in reply to Re: Event module and threads/fork segfault on Win32
in thread Event module and threads/fork segfault on Win32

You're right. Here it is the quintessence of the problem: 
#!/usr/bin/perl 
use strict;
use warnings;

use threads;


sub create_thread {
    warn "in create_thread() before async()\n";
    my $thr = async { warn "in thread\n"; };
    warn "in create_thread() before join()\n";
    sleep 2;
    $thr->join;
    warn "in create_thread() before return\n";
    return;
}


use Event;
my $timer = Event->timer( cb => \&create_thread, after => 2);

warn "starting event loop\n";
Event::loop;

[download]

When create_thread() returns, the interpreter segfaults.
The problem seems to be caused when join() is called, if you comment the call, the event loop ends and you get the warning: 
Perl exited with active threads:
        0 running and unjoined
        1 finished and unjoined
        0 running and detached

[download]

Replies are listed 'Best First'.
Re^3: Event module and threads/fork segfault on Win32
by BrowserUk (Patriarch) on Jun 05, 2007 at 03:37 UTC

    The next question is which versions of Perl, threads and Event are you using?

    I just ran your script under AS811/5.8.6 with threads v1.05 and Event v1.09 and it completed without a segfault, and just a warning for a leaked scalar? 

    C:\test>619259.pl
starting event loop
in create_thread() before async()
in create_thread() before join()
in thread
in create_thread() before return
Scalars leaked: 1

C:\test>perl -v

This is perl, v5.8.6 built for MSWin32-x86-multi-thread
(with 3 registered patches, see perl -V for more detail)

Copyright 1987-2004, Larry Wall

Binary build 811 provided by ActiveState Corp. http://www.ActiveState.
+com
ActiveState is a division of Sophos.
Built Dec 13 2004 09:52:01

Perl may be copied only under the terms of either the Artistic License
+ or the
GNU General Public License, which may be found in the Perl 5 source ki
+t.

Complete documentation for Perl, including FAQ lists, should be found 
+on
this system using `man perl' or `perldoc perl'.  If you have access to
+ the
Internet, point your browser at http://www.perl.org/, the Perl Home Pa
+ge.


C:\test>perl -Mthreads -e"warn "$threads::VERSION"
1.05 at -e line 1.

C:\test>perl -MEvent -e"warn "$Event::VERSION"
1.09 at -e line 1.

    [download]
    Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error.
    "Science is about questioning the status quo. Questioning authority".
    In the absence of evidence, opinion is indistinguishable from prejudice.
[reply]
[d/l]
      This is interesting... I'm running newer versions of everything...
      C:\>perl -v

This is perl, v5.8.8 built for MSWin32-x86-multi-thread
(with 50 registered patches, see perl -V for more detail)

Copyright 1987-2006, Larry Wall

Binary build 820 [274739] provided by ActiveState http://www.ActiveSta
+te.com
Built Jan 23 2007 15:57:46

Perl may be copied only under the terms of either the Artistic License
+ or the
GNU General Public License, which may be found in the Perl 5 source ki
+t.

Complete documentation for Perl, including FAQ lists, should be found 
+on
this system using "man perl" or "perldoc perl".  If you have access to
+ the
Internet, point your browser at http://www.perl.org/, the Perl Home Pa
+ge.


C:\>perl -MEvent -e"print qq($Event::VERSION\n)"
1.08

C:\>perl -Mthreads -e"print qq($threads::VERSION\n)"
1.58

      [download]
      The OS is WinXP Professional SP2 running in a VMWare VM v5.5.51 build-19175

      The real code starts threads in one watcher and joins them in another to collect their exit statuses. In that situation it gets the "leaked scalar" before the segfault.
[reply]
[d/l]

        To resolve this you're probably going to have to track down which version of what caused things to break and then raise a specific bug report against the specific component. And that will likely be a PITA, as there are so many versions and combinations of versions to test. For example; threads appears to have gone through around 20+ version between where you are and I am.

