Re: c/7661: gcc-3.0 optimization bug on debian GNULinux on x86 with very simple program

Re: c/7661: gcc-3.0 optimization bug on debian GNULinux on x86 with very simple program

[Martin Sebor]

The following reply was made to PR c/7661; it has been noted by GNATS.

From: Martin Sebor <sebor@roguewave.com>
To: gcc-gnats@gcc.gnu.org
Cc:  
Subject: Re: c/7661: gcc-3.0 optimization bug on debian GNULinux on x86 with very simple program
Date: Tue, 20 Aug 2002 14:18:55 -0600

 Thomas Deselaers wrote:
 
 ...
 
 >>Description:
 >>
 > 
 > OK, while playing around(together with Stefan Jacobs, he wanted to be 
 > mentioned) I found out, that this small programm:
 > 
 > -------file:maxint.cc------------
 > main() {
 >       int i=0;
 >       int oldi=0;;
 >       while(++i>oldi) oldi=i;
 >       return oldi;
 > }
 > ---------------------------------
 > 
 > does work absolutely fine, when compiled without any optimization, 
 > but does not terminate at all, when using any optimization-options:
 
 
 FYI: this looks like the same thing as PR #4371, i.e., signed integer
 overflow:
 
 http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view&pr=4371&database=gcc
 
 Regards
 Martin
 

Re: c/7661: gcc-3.0 optimization bug on debian GNULinux on x86 with very simple program

[Carlos O'Donell]

The following reply was made to PR c/7661; it has been noted by GNATS.

From: Carlos O'Donell <carlos@baldric.uwo.ca>
To: Andrew Pinski <pinskia@physics.uc.edu>
Cc: Thomas Deselaers <thomas@deselaers.de>, gcc-gnats@gcc.gnu.org,
	debian-gcc@lists.debian.org
Subject: Re: c/7661: gcc-3.0 optimization bug on debian GNULinux on x86 with very simple program
Date: Tue, 20 Aug 2002 19:39:36 -0400

 > I think this program should not terminate at all because i will 
 > always be one greater than oldi.
 > I think gcc3.0 has a problem with no optimization then but since 
 > there is later version that works gcc 3.1.1, upgrade.
 > 
 > Thanks,
 > Andrew Pinski
 > 
 
 Agreed. Infact it doesn't terminate on all the systems I've tried (3).
 But for the sake of presenting a test case :)
 
 ---
 Test System: Dell XPS-T 450Mhz PIII (Debian testing/unstable)
 ---
 Reading specs from /usr/lib/gcc-lib/i386-linux/3.0.4/specs
 Configured with: ../src/configure -v --enable-languages=c,c++,java,f77,objc --prefix=/usr --infodir=/share/info --mandir=/share/man --enable-shared --with-gnu-as --with-gnu-ld --with-system-zlib --enable-long-long --enable-nls --without-included-gettext --disable-checking --enable-threads=posix --enable-java-gc=boehm --with-cpp-install-dir=bin --enable-objc-gc i386-linux
 Thread model: posix
 gcc version 3.0.4
 ---
 
 carlos@systemhalted:~$ cat test.c 
 main() {                                                                                                      int i=0;                                                                                                int oldi=0;;                                                                                            while(++i>oldi) oldi=i;                                                                                 return oldi;                                                                                       }                 
 carlos@systemhalted:~$ gcc-3.0 -O0 -o test test.c
 carlos@systemhalted:~$ ./test
 (Attach gdb, look at stuff, it's not stopping) ^C
 carlos@systemhalted:~$ gcc-3.0 -o test test.c
 carlos@systemhalted:~$ ./test
 (Attach gdb, look at stuff, it's not stopping) ^C
 ---
 
 Maybe you've been bad to your gcc ;)
 
 In the optimized case (-O9) I see:
 080483f8 <main>:
  80483f8:       55                      push   %ebp
  80483f9:       89 e5                   mov    %esp,%ebp
  80483fb:       83 ec 08                sub    $0x8,%esp
  80483fe:       83 e4 f0                and    $0xfffffff0,%esp
  8048401:       8d 76 00                lea    0x0(%esi),%esi
  8048404:       eb fe                   jmp    8048404 <main+0xc>
  8048406:       90                      nop    
  8048407:       90                      nop    
 
 Which never terminates... and gcc knew it.
 
