Are there any debug tools which can detect program's hang-up location?

[liping guo]

I am debugging a daemon C++ program which is locked up randomly for some reason. Are
there any tools which can trace the process and detect the hang-up location. (such as which
C++ statement or function the program is locked up?).

OS: hp-ux 10.20. Language C++

Thanks,

Liping

[Sachin Patel]

I guess that will be "tusc"

Go to following site and search for tusc

http://hpux.cs.utah.edu/


Sachin

[Olav Baadsvik · Accepted solution]

Hi,

One thing you could try is the following:


. when the program seems to hang, send
a sigabrt to the process
( kill -6 PID )
this will produce a core file.

. Use a debugger to get the call-stack.
With gdb this can be done as follows:

gdb -tui core
(gdb) info stack


Another thing you could try is to use
the debugger to attach to the running process.

Regards
Olav

[Christopher Caldwell]

DDE will allow you to seize a running process, set break points, examine variable, single step, etc.

http://docs.hp.com/cgi-bin/onlinedocs.py?mpn=B3476-90015&service=hpux&path=00/00/1&title=HP/DDE%20Debugger%20User%27s%20Guide

[A. Clay Stephenson]

I am reading into this that you have the source code for this daemon. If so, my favorite technique (and one that I always do anyway) is to place assert()'s at the beginning and return(s) of each function. You pay no penalty if you complile with -NDEBUG and otherwise you have a log of each function writen to stderr.

Again, this is my standard practice and has served me well over the years in quickly finding problems like yours.

[liping guo]

Hi, Dear friends,
Your information is really helpful. I am going to try them.

Thanks,

Liping