perl segmentation fault report #788
Comments
From lvirden@cas.org----------------------------------------------------------------- $ gdb $(whence perl) core The script in question is: #!/usr/bin/perl ################################################################################# ############################################################## my $Parser = "yahoo"; ParseParams (); chdir ( $ConfigPath ); LoadConfig (); if ( $FlipReg == 1 ) # pause (); if ( ! -d $DestDir ) pause (); if ( $#URLs == -1 ) pause (); if ( $TheaterList ne "" ) if ( # # my $nURL = 0; my %hDates = (); print "\n\n"; foreach my $URL ( @URLs ) { # print "Grabbing URL if ( $Parser eq "yahoo" ) { # # # # # # if ( $FlipReg ) # print "\n\nScript Complete.\n\n"; ##################### exit 1; foreach my $line ( <CONFIG> ) chomp $line; ( my $key, my $value ) = split ( /\s+/, $line, 2 ); SWITCH: { close ( CONFIG ); # sub grabpage { # my $response = $ua->request ( $request ); sub WritePDBHeader print "Writing pdb $DBName with $nRecords records.\n"; my $attributes = 0x8000; my $now = time (); my $creationdate = $now; sub WriteDatePDB # Write out the headers syswrite ( PDB, pack ( "N", $nOffset ), 4 ); # Offset syswrite ( PDB, pack ( "xxx" ), 3 ); # Unique ID, leave zero $nOffset += GetDateSize ( $date ); # Pad two bytes # Write out the data sub GetDateSize my $size = 4; # one int return $size; sub WriteTheaterPDB # Write out the headers syswrite ( PDB, pack ( "N", $nOffset ), 4 ); # Offset syswrite ( PDB, pack ( "xxx" ), 3 ); # Unique ID, leave zero $nOffset += GetTheaterSize ( $theater ); # Pad two bytes # Write out the data # Write the info sub GetTheaterSize my $size = 4; # Two strings, two nulls, one offset return $size; sub WriteMoviePDB syswrite ( PDB, pack ( "N", $nOffset ), 4 ); # Offset syswrite ( PDB, pack ( "xxx" ), 3 ); # Unique ID, leave zero $nOffset += GetMovieSize ( $movie ); # Pad two bytes # Write out the data # Write the rating, if it exists sub GetMovieSize my $size = 7; # Three strings, three nulls, two offsets return $size; sub WriteShowtimePDB syswrite ( PDB, pack ( "xxx" ), 3 ); # Unique ID, leave zero $nOffset += GetShowtimeSize ( $showtime ); # Pad two bytes # Write out the data sub GetShowtimeSize my $size = 8; # four shorts return $size; sub WriteStringPDB syswrite ( PDB, pack ( "xxx" ), 3 ); # Unique ID, leave zero $nOffset += GetStringSize ( $string ); # Pad two bytes # Write out the data sub GetStringSize my $size = 1; # one string, one null return $size; sub syswritestr sub errormsg sub printmsg print "$_[0]\n"; sub pause sub ucwords sub AlloCineParse # Translate linefeeds # Remove blank lines $daynum++; my $mode = ""; if ( $mode eq "SALLES" ) #translate next if ( IgnoringTheater ( $theatername )); my $theaterinfo = sprintf ( "%s, %s, %s", $address, $zip, $city ); $hTheaters{ $TID }{ "Name" } = $theatername; next if ( !exists ( $hTheaters{ $TID })); my %daytimes = (); my $lastcnk = "time"; @incdays = $cnk; $lastcnk = "dow"; push @times, $cnk; $lastcnk = "time"; filltimes ( \%daytimes, \@incdays, \@excdays, \@times ); foreach my $day ( keys %daytimes ) #translate # Make sure there is showtime data for this movie $hMovies{ $MID }{ "Name" } = $moviename; last; #### auYahooParse contributed by Duncan Sargeant # " (to balance my colouriser :-) sub auYahooParse my ($content, $URL) = @_; # Figure out day offsets # $content is the list of cinemas in the selected area. foreach (@theaters) { print " Getting theater $theatername...\n"; my $foundtoday = 0; $foundtoday = 1 if ( $days { $day } == $DOW ); print " Getting day $day...