 If there were some examples of gcc-3.0 mucking 
 up optimizations... this isn't it.
 
 c.
 

Re: c/7661: gcc-3.0 optimization bug on debian GNULinux on x86 with very simple program

[Andrew Pinski]

The following reply was made to PR c/7661; it has been noted by GNATS.

From: Andrew Pinski <pinskia@physics.uc.edu>
To: Thomas Deselaers <thomas@deselaers.de>
Cc: gcc-gnats@gcc.gnu.org, debian-gcc@lists.debian.org
Subject: Re: c/7661: gcc-3.0 optimization bug on debian GNULinux on x86 with very simple program
Date: Tue, 20 Aug 2002 13:26:42 -0400

 I think this program should not terminate at all because i will 
 always be one greater than oldi.
 I think gcc3.0 has a problem with no optimization then but since 
 there is later version that works gcc 3.1.1, upgrade.
 
 Thanks,
 Andrew Pinski
 
 trace
 i    oldi
 0    0
 1    0    check here
 1    1
 2    1    check
 2    2
    .
    .
    .
 
 
 On Tuesday, August 20, 2002, at 01:11 , Thomas Deselaers wrote:
 
 > main() {
 >       int i=0;
 >       int oldi=0;;
 >       while(++i>oldi) oldi=i;
 >       return oldi;
 > }
 

c/7661: gcc-3.0 optimization bug on debian GNULinux on x86 with very simple program

[Thomas Deselaers]

>Number:         7661
>Category:       c
>Synopsis:       gcc-3.0 optimization bug on debian GNU/Linux on x86 with very simple program
>Confidential:   no
>Severity:       non-critical
>Priority:       medium
>Responsible:    unassigned
>State:          open
>Class:          wrong-code
>Submitter-Id:   net
>Arrival-Date:   Tue Aug 20 10:16:00 PDT 2002
>Closed-Date:
>Last-Modified:
>Originator:     Thomas Deselaers
>Release:        3.0.4 (Debian testing/unstable)
>Organization:
>Environment:
System: Linux racker 2.4.19 #3 Tue Aug 6 23:01:23 CEST 2002 i686 unknown unknown GNU/Linux
Architecture: i686

Dell Inspiron 8200, Pentium 4 mobile 1600
OS: Linux 2.4.19 unpatched

host: i386-pc-linux-gnu
build: i386-pc-linux-gnu
target: i386-pc-linux-gnu
configured with: ../src/configure -v --enable-languages=c,c++,java,f77,proto,objc --prefix=/usr --infodir=/share/info --mandir=/share/man --enable-shared --with-gnu-as --with-gnu-ld --with-system-zlib --enable-long-long --enable-nls --without-included-gettext --disable-checking --enable-threads=posix --enable-java-gc=boehm --with-cpp-install-dir=bin --enable-objc-gc i386-linux
>Description:

OK, while playing around(together with Stefan Jacobs, he wanted to be 
mentioned) I found out, that this small programm:

-------file:maxint.cc------------
main() {
      int i=0;
      int oldi=0;;
      while(++i>oldi) oldi=i;
      return oldi;
}
---------------------------------

does work absolutely fine, when compiled without any optimization, 
but does not terminate at all, when using any optimization-options:

with "gcc -o maxint maxint.cc" the maxint-program works fine, but
with "gcc -o maxint maxint.cc -O$X" with $X being nothing, or one of 
the numbers from 1 to 9, the maxint program does not terminate.


>How-To-Repeat:

copy the code-excerpt to a file and compile it

>Fix:
Workaround: not use -O$X
>Release-Note:
>Audit-Trail:
>Unformatted: