PerlIO::via isn't thread safe #8713

p5pRT · 2006-12-18T17:58:58Z

Migrated from rt.perl.org#41106 (status was 'open')

Searchable as RT41106$

p5pRT · 2006-12-18T17:59:12Z

From blgl@hagernas.com

Created by blgl@hagernas.com

This outputs "Hello=3DWorld!" as expected but then crashes.

### begin example program
#! /usr/unstable/bin/perl

use strict;
use warnings;
use threads;
use PerlIO::via::QuotedPrint;

open(my $foo,'>&:via(QuotedPrint)',*STDOUT)
or die "open: $!";
print $foo "Hello=World!\n"
or die "print: $!";
async {
print $foo "Hello=from=Thread!\n";
}->join();
### end example program

### begin debugger log
Program received signal EXC_BAD_ACCESS, Could not access memory.
Reason: KERN_INVALID_ADDRESS at address: 0xabababaf
0x0008f580 in Perl_gv_fetchpvn_flags (my_perl=0x1811400,
nambeg=0xbfffe520 "UNIVERSAL::", full_len=11, flags=0, sv_type=12) at
gv.c:815
815 if (!stash || !SvREFCNT(stash)) /* symbol table
under destruction */
(gdb) print stash
$1 = (HV *) 0xabababab
(gdb) bt
#0 0x0008f580 in Perl_gv_fetchpvn_flags (my_perl=0x1811400,
nambeg=0xbfffe520 "UNIVERSAL::", full_len=11, flags=0, sv_type=12) at
gv.c:815
#1 0x0008ece0 in Perl_gv_stashpvn (my_perl=0x1811400, name=0x2d3bf0
"UNIVERSAL", namelen=11, create=0) at gv.c:736
#2 0x0008c410 in Perl_gv_fetchmeth (my_perl=0x1811400,
stash=0x182af64, name=0x71cf34 "GETARG", len=6, level=0) at gv.c:410
#3 0x007167b8 in PerlIOVia_fetchmethod (my_perl=0x1811400,
s=0x1137e20, method=0x71cf34 "GETARG", save=0x1137e58) at via.xs:46
#4 0x00716948 in PerlIOVia_method (my_perl=0x1811400, f=0x1101954,
method=0x71cf34 "GETARG", save=0x1137e58, flags=0) at via.xs:72
#5 0x0071ba6c in PerlIOVia_getarg (my_perl=0x1811400, f=0x1101954,
param=0xbfffebd8, flags=1) at via.xs:576
#6 0x00008ea4 in PerlIOBase_dup (my_perl=0x1811400, f=0x1142a94,
o=0x1101954, param=0xbfffebd8, flags=1) at perlio.c:2250
#7 0x0071bac8 in PerlIOVia_dup (my_perl=0x1811400, f=0x1142a94,
o=0x1101954, param=0xbfffebd8, flags=1) at via.xs:583
#8 0x000031ec in PerlIO_fdupopen (my_perl=0x1811400, f=0x1101954,
param=0xbfffebd8, flags=1) at perlio.c:561
#9 0x001b10f4 in Perl_fp_dup (my_perl=0x1811400, fp=0x1101954,
type=0 '\0', param=0xbfffebd8) at sv.c:9499
#10 0x00003920 in PerlIO_clone (my_perl=0x1811400, proto=0x1800400,
param=0xbfffebd8) at perlio.c:673
#11 0x001b9124 in perl_clone (proto_perl=0x1800400, flags=2) at sv.c:
10868
#12 0x005d70d0 in S_ithread_create (my_perl=0x1800400,
init_function=0x18354b4, stack_size=0, gimme=0, exit_opt=0,
params=0x18357e8) at threads.xs:649
#13 0x005da560 in XS_threads_create (my_perl=0x1800400, cv=0x182ad60)
at threads.xs:930
#14 0x00138fd4 in Perl_pp_goto (my_perl=0x1800400) at pp_ctl.c:2415
#15 0x0027dad0 in Perl_runops_debug (my_perl=0x1800400) at dump.c:1874
#16 0x00056db4 in S_run_body (my_perl=0x1800400, oldscope=1) at
perl.c:2403
#17 0x000560b4 in perl_run (my_perl=0x1800400) at perl.c:2323
#18 0x00002a28 in main (argc=4, argv=0xbffff804, env=0xbffff818) at
perlmain.c:113
### end debugger log

Perl Info


Flags:
     category=library
     severity=high

Site configuration information for perl 5.9.5:

Configured by blgl at Sat Dec  9 12:49:18 CET 2006.

