Get Backtrace
 -------------

  Written by Athanasius <Athanasius@miggy.org>
  Help file by Silver <silver@ewtoo.org>

 ## This code requires "expect" - the RedHat RPM of this can be found
 ## from the RedHat/RPMS directory of any redhat linux site with the
 ## filename 'expect-5.26-20.i386.rpm'


 How to set up
 -------------

 Cut out the code between the BEGIN and the ENDs and save them in your
'pgplus/bin' directory under the names specified. Be sure to edit the
'get-bt.exp' file and change the following variables:

     basedir = the directory pointing to the talker
     pidfile = the location of the PID file
     logfile = the location of the file where the logs will go

 Compile your talker with DEBUGGING turned *on*. and USE_LIBG turned *on*.
If these are undefined then the logs will be useless. Compile and boot your
talker.

 Run the get-bt file with:

    nohup get-bt.exp &

 a log file will appear which will be blank (at the beginning). When the code
crashes, this script with do a "bt" (backtrace), store this information and
then do "c" (a continue). Therefore to see what code is crashing your
talker, just look at this file.

 To get rid of get-bt, just kill the process. If you wish to reboot the
talker (either hard or soft) kill get-bt first.

 NOTE: It is not recommended to run get-bt as a general purpose debugging
       tool since the use of DEBUGGING and LIBG is required which increase
       the size of the executables drastically. Once you have found and
       fixed the bugs, uncomment DEBUGGING, recompile and reboot otherwise
       your systems administrator may complain!

--BEGIN  get-bt.exp ----------------------------------------------------
#!/usr/local/bin/expect -f
#
#  Attach a gdb to a running talker executeable:
#
#       If it crashes get a 'bt' then 'continue'.
#       Attach to the NEW server process, repeat
#

# Base directory where the talker runs
set basedir /home/username/pgplus
# File holding PID of running talker server
set pidfile junk/PID
# Where we keep our log
set logfile logs/get-bt.log

# A loop:
#       Get talker pid
#       Attach gdb to it, and 'continue'
#       Wait for a crash, 'bt', 'continue'
#       Exit gdb, loop

# Our log of what happened
log_file $logfile

# match_max 10000
# for start test next body
for {} 1 {} {
        # Grab the pid of a running talker server
        while {![file readable $pidfile]} {
                sleep 5
        }
        set file [open $pidfile]
        gets $file process_id
        
        # Fire up gdb
        set timeout -1
        for {spawn gdb bin/surfers $process_id} 1 {} {
                send_log [timestamp -format "%d/%m/%y - %H:%M:%S"]
                expect_before "Program received signal SIGSEGV" { send "bt\r" ; 
send "continue\r" }
                expect {
                        "Program exited " { send "quit\r" ; break }
                        "The program is not being run" { send "quit\r" ; break }
                        "(gdb) bt" ""
                        "(gdb) continue" ""
                        "(gdb)" { send "continue\r" }
                }
        }
        send_log [timestamp -format "%d/%m/%y - %H:%M:%S"]
        set timeout 10
        expect "Program exited " { send "quit\r" ; break }
        wait
        sleep 5
}
--END    get-bt.exp ----------------------------------------------------

   And in the directory you run this from:

--BEGIN .gdbinit -------------------------------------------------------
handle SIGPIPE nostop
handle SIGPIPE noprint
handle SIGWINCH nostop
handle SIGWINCH noprint
handle SIGUSR1 nostop
handle SIGUSR1 noprint
handle SIGUSR2 nostop
handle SIGUSR2 noprint
--END   .gdbinit -------------------------------------------------------