        A few things that might help:

        • I downloaded the latest version (v109) of Event that was available from CPAN. I'm not sure how/where you obtained a later version?

          It's also worth poiting out that the POD carries the following warnings:

          The scope of events is pretty strange compared to most other perl objects. I'm not sure if this is a bug or a feature (OK, probably it was a mistake). ...

          There seems to be some sort of bug in the global destruction phase: 

            Attempt to free unreferenced scalar during global destruction.
  Use of uninitialized value during global destruction.
  Explicit blessing to '' (assuming package main) during global
  destruction.

          [download]

          The latter may be, (is probably), the source of the "1 leaked scalar" message issued. Although this message is associated with threads, it may well be that it is simply that it bothers to check.

          Ie. The scalar could be leaked whenever you use Event, but is only detected and reported when you use Event in conjunction with threads.

          The former may mean that the two modules may never properly coexist?

        • You might be able to verify the source of the problem as being in the eevent module, by using them together within the same program, but but keeping their respective code separate.

          Ie. Start and join a thread from the main thread, but do not use Event within that thread. Then create an Event timer callback, that does nothing with threads; run the dispatch loop and allow the program to terminate. If you still see one or the other of the symptoms, it will indicate...something :)

        • I've downloaded, but haven't yet installed the latest CPAN version of threads.

          I'm a litte reluctant to risk screwing my main installation of Perl, but I may try backing up my libraries for quick reversion, and installing it later tonight.

        HTH some. Good luck.

        Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error.
        "Science is about questioning the status quo. Questioning authority".
        In the absence of evidence, opinion is indistinguishable from prejudice.
[reply]
[d/l]

        It would appear to be 5.8.8 that is the deciding factor as to whether it segfaults or not. Having installed the latest versions threads:v1.62 and Event:v1.09; I get no trap from your code under 5.8.6(AS811), but I do under 5.8.8(AS817).

        However, if I do not nest the threads stuff inside the Event callback, no trap occurs either. HTH.

        Update: I finally got the segfault to occur under the auspices of a debugger.

        The trap is occurring in ev.c:187, when the if statement attempts to access WvACTIVE(wa). That doesn't explain why it happens, but it might give somebody a starting point. 

        Disassembly of Function Event.dll!pe_event_postCB (0x00363D57)
  ;*******************************************************************
+*************
  SYM:pe_event_postCB <Void>
  ; ev.c - Line 183
  ; static void pe_event_postCB(pe_cbframe *fp) {
  0x363D57: PUSH     EBP                                  
  0x363D58: MOV      EBP,ESP                              
  0x363D5A: SUB      ESP,0xC                              
  ; ev.c - Line 184
  ;     pe_event *ev = fp->ev;
  0x363D5D: MOV      EAX,DWORD PTR [EBP+0x8]              ; ARG:fp <pe
+_cbframe *>
  0x363D60: MOV      EAX,DWORD PTR [EAX]                  
  0x363D62: MOV      DWORD PTR [EBP-0x4],EAX              ; VAR:ev <pe
+_event *>
  ; ev.c - Line 185
  ;     pe_watcher *wa = ev->up;
  0x363D65: MOV      EAX,DWORD PTR [EBP-0x4]              ; VAR:ev <pe
+_event *>
  0x363D68: MOV      EAX,DWORD PTR [EAX+0x8]              
  0x363D6B: MOV      DWORD PTR [EBP-0x8],EAX              ; VAR:wa <pe
+_watcher *>
  ; ev.c - Line 186
  ;     --CurCBFrame;
  0x363D6E: MOV      EAX,DWORD PTR [0x10012020]           
  0x363D73: DEC      EAX                                  
  0x363D74: MOV      DWORD PTR [0x10012020],EAX           
  ; ev.c - Line 187
  ;     if (WaACTIVE(wa) && WaINVOKE1(wa) && WaREPEAT(wa))
  0x363D79: MOV      EAX,DWORD PTR [EBP-0x8]              ; VAR:wa <pe
+_watcher *>
+ 0x363D7C: MOV      EAX,DWORD PTR [EAX+0x20]             ; TRAP OCCUR
+S HERE <<<<<<<<<<<<<<<<<
  0x363D7F: AND      EAX,0x1                              
  0x363D82: TEST     EAX,EAX                              
  0x363D84: JZ       0x363DB0                             ; (*+0x2C) 
  0x363D86: MOV      EAX,DWORD PTR [EBP-0x8]              ; VAR:wa <pe
+_watcher *>
  0x363D89: MOV      EAX,DWORD PTR [EAX+0x20]             
  0x363D8C: AND      EAX,0x4000                           
  0x363D91: TEST     EAX,EAX                              
  0x363D93: JZ       0x363DB0                             ; (*+0x1D) 
  0x363D95: MOV      EAX,DWORD PTR [EBP-0x8]              ; VAR:wa <pe
+_watcher *>
  0x363D98: MOV      EAX,DWORD PTR [EAX+0x20]             
  0x363D9B: AND      EAX,0x2000                           
  0x363DA0: TEST     EAX,EAX                              
  0x363DA2: JZ       0x363DB0                             ; (*+0xE) 
  ; ev.c - Line 188
  ;     pe_watcher_on(wa, 1);
  0x363DA4: PUSH     0x1                                  
  0x363DA6: PUSH     DWORD PTR [EBP-0x8]                  ; VAR:wa <pe
+_watcher *>
  0x363DA9: CALL     0x3642EA                             ; SYM:pe_wat
+cher_on <Char *>
  0x363DAE: POP      ECX                                  
  0x363DAF: POP      ECX                                  
  ; ev.c - Line 189
  ;     if (Estat.on) {
  0x363DB0: CMP      DWORD PTR [0x10014268],0x0           ; <==0x00363
+DA2(*-0xE), 0x00363D93(*-0x1D), 0x00363D84(*-0x2C)
  0x363DB7: JZ       0x363E2C                             ; (*+0x75) 
  ; ev.c - Line 190
  ;     if (fp->stats) {
  0x363DB9: MOV      EAX,DWORD PTR [EBP+0x8]              ; ARG:fp <pe
+_cbframe *>
  0x363DBC: CMP      DWORD PTR [EAX+0x8],0x0              
  0x363DC0: JZ       0x363DDA                             ; (*+0x1A) 
  ; ev.c - Line 191
  ;         Estat.scrub(fp->stats, wa);
  0x363DC2: PUSH     DWORD PTR [EBP-0x8]                  ; VAR:wa <pe
+_watcher *>
  0x363DC5: MOV      EAX,DWORD PTR [EBP+0x8]              ; ARG:fp <pe
+_cbframe *>
  0x363DC8: PUSH     DWORD PTR [EAX+0x8]                  
  0x363DCB: CALL     DWORD PTR [0x1001427C]               
  0x363DD1: POP      ECX                                  
  0x363DD2: POP      ECX                                  
  ; ev.c - Line 192
  ;         fp->stats = 0;
  0x363DD3: MOV      EAX,DWORD PTR [EBP+0x8]              ; ARG:fp <pe
+_cbframe *>
