-Why:
    9 times out of 10, we never hear from these people again.  Please just close ill-formed reports that refer to mega-packages...
    
    General Notes: A compiled program crashing at -O2 but not -O
    does not imply a compiler bug.  This is why someone (you) will
    need to produce a small test case.

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=10030

optimization/10030: gcc 3.2(.1) produces broken code with -O2, but not -O

[dan.naumov]

>Number:         10030
>Category:       optimization
>Synopsis:       gcc 3.2(.1) produces broken code with -O2, but not -O
>Confidential:   no
>Severity:       critical
>Priority:       medium
>Responsible:    unassigned
>State:          open
>Class:          wrong-code
>Submitter-Id:   net
>Arrival-Date:   Tue Mar 11 19:56:00 UTC 2003
>Closed-Date:
>Last-Modified:
>Originator:     Dan Naumov
>Release:        gcc version 3.2.1 [FreeBSD] 20021119 (release)
>Organization:
>Environment:
FreeBSD 5.0-p4
>Description:
When compiling AbiWord (http://www.abisource.com/) using CPUTYPE=athlon-tbird and CFLAGS= -O2 -pipe, the program compiles fine, but will always segfault upon exiting. Using -O instead solves the problem.
>How-To-Repeat:
1) obtain AbiWord 1.0.4
2) compile with -O2 and CPUTYPE=athlon-tbird
3) run and exit AbiWord
>Fix:

>Release-Note:
>Audit-Trail:
>Unformatted: