This is the mail archive of the gdb@sourceware.org mailing list for the GDB project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

hook-stop sees dummy frames, bug or feature?


I was looking at fixing PR9664 <http://sourceware.org/bugzilla/show_bug.cgi?id=9664>,
which was a regression I introduced here:

 http://sourceware.org/ml/gdb-patches/2008-09/msg00193.html

The regression is about the fact that the hook-stop in normal_stop
is now run after frame printing, while it is documented as being run before.

 "(@samp{hook-stop}) makes the associated commands execute every time
 execution stops in your program: before breakpoint commands are run,
 displays are printed, or the stack frame is printed."

The current sequence is roughtly:

normal_stop ()
{
 ...
 - print frame
 - save `return' or infcall registers
 - pop dummy frame
 - hook-stop
 ...
}

Clearly, I need to put running the hook-stop higher up again.

In gdb <= 6.8, the hook-stop was run before we poped off the dummy
frame:

normal_stop ()
{
 ...
 - hook-stop
 - print frame
 - save `return' or infcall registers
 - pop dummy frame
 ...
}

 The case I'm considering is this:

(basically:
 define hook-stop
  frame
 end
 p func()
)

 >gdb-6.8 ./gdb
 (top-gdb) start
 Starting program: /home/pedro/gdb/normal_stop/build/gdb/gdb
 main (argc=1, argv=0x7ffff00043f8) at ../../src/gdb/gdb.c:28
 28        memset (&args, 0, sizeof args);
 (top-gdb) define hook-stop
 Type commands for definition of "hook-stop".
 End with a line saying just "end".
 >frame
 >end
 (top-gdb) p malloc (0)
 #0  <function called from gdb>          <<<<<< "frame" in hook-stop
 Current language:  auto; currently asm
 $1 = 11407376                           <<<<<< 'malloc (0)' result
 Current language:  auto; currently c
 (top-gdb) frame                         <<<<<< "frame" again shows something else.
 #0  main (argc=1, argv=0x7ffff00043f8) at ../../src/gdb/gdb.c:28
 28        memset (&args, 0, sizeof args);

Is the fact that the dummy frame is visible in a hook-stop a
bug, or a feature?

This is the current mainline behaviour:

 (top-gdb) p malloc (0)
 #0  main (argc=1, argv=0x7fffffffe408) at ../../src/gdb/gdb.c:28   <<< "frame" in hook-stop.
 28        memset (&args, 0, sizeof args);
 $1 = 11407440
 (top-gdb)

I'm inclined to consider it a bug to show the dummy frame, but I may be
missing some use cases.

Say, change normal_stop to:

 normal_stop ()
 {
  ...
  - pop dummy frame
  - hook-stop
  - print frame
  - save `return' or infcall registers
  ...
 }

(saving infcall registers could also move up, I think, but I'll leave
that out for now)

What do you guys think?

-- 
Pedro Alves


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]