ProcessRecord problem with recursion

ProcessRecord problem with recursion

[Marc Khouzam]

Hi,

I'm having problems with ProcessRecord and recursion.
It looks like the reverse-next operation behaves like
reverse-step when dealing with a recursive method.

I have GDB HEAD from the 18th of March, with the patches included in
http://sourceware.org/ml/gdb-patches/2009-03/msg00375.html
http://sourceware.org/ml/gdb-patches/2009-03/msg00005.html
http://sourceware.org/ml/gdb-patches/2009-01/msg00444.html

Here is a program and session that shows the problem:

GNU gdb (GDB) 6.8.50.20090318-cvs
[...]
(gdb) l
1       int factorial(int x) {
2          if (x == 1) return 1;
3          int result = x * factorial(x-1);
4          return result;
5       }
6
7       int main() {
8           factorial(5);
9           return 0;
10      }
(gdb) start
Temporary breakpoint 1 at 0x804847b: file a.cc, line 8.
Starting program: /local/home/lmckhou/testing/a.out 

Temporary breakpoint 1, main () at a.cc:8
8           factorial(5);
(gdb) rec
(gdb) s
factorial (x=5) at a.cc:2
2          if (x == 1) return 1;
(gdb) n
3          int result = x * factorial(x-1);
(gdb) n
4          return result;
(gdb) rn
factorial (x=4) at a.cc:5
5       }

Notice how the reverse-next(rn) command jumped to the end
of the next factorial method on the stack (x=4 instead of x=5).

Thanks

Marc

Re: ProcessRecord problem with recursion

[Hui Zhu]

Thanks Marc,

That is a very cool test code.

4	  return result;
0x0804844e <_Z9factoriali+42>:	mov    -0x4(%ebp),%eax
0x08048451 <_Z9factoriali+45>:	mov    %eax,-0x14(%ebp)
0x08048454 <_Z9factoriali+48>:	mov    -0x14(%ebp),%eax
(gdb) rn
infrun: clear_proceed_status_thread (process 9955)
infrun: proceed (addr=0xffffffff, signal=144, step=1)
infrun: resume (step=1, signal=0), trap_expected=0
infrun: wait_for_inferior (treat_exec_as_sigtrap=0)
infrun: target_wait (-1, status) =
infrun:   9955 [process 9955],
infrun:   status->kind = stopped, signal = SIGTRAP
infrun: infwait_normal_state
infrun: TARGET_WAITKIND_STOPPED
infrun: stop_pc = 0x804844b
infrun: keep going
infrun: resume (step=1, signal=0), trap_expected=0
infrun: prepare_to_wait
infrun: target_wait (-1, status) =
infrun:   9955 [process 9955],
infrun:   status->kind = stopped, signal = SIGTRAP
infrun: infwait_normal_state
infrun: TARGET_WAITKIND_STOPPED
infrun: stop_pc = 0x8048447
infrun: stepping inside range [0x8048439-0x804844e]
infrun: resume (step=1, signal=0), trap_expected=0
infrun: prepare_to_wait
infrun: target_wait (-1, status) =
infrun:   9955 [process 9955],
infrun:   status->kind = stopped, signal = SIGTRAP
infrun: infwait_normal_state
infrun: TARGET_WAITKIND_STOPPED
infrun: stop_pc = 0x8048458
infrun: keep going
infrun: resume (step=1, signal=0), trap_expected=0
infrun: prepare_to_wait
infrun: target_wait (-1, status) =
infrun:   9955 [process 9955],
infrun:   status->kind = stopped, signal = SIGTRAP
infrun: infwait_normal_state
infrun: TARGET_WAITKIND_STOPPED
infrun: stop_pc = 0x8048457
infrun: stepping inside range [0x8048457-0x804845a]
infrun: stop_stepping
factorial (x=4) at b.cc:5
5	}
0x08048457 <_Z9factoriali+51>:	leave
0x08048458 <_Z9factoriali+52>:	ret
0x08048459:	nop


I had reproduced it.  I think this issue is about reverse execute.  I
will try to deal with it.