\n"; $content = grabpage ("$URL/$dir/$day.html"); # I think is has been fixed -jrray my $time = timegm ( 0, 0, 0, $date, $month, $year ); # If we are running on a Mac, we need to fudge the time and my $DID = -1; $content =~ />([^>][^>][^>]+)\n/s; my $TID = -1; my @movies = split (/<li>/i, $content); my ($moviename, $showtimes) = split (/ - /, $movie); my $MID = -1; # Check that this ID combo hasn't been stored already # sub YahooParse my $content = $_[0]; # # Translate metaspaces into newlines $time = timegm ( 0, 0, 0, $day, $month, $year ); # If we are running on a Mac, we need to fudge the time and my $DID = -1; if ( IgnoringTheater ( $theatername )) my $TID = -1; $hTheaters{ $TID }{ "Name" } = $theatername; #querymovie ( $moviename ); # Check that this ID combo hasn't been stored already # Lop of extra garbage at the end # Remove the trailing linefeed my $SID = -1; sub ScootParse #### GARETH'S SCOOT CODE STARTS HERE ### my %months = ( 'January' => 0, # If we are running on a Mac, we need to fudge the time and $DID = -1; if ($line =~ /r=CINOC[^>]*>(.*?)<\//) { # If this theater is in our kill file, ignore it -- jrray $tline=1; ##################### # Get rid of those pesky ,;'s # # Check that this ID combo hasn't been stored already sub IgnoringTheater if ( exists $TheaterIgnore{ $theatername }) return 0; sub printspaces sub querymovie $title =~ s/\ /\+/g; my $URL = "http://us.imdb.com/Tfuzzy?title=" . $title . "&type=fuzzy&sort=chrono&tv=off"; #my $URL = "http://us.imdb.com/Plot?" . $title; print $URL . "\n"; #my $request = new HTTP::Request ( 'GET', $URL ); sub filltimes if ( $#{$_[3]} >= 0 ) if ( exists ( ${_[0]}{ $day })) my @times = split ( /,\s*/, ${$_[0]}{ $day }); sub ParseParams sub PrintUsage pause (); Perl Info |
From [Unknown Contact. See original ticket]I recompiled perl with -g and here is the stack trace I am getting. program terminated by signal SEGV (no mapping at the fault address) |
From [Unknown Contact. See original ticket]Larry W. Virden writes:
Edit "cflags" file. I have #optdebug='' # ensure -g used if building a -DDEBUGGING libperl
Well, having something less than 1500+ lines long script would be Ilya |
From [Unknown Contact. See original ticket]Larry W. Virden writes:
And may I ask what is the reason for segfault? What is the contents #define CHR_SVLEN(sv) (UTF ? sv_len_utf8(sv) : SvCUR(sv)) and I suspect you do not use utf8, right?
Hey, I have seen this string (rendered ;-) yesterday! How did you Ilya |
From [Unknown Contact. See original ticket]From: Ilya Zakharevich <ilya@math.ohio-state.edu> Larry W. Virden writes: Current function is Perl_re_intuit_start dbx: see `help scope' for details Unfortunately, I'm not familiar enough with Perl's internals to -- |
From [Unknown Contact. See original ticket]On Tue, Oct 26, 1999 at 02:35:01PM -0400, Larry W. Virden wrote:
Well, it is not rocket science to resolve macros (at least if you run #define check_substr substrs->data[2].substr
Well, I think you know it already. *((GV*)sv)->sv_any is the simplest Ilya |
From [Unknown Contact. See original ticket]From: Ilya Zakharevich <ilya@math.ohio-state.edu> Well, I think you know it already. *((GV*)sv)->sv_any is the print *((GV*)sv)->sv_any -- |
From [Unknown Contact. See original ticket]Larry W. Virden writes:
This is all pretty nice, but I want to know why *segfault* happens. Ilya |
From [Unknown Contact. See original ticket]Larry W. Virden writes:
This is a plug only. It fixes a bug, but uncovers yet another Enjoy, Inline Patch--- ./t/op/pat.t~ Tue Oct 5 23:20:31 1999
