I am attaching a backtrace...
I am using HP-ux 11i, mico 2.3.7 compiled with aCC. micod is started with
-ORBIIOPAddr inet:myMachine:1221.
.micorc file...
-ORBImplRepoAddr inet:myMachine:1221
-ORBNamingAddr inet:myMachine:1221
-ORBBindAddr inet:myMachine:1221
Also using imr as a lookup for my clients and a way to start the servers.
backtrace
#0 MICO::TCPTransport::wselect () at /mico/orb/transport.cc:150
#1 in MICO::GIOPConn::check_busy() at /mico/orb/iop.cc:2454
#2 in MICO::GIOPConn::output () at /mico/orb/iop.cc:2506
#3 in MICO::IIOPProxy::invoke () at /mico/orb/iop.cc:3038
#4 in CORBA::ORB::invoke_async () at /mico/orb/orb.cc 1944
#5 in CORBA::StaticRequest::oneway () at /mico/orb/static.cc:2248
#6 in MyModule::MyInterface_stub::put () at MyModule.cc
#7 - #12 UIMX wrapper from X windows Gui calls
#13 in XtCallCallbackList () from /usr/lib/libXt.3
#14 in ActivateCommon () from /usr/lib/libXm.4
#15 in Activate () from /usr/lib/libXm.4
#16 in HandleActions () from /usr/lib/libXt.3
#17 in HandleComplexState () from /usr/lib/libXt.3
#18 in _XtTranslateEvent () from /usr/lib/libXt.3
#19 in XtDispatchEventToWidget () from /usr/lib/libXt.3
#20 in _XtDefaultDispatcher () from /usr/lib/libXt.3
#21 in XtDispatchEvent () from /usr/lib/libXt.3
#22 in XtAppProcessEvent () from /usr/lib/libXt.3
#23 in X11Dispatcher::run() at x11.cc:143
#24 in CORBA::ORB::run() at /mico/orb/orb.cc:1346
#25 in main () at main.C
What I found is if iop.cc line 2497 "if (b->length()==0)" is true, then
everything works fine. If this fails, then execution gets to iop.cc line 2506
(backtrace #3 above), which causes the sigseg.
If I break the program at iop.cc line 2496 "_transp->write (*b,
b->length());" and then run - it works fine. So it seems like there is a
timing issue.
This problem is not consistent - in that if I start everything up, sometimes
it happens, some times it doesn't. However, once the problem occurs, it will
happen again and again until I restart micod.
I ran into the same exact problem on mico 2.3.11, but the line numbers were
different (haven't tried on 2.3.12 yet).
Thanks for any help that you have to offer - I'm stumped and I'm not getting
very far with my guesses. Mark
Karel Gardas <[EMAIL PROTECTED]> wrote:
Mark,
it would great if you experience any crashes to obtain a backtrace for
us. Also the way how did you configured MICO and what exactly platform
(OS + C++ compiler name and version) do you use might also be important.
Anyway, I always recommend to try to duplicate any bug in obsolete
releases with the last release. Perhaps it's already solved?
Thanks,
Karel
Mark Richardson wrote:
> Hi all, You've all been a great help in the past, so I thought I'd
> see if yout could help me with this.
>
> We've been running HP-Ux 11i with mico 2.3.7 (I'll get to the update
> in a minute). It's been running great with a couple of exceptions.
> One particular Gui client crashes, and after debugging I found that
> mico is crashing on MICO::TCPTransport::write. Specifically, the
> MICO::GIOPConn::Output which I found to crash when the tail pointer
> doesn't end up at the head pointer after a write. Other machines
> never have the tail pointer not equal to the head pointer after a
> write.
>
> This is an odd failure because it doesn't happen on other machines,
> and it's sporatic on this particular machine. This machine is
> running X11 and most of the others are running X11 version 1.
>
> So I thought I'd upgrade to mico 2.3.11 (I've found mico 2.3.12+ has
> too many changes to go that far yet). After this, the Gui crashes
> during an orb->bind call. But I don't think it's the speicific call
> to bind, because the debugger shows that its after the X11Dispatcher
> on a callback.
>
> Interesting data point is when I ran this gui on other machines it
> worked with just mico2.3.11 and micocoss2.3.11 lib. While this
> problem machine also required micox2.3.11 lib.
>
> So the question comes down to this - Is there a known issue with X11
> vs X11 version 1 and event processing when working with mico2.3.7 or
> mico2.3.11?
>
> Thanks for any help/ideas/suggestions in advance, Mark
>
>
> --------------------------------- We won't tell. Get more on shows
> you hate to love (and love to hate): Yahoo! TV's Guilty Pleasures
> list.
>
>
> ------------------------------------------------------------------------
>
>
> _______________________________________________ Mico-devel mailing
> list Mico-devel@mico.org
> http://www.mico.org/mailman/listinfo/mico-devel
--
Karel Gardas [EMAIL PROTECTED]
ObjectSecurity Ltd. http://www.objectsecurity.com
---------------------------------
Get your own web address.
Have a HUGE year through Yahoo! Small Business.
_______________________________________________
Mico-devel mailing list
Mico-devel@mico.org
http://www.mico.org/mailman/listinfo/mico-devel
