[Bug breakpoints/16280] New: Step into an inline function with varargs doesn't stop

[Bug breakpoints/16280] New: Step into an inline function with varargs doesn't stop

[psmith at gnu dot org]

https://sourceware.org/bugzilla/show_bug.cgi?id=16280

            Bug ID: 16280
           Summary: Step into an inline function with varargs doesn't stop
           Product: gdb
           Version: 7.6
            Status: NEW
          Severity: normal
          Priority: P2
         Component: breakpoints
          Assignee: unassigned at sourceware dot org
          Reporter: psmith at gnu dot org

Created attachment 7311
  --> https://sourceware.org/bugzilla/attachment.cgi?id=7311&action=edit
Repro case compressed tarball

I am using GCC 4.8.1 and GDB 7.6.1 built with Python 2.7.6 on my GNU/Linux
system.  In a C++ inline function that uses va_list, if I try to step into that
inline function it will not stop and just keep running.  If I use "next" to
skip over the function, that works properly.  If I use "si" to step by
instructions until I get into the inline function, then I can continue to use
normal step/next commands and it works properly.

I've also tried this with GCC 4.8.0 / GDB 7.6 / Python 2.7.5, with the same
result.

Attached is a tar file containing a repro case.  Compile it like this:

    gcc -g -o step main.cpp Step.cpp

Now debugging:

  $ gdb ./step
  GNU gdb (GDB) 7.6.1
  ...
  Reading symbols from ./step...done.
  (gdb) l
  1       #include "Step.h"
  2       
  3       int main(int argc, const char** argv)
  4       {
  5           Step step;
  6           step.show("hi %s\n", "there");
  7           return 0;
  8       }

  (gdb) br 6
  Breakpoint 1 at 0x4006ff: file main.cpp, line 6.

  (gdb) run
  Starting program: ./step 

  Breakpoint 1, main (argc=1, argv=0x7fffffffe638) at main.cpp:6
  6           step.show("hi %s\n", "there");

  (gdb) step
  hi there
  [Inferior 1 (process 24724) exited normally]

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug breakpoints/16280] Step into an inline function with varargs doesn't stop

[palves at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=16280

Pedro Alves <palves at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |palves at redhat dot com

--- Comment #1 from Pedro Alves <palves at redhat dot com> ---
Temporary breakpoint 1, main (argc=1, argv=0x7fffffffdb38) at main.cpp:6
6           step.show("hi %s\n", "there");
(gdb) set debug infrun 1
(gdb) s
infrun: clear_proceed_status_thread (process 20301)
infrun: proceed (addr=0xffffffffffffffff, signal=GDB_SIGNAL_DEFAULT, step=1)
infrun: resume (step=1, signal=GDB_SIGNAL_0), trap_expected=0, current thread
[process 20301] at 0x4006bb
infrun: wait_for_inferior ()
infrun: target_wait (-1, status) =
infrun:   20301 [process 20301],
infrun:   status->kind = stopped, signal = GDB_SIGNAL_TRAP
infrun: infwait_normal_state
infrun: TARGET_WAITKIND_STOPPED
infrun: stop_pc = 0x4006c0
infrun: stepping inside range [0x4006bb-0x4006cf]
infrun: resume (step=1, signal=GDB_SIGNAL_0), trap_expected=0, current thread
[process 20301] at 0x4006c0
infrun: prepare_to_wait
infrun: target_wait (-1, status) =
infrun:   20301 [process 20301],
infrun:   status->kind = stopped, signal = GDB_SIGNAL_TRAP
infrun: infwait_normal_state
infrun: TARGET_WAITKIND_STOPPED
infrun: stop_pc = 0x4006c5
infrun: stepping inside range [0x4006bb-0x4006cf]
infrun: resume (step=1, signal=GDB_SIGNAL_0), trap_expected=0, current thread
[process 20301] at 0x4006c5
infrun: prepare_to_wait
infrun: target_wait (-1, status) =
infrun:   20301 [process 20301],
infrun:   status->kind = stopped, signal = GDB_SIGNAL_TRAP
infrun: infwait_normal_state
infrun: TARGET_WAITKIND_STOPPED
infrun: stop_pc = 0x4006ca
infrun: stepping inside range [0x4006bb-0x4006cf]
infrun: resume (step=1, signal=GDB_SIGNAL_0), trap_expected=0, current thread
[process 20301] at 0x4006ca
infrun: prepare_to_wait
infrun: target_wait (-1, status) =
infrun:   20301 [process 20301],
infrun:   status->kind = stopped, signal = GDB_SIGNAL_TRAP
infrun: infwait_normal_state
infrun: TARGET_WAITKIND_STOPPED
infrun: stop_pc = 0x4006e0
infrun: stepped into subroutine
infrun: inserting step-resume breakpoint at 0x400712
infrun: resume (step=0, signal=GDB_SIGNAL_0), trap_expected=0, current thread
[process 20301] at 0x4006e0
infrun: prepare_to_wait
hi there
infrun: target_wait (-1, status) =
infrun:   20301 [process 20301],
infrun:   status->kind = exited, status = 0
infrun: infwait_normal_state
infrun: TARGET_WAITKIND_EXITED
[Inferior 1 (process 20301) exited normally]
infrun: stop_stepping
(gdb)

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug breakpoints/16280] Step into an inline function with varargs doesn't stop

[palves at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=16280

--- Comment #2 from Pedro Alves <palves at redhat dot com> ---
> infrun: stop_pc = 0x4006e0
> infrun: stepped into subroutine
> infrun: inserting step-resume breakpoint at 0x400712

GDB detected the program stepped into Step::show, and tried to put a breakpoint
past the prologue (0x400712), and run to it.  But, the breakpoint is never hit:

> infrun: resume (step=0, signal=GDB_SIGNAL_0), trap_expected=0, current thread > [process 20301] at 0x4006e0
> infrun: prepare_to_wait
> hi there
> infrun: target_wait (-1, status) =
> infrun:   20301 [process 20301],
> infrun:   status->kind = exited, status = 0
> infrun: infwait_normal_state
> infrun: TARGET_WAITKIND_EXITED
> [Inferior 1 (process 20301) exited normally]
> infrun: stop_stepping
> (gdb)

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug breakpoints/16280] Step into an inline function with varargs doesn't stop

[palves at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=16280

--- Comment #3 from Pedro Alves <palves at redhat dot com> ---
So, a simpler reproducer is to try to set a breakpoint in the function, and run
to it, no stepping involved, which fails as well:

...
Temporary breakpoint 4, main (argc=1, argv=0x7fffffffdb38) at main.cpp:6
6           step.show("hi %s\n", "there");
(gdb) b Step::show
Breakpoint 2 at 0x400712: file Step.h, line 15.
(gdb) c
Continuing.
infrun: clear_proceed_status_thread (process 20347)
infrun: proceed (addr=0xffffffffffffffff, signal=GDB_SIGNAL_DEFAULT, step=0)
infrun: resume (step=0, signal=GDB_SIGNAL_0), trap_expected=0, current thread
[process 20347] at 0x4006bb
hi there
infrun: wait_for_inferior ()
infrun: target_wait (-1, status) =
infrun:   20347 [process 20347],
infrun:   status->kind = exited, status = 0
infrun: infwait_normal_state
infrun: TARGET_WAITKIND_EXITED
[Inferior 1 (process 20347) exited normally]
infrun: stop_stepping
(gdb)

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug breakpoints/16280] Step into an inline function with varargs doesn't stop

