Re: Weird reverse problems

[teawater <teawater@gmail.com>]

Hi,

I make a patch for it in
http://sourceware.org/ml/gdb-patches/2009-01/msg00444.html

Thanks,
Hui

On Thu, Jan 22, 2009 at 14:05, teawater <teawater@gmail.com> wrote:
> Hi Marc,
>
> I had reproduced it.
>
> 13           printf("A\n");
> (gdb) rn
> infrun: clear_proceed_status_thread (process 21364)
> 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: infwait_normal_state
> infrun: TARGET_WAITKIND_STOPPED
> infrun: stop_pc = 0xb7f01af8
> infrun: stepped into subroutine
> infrun: inserting step-resume breakpoint at 0xb7f01980
> infrun: resume (step=0, signal=0), trap_expected=0
> infrun: prepare_to_wait
> infrun: infwait_normal_state
> infrun: TARGET_WAITKIND_STOPPED
> infrun: stop_pc = 0xb7f01980
> infrun: BPSTAT_WHAT_STEP_RESUME
> infrun: resume (step=1, signal=0), trap_expected=1
> infrun: prepare_to_wait
> infrun: infwait_normal_state
> infrun: TARGET_WAITKIND_STOPPED
> infrun: stop_pc = 0x80482f8
> infrun: stepped into subroutine
> infrun: inserting step-resume breakpoint at 0x80482f8
> infrun: resume (step=0, signal=0), trap_expected=0
> infrun: prepare_to_wait
> infrun: infwait_normal_state
> infrun: TARGET_WAITKIND_STOPPED
> infrun: stop_pc = 0x80482f8
> infrun: BPSTAT_WHAT_STEP_RESUME
> infrun: resume (step=1, signal=0), trap_expected=1
> infrun: prepare_to_wait
> infrun: infwait_normal_state
> infrun: TARGET_WAITKIND_STOPPED
> infrun: stop_pc = 0x80483dc
> infrun: keep going
> infrun: resume (step=1, signal=0), trap_expected=0
> infrun: prepare_to_wait
> infrun: infwait_normal_state
> infrun: TARGET_WAITKIND_STOPPED
> infrun: stop_pc = 0x80483d5
> infrun: stepping inside range [0x80483d5-0x80483e1]
> infrun: stop_stepping
> 11         sleep(0);//3
> (gdb)
>
>
>
> I think this is a bug of reverse code in infrun.c.  That is because it
> deal with plt not very well.
>
> Michael, could you please take a look at this bug?
> And I will try to deal with it too. :)
>
> Thanks,
> Hui
>
> On Thu, Jan 22, 2009 at 02:52, Marc Khouzam <marc.khouzam@ericsson.com> wrote:
>> Hi,
>>
>> so I'm trying Reverse Debugging by having applied teawater's patches
>> on top of HEAD which contains Michael S reverse infrastructure.
>>
>> I ran into some bizarre behavior when reverse stepping.
>> I found that with (I'm guessing) library functions, a reverse next
>> will skip a line.  If you look at the simple session I ran below
>> (a bit long, but very simple) you can see that on the reverse,
>> every two printf gets skipped.
>> Also, on a reverse step, multiple sleep() can be skipped.
>>
>> Weird no?
>>
>> I tried to look at the code, but performing steps gets pretty
>> low level, so I gave up.
>>
>> Thanks
>>
>> Marc
>>
>> GNU gdb (GDB) 6.8.50.20090113-cvs
>> Copyright (C) 2009 Free Software Foundation, Inc.
>> License GPLv3+: GNU GPL version 3 or later
>> <http://gnu.org/licenses/gpl.html>
>> This is free software: you are free to change and redistribute it.
>> There is NO WARRANTY, to the extent permitted by law.  Type "show
>> copying"
>> and "show warranty" for details.
>> This GDB was configured as "i686-pc-linux-gnu".
>> For bug reporting instructions, please see:
>> <http://www.gnu.org/software/gdb/bugs/>...
>> (gdb) l
>> 1       #include <stdio.h>
>> 2       #include <unistd.h>
>> 3
>> 4       int foo(int i) {
>> 5           return i;
>> 6       }
>> 7       int main()
>> 8       {
>> 9           sleep(0);//1
>> 10          sleep(0);//2
>> (gdb)
>> 11          sleep(0);//3
>> 12          sleep(0);//4
>> 13          printf("A\n");
>> 14          printf("B\n");
>> 15          printf("C\n");
>> 16          printf("D\n");
>> 17          int a = 0;
>> 18          a++;
>> 19          printf("E\n");
>> 20          printf("F\n");
>> (gdb)
>> 21          printf("G\n");
>> 22          printf("H\n");
>> 23          return a;
>> 24      }
>> (gdb) b main
>> Breakpoint 1 at 0x80484bd: file a.cc, line 9.
>> (gdb) r
>> Starting program: /local/home/lmckhou/testing/a.out
>>
>> Breakpoint 1, main () at a.cc:9
>> 9           sleep(0);//1
>> (gdb) record
>> (gdb) b 24
>> Breakpoint 2 at 0x804855b: file a.cc, line 24.
>> (gdb) c
>> Continuing.
>> A
>> B
>> C
>> D
>> E
>> F
>> G
>> H
>>
>> Breakpoint 2, main () at a.cc:24
>> 24      }
>> (gdb) rn
>> 23          return a;
>> (gdb) rn
>> 21          printf("G\n");    <--- Missed a printf here!
>> (gdb) rn
>> 19          printf("E\n");    <--- Missed a printf here!
>> (gdb) rn
>> 18          a++;
>> (gdb) rn
>> 17          int a = 0;
>> (gdb) rn
>> 15          printf("C\n");    <--- Missed a printf here!
>> (gdb) rn
>> 13          printf("A\n");    <--- Missed a printf here!
>> (gdb) rn
>> 11          sleep(0);//3    <--- Missed a sleep here!
>> (gdb) rn
>>
>> No more reverse-execution history.
>> main () at a.cc:9
>> 9           sleep(0);//1
>> (gdb) n
>> 10          sleep(0);//2
>> (gdb) n
>> 11          sleep(0);//3
>> (gdb) n
>> 12          sleep(0);//4
>> (gdb) n
>> 13          printf("A\n");
>> (gdb) rs
>> 12          sleep(0);//4
>> (gdb) rs
>> 11          sleep(0);//3
>> (gdb) rs
>> 10          sleep(0);//2
>> (gdb) n
>> 11          sleep(0);//3
>> (gdb) n
>> 12          sleep(0);//4
>> (gdb) n
>> 13          printf("A\n");
>> (gdb) n
>> 14          printf("B\n");
>> (gdb) n
>> 15          printf("C\n");
>> (gdb) rs
>> 14          printf("B\n");
>> (gdb)
>>
>> No more reverse-execution history.
>> main () at a.cc:9
>> 9           sleep(0);//1       <--- Woah, missed a bunch of stuff on the
>> reverse step!
>>
>>
>>
>>
>