threaded regex problem in _57

[p5pRT]

Migrated from rt.perl.org#974 (status was 'resolved')

Searchable as RT974$

[p5pRT]

From bl@incyte.com

The following script reaches the 'ERROR' part on a multi-processor
box. It doesn't fail on a uni-proc box.

#!/opt/perl5.005_57/bin/perl

use Thread;

my $msg = 'aaa';
my @​thrds;
my $cnt=100;
foreach my $n ( 0..$cnt )
{
  $thrds[$n] = new Thread \&loop, $msg;
}
foreach $thrd ( @​thrds )
{
  $thrd->join;
}

sub loop
{
  my $string = shift;
  while ( 1 )
  {
  my ( $i ) = ( $string =~ /^.(.).$/ );
  unless ( $i )
  {
  print "ERROR​: string​: '$string'\t\ti​: '$i'.\n";
  kill 9 => $$;
  }
  }
}

It usually takes less than a second to die.
Looks like $1..$9 aren't sufficiently protected between threads.

Tested under IRIX and Solaris.

bl

Perl Info

Site configuration information for perl 5.00557:

Configured by bl at Wed Jul  7 15:23:53 PDT 1999.

Summary of my perl5 (revision 5.0 version 5 subversion 57) configuration:
  Platform:
    osname=irix, osvers=6.5, archname=IP30-irix-thread
    uname='irix64 blah 6.5 04151556 ip30 mips '
    config_args='-Dprefix=/opt/perl5.005_57 -Dusethreads -Doptimize=-g'
    hint=recommended, useposix=true, d_sigaction=define
    usethreads=define useperlio=undef d_sfio=undef
    use64bits=undef usemultiplicity=undef
  Compiler:
    cc='cc -n32', optimize='-g', gccversion=
    cppflags='-D_BSD_TYPES -D_BSD_TIME -OPT:Olimit=0:space=ON -DDEBUGGING -I/usr/local/include -I/usr/gnu/include -DLANGUAGE_C'
    ccflags ='-D_BSD_TYPES -D_BSD_TIME -woff 1009,1110,1184 -OPT:Olimit=0:space=ON -DDEBUGGING -I/usr/local/include -I/usr/gnu/include -DLANGUAGE_C'
    stdchar='unsigned char', d_stdstdio=define, usevfork=false
    intsize=4, longsize=4, ptrsize=4, doublesize=8
    d_longlong=define, longlongsize=8, d_longdbl=define, longdblsize=16
    alignbytes=8, usemymalloc=n, prototype=define
  Linker and Libraries:
    ld='cc -n32', ldflags =' -L/usr/local/lib32 -L/usr/local/lib -L/usr/gnu/lib'
    libpth=/usr/local/lib /usr/gnu/lib /usr/lib32 /lib32 /lib /usr/lib
    libs=-ldb -lm -lrt -lpthread
    libc=/usr/lib32/libc.so, so=so, useshrplib=false, libperl=libperl.a
  Dynamic Linking:
    dlsrc=dl_dlopen.xs, dlext=so, d_dlsymun=undef, ccdlflags=' '
    cccdlflags=' ', lddlflags='-n32 -shared -L/usr/local/lib32 -L/usr/local/lib -L/usr/gnu/lib'

Locally applied patches:
    


@INC for perl 5.00557:
    /opt/perl5.005_57/lib/5.00557/IP30-irix-thread
    /opt/perl5.005_57/lib/5.00557
    /opt/perl5.005_57/lib/site_perl/5.00557/IP30-irix-thread
    /opt/perl5.005_57/lib/site_perl/5.00557
    .


Environment for perl 5.00557:
    HOME=/u/cuss/bl
    LANG=
    LANGUAGE (unset)
    LD_LIBRARY_PATH=/usr/X11/xview:/usr/local/lib:/usr/freeware/lib:/usr/sgitcl/lib:/u/cuss/bl/lib/sgi
    LOGDIR (unset)
    PATH=/usr/cdr/bin/SGI:/u/cuss/bl/bin:/u/cuss/bl/bin/irix:/usr/local/bin:/opt/MIPSpro/bin:/usr/freeware/bin:/usr/Cadmin/bin:/usr/Motif-1.2/bin:/usr/Motif-2.1/bin:/usr/OnRamp/bin:/usr/SGImeeting/bin:/usr/ToolTalk/bin:/usr/WebFace/bin:/usr/demos/bin:/usr/diags/bin:/usr/gnu/bin:/usr/java/bin:/usr/pcp/bin:/usr/pkg/bin:/usr/sgitcl/bin:/usr/ssm/bin:/usr/sysadm/bin:/sbin:/usr/sbin:/bin:/usr/bin:/usr/bsd:/bin/X11:/usr/etc:/etc:/usr/lib
    PERL_BADLANG (unset)
    SHELL=/bin/tcsh

[p5pRT]

From [Unknown Contact. See original ticket]

Hi,

in my search for 'the bug' in our code, I stumbled upon the "real"
problem ... which was identified as bug 19990707.008. as far as I can
see this problem doesn't seem to be fixed. can anyone tell me wrong ? or
what's the status on this one ?

the original msg :
http​://bugs.perl.org/perlbug.cgi?req=tidmids&tidmids=19990707.008&range=32405&format=h

a proposed solution :
http​://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/1999-11/msg00809.html

Christophe VG

[p5pRT]

From @gsar

On Mon, 03 Jul 2000 18​:08​:48 +0200, Christophe VG wrote​:
[about the match-variables-aren't-thread-safe issue]

in my search for 'the bug' in our code, I stumbled upon the "real"
problem ... which was identified as bug 19990707.008. as far as I can
see this problem doesn't seem to be fixed. can anyone tell me wrong ? or
what's the status on this one ?

Harry Wolfson <HarryWolfson@​ll.mit.edu> worked on an implementation,
but had to give up on it for lack of time. IIRC, his partial
implementation still had problems with the rather seriously broken
implementation of (?{}) and (??{}) (which uses a fake PL_curpad
without ensuring that it is a proper clone of the real pad).

Please talk to Harry to find out more if you're interested in
working on this.

Sarathy
gsar@​ActiveState.com

[p5pRT]

@gbarr - Status changed from 'open' to 'resolved'

[p5pRT]

From @gbarr

Bug was with 5.005 threads which are no longer supported