libstdc++/10809: [3.3 regression] mips-sgi-irix6.5 libstdc++-v3 testsuite execution failure in 27_io/istream_unformatted.cc

libstdc++/10809: [3.3 regression] mips-sgi-irix6.5 libstdc++-v3 testsuite execution failure in 27_io/istream_unformatted.cc

[ghazi]

>Number:         10809
>Category:       libstdc++
>Synopsis:       [3.3 regression] mips-sgi-irix6.5 libstdc++-v3 testsuite execution failure in 27_io/istream_unformatted.cc
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    unassigned
>State:          open
>Class:          wrong-code
>Submitter-Id:   net
>Arrival-Date:   Fri May 16 01:46:00 UTC 2003
>Closed-Date:
>Last-Modified:
>Originator:     Kaveh Ghazi
>Release:        gcc 3.3 (release)
>Organization:
>Environment:
mips-sgi-irix6.5
mips-sgi-irix6.2
>Description:
I noticed that there is a testsuite regression on irix6 in 27_io/istream_unformatted.cc.  See:
irix6.2: http://gcc.gnu.org/ml/gcc-testresults/2003-05/msg00943.html
irix6.5: http://gcc.gnu.org/ml/gcc-testresults/2003-05/msg00942.html

Looking back through the testresult archives, it looks like the regression appeared around two weeks ago in the May 2nd snapshot.
May 1: http://gcc.gnu.org/ml/gcc-testresults/2003-05/msg00066.html
May 2: http://gcc.gnu.org/ml/gcc-testresults/2003-05/msg00125.html

It's an execution failure.  Running the binary I get:

Assertion failed: EX, file /caip/u58/ghazi/gcc-testing/branch/gcc-3.3/libstdc++-v3/testsuite/27_io/istream_unformatted.cc, line 606

Unfortunately I can't make sense of why it does this though.
>How-To-Repeat:
It's an execute failure, compile and run 27_io/istream_unformatted.cc.  In my tree the command line to build it was the following, but it's very path dependent so you'll probably have to run the libstdc++-v3 testsuite on irix6 yourself:

/caip/u58/ghazi/gcc-testing/branch/build/gcc/g++ -shared-libgcc -B/caip/u58/ghazi/gcc-testing/branch/build/gcc/ -nostdinc++ -L/caip/u58/ghazi/gcc-testing/branch/build/mips-sgi-irix6.5/libstdc++-v3/src -L/caip/u58/ghazi/gcc-testing/branch/build/mips-sgi-irix6.5/libstdc++-v3/src/.libs -B/usr/local/mips-sgi-irix6.5/bin/ -B/usr/local/mips-sgi-irix6.5/lib/ -isystem /usr/local/mips-sgi-irix6.5/include -g -ffunction-sections -fdata-sections -fmessage-length=0 -DDEBUG_ASSERT -DLOCALEDIR="/caip/u58/ghazi/gcc-testing/branch/build/mips-sgi-irix6.5/libstdc++-v3/po/share/locale" -nostdinc++ -I/caip/u58/ghazi/gcc-testing/branch/build/mips-sgi-irix6.5/libstdc++-v3/include/mips-sgi-irix6.5 -I/caip/u58/ghazi/gcc-testing/branch/build/mips-sgi-irix6.5/libstdc++-v3/include -I/caip/u58/ghazi/gcc-testing/branch/gcc-3.3/libstdc++-v3/libsupc++ -I/caip/u58/ghazi/gcc-testing/branch/gcc-3.3/libstdc++-v3/libio -I/caip/u58/ghazi/gcc-testing/branch/gcc-3.3/libstdc++-v3/include/backward -I/caip/u58/ghazi/gcc-testing/branch/gcc-3.3/libstdc++-v3/testsuite /caip/u58/ghazi/gcc-testing/branch/gcc-3.3/libstdc++-v3/testsuite/27_io/istream_unformatted.cc -g -O2 -DDEBUG_ASSERT -lv3test -L/caip/u58/ghazi/gcc-testing/branch/build/mips-sgi-irix6.5//libstdc++-v3/testsuite -lm -o ./istream_unformatted.exe
>Fix:

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

Re: libstdc++/10809: [3.3 regression] mips-sgi-irix6.5 libstdc++-v3 testsuite execution failure in 27_io/istream_unformatted.cc

[Eric Botcazou]

The following reply was made to PR libstdc++/10809; it has been noted by GNATS.