  0x363DD6: AND      DWORD PTR [EAX+0x8],0x0              
  ; ev.c - Line 194
  ;     }
  ;     if (CurCBFrame >= 0) {
  0x363DDA: CMP      DWORD PTR [0x10012020],0x0           ; <==0x00363
+DC0(*-0x1A)
  0x363DE1: JL       0x363E2C                             ; (*+0x4B) 
  ; ev.c - Line 195
  ;         pe_cbframe *pfp = CBFrame + CurCBFrame;
  0x363DE3: MOV      EAX,DWORD PTR [0x10012020]           
  0x363DE8: IMUL     EAX,EAX,0xC                          
  0x363DEB: ADD      EAX,0x10014438                       
  0x363DF0: MOV      DWORD PTR [EBP-0xC],EAX              ; VAR:pfp <p
+e_cbframe *>
  ; ev.c - Line 196
  ;         if (!pfp->stats)
  0x363DF3: MOV      EAX,DWORD PTR [EBP-0xC]              ; VAR:pfp <p
+e_cbframe *>
  0x363DF6: CMP      DWORD PTR [EAX+0x8],0x0              
  0x363DFA: JNZ      0x363E1F                             ; (*+0x25) 
  ; ev.c - Line 197
  ;         pfp->stats = Estat.enter(CurCBFrame, pfp->ev->up->max_cb_t
+m);
  0x363DFC: MOV      EAX,DWORD PTR [EBP-0xC]              ; VAR:pfp <p
+e_cbframe *>
  0x363DFF: MOV      EAX,DWORD PTR [EAX]                  
  0x363E01: MOV      EAX,DWORD PTR [EAX+0x8]              
  0x363E04: MOVSX    EAX,DWORD PTR [EAX+0x48]             
  0x363E08: PUSH     EAX                                  
  0x363E09: PUSH     DWORD PTR [0x10012020]               
  0x363E0F: CALL     DWORD PTR [0x1001426C]               
  0x363E15: POP      ECX                                  
  0x363E16: POP      ECX                                  
  0x363E17: MOV      ECX,DWORD PTR [EBP-0xC]              ; VAR:pfp <p
+e_cbframe *>
  0x363E1A: MOV      DWORD PTR [ECX+0x8],EAX              
  ; ev.c - Line 198
  ;         else
  0x363E1D: JMP      0x363E2C                             ; (*+0xF) 
  ; ev.c - Line 199
  ;         Estat.resume(pfp->stats);
  0x363E1F: MOV      EAX,DWORD PTR [EBP-0xC]              ; VAR:pfp <p
+e_cbframe *>; <==0x00363DFA(*-0x25)
  0x363E22: PUSH     DWORD PTR [EAX+0x8]                  
  0x363E25: CALL     DWORD PTR [0x10014274]               
  0x363E2B: POP      ECX                                  
  ; ev.c - Line 203
  ;     }
  ;     }
  ;     /* this must be last because it can destroy the watcher */
  ;     pe_event_release(ev);
  0x363E2C: PUSH     DWORD PTR [EBP-0x4]                  ; VAR:ev <pe
+_event *>; <==0x00363DE1(*-0x4B), 0x00363E1D(*-0xF), 0x00363DB7(*-0x7
+5)
  0x363E2F: CALL     0x363E37                             ; SYM:pe_eve
+nt_release <Void>
  0x363E34: POP      ECX                                  
  ; ev.c - Line 204
  ; }
  0x363E35: LEAVE                                         
  0x363E36: RET                                           
  ;*******************************************************************
+*************

        [download]

        And a stack trace:

        PID: 4020 TID: 4192 - Stack Contents for 0x00363D7C
0x00363D7C: Event.dll:pe_event_postCB + 0x0025 <Void>
0x00363C6F: Event.dll:pe_event_invoke + 0x0738 <Void>
0x00364EA1: Event.dll:pe_empty_queue + 0x0031 <Int>
0x00364ACC: Event.dll:one_event + 0x00ED <Int>
0x280933F9: Perl_get_context + 0x001F
0x00365A0E: Event.dll:XS_Event__loop + 0x011B <Void>
0x2804002B: Perl_sv_compile_2op + 0x7B9F
0x2805D4AF: Perl_runops_standard + 0x000C
0x28088A87: RunPerl + 0x0086
0x77F53CB3: RtlDosPathNameToNtPathName_U + 0x02E3
0x00401012: 
0x77E814C7: GetCurrentDirectoryW + 0x0044

        [download]

        I'm intrigued by what you are trying to achieve by mixing these two anyway?

        Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error.
        "Science is about questioning the status quo. Questioning authority".
        In the absence of evidence, opinion is indistinguishable from prejudice.
[reply]
[d/l]
[select]

In Section Seekers of Perl Wisdom