Thanks,
Hui



On Fri, Mar 20, 2009 at 09:13, Marc Khouzam <marc.khouzam@ericsson.com> wrote:
> Hi,
>
> I'm having problems with ProcessRecord and recursion.
> It looks like the reverse-next operation behaves like
> reverse-step when dealing with a recursive method.
>
> I have GDB HEAD from the 18th of March, with the patches included in
> http://sourceware.org/ml/gdb-patches/2009-03/msg00375.html
> http://sourceware.org/ml/gdb-patches/2009-03/msg00005.html
> http://sourceware.org/ml/gdb-patches/2009-01/msg00444.html
>
> Here is a program and session that shows the problem:
>
> GNU gdb (GDB) 6.8.50.20090318-cvs
> [...]
> (gdb) l
> 1       int factorial(int x) {
> 2          if (x == 1) return 1;
> 3          int result = x * factorial(x-1);
> 4          return result;
> 5       }
> 6
> 7       int main() {
> 8           factorial(5);
> 9           return 0;
> 10      }
> (gdb) start
> Temporary breakpoint 1 at 0x804847b: file a.cc, line 8.
> Starting program: /local/home/lmckhou/testing/a.out
>
> Temporary breakpoint 1, main () at a.cc:8
> 8           factorial(5);
> (gdb) rec
> (gdb) s
> factorial (x=5) at a.cc:2
> 2          if (x == 1) return 1;
> (gdb) n
> 3          int result = x * factorial(x-1);
> (gdb) n
> 4          return result;
> (gdb) rn
> factorial (x=4) at a.cc:5
> 5       }
>
> Notice how the reverse-next(rn) command jumped to the end
> of the next factorial method on the stack (x=4 instead of x=5).
>
> Thanks
>
> Marc
>
>

Re: ProcessRecord problem with recursion

[Hui Zhu]

Hi Marc,

I had post a patch for it in
http://sourceware.org/ml/gdb-patches/2009-03/msg00428.html

Thanks,
Hui

On Fri, Mar 20, 2009 at 09:13, Marc Khouzam <marc.khouzam@ericsson.com> wrote:
> Hi,
>
> I'm having problems with ProcessRecord and recursion.
> It looks like the reverse-next operation behaves like
> reverse-step when dealing with a recursive method.
>
> I have GDB HEAD from the 18th of March, with the patches included in
> http://sourceware.org/ml/gdb-patches/2009-03/msg00375.html
> http://sourceware.org/ml/gdb-patches/2009-03/msg00005.html
> http://sourceware.org/ml/gdb-patches/2009-01/msg00444.html
>
> Here is a program and session that shows the problem:
>
> GNU gdb (GDB) 6.8.50.20090318-cvs
> [...]
> (gdb) l
> 1       int factorial(int x) {
> 2          if (x == 1) return 1;
> 3          int result = x * factorial(x-1);
> 4          return result;
> 5       }
> 6
> 7       int main() {
> 8           factorial(5);
> 9           return 0;
> 10      }
> (gdb) start
> Temporary breakpoint 1 at 0x804847b: file a.cc, line 8.
> Starting program: /local/home/lmckhou/testing/a.out
>
> Temporary breakpoint 1, main () at a.cc:8
> 8           factorial(5);
> (gdb) rec
> (gdb) s
> factorial (x=5) at a.cc:2
> 2          if (x == 1) return 1;
> (gdb) n
> 3          int result = x * factorial(x-1);
> (gdb) n
> 4          return result;
> (gdb) rn
> factorial (x=4) at a.cc:5
> 5       }
>
> Notice how the reverse-next(rn) command jumped to the end
> of the next factorial method on the stack (x=4 instead of x=5).
>
> Thanks
>
> Marc
>
>

ProcessRecord problem with recursion

[jian shen]

I find an issue with record in recursion function.  And I notice
there's a patch long ago.
https://sourceware.org/ml/gdb/2009-03/msg00134.html
But I still could reproduce the issue with the latest gdb7.7.

Now the code is different and the patch could not be applied.

Is it a missed bug?  Is there any patch for the latest code?

Thanks,
Jian