+++ ./t/op/pat.t Mon Nov 1 18:08:47 1999
@@ -4,7 +4,7 @@
# the format supported by op/regexp.t. If you want to add a test
# that does fit that format, add it to op/re_tests, not here.
-print "1..193\n";
+print "1..194\n";
BEGIN {
chdir 't' if -d 't';
@@ -891,5 +891,10 @@ $test++;
$text = "aaXbXcc";
pos($text)=0;
$text =~ /\GXb*X/g and print 'not ';
+print "ok $test\n";
+$test++;
+
+$text = "xA\n" x 500;
+$text =~ /^\s*A/m and print 'not ';
print "ok $test\n";
$test++;
--- ./regexec.c~ Sun Oct 10 15:25:52 1999
+++ ./regexec.c Mon Nov 1 18:02:03 1999
@@ -642,6 +642,9 @@ Perl_re_intuit_start(pTHX_ regexp *prog,
prog->check_substr = Nullsv; /* disable */
prog->float_substr = Nullsv; /* clear */
s = strpos;
+ /* XXXX This is a remnant of the old implementation. It
+ looks wasteful, since now INTUIT can use many
+ other heuristics too. */
prog->reganch &= ~RE_USE_INTUIT;
}
else
@@ -804,9 +807,13 @@ Perl_regexec_flags(pTHX_ register regexp
after_try:
if (s >= end)
goto phooey;
- s = re_intuit_start(prog, sv, s + 1, strend, flags, NULL);
- if (!s)
- goto phooey;
+ if (prog->reganch & RE_USE_INTUIT) {
+ s = re_intuit_start(prog, sv, s + 1, strend, flags, NULL);
+ if (!s)
+ goto phooey;
+ }
+ else
+ s++;
}
} else {
if (s > startpos) |
From [Unknown Contact. See original ticket]Thank you for the patch. Unfortunately, after applying that patch and pragma/warnings.....PROG: |
From [Unknown Contact. See original ticket]On Tue, Nov 02, 1999 at 06:47:57AM -0500, Larry W. Virden wrote:
Was not that seen before? I do not get this warning. Are you running Ilya |
From [Unknown Contact. See original ticket]Re: was the error seen before? No, this is the first time I have seen this error. Re: am I running with PERL_DESTRUCT_LEVEL > 3? I have no such environment variable set - does the make test set -- |
From [Unknown Contact. See original ticket]On Tue, Nov 02, 1999 at 03:37:24PM -0500, Larry W. Virden wrote:
On p5p?
It is set to 2 if not defined. The additional message "Scalars Ilya |
From [Unknown Contact. See original ticket]On Nov 2, 17:16, Ilya Zakharevich wrote: I've never seen this error in person or on p5p <smile>... --> > Re: am I running with PERL_DESTRUCT_LEVEL > 3? Interesting... -- |
From [Unknown Contact. See original ticket]I'm continuing to see this segmentation fault. The last suggestion from |
From @andk
Ilya said, that what you saw could typically only be seen with -- |
From [Unknown Contact. See original ticket]Andreas J. Koenig writes:
Nope. The whole running program is usable. "Scalar leaked:" message Ilya |
From @gsarOn Thu, 18 Nov 1999 11:04:25 EST, "Larry W. Virden" wrote:
The test harness enables PERL_DESTRUCT_LEVEL for all tests not Sarathy |
From [Unknown Contact. See original ticket]Gurusamy Sarathy writes:
Nope. The test in question *already has* "we have a known leak" clause. Ilya |
From @gsarOn Thu, 18 Nov 1999 14:13:36 EST, Ilya Zakharevich wrote:
Is the "we have a known leak" clause in a BEGIN block? That would Sarathy |
From [Unknown Contact. See original ticket]On Thu, Nov 18, 1999 at 11:33:33AM -0800, Gurusamy Sarathy wrote:
IIRC it was a the misc test failing during compilation indeed. But Ilya |
Migrated from rt.perl.org#1715 (status was 'resolved')
Searchable as RT1715$
The text was updated successfully, but these errors were encountered: