bin/74020: regexec() hangs with UTF-8 locales

[Jean-Yves Lefort]

>Number:         74020
>Category:       bin
>Synopsis:       regexec() hangs with UTF-8 locales
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    freebsd-bugs
>State:          open
>Quarter:        
>Keywords:       
>Date-Required:
>Class:          sw-bug
>Submitter-Id:   current-users
>Arrival-Date:   Tue Nov 16 23:40:23 GMT 2004
>Closed-Date:
>Last-Modified:
>Originator:     Jean-Yves Lefort
>Release:        FreeBSD 5.3-RELEASE i386
>Organization:
>Environment:
System: FreeBSD jsite.lefort.net 5.3-RELEASE FreeBSD 5.3-RELEASE #0: Fri Nov 12 15:27:39 CET 2004 jylefort at jsite.lefort.net:/usr/obj/usr/src/sys/JSITE i386
>Description:
In some situations, regexec() hangs.
>How-To-Repeat:
Compile this:

--- cut ---
#include <locale.h>
#include <sys/types.h>
#include <regex.h>
#include <assert.h>

int
main (int argc, char **argv)
{
  int status;
  regex_t test_re;
  regmatch_t pmatch[3];

  setlocale(LC_ALL, "");

  status = regcomp(&test_re, "foo=(.*) bar=(.*)", REG_EXTENDED);
  assert(status == 0);

  /* if the locale encoding is UTF-8, this call hangs */
  regexec(&test_re, "foo=one bar=two\302\251", test_re.re_nsub + 1, pmatch, 0);

  return 0;
}
--- cut ---

Works fine when executed with a non UTF-8 locale:

	$ LANG=en_US.ISO8859-1 ./test
	$

Hangs when executed with an UTF-8 locale:

	$ LANG=en_US.UTF-8 ./test
	<yawn>
>Fix:
>Release-Note:
>Audit-Trail:
>Unformatted: