B::CC cannot sort

[p5pRT]

Migrated from rt.perl.org#2900 (status was 'resolved')

Searchable as RT2900$

[p5pRT]

This simple program:

for $k (sort { length $ENV{$b} <=> length $ENV{$a} } keys %ENV) {
print "$k=$ENV{$k}\n";
}

Under B::Bytecode, produces correct results, then SEGVs

Under B::C, produces correct results, then exits correctly.

Under B::CC, produces no results, but goes into recursive
doom in the sort, eventually SEGVing for lack of memory
in stack death.

--tom

#0 0x4015825e in tcgetattr ()
#1 0x40158535 in tcgetattr ()
#2 0x40158d6b in malloc ()
#3 0x40061e50 in Perl_safemalloc ()
#4 0x40076cd3 in S_more_xpv ()
#5 0x40076c7c in S_new_xpv ()
#6 0x400775f1 in L273 ()
#7 0x4007adfd in Perl_sv_setpvn ()
#8 0x4007ddd6 in Perl_newSVpvn ()
#9 0x4006d160 in Perl_hv_iterkeysv ()
#10 0x400aa655 in Perl_do_kv ()
#11 0x4008926c in Perl_pp_keys ()
#12 0x69e0 in pp_main () at /tmp/sort.p.c:1152
#13 0x4006f719 in Perl_runops_standard ()
#14 0x4009b6bd in sortcv ()
#15 0x4009ae99 in S_qsortsv ()
#16 0x40093590 in Perl_pp_sort ()
#17 0x6a0f in pp_main () at /tmp/sort.p.c:1155
#18 0x4006f719 in Perl_runops_standard ()
#19 0x4009b6bd in sortcv ()
#20 0x4009ae99 in S_qsortsv ()
#21 0x40093590 in Perl_pp_sort ()
#22 0x6a0f in pp_main () at /tmp/sort.p.c:1155
#23 0x4006f719 in Perl_runops_standard ()
#24 0x4009b6bd in sortcv ()
#25 0x4009ae99 in S_qsortsv ()
#26 0x40093590 in Perl_pp_sort ()
#27 0x6a0f in pp_main () at /tmp/sort.p.c:1155
#28 0x4006f719 in Perl_runops_standard ()
#29 0x4009b6bd in sortcv ()
#30 0x4009ae99 in S_qsortsv ()
#31 0x40093590 in Perl_pp_sort ()
#32 0x6a0f in pp_main () at /tmp/sort.p.c:1155
#33 0x4006f719 in Perl_runops_standard ()
#34 0x4009b6bd in sortcv ()
#35 0x4009ae99 in S_qsortsv ()
#36 0x40093590 in Perl_pp_sort ()
#37 0x6a0f in pp_main () at /tmp/sort.p.c:1155
#38 0x4006f719 in Perl_runops_standard ()
#39 0x4009b6bd in sortcv ()
#40 0x4009ae99 in S_qsortsv ()
#41 0x40093590 in Perl_pp_sort ()
#42 0x6a0f in pp_main () at /tmp/sort.p.c:1155
#43 0x4006f719 in Perl_runops_standard ()
#44 0x4009b6bd in sortcv ()
#45 0x4009ae99 in S_qsortsv ()
#46 0x40093590 in Perl_pp_sort ()
#47 0x6a0f in pp_main () at /tmp/sort.p.c:1155
#48 0x4006f719 in Perl_runops_standard ()
#49 0x4009b6bd in sortcv ()
#50 0x4009ae99 in S_qsortsv ()
#51 0x40093590 in Perl_pp_sort ()
#52 0x6a0f in pp_main () at /tmp/sort.p.c:1155
#53 0x4006f719 in Perl_runops_standard ()
#54 0x4009b6bd in sortcv ()
#55 0x4009ae99 in S_qsortsv ()
#56 0x40093590 in Perl_pp_sort ()
#57 0x6a0f in pp_main () at /tmp/sort.p.c:1155
#58 0x4006f719 in Perl_runops_standard ()
#59 0x4009b6bd in sortcv ()
#60 0x4009ae99 in S_qsortsv ()
#61 0x40093590 in Perl_pp_sort ()
#62 0x6a0f in pp_main () at /tmp/sort.p.c:1155
#63 0x4006f719 in Perl_runops_standard ()

[the following 25000 identical stackframes deleted]

[p5pRT]

From tchrist@chthon.perl.com

This simple program​:

  for $k (sort { length $ENV{$b} <=> length $ENV{$a} } keys %ENV) {
  print "$k=$ENV{$k}\n";
  }

Under B​::Bytecode, produces correct results, then SEGVs

Under B​::C, produces correct results, then exits correctly.

Under B​::CC, produces no results, but goes into recursive
  doom in the sort, eventually SEGVing for lack of memory
  in stack death.

--tom

#0 0x4015825e in tcgetattr ()
#1 0x40158535 in tcgetattr ()
#2 0x40158d6b in malloc ()
#3 0x40061e50 in Perl_safemalloc ()
#4 0x40076cd3 in S_more_xpv ()
#5 0x40076c7c in S_new_xpv ()
#6 0x400775f1 in L273 ()
#7 0x4007adfd in Perl_sv_setpvn ()
#8 0x4007ddd6 in Perl_newSVpvn ()
#9 0x4006d160 in Perl_hv_iterkeysv ()
#10 0x400aa655 in Perl_do_kv ()
#11 0x4008926c in Perl_pp_keys ()
#12 0x69e0 in pp_main () at /tmp/sort.p.c​:1152
#13 0x4006f719 in Perl_runops_standard ()
#14 0x4009b6bd in sortcv ()
#15 0x4009ae99 in S_qsortsv ()
#16 0x40093590 in Perl_pp_sort ()
#17 0x6a0f in pp_main () at /tmp/sort.p.c​:1155
#18 0x4006f719 in Perl_runops_standard ()
#19 0x4009b6bd in sortcv ()
#20 0x4009ae99 in S_qsortsv ()
#21 0x40093590 in Perl_pp_sort ()
#22 0x6a0f in pp_main () at /tmp/sort.p.c​:1155
#23 0x4006f719 in Perl_runops_standard ()
#24 0x4009b6bd in sortcv ()
#25 0x4009ae99 in S_qsortsv ()
#26 0x40093590 in Perl_pp_sort ()
#27 0x6a0f in pp_main () at /tmp/sort.p.c​:1155
#28 0x4006f719 in Perl_runops_standard ()
#29 0x4009b6bd in sortcv ()
#30 0x4009ae99 in S_qsortsv ()
#31 0x40093590 in Perl_pp_sort ()
#32 0x6a0f in pp_main () at /tmp/sort.p.c​:1155
#33 0x4006f719 in Perl_runops_standard ()
#34 0x4009b6bd in sortcv ()
#35 0x4009ae99 in S_qsortsv ()
#36 0x40093590 in Perl_pp_sort ()
#37 0x6a0f in pp_main () at /tmp/sort.p.c​:1155
#38 0x4006f719 in Perl_runops_standard ()
#39 0x4009b6bd in sortcv ()
#40 0x4009ae99 in S_qsortsv ()
#41 0x40093590 in Perl_pp_sort ()
#42 0x6a0f in pp_main () at /tmp/sort.p.c​:1155
#43 0x4006f719 in Perl_runops_standard ()
#44 0x4009b6bd in sortcv ()
#45 0x4009ae99 in S_qsortsv ()
#46 0x40093590 in Perl_pp_sort ()
#47 0x6a0f in pp_main () at /tmp/sort.p.c​:1155
#48 0x4006f719 in Perl_runops_standard ()
#49 0x4009b6bd in sortcv ()
#50 0x4009ae99 in S_qsortsv ()
#51 0x40093590 in Perl_pp_sort ()
#52 0x6a0f in pp_main () at /tmp/sort.p.c​:1155
#53 0x4006f719 in Perl_runops_standard ()
#54 0x4009b6bd in sortcv ()
#55 0x4009ae99 in S_qsortsv ()
#56 0x40093590 in Perl_pp_sort ()
#57 0x6a0f in pp_main () at /tmp/sort.p.c​:1155
#58 0x4006f719 in Perl_runops_standard ()
#59 0x4009b6bd in sortcv ()
#60 0x4009ae99 in S_qsortsv ()
#61 0x40093590 in Perl_pp_sort ()
#62 0x6a0f in pp_main () at /tmp/sort.p.c​:1155
#63 0x4006f719 in Perl_runops_standard ()

