TeamForge : Post

Forum Topic - STOPPED: (10 Items)

View: as

Mario Charest

01/04/2008 11:32 AM

post4052

Sometime when one of my program terminates normally, by returning from main(), it gets in a STOPPED stated.

What can cause a program to go into a STOPPED stated on its own?  I'm guessing is caused by a destructor of some C++ 
global object.

I can I kill a program in STOPPED stated.  Only solution I found is to reboot the computer.

David Gibbs

01/08/2008 2:50 PM

post4126

Re: STOPPED

If stopped --- try hitting with with some death signal (SIGTERM, SIGKILL) followed by a SIGCONT.  I usually see this as 
a result of a debugger trap.

Mario Charest

01/08/2008 3:18 PM

post4130

Re: STOPPED

> If stopped --- try hitting with with some death signal (SIGTERM, SIGKILL) 
> followed by a SIGCONT.  I usually see this as a result of a debugger trap.


I think I already tried that but will try again next time it happens.

Mario Charest

01/10/2008 10:44 AM

post4181

Re: STOPPED

> > If stopped --- try hitting with with some death signal (SIGTERM, SIGKILL) 
> > followed by a SIGCONT.  I usually see this as a result of a debugger trap.
> 
> 
> I think I already tried that but will try again next time it happens.

I doesn't work. I hit the program with SIGKILL and then SIGCONT but nothing.  Only solution was to reboot.

Mario Charest

01/21/2008 8:53 AM

post4459

Re: STOPPED

> > > If stopped --- try hitting with with some death signal (SIGTERM, SIGKILL) 
> 
> > > followed by a SIGCONT.  I usually see this as a result of a debugger trap.
> 
> > 
> > 
> > I think I already tried that but will try again next time it happens.
> 
> I doesn't work. I hit the program with SIGKILL and then SIGCONT but nothing.  
> Only solution was to reboot.
> 


Comments?

Colin Burgess(deleted)

01/21/2008 9:33 AM

post4462

Re: STOPPED

Hi Mario, sorry this slipped through the radar.

We do have a similar PR open on this one, it's on my todo list to get to it.

In the meantime, how reproducable is this?  Do you think it would be
possible
to capture a trace?

You could add an event handler for NTO_TRACE_THSTOPPED and do a
TraceEvent(_NTO_TRACE_STOP),
with the tracelogger in ring buffer mode.

Cheers,

Colin

Mario Charest wrote: 

> > > If stopped --- try hitting with with some death signal (SIGTERM,
SIGKILL) 
> 
> > > followed by a SIGCONT.  I usually see this as a result of a debugger
trap. 
> 
> > 
> > 
> > I think I already tried that but will try again next time it happens. 
> 
> I doesn't work. I hit the program with SIGKILL and then SIGCONT but
nothing.  
> Only solution was to reboot. 
> 

Comments? 

_______________________________________________ 
OSTech 
http://community.qnx.com/sf/go/post4459
<http://community.qnx.com/sf/go/post4459>;  

-- 

cburgess@qnx.com <mailto:cburgess@qnx.com>;

Mario Charest

01/21/2008 6:55 PM

post4486

Re: STOPPED

> Hi Mario, sorry this slipped through the radar.
> 
> We do have a similar PR open on this one, it's on my todo list to get to it.
> 
> In the meantime, how reproducable is this?  Do you think it would be
> possible
> to capture a trace?
> 
> You could add an event handler for NTO_TRACE_THSTOPPED and do a
> TraceEvent(_NTO_TRACE_STOP),
> with the tracelogger in ring buffer mode.
> 
I can try.  Unfortunately it's not reproducible, it seems related to some of the threads crashing. but when in stopped 
state there is only one thread left.

At first I though it could be related to dumper, but I killed dumper will my program was in stopped state and that had 
no affect.