[palves at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=16280

--- Comment #4 from Pedro Alves <palves at redhat dot com> ---
(gdb) disassemble Step::show
Dump of assembler code for function Step::show(char const*, ...):
   0x00000000004006e0 <+0>:     push   %rbp
   0x00000000004006e1 <+1>:     mov    %rsp,%rbp
   0x00000000004006e4 <+4>:     sub    $0xe0,%rsp
   0x00000000004006eb <+11>:    mov    %rsi,-0xa8(%rbp)
   0x00000000004006f2 <+18>:    mov    %rdx,-0xa0(%rbp)
   0x00000000004006f9 <+25>:    mov    %rcx,-0x98(%rbp)
   0x0000000000400700 <+32>:    mov    %r8,-0x90(%rbp)
   0x0000000000400707 <+39>:    mov    %r9,-0x88(%rbp)
   0x000000000040070e <+46>:    test   %al,%al
   0x0000000000400710 <+48>:    je     0x400732 <Step::show(char const*,
...)+82>
   0x0000000000400712 <+50>:    movaps %xmm0,-0x80(%rbp)
   0x0000000000400716 <+54>:    movaps %xmm1,-0x70(%rbp)
   0x000000000040071a <+58>:    movaps %xmm2,-0x60(%rbp)
   0x000000000040071e <+62>:    movaps %xmm3,-0x50(%rbp)
   0x0000000000400722 <+66>:    movaps %xmm4,-0x40(%rbp)
   0x0000000000400726 <+70>:    movaps %xmm5,-0x30(%rbp)
   0x000000000040072a <+74>:    movaps %xmm6,-0x20(%rbp)
   0x000000000040072e <+78>:    movaps %xmm7,-0x10(%rbp)
   0x0000000000400732 <+82>:    mov    %rdi,-0xd8(%rbp)
   0x0000000000400739 <+89>:    movl   $0x8,-0xc8(%rbp)
   0x0000000000400743 <+99>:    movl   $0x30,-0xc4(%rbp)
   0x000000000040074d <+109>:   lea    0x10(%rbp),%rax
   0x0000000000400751 <+113>:   mov    %rax,-0xc0(%rbp)
   0x0000000000400758 <+120>:   lea    -0xb0(%rbp),%rax
   0x000000000040075f <+127>:   mov    %rax,-0xb8(%rbp)
   0x0000000000400766 <+134>:   lea    -0xc8(%rbp),%rdx
   0x000000000040076d <+141>:   mov    -0xd8(%rbp),%rax
   0x0000000000400774 <+148>:   mov    %rdx,%rsi
   0x0000000000400777 <+151>:   mov    %rax,%rdi
   0x000000000040077a <+154>:   callq  0x40078c <Step::vlog(char const*,
__va_list_tag*)>
   0x000000000040077f <+159>:   jmp    0x400789 <Step::show(char const*,
...)+169>
   0x0000000000400781 <+161>:   mov    %rax,%rdi
   0x0000000000400784 <+164>:   callq  0x400590 <_Unwind_Resume@plt>
   0x0000000000400789 <+169>:   leaveq 
   0x000000000040078a <+170>:   retq   
End of assembler dump.
(gdb)

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug breakpoints/16280] Step into an inline function with varargs doesn't stop

[palves at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=16280

--- Comment #5 from Pedro Alves <palves at redhat dot com> ---
And above we can see why.  0x0000000000400712 is just after the je at
0x0000000000400710.

(gdb) info line Step::show
Line 15 of "Step.h" starts at address 0x4006e0 <Step::show(char const*, ...)>
and ends at 0x400712 <Step::show(char const*, ...)+50>.


I'm almost certain I've seen a gcc bug open for this before.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug breakpoints/16280] Step into an inline function with varargs doesn't stop

[palves at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=16280

--- Comment #6 from Pedro Alves <palves at redhat dot com> ---
> I'm almost certain I've seen a gcc bug open for this before.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54533

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug breakpoints/16280] Step into an inline function with varargs doesn't stop

[palves at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=16280

--- Comment #7 from Pedro Alves <palves at redhat dot com> ---
BTW, I don't see an issue with -m32.  Please confirm you're seeing this on
x86_64.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug breakpoints/16280] Step into an inline function with varargs doesn't stop

[palves at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=16280

--- Comment #8 from Pedro Alves <palves at redhat dot com> ---
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47471

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55586

https://sourceware.org/bugzilla/show_bug.cgi?id=12435

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug breakpoints/16280] Step into an inline function with varargs doesn't stop

[psmith at gnu dot org]

https://sourceware.org/bugzilla/show_bug.cgi?id=16280

--- Comment #9 from psmith at gnu dot org ---
Yes, x86_64.  Sorry for omitting that information.

Thanks Pedro; I'll look at those bugs.

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug breakpoints/16280] Step into an inline function with varargs doesn't stop

[gbenson at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=16280

Gary Benson <gbenson at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |gbenson at redhat dot com

-- 
You are receiving this mail because:
You are on the CC list for the bug.