PDA

View Full Version : Waiting for process to stop



strobins
08-21-2007, 07:40 AM
We're using TotalView on a large program with several processes, including some that use shared memory for IPC. Every now and then we get a dialog box that says "Waiting for process to stop" and a <Cancel> button. All processes appear to hang until the Cancel button is pressed; at which time the error output (apparently from the debugger) repeats the following:

Error: Forcing process 123456 into a known state.
Error: Timeout period truncated by user interrupt
...repeat several times

Has anyone met this sequence of output before? Any ideas?
Thanks in advance!

Josh-TotalView-Tech
08-24-2007, 09:56 AM
Hi,

Normally, you get the "Waiting for process to stop" dialog when you perform an interactive command in the debugger and it needs to wait for it to complete. For example, if you single step over a blocking function call, it may not return right away. One common scenario is that you have a thread stopped at a breakpoint and you single step another thread over a function that attempts to lock a mutex held by the other thread that holds the mutex. I would suggest looking at the status of your other threads in the Threads pane or Root window to see what it says concerning the states of your various threads.

Another suggestion I have is to try using the option '-folow_clone 2' (can be set by adding 'dset TV::follow_clone 2' to your .tvdrc). We have seen some issues similiar to this that are resolved when setting this to 2.

If this does not resolve the problem can you please report this to Technical Support? Please include your license information, the platform and OS details you are using (uname -a should do it), and all the output emitted by TotalView. Please also add the log file generated by executing the CLI command TV::Private::diagnostics after you cancel the said dialog.

eleffant9
07-07-2008, 11:25 AM
Hi Josh. Did you ever find the reason that strobins program was hanging? I have the exact same situation.
Error: Forcing process 10150 into a known state.
Error: Timeout period truncated by user interrupt.

Thanks.

Josh-TotalView-Tech
07-07-2008, 11:33 AM
We did not hear further from strobins. Please do report this issue to support@totalviewtech.com so that it can be properly addressed.

Thanks

eleffant9
07-08-2008, 08:16 AM
OK. Thanks for your reply.