From: Eric Botcazou <ebotcazou@libertysurf.fr>
To: ghazi@caip.rutgers.edu
Cc: gcc-gnats@gcc.gnu.org,
 mark@codesourcery.com
Subject: Re: libstdc++/10809: [3.3 regression] mips-sgi-irix6.5 libstdc++-v3 testsuite execution failure in 27_io/istream_unformatted.cc
Date: Fri, 16 May 2003 09:39:08 +0200

 > I noticed that there is a testsuite regression on irix6 in
 > 27_io/istream_unformatted.cc.  See: irix6.2:
 > http://gcc.gnu.org/ml/gcc-testresults/2003-05/msg00943.html irix6.5:
 > http://gcc.gnu.org/ml/gcc-testresults/2003-05/msg00942.html
 
 Same problem on Sparc/Solaris 2.5.1 and 2.6, see PR libstdc++/10703.
 
 -- 
 Eric Botcazou

Re: libstdc++/10809: [3.3 regression] mips-sgi-irix6.5 libstdc++-v3 testsuite execution failure in 27_io/istream_unformatted.cc

[paolo]

Synopsis: [3.3 regression] mips-sgi-irix6.5 libstdc++-v3 testsuite execution failure in 27_io/istream_unformatted.cc

Responsible-Changed-From-To: unassigned->paolo
Responsible-Changed-By: paolo
Responsible-Changed-When: Fri May 16 17:25:57 2003
Responsible-Changed-Why:
    Analyzing.
State-Changed-From-To: open->feedback
State-Changed-By: paolo
State-Changed-When: Fri May 16 17:25:57 2003
State-Changed-Why:
    Hi guys. I'm suspecting this commit (of mine ;):
    
    2003-05-01  Paolo Carlini  <pcarlini@unitus.it>
                Nathan Myers  <ncm@cantrip.org>
    
    	Backport from mainline fix of 2003-03-28.
    	PR libstdc++/9533
    
    I would appreciate if you could report on inadequacies
    for your favorite platforms of the various system call
    used in  __basic_file<char>::showmanyc_helper
    (in config/io/basic_file_stdio.cc). In particular: does
    the autoconf test for 'fstat' succeed? If not, why?
    Thanks for your collaboration, Paolo.

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=10809

Re: libstdc++/10809: [3.3 regression] mips-sgi-irix6.5 libstdc++-v3 testsuite execution failure in 27_io/istream_unformatted.cc

[Eric Botcazou]

The following reply was made to PR libstdc++/10809; it has been noted by GNATS.

From: Eric Botcazou <ebotcazou@libertysurf.fr>
To: paolo@gcc.gnu.org
Cc: gcc-bugs@gcc.gnu.org,
 ghazi@caip.rutgers.edu,
 mark@codesourcery.com,
 nobody@gcc.gnu.org,
 gcc-gnats@gcc.gnu.org
Subject: Re: libstdc++/10809: [3.3 regression] mips-sgi-irix6.5 libstdc++-v3 testsuite execution failure in 27_io/istream_unformatted.cc
Date: Sat, 17 May 2003 15:59:48 +0200

 >     I would appreciate if you could report on inadequacies
 >     for your favorite platforms of the various system call
 >     used in  __basic_file<char>::showmanyc_helper
 >     (in config/io/basic_file_stdio.cc). In particular: does
 >     the autoconf test for 'fstat' succeed? If not, why?
 >     Thanks for your collaboration, Paolo.
 