Summary of my perl5 (revision 5 version 9 subversion 5 patch 29492)  
configuration:
   Platform:
     osname=darwin, osvers=8.8.0, archname=darwin-thread- 
multi-64int-2level
     uname='darwin sunwukung.local 8.8.0 darwin kernel version 8.8.0:  
fri sep 8 17:18:57 pdt 2006; root:xnu-792.12.6.obj~1release_ppc power  
macintosh powerpc '
     config_args=''
     hint=previous, useposix=true, d_sigaction=define
     useithreads=define, usemultiplicity=define
     useperlio=define, d_sfio=undef, uselargefiles=define,  
usesocks=undef
     use64bitint=define, use64bitall=undef, uselongdouble=undef
     usemymalloc=n, bincompat5005=undef
   Compiler:
     cc='cc', ccflags ='-fno-common -DPERL_DARWIN -no-cpp-precomp - 
DDEBUGGING -fno-strict-aliasing -pipe -I/usr/local/include',
     optimize='-g',
     cppflags='-no-cpp-precomp -fno-common -DPERL_DARWIN -no-cpp- 
precomp -DDEBUGGING -fno-strict-aliasing -pipe -I/usr/local/include - 
fno-common -DPERL_DARWIN -no-cpp-precomp -DDEBUGGING -fno-strict- 
aliasing -pipe -I/usr/local/include'
     ccversion='', gccversion='4.0.1 (Apple Computer, Inc. build  
5363)', gccosandvers=''
     intsize=4, longsize=4, ptrsize=4, doublesize=8, byteorder=87654321
     d_longlong=define, longlongsize=8, d_longdbl=define, longdblsize=16
     ivtype='long long', ivsize=8, nvtype='double', nvsize=8,  
Off_t='off_t', lseeksize=8
     alignbytes=8, prototype=define
   Linker and Libraries:
     ld='env MACOSX_DEPLOYMENT_TARGET=10.3 cc', ldflags =' -L/usr/ 
local/lib'
     libpth=/usr/local/lib /usr/lib
     libs=-ldbm -ldl -lm -lc
     perllibs=-ldl -lm -lc
     libc=/usr/lib/libc.dylib, so=dylib, useshrplib=false,  
libperl=libperl.a
     gnulibc_version=''
   Dynamic Linking:
     dlsrc=dl_dlopen.xs, dlext=bundle, d_dlsymun=undef, ccdlflags=' '
     cccdlflags=' ', lddlflags=' -bundle -undefined dynamic_lookup -L/ 
usr/local/lib'

Locally applied patches:
     DEVEL


@INC for perl 5.9.5:
     /usr/unstable/lib/perl5/5.9.5/darwin-thread-multi-64int-2level
     /usr/unstable/lib/perl5/5.9.5
     /usr/unstable/lib/perl5/site_perl/5.9.5/darwin-thread- 
multi-64int-2level
     /usr/unstable/lib/perl5/site_perl/5.9.5
     .


Environment for perl 5.9.5:
     DYLD_LIBRARY_PATH (unset)
     HOME=/Users/blgl
     LANG (unset)
     LANGUAGE (unset)
     LD_LIBRARY_PATH (unset)
     LOGDIR (unset)
     PATH=/Users/blgl/bin:/Users/blgl/bin/powerpc-apple-darwin:/usr/ 
local/bin:/usr/local/sbin:/bin:/sbin:/usr/bin:/usr/sbin:/usr/arla/bin
     PERL_BADLANG (unset)
     SHELL=/bin/tcsh

p5pRT · 2006-12-21T15:32:36Z

From blgl@hagernas.com

It turns out that PerlIO::via and PerlIO::encoding are both thread
unsafe for exactly the same reason: trying to execute Perl code
in the new interpreter which is in the middle of being cloned.
Attempted call graph:

The least ugly solution I can think of involves adding a helper
function which does everything PerlIOBase_dup does, except for
calling virtual functions. PerlIOVia_dup and PerlIOEncode_dup
would (in the PERLIO_DUP_CLONE case) call this and then copy
any extra data from the old instance to the new one with sv_dup
as appropriate.

Any better ideas?

/Bo Lindbergh

p5pRT · 2016-07-14T19:15:11Z

From @dcollinsn

dcollins@nightshade64:~/toolchain$ perl5.8.8-thread-multi 41106.pl
Hello=3DWorld!
Bus error
dcollins@nightshade64:~/toolchain$ perl5.25.2-thread-multi 41106.pl
Hello=3DWorld!
No package specified at 41106.pl line 12.
No package specified at 41106.pl line 12.
Filehandle $foo opened only for input at 41106.pl line 11.
Unbalanced string table refcount: (1) for "GETARG" during global destruction.
Attempt to free nonexistent shared string 'GETARG', Perl interpreter: 0x1f4a010 during global destruction.

--
Respectfully,
Dan Collins

p5pRT · 2016-12-31T14:56:05Z

From @jkeenan

On Mon, 18 Dec 2006 17:59:12 GMT, blgl@hagernas.com wrote:

This is a bug report for perl from blgl@hagernas.com,
generated with the help of perlbug 1.35 running under perl 5.9.5.

-----------------------------------------------------------------
[Please enter your report here]

This outputs "Hello=3DWorld!" as expected but then crashes.

### begin example program
#! /usr/unstable/bin/perl

use strict;
use warnings;
use threads;
use PerlIO::via::QuotedPrint;

open(my $foo,'>&:via(QuotedPrint)',*STDOUT)
or die "open: $!";
print $foo "Hello=World!\n"
or die "print: $!";
async {
print $foo "Hello=from=Thread!\n";
}->join();
### end example program

### begin debugger log
Program received signal EXC_BAD_ACCESS, Could not access memory.
Reason: KERN_INVALID_ADDRESS at address: 0xabababaf
0x0008f580 in Perl_gv_fetchpvn_flags (my_perl=0x1811400,
nambeg=0xbfffe520 "UNIVERSAL::", full_len=11, flags=0, sv_type=12) at
gv.c:815
815 if (!stash || !SvREFCNT(stash)) /* symbol table
under destruction */
(gdb) print stash
$1 = (HV *) 0xabababab
(gdb) bt
#0 0x0008f580 in Perl_gv_fetchpvn_flags (my_perl=0x1811400,
nambeg=0xbfffe520 "UNIVERSAL::", full_len=11, flags=0, sv_type=12) at
gv.c:815
#1 0x0008ece0 in Perl_gv_stashpvn (my_perl=0x1811400, name=0x2d3bf0
"UNIVERSAL", namelen=11, create=0) at gv.c:736
#2 0x0008c410 in Perl_gv_fetchmeth (my_perl=0x1811400,
stash=0x182af64, name=0x71cf34 "GETARG", len=6, level=0) at gv.c:410
#3 0x007167b8 in PerlIOVia_fetchmethod (my_perl=0x1811400,
s=0x1137e20, method=0x71cf34 "GETARG", save=0x1137e58) at via.xs:46
#4 0x00716948 in PerlIOVia_method (my_perl=0x1811400, f=0x1101954,
method=0x71cf34 "GETARG", save=0x1137e58, flags=0) at via.xs:72
#5 0x0071ba6c in PerlIOVia_getarg (my_perl=0x1811400, f=0x1101954,
param=0xbfffebd8, flags=1) at via.xs:576
#6 0x00008ea4 in PerlIOBase_dup (my_perl=0x1811400, f=0x1142a94,
o=0x1101954, param=0xbfffebd8, flags=1) at perlio.c:2250
#7 0x0071bac8 in PerlIOVia_dup (my_perl=0x1811400, f=0x1142a94,
o=0x1101954, param=0xbfffebd8, flags=1) at via.xs:583
#8 0x000031ec in PerlIO_fdupopen (my_perl=0x1811400, f=0x1101954,
param=0xbfffebd8, flags=1) at perlio.c:561
#9 0x001b10f4 in Perl_fp_dup (my_perl=0x1811400, fp=0x1101954,
type=0 '\0', param=0xbfffebd8) at sv.c:9499
#10 0x00003920 in PerlIO_clone (my_perl=0x1811400, proto=0x1800400,
param=0xbfffebd8) at perlio.c:673
#11 0x001b9124 in perl_clone (proto_perl=0x1800400, flags=2) at sv.c:
10868
#12 0x005d70d0 in S_ithread_create (my_perl=0x1800400,
init_function=0x18354b4, stack_size=0, gimme=0, exit_opt=0,
params=0x18357e8) at threads.xs:649
#13 0x005da560 in XS_threads_create (my_perl=0x1800400, cv=0x182ad60)
at threads.xs:930
#14 0x00138fd4 in Perl_pp_goto (my_perl=0x1800400) at pp_ctl.c:2415
#15 0x0027dad0 in Perl_runops_debug (my_perl=0x1800400) at dump.c:1874
#16 0x00056db4 in S_run_body (my_perl=0x1800400, oldscope=1) at
perl.c:2403
#17 0x000560b4 in perl_run (my_perl=0x1800400) at perl.c:2323
#18 0x00002a28 in main (argc=4, argv=0xbffff804, env=0xbffff818) at
perlmain.c:113
### end debugger log

The way perl handles this appears to have been mitigated over the years since perl-5.9.5.

I added a one-line 'print' statement to your original program (see attachment).

I built threaded perls at several tags and ran your program at each.

#####
[v5.12.5] $ ./bin/perl -Ilib ~/learn/perl/p5p/41106-threads-quoted-print.pl
Hello=3DWorld!
No package specified at /home/jkeenan/learn/perl/p5p/41106-threads-quoted-print.pl line 12.
No package specified at /home/jkeenan/learn/perl/p5p/41106-threads-quoted-print.pl line 12.
Filehandle $foo opened only for input at /home/jkeenan/learn/perl/p5p/41106-threads-quoted-print.pl line 11.
Unbalanced string table refcount: (1) for "/home/jkeenan/learn/perl/p5p/41106-threads-quoted-print.pl" during global destruction.
Unbalanced string table refcount: (1) for "_GEN_0" during global destruction.
Unbalanced string table refcount: (1) for "GETARG" during global destruction.
Scalars leaked: 1
Finished
Segmentation fault (core dumped)

[v5.16.3] $ ./bin/perl -Ilib ~/learn/perl/p5p/41106-threads-quoted-print.pl
Hello=3DWorld!
No package specified at /home/jkeenan/learn/perl/p5p/41106-threads-quoted-print.pl line 12.
No package specified at /home/jkeenan/learn/perl/p5p/41106-threads-quoted-print.pl line 12.
Filehandle $foo opened only for input at /home/jkeenan/learn/perl/p5p/41106-threads-quoted-print.pl line 11.
Unbalanced string table refcount: (1) for "GETARG" during global destruction.
Finished
Attempt to free nonexistent shared string '0', Perl interpreter: 0x260b010 during global destruction.

[v5.20.3] $ ./bin/perl -Ilib ~/learn/perl/p5p/41106-threads-quoted-print.pl
Hello=3DWorld!
No package specified at /home/jkeenan/learn/perl/p5p/41106-threads-quoted-print.pl line 12.
No package specified at /home/jkeenan/learn/perl/p5p/41106-threads-quoted-print.pl line 12.
Filehandle $foo opened only for input at /home/jkeenan/learn/perl/p5p/41106-threads-quoted-print.pl line 11.
Unbalanced string table refcount: (1) for "GETARG" during global destruction.
Finished

[blead: commit 8df0224] same results as v5.20.3
#####

So the segfault disappeared somewhere between 5.12.5 and 5.16.3. From some point between 5.16.3 and 5.20.3, we start to get fairly sane warnings.

Is there any reason to keep this ticket open?

Thank you very much.

--
James E Keenan (jkeenan@cpan.org)

p5pRT · 2016-12-31T14:56:05Z

From @jkeenan

41106-threads-quoted-print.pl

p5pRT · 2016-12-31T14:56:05Z

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

p5pRT · 2017-01-02T17:02:14Z

From @iabyn

On Sat, Dec 31, 2016 at 06:56:05AM -0800, James E Keenan via RT wrote:

[v5.20.3] $ ./bin/perl -Ilib ~/learn/perl/p5p/41106-threads-quoted-print.pl
....
Unbalanced string table refcount: (1) for "GETARG" during global destruction.

[blead: commit 8df0224] same results as v5.20.3

So the segfault disappeared somewhere between 5.12.5 and 5.16.3. From
some point between 5.16.3 and 5.20.3, we start to get fairly sane
warnings.

The 'Unbalanced string table refcount' error indicates that there is still at
least one bug.

--
Little fly, thy summer's play my thoughtless hand
has terminated with extreme prejudice.
(with apologies to William Blake)

p5pRT added Severity High distro-All type-ithreads type-library labels Oct 18, 2019

xenu removed affects-5.9 labels Nov 19, 2021

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

PerlIO::via isn't thread safe #8713

PerlIO::via isn't thread safe #8713

p5pRT commented Dec 18, 2006

p5pRT commented Dec 18, 2006

p5pRT commented Dec 21, 2006

p5pRT commented Jul 14, 2016

p5pRT commented Dec 31, 2016

p5pRT commented Dec 31, 2016

p5pRT commented Dec 31, 2016

p5pRT commented Jan 2, 2017

PerlIO::via isn't thread safe #8713

PerlIO::via isn't thread safe #8713

Comments

p5pRT commented Dec 18, 2006

p5pRT commented Dec 18, 2006

From blgl@hagernas.com

Created by blgl@hagernas.com

p5pRT commented Dec 21, 2006

From blgl@hagernas.com

p5pRT commented Jul 14, 2016

From @dcollinsn

p5pRT commented Dec 31, 2016

From @jkeenan

p5pRT commented Dec 31, 2016

From @jkeenan

p5pRT commented Dec 31, 2016

p5pRT commented Jan 2, 2017

From @iabyn