Stepping through code with breakpoints on adjacent address lines.

Stepping through code with breakpoints on adjacent address lines.

[Matt Fyles]

I just wanted to clear up the expected behavior when stepping code that has
breakpoints on adjacent address lines.

In this simple example.

0x0	instruction A
0x4   instruction B
0x8   instruction C

If I break on addresses 0x0, 0x4 and 0x8 with the break *(address)
functionality and then continue to 0x0 and step the other 2 instructions I
only ever see the breakpoint trigger on address 0x0. GDB steps to the next 2
addresses but does not report a breakpoint at either location. Is this the
expected behavior.

It looks from the source in infrun.c (handle_inferior_event()) that if the
trap_expected flag is set then it discards any breakpoint matching at the
current address.

Any thoughts?

Re: Stepping through code with breakpoints on adjacent address lines.

[Andrew Cagney]

Matt Fyles wrote:
> I just wanted to clear up the expected behavior when stepping code that has
> breakpoints on adjacent address lines.
> 
> In this simple example.
> 
> 0x0	instruction A
> 0x4   instruction B
> 0x8   instruction C
> 
> If I break on addresses 0x0, 0x4 and 0x8 with the break *(address)
> functionality and then continue to 0x0 and step the other 2 instructions I
> only ever see the breakpoint trigger on address 0x0. GDB steps to the next 2
> addresses but does not report a breakpoint at either location. Is this the
> expected behavior.
> 
> It looks from the source in infrun.c (handle_inferior_event()) that if the
> trap_expected flag is set then it discards any breakpoint matching at the
> current address.

Smells like a broken edge case.  Here's similar equally wierd behavior:

(top-gdb) info break
Num Type           Disp Enb Address    What
4   breakpoint     keep y   0x1004e83c in main
                                        at 
/home/scratch/GDB/src/gdb/gdb.c:30
         breakpoint already hit 1 time
5   breakpoint     keep y   0x1004e83c in main
                                        at 
/home/scratch/GDB/src/gdb/gdb.c:31
         breakpoint already hit 1 time
6   breakpoint     keep y   0x1004e844 in main
                                        at 
/home/scratch/GDB/src/gdb/gdb.c:33
7   breakpoint     keep y   0x1004e840 in main
                                        at 
/home/scratch/GDB/src/gdb/gdb.c:32
8   breakpoint     keep y   0x1004e84c in main
                                        at 
/home/scratch/GDB/src/gdb/gdb.c:34
         breakpoint already hit 1 time
((top-gdb) run
Breakpoint 4, main (argc=0x1, argv=0x7ffff954)
     at /home/scratch/GDB/src/gdb/gdb.c:31
31        args.argc = argc;
(top-gdb) s
32        args.argv = argv;
(top-gdb)
33        args.use_windows = 0;
(top-gdb)

Breakpoint 8, main (argc=0x1, argv=0x7ffff954)
     at /home/scratch/GDB/src/gdb/gdb.c:34
34        args.interpreter_p = INTERP_CONSOLE;
(top-gdb) info break

Would you be able to come up with an automated testcase?

Andrew