> Cheers,
> 
> Colin
> 
> Mario Charest wrote: 
> 
> > > > If stopped --- try hitting with with some death signal (SIGTERM,
> SIGKILL) 
> > 
> > > > followed by a SIGCONT.  I usually see this as a result of a debugger
> trap. 
> > 
> > > 
> > > 
> > > I think I already tried that but will try again next time it happens. 
> > 
> > I doesn't work. I hit the program with SIGKILL and then SIGCONT but
> nothing.  
> > Only solution was to reboot. 
> > 
> 
> 
> Comments? 
> 
> _______________________________________________ 
> OSTech 
> http://community.qnx.com/sf/go/post4459
> <http://community.qnx.com/sf/go/post4459>;  
> 
> 
> -- 
> 
> cburgess@qnx.com <mailto:cburgess@qnx.com>;

Colin Burgess(deleted)

01/21/2008 10:05 PM

post4489

Re: STOPPED

Hmmm, it could be related to process cleanup.    How often does it happen?

Mario Charest wrote: 

> Hi Mario, sorry this slipped through the radar. 
> 
> We do have a similar PR open on this one, it's on my todo list to get to
it. 
> 
> In the meantime, how reproducable is this?  Do you think it would be 
> possible 
> to capture a trace? 
> 
> You could add an event handler for NTO_TRACE_THSTOPPED and do a 
> TraceEvent(_NTO_TRACE_STOP), 
> with the tracelogger in ring buffer mode. 
> 
I can try.  Unfortunately it's not reproducible, it seems related to some of
the threads crashing. but when in stopped state there is only one thread
left.

At first I though it could be related to dumper, but I killed dumper will my
program was in stopped state and that had no affect.

> Cheers, 
> 
> Colin 
> 
> Mario Charest wrote: 
> 
> > > > If stopped --- try hitting with with some death signal (SIGTERM, 
> SIGKILL) 
> > 
> > > > followed by a SIGCONT.  I usually see this as a result of a debugger

> trap. 
> > 
> > > 
> > > 
> > > I think I already tried that but will try again next time it happens. 
> > 
> > I doesn't work. I hit the program with SIGKILL and then SIGCONT but 
> nothing.  
> > Only solution was to reboot. 
> > 
> 
> 
> Comments? 
> 
> _______________________________________________ 
> OSTech 
> http://community.qnx.com/sf/go/post4459
<http://community.qnx.com/sf/go/post4459>;  
> <http://community.qnx.com/sf/go/post4459
<http://community.qnx.com/sf/go/post4459>; >  
> 
> 
> -- 
> 
> cburgess@qnx.com <mailto:cburgess@qnx.com>;  <mailto:cburgess@qnx.com
<mailto:cburgess@qnx.com>; > 




_______________________________________________ 
OSTech 
http://community.qnx.com/sf/go/post4486
<http://community.qnx.com/sf/go/post4486>;  


-- 

cburgess@qnx.com <mailto:cburgess@qnx.com>;

Mario Charest

02/08/2008 2:04 PM

post4876

Re: STOPPED

> Hmmm, it could be related to process cleanup.    How often does it happen?
> 

I'm not sure cause it has happend for a while but I think it's because the process uses lot of ram (400Meg) and it 
crashed on termination and is being dumped by dumper.  

Given our usage of slow flash drive it probably takes a minute or 2 for the dumper to do its thing. Long enough for me 
to get confused ;-)

I have fixed some bug on process termination which probably explain why I haven't see this problem for a while.


May I suggest a new process/thread state call dumper DUMPING.

Colin Burgess(deleted)

02/08/2008 2:09 PM

post4877

Re: STOPPED

At least pidin should check the process flags for _NTO_PF_COREDUMP

Mario Charest wrote:
>  > Hmmm, it could be related to process cleanup.    How often does it 
> happen?
>  >
> 
> I'm not sure cause it has happend for a while but I think it's because 
> the process uses lot of ram (400Meg) and it crashed on termination and 
> is being dumped by dumper. 
> 
> Given our usage of slow flash drive it probably takes a minute or 2 for 
> the dumper to do its thing. Long enough for me to get confused ;-)
> 
> I have fixed some bug on process termination which probably explain why 
> I haven't see this problem for a while.
> 
> 
> May I suggest a new process/thread state call dumper DUMPING.
> 
> 
> 
> _______________________________________________
> OSTech
> http://community.qnx.com/sf/go/post4876
> 

-- 
cburgess@qnx.com

Return