 Solaris 2.5.1:
 
 sunshine% cd sparc-sun-solaris2.5.1/libstdc++-v3
 sunshine% grep "HAVE_SYS_IOCTL_H" *.h
 #define HAVE_SYS_IOCTL_H 1
 sunshine% grep "HAVE_SYS_FILIO_H" *.h
 #define HAVE_SYS_FILIO_H 1
 sunshine% grep "HAVE_POLL" *.h
 #define HAVE_POLL 1
 sunshine% grep "HAVE_S_ISREG" *.h
 #define HAVE_S_ISREG 1
 sunshine% grep "HAVE_S_IFREG" *.h
 /* #undef HAVE_S_IFREG */
 
 Solaris 2.6:
 
 beeba% cd sparc-sun-solaris2.6/libstdc++-v3
 beeba% grep "HAVE_SYS_IOCTL_H" *.h
 #define HAVE_SYS_IOCTL_H 1
 beeba% grep "HAVE_SYS_FILIO_H" *.h
 #define HAVE_SYS_FILIO_H 1
 beeba% grep "HAVE_POLL" *.h
 #define HAVE_POLL 1
 beeba% grep "HAVE_S_ISREG" *.h
 #define HAVE_S_ISREG 1
 beeba% grep "HAVE_S_IFREG" *.h
 /* #undef HAVE_S_IFREG */
 
 -- 
 Eric Botcazou

Re: libstdc++/10809: [3.3 regression] mips-sgi-irix6.5 libstdc++-v3 testsuite execution failure in 27_io/istream_unformatted.cc

[Paolo Carlini]

The following reply was made to PR libstdc++/10809; it has been noted by GNATS.

From: Paolo Carlini <pcarlini@unitus.it>
To: Eric Botcazou <ebotcazou@libertysurf.fr>
Cc: paolo@gcc.gnu.org,  gcc-bugs@gcc.gnu.org,  ghazi@caip.rutgers.edu, 
 nobody@gcc.gnu.org,  gcc-gnats@gcc.gnu.org
Subject: Re: libstdc++/10809: [3.3 regression] mips-sgi-irix6.5 libstdc++-v3
 testsuite execution failure in 27_io/istream_unformatted.cc
Date: Sun, 18 May 2003 00:44:03 +0200

 Thanks Eric.
 
 Now, in order to make further progress, I have to ask you to debug a bit 
 that call (*): perhaps 'ioctl' returns zero (success) but __num is 
 unexpectedly zero too? Or is 'poll' to fail? Or, finally, 'fstat' 
 behaves unexpectedly?
 
 Thanks in advance,
 Paolo.
 
 (*) Just copy the whole test file (together with the required 
 'istream_unformatted-1.tst') and replace <testsuite_hooks.h> by 
 <cassert> and each VERIFY by assert. Then build and step into that 
 in_avail call.
 
 
 
 

Re: libstdc++/10809: [3.3 regression] mips-sgi-irix6.5 libstdc++-v3 testsuite execution failure in 27_io/istream_unformatted.cc

[Eric Botcazou]

The following reply was made to PR libstdc++/10809; it has been noted by GNATS.

From: Eric Botcazou <ebotcazou@libertysurf.fr>
To: Paolo Carlini <pcarlini@unitus.it>
Cc: gcc-bugs@gcc.gnu.org,
 ghazi@caip.rutgers.edu,
 nobody@gcc.gnu.org,
 gcc-gnats@gcc.gnu.org
Subject: Re: libstdc++/10809: [3.3 regression] mips-sgi-irix6.5 libstdc++-v3 testsuite execution failure in 27_io/istream_unformatted.cc
Date: Mon, 19 May 2003 10:32:49 +0200

 > (*) Just copy the whole test file (together with the required
 > 'istream_unformatted-1.tst') and replace <testsuite_hooks.h> by
 > <cassert> and each VERIFY by assert. Then build and step into that
 > in_avail call.
 
 It's getting weirder: the log file contains
 
 Assertion failed: i != 0, file 
 /opt/build/eric/gcc-3_3-branch/src/libstdc++-v3/t
 estsuite/27_io/istream_unformatted.cc, line 606
 
 Now if I manually compile and run the test as you suggested, I get:
 
 Assertion failed: sum != 0, file istream_unformatted.cc, line 619
 
 and
 
 i = 4294967295 
 extracted = 0
 
 I get random bus errors with both GDB 5.2.1 and GDB 5.3. And I can't 
 step into in_avail (GDB jumps to the next line instead, after a little 
 delay), not too sure why.
 
 -- 
 Eric Botcazou