Bleadperl v5.19.6-22-gebdc880 breaks GWILLIAMS/RDF-Trine-1.007.tar.gz

p5pRT commented 10 years ago

Migrated from rt.perl.org#120626 (status was 'resolved')

Searchable as RT120626$

p5pRT commented 10 years ago

From @andk

git bisect

commit ebdc88085efa6fca8a1b0afaa388f0491bdccd5a Author: Nicholas Clark \nick@ccl4\.org Date: Fri Oct 4 14:54:00 2013 +0200

S_process_special_blocks() should use a new stack for BEGIN blocks.

diagnostics

It happens only when XML::LibXML is installed. If XML::LibXML is not installed\, I typically get a pass. Sample fail report:

http://www.cpantesters.org/cpan/report/917e0d8a-546a-11e3-9195-4aca9d5baa37

perl -V

Summary of my perl5 (revision 5 version 19 subversion 7) configuration:
Commit id: ebdc88085efa6fca8a1b0afaa388f0491bdccd5a Platform: osname=linux\, osvers=3.10-3-amd64\, archname=x86_64-linux-ld uname='linux k83 3.10-3-amd64 #1 smp debian 3.10.11-1 (2013-09-10) x86_64 gnulinux ' config_args='-Dprefix=/home/sand/src/perl/repoperls/installed-perls/perl/v5.19.6-22-gebdc880/127e -Dmyhostname=k83 -Dinstallusrbinperl=n -Uversiononly -Dusedevel -des -Ui_db -Uuseithreads -Duselongdouble -DDEBUGGING=-g' hint=recommended\, useposix=true\, d_sigaction=define useithreads=undef\, usemultiplicity=undef useperlio=define\, d_sfio=undef\, uselargefiles=define\, usesocks=undef use64bitint=define\, use64bitall=define\, uselongdouble=define usemymalloc=n\, bincompat5005=undef Compiler: cc='cc'\, ccflags ='-fno-strict-aliasing -pipe -fstack-protector -I/usr/local/include -D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64'\, optimize='-O2 -g'\, cppflags='-fno-strict-aliasing -pipe -fstack-protector -I/usr/local/include' ccversion=''\, gccversion='4.8.2'\, gccosandvers='' intsize=4\, longsize=8\, ptrsize=8\, doublesize=8\, byteorder=12345678 d_longlong=define\, longlongsize=8\, d_longdbl=define\, longdblsize=16 ivtype='long'\, ivsize=8\, nvtype='long double'\, nvsize=16\, Off_t='off_t'\, lseeksize=8 alignbytes=16\, prototype=define Linker and Libraries: ld='cc'\, ldflags =' -fstack-protector -L/usr/local/lib' libpth=/usr/local/lib /lib/x86_64-linux-gnu /lib/../lib /usr/lib/x86_64-linux-gnu /usr/lib/../lib /lib /usr/lib libs=-lnsl -lgdbm -ldb -ldl -lm -lcrypt -lutil -lc -lgdbm_compat perllibs=-lnsl -ldl -lm -lcrypt -lutil -lc libc=\, so=so\, useshrplib=false\, libperl=libperl.a gnulibc_version='2.17' Dynamic Linking: dlsrc=dl_dlopen.xs\, dlext=so\, d_dlsymun=undef\, ccdlflags='-Wl\,-E' cccdlflags='-fPIC'\, lddlflags='-shared -O2 -g -L/usr/local/lib -fstack-protector'

Characteristics of this binary (from libperl): Compile-time options: HAS_TIMES PERLIO_LAYERS PERL_DONT_CREATE_GVSV PERL_HASH_FUNC_ONE_AT_A_TIME_HARD PERL_MALLOC_WRAP PERL_NEW_COPY_ON_WRITE PERL_PRESERVE_IVUV PERL_USE_DEVEL USE_64_BIT_ALL USE_64_BIT_INT USE_LARGE_FILES USE_LOCALE USE_LOCALE_COLLATE USE_LOCALE_CTYPE USE_LOCALE_NUMERIC USE_LONG_DOUBLE USE_PERLIO USE_PERL_ATOF Built under linux Compiled at Nov 25 2013 15:13:59 @INC: /home/sand/src/perl/repoperls/installed-perls/perl/v5.19.6-22-gebdc880/127e/lib/site_perl/5.19.7/x86_64-linux-ld /home/sand/src/perl/repoperls/installed-perls/perl/v5.19.6-22-gebdc880/127e/lib/site_perl/5.19.7 /home/sand/src/perl/repoperls/installed-perls/perl/v5.19.6-22-gebdc880/127e/lib/5.19.7/x86_64-linux-ld /home/sand/src/perl/repoperls/installed-perls/perl/v5.19.6-22-gebdc880/127e/lib/5.19.7

-- andreas

p5pRT commented 10 years ago

From @nwc10

On Mon\, Nov 25\, 2013 at 06:48:57AM -0800\, Andreas J. Koenig via RT wrote:

# New Ticket Created by (Andreas J. Koenig) # Please include the string: [perl #120626] # in the subject line of all future correspondence about this issue. # \<URL: https://rt-archive.perl.org/perl5/Ticket/Display.html?id=120626 >

git bisect ---------- commit ebdc88085efa6fca8a1b0afaa388f0491bdccd5a Author: Nicholas Clark \nick@ccl4\.org Date: Fri Oct 4 14:54:00 2013 +0200
S\_process\_special\_blocks should use a new stack for BEGIN blocks\.
diagnostics ----------- It happens only when XML::LibXML is installed. If XML::LibXML is not installed\, I typically get a pass. Sample fail report:

http://www.cpantesters.org/cpan/report/917e0d8a-546a-11e3-9195-4aca9d5baa37

I'm not going to get a chance to look at this (heck\, I didn't get time myself to get that commit into blead)\, but I think that the key clue is this:

Attempt to reload XML/LibXML.pm aborted. Compilation failed in require at /home/sand/src/perl/repoperls/installed-perls/perl/v5.19.6-26-ga2235d7/9980/lib/site_perl/5.19.7/x86_64-linux-thread-multi/XML/LibXML/SAX.pm line 17. BEGIN failed--compilation aborted at /home/sand/src/perl/repoperls/installed-perls/perl/v5.19.6-26-ga2235d7/9980/lib/site_perl/5.19.7/x86_64-linux-thread-multi/XML/LibXML/SAX.pm line 17. Compilation failed in require at (eval 347) line 1. # Looks like your test exited with 255 before it could output anything.

Why is it now attempting a reload? Presumably it didn't before.

Nicholas Clark

p5pRT commented 10 years ago

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

p5pRT commented 10 years ago

From @andk

also affected: GSHANK/HTML-FormHandler-0.40053.tar.gz

sample fail report: http://www.cpantesters.org/cpan/report/d85bd3a2-5561-11e3-a024-97dd9d5baa37

Running the failing test verbosely:

PERL_DL_NONLAZY=1 /home/sand/src/perl/repoperls/installed-perls/perl/v5.19.6-22-gebdc880/127e/bin/perl "-MExtUtils::Command::MM" "-MTest::Harness" "-e" "undef *Test::Harness::Switches; test_harness(1\, 'blib/lib'\, 'blib/arch')" t/fields/repeatable.t t/fields/repeatable.t .. ok 1 - use HTML::FormHandler::Field::Repeatable; ok 2 - field built ok 3 - attribute set perl: malloc.c:3544: _int_malloc: Assertion `victim->fd_nextsize->bk_nextsize == victim' failed. All 3 subtests passed

Test Summary Report

t/fields/repeatable.t (Wstat: 6 Tests: 3 Failed: 0) Non-zero wait status: 6 Parse errors: No plan found in TAP output Files=1\, Tests=3\, 1 wallclock secs ( 0.03 usr 0.00 sys + 0.53 cusr 0.04 csys = 0.60 CPU)

-- andreas

p5pRT commented 10 years ago

From @cpansprout

On Mon Nov 25 06:48:56 2013\, andreas.koenig.7os6VVqR@franz.ak.mind.de wrote:

git bisect ---------- commit ebdc88085efa6fca8a1b0afaa388f0491bdccd5a Author: Nicholas Clark \nick@ccl4\.org Date: Fri Oct 4 14:54:00 2013 +0200

S_process_special_blocks() should use a new stack for BEGIN blocks.

diagnostics ----------- It happens only when XML::LibXML is installed. If XML::LibXML is not installed\, I typically get a pass. Sample fail report:

http://www.cpantesters.org/cpan/report/917e0d8a-546a-11e3-9195- 4aca9d5baa37

It is not failing for me under current bleadperl. Does it fail for you still?

I wonder whether 901ee108fe1 might have fixed it\, but it would take me a while to find out with my current setup.

--

Father Chrysostomos

p5pRT commented 10 years ago

From @andk

"Father Chrysostomos via RT" \perlbug\-followup@perl\.org writes:

It is not failing for me under current bleadperl. Does it fail for you still?

Yes\, but I had to determine the conditions under which to reproduce more carefully. Last time I said\, XML::LibXML must be installed. But now I see another condition must be met which I did not realize last time:

- XML::SAX::Expat is not installed

--or--

- XML::SAX::Expat is also installed and XML/SAX/ParserDetails.ini is listing Expat before LibXML

After this finding I checked whether 5.19.6 is indeed passing when exactly those conditions are met\, and it is.

-- andreas

p5pRT commented 10 years ago

From @andk

Now also affected: SREZIC/Tk-804.032.tar.gz

http://www.cpantesters.org/cpan/report/776f4750-86ac-11e3-95c9-a5a95d7e0655 -- andreas

p5pRT commented 10 years ago

From @andk

Andreas Koenig \andreas\.koenig\.7os6VVqR@franz\.ak\.mind\.de writes:

Now also affected: SREZIC/Tk-804.032.tar.gz

Program terminated with signal 6\, Aborted. #0 0x00002ad6b01821d5 in __GI_raise (sig=sig@entry=6) at ../nptl/sysdeps/unix/sysv/linux/raise.c:56 56 ../nptl/sysdeps/unix/sysv/linux/raise.c: No such file or directory. (gdb) bt #0 0x00002ad6b01821d5 in __GI_raise (sig=sig@entry=6) at ../nptl/sysdeps/unix/sysv/linux/raise.c:56 #1 0x00002ad6b0185388 in __GI_abort () at abort.c:90 #2 0x00002ad6b166ff89 in InterpHv (interp=0x27c5660\, fatal=\) at tkGlue.c:516 #3 0x00002ad6b166ffce in FindXv (my_perl=0x26e1010\, interp=interp@entry=0x27c5660\, create=create@entry=1\, key=key@entry=0x2ad6b16e99f9 "_TK_RESULT_"\, type=type@entry=0\, createProc=createProc@entry=0x2ad6b166e570 \\, who=\) at tkGlue.c:528 #4 0x00002ad6b16705a4 in FindSv (who=\\, key=0x2ad6b16e99f9 "_TK_RESULT_"\, create=1\, hv=0x27c5660\, my_perl=\) at tkGlue.c:610 #5 Tcl_GetObjResult (interp=interp@entry=0x27c5660) at tkGlue.c:622 #6 0x00002ad6b1680d99 in Tcl_GetStringResult (interp=interp@entry=0x27c5660) at strGlue.c:99 #7 0x00002ad6b167b063 in XS_Tk__MainWindow_Create (my_perl=0x1\, cv=\) at tkGlue.c:2382 #8 0x00000000004ae302 in Perl_pp_entersub (my_perl=0x26e1010) at pp_hot.c:2760 #9 0x00000000004a6af6 in Perl_runops_standard (my_perl=0x26e1010) at run.c:42 #10 0x000000000043c044 in S_run_body (oldscope=1\, my_perl=0x26e1010) at perl.c:2433 #11 perl_run (my_perl=0x26e1010) at perl.c:2349 #12 0x000000000041bc3b in main (argc=79\, argv=0x7fff63ed9108\, env=0x7fff63ed9388) at perlmain.c:112

-- andreas

p5pRT commented 10 years ago

From @wolfsage

With ebdc88085\, I can reproduce the HTML::FormHandler test crash\, and I get this valgrind output:

mhorsfall@tworivers:\~/.cpanm/work/1390878740.27975/HTML-FormHandler-0.40055$ valgrind ~/badcommit/bin/perl -Iblib/lib -Iblib/arch t/fields/repeatable.t 2>&1 ==28093== Memcheck\, a memory error detector ==28093== Copyright (C) 2002-2011\, and GNU GPL'd\, by Julian Seward et al. ==28093== Using Valgrind-3.7.0 and LibVEX; rerun with -h for copyright info ==28093== Command: /home/mhorsfall/badcommit/bin/perl -Iblib/lib -Iblib/arch t/fields/repeatable.t ==28093== ok 1 - use HTML::FormHandler::Field::Repeatable; ok 2 - field built ==28093== Invalid write of size 8 ==28093== at 0xA472891: XS_Data__Clone_clone (Data-Clone.xs:395) ==28093== by 0x4F0214: Perl_pp_entersub (pp_hot.c:2760) ==28093== by 0x4E50D1: Perl_runops_standard (run.c:42) ==28093== by 0x43F792: S_run_body (perl.c:2433) ==28093== by 0x43F371: perl_run (perl.c:2349) ==28093== by 0x41CD3B: main (perlmain.c:112) ==28093== Address 0x592fd20 is 16 bytes inside a block of size 1\,024 free'd ==28093== at 0x4C2B7B2: realloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) ==28093== by 0x4BAD8C: Perl_safesysrealloc (util.c:173) ==28093== by 0x4E2F7E: Perl_av_extend_guts (av.c:154) ==28093== by 0x4E2CF8: Perl_av_extend (av.c:82) ==28093== by 0x5285CC: Perl_stack_grow (scope.c:38) ==28093== by 0x55BDC2: Perl_do_kv (doop.c:1268) ==28093== by 0x4E8E76: Perl_pp_rv2av (pp_hot.c:934) ==28093== by 0x4E50D1: Perl_runops_standard (run.c:42) ==28093== by 0x43FF64: Perl_call_sv (perl.c:2732) ==28093== by 0xA471F53: dc_call_sv1 (Data-Clone.xs:176) ==28093== by 0xA4721E1: dc_clone_object (Data-Clone.xs:227) ==28093== by 0xA472356: clone_rv (Data-Clone.xs:266) ==28093== ok 3 - attribute set ok 4 - form built ok 5 - worked in form 1..5

(Though it doesn't crash when run under valgrind. Neat.):

mhorsfall@tworivers:\~/.cpanm/work/1390878740.27975/HTML-FormHandler-0.40055$ ~/badcommit/bin/perl -Iblib/lib -Iblib/arch t/fields/repeatable.t 2>&1 ok 1 - use HTML::FormHandler::Field::Repeatable; ok 2 - field built ok 3 - attribute set perl: malloc.c:3729: _int_malloc: Assertion `victim->fd_nextsize->bk_nextsize == victim' failed.

With ea35f8011 (ebdc88085~1) I cannot reproduce the crash\, and the valgrind output is "clean". (Valgrind reports some definitely lost bytes thought).

With latest blead (b37339699) I still get the crash with the same valgrind output.

Each time\, I built perl with grindperl\, for example:

grindperl --prefix=/home/mhorsfall/blead -j 5 --install --no-debugging --no-threads -D optimize=-g

-- Matthew Horsfall (alh)

p5pRT commented 10 years ago

From perl5-porters@perl.org

Matthew Horsfall wrote:

==28093== Invalid write of size 8 ==28093== at 0xA472891: XS_Data__Clone_clone (Data-Clone.xs:395)

Could you get a Perl backtrace at that point?

E.g.\, set a breakpoint on XS_Data__Clone_clone and then use:

gdb) p Perl_warn("") /* no threads */ gdb) p Perl_warn(my_perl\,"") /* threads */

Then edit the cited sourced file and put Carp::cluck() where appropriate.

Then re-run the whole thing through valgrind.

The output of that would be much appreciated.

==28093== Address 0x592fd20 is 16 bytes inside a block of size 1\,024 free'd ... ==28093== by 0x5285CC: Perl_stack_grow (scope.c:38) ==28093== by 0x55BDC2: Perl_do_kv (doop.c:1268)

That line is an EXTEND. So somehow the entersub call above is using a reallocated stack\, which is weird.

p5pRT commented 10 years ago

From @wolfsage

I think I've got this right:

ok 1 - use HTML::FormHandler::Field::Repeatable; ok 2 - field built ==3474== Invalid write of size 8 ==3474== at 0xA873891: XS_Data__Clone_clone (Data-Clone.xs:395) ==3474== by 0x4F24CA: Perl_pp_entersub (pp_hot.c:2768) ==3474== by 0x4E7399: Perl_runops_standard (run.c:42) ==3474== by 0x43FC9C: S_run_body (perl.c:2446) ==3474== by 0x43F87B: perl_run (perl.c:2362) ==3474== by 0x41CD9B: main (perlmain.c:112) ==3474== Address 0x592fd20 is 16 bytes inside a block of size 1\,024 free'd ==3474== at 0x4C2B7B2: realloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) ==3474== by 0x4BD3F7: Perl_safesysrealloc (util.c:244) ==3474== by 0x4E5246: Perl_av_extend_guts (av.c:154) ==3474== by 0x4E4FC0: Perl_av_extend (av.c:82) ==3474== by 0x52AB5C: Perl_stack_grow (scope.c:38) ==3474== by 0x55E26E: Perl_do_kv (doop.c:1268) ==3474== by 0x4EB13E: Perl_pp_rv2av (pp_hot.c:934) ==3474== by 0x4E7399: Perl_runops_standard (run.c:42) ==3474== by 0x44046E: Perl_call_sv (perl.c:2746) ==3474== by 0xA872F53: dc_call_sv1 (Data-Clone.xs:176) ==3474== by 0xA8731E1: dc_clone_object (Data-Clone.xs:227) ==3474== by 0xA873356: clone_rv (Data-Clone.xs:266) ==3474== Squawk at blib/lib/HTML/FormHandler/Merge.pm line 56. HTML::FormHandler::Merge::merge_hashes(HASH(0x9ccc120)\, HASH(0x826db90)) called at blib/lib/HTML/FormHandler/Merge.pm line 23 HTML::FormHandler::Merge::__ANON__(HASH(0x9ccc120)\, HASH(0x826db90)) called at blib/lib/HTML/FormHandler/Merge.pm line 40 HTML::FormHandler::Merge::merge(HASH(0x595afe0)\, HASH(0x826db90)) called at blib/lib/HTML/FormHandler/Field/Repeatable.pm line 75 HTML::FormHandler::Field::Repeatable::create_element(HTML::FormHandler::Field::Repeatable=HASH(0x5947578)) called at blib/lib/HTML/FormHandler/Field/Repeatable.pm line 55 HTML::FormHandler::Field::Repeatable::init_state(HTML::FormHandler::Field::Repeatable=HASH(0x5947578)) called at t/fields/repeatable.t line 15 Squawk at blib/lib/HTML/FormHandler/Merge.pm line 56. HTML::FormHandler::Merge::merge_hashes(HASH(0x9ccc120)\, HASH(0x826db90)) called at blib/lib/HTML/FormHandler/Merge.pm line 23 HTML::FormHandler::Merge::__ANON__(HASH(0x9ccc120)\, HASH(0x826db90)) called at blib/lib/HTML/FormHandler/Merge.pm line 40 HTML::FormHandler::Merge::merge(HASH(0x595afe0)\, HASH(0x826db90)) called at blib/lib/HTML/FormHandler/Field/Repeatable.pm line 75 HTML::FormHandler::Field::Repeatable::create_element(HTML::FormHandler::Field::Repeatable=HASH(0x5947578)) called at blib/lib/HTML/FormHandler/Field/Repeatable.pm line 55 HTML::FormHandler::Field::Repeatable::init_state(HTML::FormHandler::Field::Repeatable=HASH(0x5947578)) called at t/fields/repeatable.t line 15 Squawk at blib/lib/HTML/FormHandler/Merge.pm line 56. HTML::FormHandler::Merge::merge_hashes(HASH(0x9ccc120)\, HASH(0x826db90)) called at blib/lib/HTML/FormHandler/Merge.pm line 23 HTML::FormHandler::Merge::__ANON__(HASH(0x9ccc120)\, HASH(0x826db90)) called at blib/lib/HTML/FormHandler/Merge.pm line 40 HTML::FormHandler::Merge::merge(HASH(0x595afe0)\, HASH(0x826db90)) called at blib/lib/HTML/FormHandler/Field/Repeatable.pm line 75 HTML::FormHandler::Field::Repeatable::create_element(HTML::FormHandler::Field::Repeatable=HASH(0x5947578)) called at blib/lib/HTML/FormHandler/Field/Repeatable.pm line 55 HTML::FormHandler::Field::Repeatable::init_state(HTML::FormHandler::Field::Repeatable=HASH(0x5947578)) called at t/fields/repeatable.t line 15 Squawk at blib/lib/HTML/FormHandler/Merge.pm line 56. HTML::FormHandler::Merge::merge_hashes(HASH(0x9ccc120)\, HASH(0x826db90)) called at blib/lib/HTML/FormHandler/Merge.pm line 23 HTML::FormHandler::Merge::__ANON__(HASH(0x9ccc120)\, HASH(0x826db90)) called at blib/lib/HTML/FormHandler/Merge.pm line 40 HTML::FormHandler::Merge::merge(HASH(0x595afe0)\, HASH(0x826db90)) called at blib/lib/HTML/FormHandler/Field/Repeatable.pm line 75 HTML::FormHandler::Field::Repeatable::create_element(HTML::FormHandler::Field::Repeatable=HASH(0x5947578)) called at blib/lib/HTML/FormHandler/Field/Repeatable.pm line 55 HTML::FormHandler::Field::Repeatable::init_state(HTML::FormHandler::Field::Repeatable=HASH(0x5947578)) called at t/fields/repeatable.t line 15 ok 3 - attribute set

-- Matthew Horsfall (alh)

p5pRT commented 10 years ago

From @wolfsage

And that looks terrible in my email client.

Attached.

p5pRT commented 10 years ago

From @wolfsage

==3474== Memcheck\, a memory error detector ==3474== Copyright (C) 2002-2011\, and GNU GPL'd\, by Julian Seward et al. ==3474== Using Valgrind-3.7.0 and LibVEX; rerun with -h for copyright info ==3474== Command: /home/mhorsfall/blead/bin/perl -Iblib/lib -Iblib/arch t/fields/repeatable.t ==3474== ok 1 - use HTML::FormHandler::Field::Repeatable; ok 2 - field built ==3474== Invalid write of size 8 ==3474== at 0xA873891: XS_Data__Clone_clone (Data-Clone.xs:395) ==3474== by 0x4F24CA: Perl_pp_entersub (pp_hot.c:2768) ==3474== by 0x4E7399: Perl_runops_standard (run.c:42) ==3474== by 0x43FC9C: S_run_body (perl.c:2446) ==3474== by 0x43F87B: perl_run (perl.c:2362) ==3474== by 0x41CD9B: main (perlmain.c:112) ==3474== Address 0x592fd20 is 16 bytes inside a block of size 1\,024 free'd ==3474== at 0x4C2B7B2: realloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) ==3474== by 0x4BD3F7: Perl_safesysrealloc (util.c:244) ==3474== by 0x4E5246: Perl_av_extend_guts (av.c:154) ==3474== by 0x4E4FC0: Perl_av_extend (av.c:82) ==3474== by 0x52AB5C: Perl_stack_grow (scope.c:38) ==3474== by 0x55E26E: Perl_do_kv (doop.c:1268) ==3474== by 0x4EB13E: Perl_pp_rv2av (pp_hot.c:934) ==3474== by 0x4E7399: Perl_runops_standard (run.c:42) ==3474== by 0x44046E: Perl_call_sv (perl.c:2746) ==3474== by 0xA872F53: dc_call_sv1 (Data-Clone.xs:176) ==3474== by 0xA8731E1: dc_clone_object (Data-Clone.xs:227) ==3474== by 0xA873356: clone_rv (Data-Clone.xs:266) ==3474== Squawk at blib/lib/HTML/FormHandler/Merge.pm line 56. HTML::FormHandler::Merge::merge_hashes(HASH(0x9ccc120)\, HASH(0x826db90)) called at blib/lib/HTML/FormHandler/Merge.pm line 23 HTML::FormHandler::Merge::__ANON__(HASH(0x9ccc120)\, HASH(0x826db90)) called at blib/lib/HTML/FormHandler/Merge.pm line 40 HTML::FormHandler::Merge::merge(HASH(0x595afe0)\, HASH(0x826db90)) called at blib/lib/HTML/FormHandler/Field/Repeatable.pm line 75 HTML::FormHandler::Field::Repeatable::create_element(HTML::FormHandler::Field::Repeatable=HASH(0x5947578)) called at blib/lib/HTML/FormHandler/Field/Repeatable.pm line 55 HTML::FormHandler::Field::Repeatable::init_state(HTML::FormHandler::Field::Repeatable=HASH(0x5947578)) called at t/fields/repeatable.t line 15 Squawk at blib/lib/HTML/FormHandler/Merge.pm line 56. HTML::FormHandler::Merge::merge_hashes(HASH(0x9ccc120)\, HASH(0x826db90)) called at blib/lib/HTML/FormHandler/Merge.pm line 23 HTML::FormHandler::Merge::__ANON__(HASH(0x9ccc120)\, HASH(0x826db90)) called at blib/lib/HTML/FormHandler/Merge.pm line 40 HTML::FormHandler::Merge::merge(HASH(0x595afe0)\, HASH(0x826db90)) called at blib/lib/HTML/FormHandler/Field/Repeatable.pm line 75 HTML::FormHandler::Field::Repeatable::create_element(HTML::FormHandler::Field::Repeatable=HASH(0x5947578)) called at blib/lib/HTML/FormHandler/Field/Repeatable.pm line 55 HTML::FormHandler::Field::Repeatable::init_state(HTML::FormHandler::Field::Repeatable=HASH(0x5947578)) called at t/fields/repeatable.t line 15 Squawk at blib/lib/HTML/FormHandler/Merge.pm line 56. HTML::FormHandler::Merge::merge_hashes(HASH(0x9ccc120)\, HASH(0x826db90)) called at blib/lib/HTML/FormHandler/Merge.pm line 23 HTML::FormHandler::Merge::__ANON__(HASH(0x9ccc120)\, HASH(0x826db90)) called at blib/lib/HTML/FormHandler/Merge.pm line 40 HTML::FormHandler::Merge::merge(HASH(0x595afe0)\, HASH(0x826db90)) called at blib/lib/HTML/FormHandler/Field/Repeatable.pm line 75 HTML::FormHandler::Field::Repeatable::create_element(HTML::FormHandler::Field::Repeatable=HASH(0x5947578)) called at blib/lib/HTML/FormHandler/Field/Repeatable.pm line 55 HTML::FormHandler::Field::Repeatable::init_state(HTML::FormHandler::Field::Repeatable=HASH(0x5947578)) called at t/fields/repeatable.t line 15 Squawk at blib/lib/HTML/FormHandler/Merge.pm line 56. HTML::FormHandler::Merge::merge_hashes(HASH(0x9ccc120)\, HASH(0x826db90)) called at blib/lib/HTML/FormHandler/Merge.pm line 23 HTML::FormHandler::Merge::__ANON__(HASH(0x9ccc120)\, HASH(0x826db90)) called at blib/lib/HTML/FormHandler/Merge.pm line 40 HTML::FormHandler::Merge::merge(HASH(0x595afe0)\, HASH(0x826db90)) called at blib/lib/HTML/FormHandler/Field/Repeatable.pm line 75 HTML::FormHandler::Field::Repeatable::create_element(HTML::FormHandler::Field::Repeatable=HASH(0x5947578)) called at blib/lib/HTML/FormHandler/Field/Repeatable.pm line 55 HTML::FormHandler::Field::Repeatable::init_state(HTML::FormHandler::Field::Repeatable=HASH(0x5947578)) called at t/fields/repeatable.t line 15 ok 3 - attribute set

p5pRT commented 10 years ago

From perl5-porters@perl.org

Matthew Horsfall wrote:

I think I've got this right:

Thank you. Sorry for putting you to the trouble. The valgrind output has just made something in my head click:

==3474== Address 0x592fd20 is 16 bytes inside a block of size 1\,024 free'd ==3474== at 0x4C2B7B2: realloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) ==3474== by 0x4BD3F7: Perl_safesysrealloc (util.c:244) ==3474== by 0x4E5246: Perl_av_extend_guts (av.c:154) ==3474== by 0x4E4FC0: Perl_av_extend (av.c:82) ==3474== by 0x52AB5C: Perl_stack_grow (scope.c:38) ==3474== by 0x55E26E: Perl_do_kv (doop.c:1268) ==3474== by 0x4EB13E: Perl_pp_rv2av (pp_hot.c:934) ==3474== by 0x4E7399: Perl_runops_standard (run.c:42) ==3474== by 0x44046E: Perl_call_sv (perl.c:2746) ==3474== by 0xA872F53: dc_call_sv1 (Data-Clone.xs:176) ==3474== by 0xA8731E1: dc_clone_object (Data-Clone.xs:227) ==3474== by 0xA873356: clone_rv (Data-Clone.xs:266)

Data::Clone is calling the perl code that reallocated the stack\, so I suspect the bug is there. Where is *it* not protecting things?

(I suspect its use of call_sv is buggy and that the commit that supposedly broke it is a red herring.)

I am trying to reproduce this on dromedary. The bad news is that I keep losing my ssh connection before I finish processing dependencies. But I seem to be getting closer each time....

p5pRT commented 10 years ago

From perl5-porters@perl.org

I wrote\, in response to Matthew Horsfall:

(I suspect its use of call_sv is buggy and that the commit that supposedly broke it is a red herring.)

It is likely that\, since require() does not use the existing stack\, the stack at the offending call to Data::Clone::clone() is smaller than it used to be\, requiring a clone callback to reallocated the stack.

Indeed\, Data::Clone does not protect against stack reallocation during callbacks.

RDF::Trine does not depend on Data::Clone\, so I do not know which of its dependencies has a similar bug. If I do not get a chance\, could you try running whichever of its tests is failing through valgrind (presuming it fails for you)?

p5pRT commented 10 years ago

From @tonycoz

On Wed\, Jan 29\, 2014 at 05:44:07AM -0000\, Father Chrysostomos wrote:

I am trying to reproduce this on dromedary. The bad news is that I keep losing my ssh connection before I finish processing dependencies. But I seem to be getting closer each time....

You can use screen(1) to deal with dodgy connections.

