Using Advanced Breakpoint Techniques

This section demonstrates some advanced techniques for using breakpoints:

• Using breakpoint counts

• Using bounded breakpoints

• Picking a useful breakpoint count

• Watchpoints

• Using breakpoint conditions

• Micro replay using pop

• Using fix and continue

This section, and the example program, are inspired by an actual bug discovered in dbx using much the same sequence described in this section.

The source code includes a sample input file named in, which triggers a bug in the example program. in contains the following code:

display nonexistent_var	# should yield an error
display var
stop in X	# will cause one "stopped" message and display
stop in Y	# will cause second "stopped" message and display
run
cont
cont
run
cont
cont

When you run the program with the input file, the output is as follows:

$ a.out < in
> display nonexistent_var 
error: Don't know about 'nonexistent_var'
> display var 
will display 'var'
> stop in X 
> stop in Y 
> run 
running ...
stopped in X
var = {
      a = '100'
      b = '101
      c = '<error>'
      d = '102
      e = '103'
      f = '104'
      }
> cont 
stopped in Y
var = {
      a = '105'
      b = '106'
      c = '<error>'
      d = '107'
      e = '108'
      f = '109'
      }
> cont 
exited
> run 
running ...
stopped in X
var = {
      a = '110'
      b = '111'

      c = '<error>'
      d = '112'
      e = '113'
      f = '114'
      }
> cont 
stopped in Y
var = {
      a = '115'
      b = '116'

      c = '<error>'
      d = '117'
      e = '118'
      f = '119'
      }
> cont 
exited
> quit
Goodby

This output might seem voluminous but the point of this example is to illustrate techniques to be used with long running, complex programs where stepping through code or tracing just are not practical.

Notice that when showing the value of field c, you get a value of <error>. Such a situation might occur if the field contains a bad address.