From siegel@zrz.tu-berlin.de

This is a bug report for perl from anno4000@​lublin.zrz.tu-berlin.de,
generated with the help of perlbug 1.34 running under perl v5.8.1.

Benchmark​::runloop calls a benchmarked coderef with two spurious
arguments (the arguments of the call to runloop itself). That
is mostly no big deal, but a sub may check its argumets. In particular,
Inline-compiled parameterless functions do this and die in Benchmark.
(That is how the subject came up on clpm.)

The fix below (against 1.06 from Perl 5.8.5) changes the call to $subref
in runloop() so that @​_ is cleared. The (presumably timing-sensitive)
call to $c is unchanged. Clearing @​_ explicitly, say after line 648
is another possibility.

Note​: perlbug was run on a different machine from the one where the
patch was prepared. The auto-included "Site configuration information"
below is irrelevant to the report.

Regards, Anno

--- orig/lib/Benchmark.pm	Mon Feb 23 19:50:27 2004
+++ perl-5.8.5/lib/Benchmark.pm	Thu Nov  4 20:21:07 2004
@@ -648,7 +648,7 @@
     # &runloop a lot, and thus reduce additive errors.
     my $tbase = Benchmark->new(0)->[1];
     while ( ( $t0 = Benchmark->new(0) )->[1] == $tbase ) {} ;
-    &$subref;
+    $subref->();
     $t1 = Benchmark->new($n);
     $td = &timediff($t1, $t0);
     timedebug("runloop:",$td);
--- orig/lib/Benchmark.t	Thu Aug  7 21:26:47 2003
+++ perl-5.8.5/lib/Benchmark.t	Thu Nov  4 20:40:19 2004
@@ -8,7 +8,7 @@
 use warnings;
 use strict;
 use vars qw($foo $bar $baz $ballast);
-use Test::More tests => 193;
+use Test::More tests => 194;
 
 use Benchmark qw(:all);
 
@@ -41,6 +41,11 @@
 
 isa_ok(timeit(5, '++$bar'), 'Benchmark', "timeit eval");
 is ($bar, 5, "benchmarked code was run 5 times");
+
+# is coderef called with spurious arguments?
+timeit( 1, sub { $foo = @_ });
+is ($foo, 0, "benchmarked code called without arguments");
+
 
 print "# Burning CPU to benchmark things will take time...\n";
 
---

Site configuration information for perl v5.8.1​:

Configured by anno4000 at Wed Oct 1 12​:15​:29 MEST 2003.

Summary of my perl5 (revision 5.0 version 8 subversion 1) configuration​:
  Platform​:
  osname=linux, osvers=2.0.36, archname=i586-linux
  uname='linux lublin.zrz.tu-berlin.de 2.0.36 #2 mon jun 26 12​:24​:02 mest 2000 i586 unknown '
  config_args='-des'
  hint=recommended, useposix=true, d_sigaction=define
  usethreads=undef use5005threads=undef useithreads=undef usemultiplicity=undef
  useperlio=define d_sfio=undef uselargefiles=define usesocks=undef
  use64bitint=undef use64bitall=undef uselongdouble=undef
  usemymalloc=n, bincompat5005=undef
  Compiler​:
  cc='cc', ccflags ='-I/usr/local/include -D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64',
  optimize='-O2',
  cppflags='-I/usr/local/include'
  ccversion='', gccversion='2.7.2.3', gccosandvers=''
  intsize=4, longsize=4, ptrsize=4, doublesize=8, byteorder=1234
  d_longlong=define, longlongsize=8, d_longdbl=define, longdblsize=12
  ivtype='long', ivsize=4, nvtype='double', nvsize=8, Off_t='off_t', lseeksize=4
  alignbytes=4, prototype=define
  Linker and Libraries​:
  ld='cc', ldflags =' -L/usr/local/lib'
  libpth=/usr/local/lib /lib /usr/lib
  libs=-lnsl -lndbm -lgdbm -ldb -ldl -lm -lcrypt -lutil -lc -lposix
  perllibs=-lnsl -ldl -lm -lcrypt -lutil -lc -lposix
  libc=/lib/libc-2.0.7.so, so=so, useshrplib=false, libperl=libperl.a
  gnulibc_version='2.0'
  Dynamic Linking​:
  dlsrc=dl_dlopen.xs, dlext=so, d_dlsymun=undef, ccdlflags='-rdynamic'
  cccdlflags='-fpic', lddlflags='-shared -L/usr/local/lib'

Locally applied patches​:
 

@​INC for perl v5.8.1​:
  /opt/home/anno4000/lib/perl
  /usr/local/lib/perl5/5.8.1/i586-linux
  /usr/local/lib/perl5/5.8.1
  /usr/local/lib/perl5/site_perl/5.8.1/i586-linux
  /usr/local/lib/perl5/site_perl/5.8.1
  /usr/local/lib/perl5/site_perl
  .

Environment for perl v5.8.1​:
  HOME=/opt/home/anno4000
  LANG (unset)
  LANGUAGE (unset)
  LD_LIBRARY_PATH=/usr/X11R6/lib​:/usr/local/LessTif/Motif1.2/lib
  LOGDIR (unset)
  PATH=/opt/home/anno4000/bin​:/sbin​:/usr/local/bin​:/bin​:/usr/bin​:/usr/sbin​:/usr/X11R6/bin​:/usr/openwin/bin​:/usr/games
  PERL5LIB=/opt/home/anno4000/lib/perl
  PERL_BADLANG (unset)
  PERL_PRIVLIB=/usr/local/lib/perl5/5.8.1
  PERL_SITELIB=/usr/local/lib/perl5/site_perl/5.8.1
  SHELL=/bin/tcsh

From @rgs

Anno Siegel (via RT) wrote​:

Benchmark​::runloop calls a benchmarked coderef with two spurious
arguments (the arguments of the call to runloop itself). That
is mostly no big deal, but a sub may check its argumets. In particular,
Inline-compiled parameterless functions do this and die in Benchmark.
(That is how the subject came up on clpm.)

The fix below (against 1.06 from Perl 5.8.5) changes the call to $subref
in runloop() so that @​_ is cleared. The (presumably timing-sensitive)
call to $c is unchanged. Clearing @​_ explicitly, say after line 648
is another possibility.

Thanks, applied to bleadperl, bumping $Benchmark​::VERSION to 1.07,
as change #23473.

The RT System itself - Status changed from 'new' to 'open'

@rgs - Status changed from 'new' to 'resolved'