Tony

p5pRT commented 10 years ago

From @khwilliamson

On 01/28/2014 10:55 PM\, Tony Cook wrote:

On Wed\, Jan 29\, 2014 at 05:44:07AM -0000\, Father Chrysostomos wrote:

I am trying to reproduce this on dromedary. The bad news is that I keep losing my ssh connection before I finish processing dependencies. But I seem to be getting closer each time....

You can use screen(1) to deal with dodgy connections.

Tony

Another possible thing to try is

ssh -oServerAliveInterval=150 users.perl5.git.perl.org

p5pRT commented 10 years ago

From @wolfsage

On Wed\, Jan 29\, 2014 at 12:44 AM\, Father Chrysostomos \sprout@cpan\.org wrote:

Thank you. Sorry for putting you to the trouble. The valgrind output has just made something in my head click:

No problem. I'm trying to set up the RDF::Trine case now.

-- Matthew Horsfall (alh)

p5pRT commented 10 years ago

From @wolfsage

On Wed\, Jan 29\, 2014 at 8:57 AM\, Matthew Horsfall (alh) \wolfsage@gmail\.com wrote:

On Wed\, Jan 29\, 2014 at 12:44 AM\, Father Chrysostomos \sprout@cpan\.org wrote:

Thank you. Sorry for putting you to the trouble. The valgrind output has just made something in my head click:

No problem. I'm trying to set up the RDF::Trine case now.

I believe I've reduced it to this:

use strict; use warnings;

use Module::Load::Conditional qw(can_load);

BEGIN { can_load( modules => { 'XML::LibXML' => 0\, } ); }

eval "use XML::LibXML;"; die $@ if $@;

If you look at %INC\, 'XML/LibXML.pm' => undef !

-- Matthew Horsfall (alh)

p5pRT commented 10 years ago

From perl5-porters@perl.org

I wrote:

Indeed\, Data::Clone does not protect against stack reallocation during callbacks.

I was implying that it should do PUSHSTACK/POPSTACK\, but actually it should not have to\, as XSUBs may reallocate the stack and pp_entersub already accounts for that. What seems to be the problem is that Data::Clone is not protecting itself. The code in question is:

void clone(SV* sv) CODE: { ST(0) = sv_clone(sv); XSRETURN(1); }

where sv_clone can reallocate the stack. That line translates into:

PL_stack_base[ax + (0)] = Data_Clone_sv_clone( (sv));

Does the C standard guarantee any execution order here? (According to Perl execution order\, that code would actually work. According to JavaScript execution order\, it is completely broken.)

p5pRT commented 10 years ago

From @tonycoz

On Thu\, Jan 30\, 2014 at 04:38:47AM -0000\, Father Chrysostomos wrote:

where sv_clone can reallocate the stack. That line translates into:
PL\_stack\_base\[ax \+ $0$\] = Data\_Clone\_sv\_clone$ \(sv$\);
Does the C standard guarantee any execution order here? (According to Perl execution order\, that code would actually work. According to JavaScript execution order\, it is completely broken.)

It doesn't.

Tony

p5pRT commented 10 years ago

From perl5-porters@perl.org

Matthew Horsfall wrote:

I believe I've reduced it to this:

use strict; use warnings;

use Module::Load::Conditional qw(can_load);

BEGIN { can_load( modules => { 'XML::LibXML' => 0\, } ); }

eval "use XML::LibXML;"; die $@ if $@;

Thank you. I was able to reproduce it with that.

XML::LibXML::bootstrap does POPMARK twice. No wonder it screws up. The version number read by Perl_xs_version_bootcheck was the wrong stack item. That it worked before was a stroke of serendipity (or not). (I do not yet fully understand why it worked.)

p5pRT commented 10 years ago

From @cpansprout

Thank you to everyone who provided diagnostics and advice. I have submitted a patch to CPAN ticket #92606 and mentioned RDF::Trine in the Known Problems section of perl5200delta in commit 1a85989.

--

Father Chrysostomos

p5pRT commented 10 years ago

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

p5pRT commented 10 years ago

From @eserte

"Father Chrysostomos via RT" \perlbug\-followup@perl\.org writes:

Thank you to everyone who provided diagnostics and advice. I have submitted a patch to CPAN ticket #92606 and mentioned RDF::Trine in the Known Problems section of perl5200delta in commit 1a85989.

Unfortunately Tk was forgotten.

-- Slaven Rezic - slaven \ rezic \ de

Berlin Perl Mongers - http://berlin.pm.org

Perl / perl5