FAQ Answerpage
QUESTION:
How do I get a "traceback" of a crash?
ANSWER:
The following detailed response will be included in the upcoming User Workbook. At that time, the details here will be omitted and replaced with a pointer to the User Workbook.
A traceback is a history of "what called what" and leads up to the routine that contained the crash. It is useful for narrowing down where to look further. You get a traceback on the debugger. On Sun and AIX, the command is:
dbx {TYPE YOUR PROGRAM NAME HERE}
On OSF, you use a different debugger and the command is:
ladebug {TYPE YOUR PROGRAM NAME HERE}
If running from workdir, your program is generally in ../bin/$BFARCH, so this command is either:
dbx ../bin/$BFARCH/BetaApp
OR
ladebug ../bin/$BFARCH/BetaApp
Once the debugger starts up (on Sun, you might have to page down through some announcements) you can run your program with the command:
run {TYPE ARGUMENTS HERE}
For example, type the command:
run ../BetaUser/xdr.tcl
You've now started the Framework, and can make your job crash just like last time. Again, as a complete example, to run Beta from .xdr under the debugger on OSF, type the commands:
ladebug ../bin/$BFARCH/BetaApp
run ../BetaUser/xdr.tcl
Once your program crashes (REMEMBER, we want it to crash!) you'll get a debugger prompt. At that point, to get a traceback, type the word:
where
A list with a lot of details will print onscreen. Basically, these details correspond to a reverse-ordered list of what calls what. Cut and paste that list and mail it to the relevant expert.
(Last updated 7/22/98)
[ SLAC Homepage] [Search] [FAQ Homepage]
If you have a FAQ (with or WITHOUT an answer), use our prototype submission form.
If you have comments or suggestions, email me at sison@slac.stanford.edu
|
