http://qs321.pair.com?node_id=617538

imp has asked for the wisdom of the Perl Monks concerning the following question:

I was revisiting Negating Regexes: Tips, Tools, And Tricks Of The Trade and caused perl to segfault with one of the patterns I tried. I reduced the pattern to he simplest form that still causes a problem. Tested with: 
This is perl, v5.8.7 built for i486-linux-gnu-thread-multi
Characteristics of this binary (from libperl):
  Compile-time options: MULTIPLICITY USE_ITHREADS USE_LARGE_FILES
                        PERL_IMPLICIT_CONTEXT

[download]

This works: 

  "1" =~ qr[(??{/^1/})]

[download]
This segfaults 
  "<" =~ qr[(??{/^</})]

[download]
Segfault re=debug output: 
Freeing REx: `","'
Compiling REx `(??{/^</})'
size 4 Got 36 bytes for offset annotations.
Compiling REx `^<'

size 4 Got 36 bytes for offset annotations.
first at 2
   1: BOL(2)
   2: EXACT <<>(4)
   4: END(0)
anchored `<' at 0 (checking anchored) anchored(BOL) minlen 1
Offsets: [4]
        1[1] 2[1] 0[0] 3[0]
first at 1
   1: LOGICAL[2](2)
   2: EVAL(4)
   4: END(0)
minlen 0 with eval
Offsets: [4]
        10[0] 10[0] 0[0] 11[0]
Matching REx `(??{/^</})' against `<'
  Setting an EVAL scope, savestack=15
   0 <> <<>               |  1:  LOGICAL[2]
   0 <> <<>               |  2:  EVAL
  re_eval 0x8153fe8
Guessing start of match, REx `^<' against `<'...
Guessed: match at offset 0
Matching REx `^<' against `<'
  Setting an EVAL scope, savestack=15
   0 <> <<>               |  1:  BOL
   0 <> <<>               |  2:  EXACT <<>
   1 <<> <>               |  4:  END
Match successful!
Compiling REx `1'
size 3 Got 28 bytes for offset annotations.
first at 1
   1: EXACT <1>(3)
   3: END(0)
anchored `1' at 0 (checking anchored isall) minlen 1
Offsets: [3]
        1[1] 0[0] 2[0]
Entering embedded `1'
  Setting an EVAL scope, savestack=21
   0 <> <<>               |  1:  EXACT <1>
                            failed...
Freeing REx: `"1"'
                          failed...
  Setting an EVAL scope, savestack=24
   1 <<> <>               |50339:  LOGICAL[2]
   1 <<> <>               |50340:  EVAL

[download]
Is this fixed in later perls?

Replies are listed 'Best First'.
Re: regex bug in perl 5.8.7
by dmitri (Priest) on May 25, 2007 at 16:44 UTC
[reply]
Re: regex bug in perl 5.8.7
by dave_the_m (Monsignor) on May 25, 2007 at 18:41 UTC
    Appears to be fixed in 5.9.4.

    Dave.

[reply]
      Damn, I was looking forward to submitting a bug report :-/
[reply]
        IIRC, demerphq has rewritten a substantial portion of the the regex system. If there are any bugs in it, they're likely new and different ones instead of the ones in earlier perls.


        Christopher E. Stith
[reply]
Re: regex bug in perl 5.8.7
by TGI (Parson) on May 25, 2007 at 16:58 UTC
    This example crashes for me with ActivePerl build 819, which is Perl 5.8.8


    TGI says moo

[reply]
Re: regex bug in perl 5.8.7
by Skeeve (Parson) on May 25, 2007 at 17:18 UTC
    Me too on PPC Mac 
    skeeve@Skeeves-PowerBook:~ $ perl -e '"<" =~ qr[(??{/^</})]'
Bus error
skeeve@Skeeves-PowerBook:~ $ perl -v

This is perl, v5.8.6 built for darwin-thread-multi-2level
(with 3 registered patches, see perl -V for more detail)

Copyright 1987-2004, Larry Wall

Perl may be copied only under the terms of either the Artistic License
+ or the
GNU General Public License, which may be found in the Perl 5 source ki
+t.

Complete documentation for Perl, including FAQ lists, should be found 
+on
this system using `man perl' or `perldoc perl'.  If you have access to
+ the
Internet, point your browser at http://www.perl.org/, the Perl Home Pa
+ge.

    [download]
    s$$([},&%#}/&/]+}%&{})*;#$&&s&&$^X.($'^"%]=\&(|?*{%
    +.+=%;.#_}\&"^"-+%*).}%:##%}={~=~:.")&e&&s""`$''`"e
[reply]
[d/l]
[select]
Re: regex bug in perl 5.8.7
by diotalevi (Canon) on May 25, 2007 at 23:22 UTC

    You need a re-entrant regexp engine to do what you tried. I don't think even the latest bleads are. Also, your pattern isn't terribly meaningful. Are you *sure* you meant to use the return value of a boolean regexp as another regexp pattern to compile and apply?

    ⠤⠤ ⠙⠊⠕⠞⠁⠇⠑⠧⠊

[reply]
      The pattern isn't meant to be meaningful. The regex I listed was already reduced to the simplest case I could find that still caused a segfault. The original form was much different.
[reply]

        In the future, you might want to use the (?{...}) form instead since that has no implications for compiling and interpolating another regexp.

        ⠤⠤ ⠙⠊⠕⠞⠁⠇⠑⠧⠊

[reply]
Re: regex bug in perl 5.8.7
by vrk (Chaplain) on May 25, 2007 at 18:22 UTC

    Confirmed, crashes here, too. 

    This is perl, v5.8.8 built for i486-linux

    [download]
      Compile-time options: PERL_MALLOC_WRAP USE_LARGE_FILES USE_PERLIO

    [download]

    --
    print "Just Another Perl Adept\n";

[reply]
[d/l]
[select]

Back to Seekers of Perl Wisdom