[the following 25000 identical stackframes deleted]

[p5pRT]

From @rspier

Dupe of #2900

[p5pRT]

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

[p5pRT]

From @schwern

[tchrist@​chthon.perl.com - Thu Mar 30 11​:30​:55 2000]​:

This simple program​:

for $k \(sort \{ length $ENV\{$b\} \<=> length $ENV\{$a\} \} keys %ENV\) \{
print "$k=$ENV\{$k\}\\n";
\} 

Under B​::Bytecode, produces correct results, then SEGVs

Now working in 5.8.6.

Under B​::C, produces correct results, then exits correctly.

Still working in 5.8.6.

Under B​::CC, produces no results, but goes into recursive
doom in the sort, eventually SEGVing for lack of memory
in stack death.

$ perlcc -O foo
/sw/bin/perlcc​: foo did not compile, which can't happen​:
Can't locate object method "_save_common_middle" via package "B​::FAKEOP"
at /sw/lib/perl5-core/5.8.6/darwin-thread-multi-2level/B/C.pm line 389.
CHECK failed--call queue aborted.

Same for bleadperl.

[p5pRT]

From guest@guest.guest.xxxxxxxx

  sub B​::OP​::_save_common_middle {
  my $op = shift;
  sprintf ("%s, %u, %u, $static, 0x%x, 0x%x",
  $op->fake_ppaddr, $op->targ, $op->type, $op->flags,
$op->private);
  }

Is the actual _save_common_middle code. Unfortuantely I don't quite
see where the inheritance is occuring for it to call that function
thusly​:

sub B​::FAKEOP​::save {
  my ($op, $level) = @​_;
  $opsect->add(sprintf("%s, %s, %s",
  $op->next, $op->sibling,
$op->_save_common_middle));
  my $ix = $opsect->index;
  $init->add(sprintf("op_list[$ix].op_ppaddr = %s;", $op->ppaddr))
  unless $optimize_ppaddr;
  return "&op_list[$ix]";
}

I tried browsing the code to see where $op-> would at all come up
relating to B​::Op, but didn't come up with anything.

[p5pRT]

@smpeters - Status changed from 'open' to 'stalled'

[p5pRT]

From @rurban

Ticket moved to [CPAN #53536] in the B-C queue. Please close it someone.
This simple pp_sort bug for B​::CC was fixed
in B-C-1.04_27, B​::CC 1.02_03

* CC.pm (1.02_03)​: fixed endless loop in optimized sort. The sortcv
  cmp needed a return NULL (test 19).

More pp_sort bugs are handled in
tests 18 and 25 and I'm working on it.

# 25 - simple sort
print sort { print $i++," "; $b <=> $a } 1..4
=> 0 1 2 3 4321

# 18 - custom key sort 18
my $h = { a=>3, b=>1 }; print sort {$h->{$a} <=> $h->{$b}} keys %$h
=> ba

# 19 - fool the sort optimizer by my $p. This pp_sort works ok now on CC
print sort { my $p; $b <=> $a } 1,4,3
431
--
Reini Urban

[p5pRT]

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

[p5pRT]

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