[RFA] Simple patch for x86-64 (can go to 5.2?)

[RFA] Simple patch for x86-64 (can go to 5.2?)

[Michal Ludvig]

[-- Attachment #1: Type: text/plain, Size: 996 bytes --]

This patch is a really simple one without any side-effects. The problem 
is, that gcc for x86-64 by default has -fomit-frame-pointer switched on 
when optimalization is requested. People don't know it and are 
complaining that (diplomaticaly said) "gdb doesn't work" (depending on 
their attitude :-). The only thing this patch does is, that it prints a 
note, when x86_64_skip_prologue doesn't see "push %rbp" as the very 
first instruction of a given function. The note is printed only once. 
That's all. I'd like to see it in 5.2 to teach people that they should 
explicitly use -fno-omit-frame-pointer when compiling a code for 
debugging. Can I commit? I'll remove this note as soon as I fix dwarf2 
reader to handle even optimized code.

2002-04-26  Michal Ludvig  <mludvig@suse.cz>
      * x86-64-tdep.c (x86_64_skip_prologue): Print note when debugging
      code without frame pointers.

Michal Ludvig
-- 
* SuSE CR, s.r.o     * mludvig@suse.cz
* +420 2 9654 5373   * http://www.suse.cz

[-- Attachment #2: omitfp1.diff --]
[-- Type: text/plain, Size: 1861 bytes --]

Index: x86-64-tdep.c
===================================================================
RCS file: /cvs/src/src/gdb/x86-64-tdep.c,v
retrieving revision 1.13
diff -c -3 -p -r1.13 x86-64-tdep.c
*** x86-64-tdep.c	6 Apr 2002 00:02:50 -0000	1.13
--- x86-64-tdep.c	26 Apr 2002 14:06:54 -0000
*************** static struct register_info x86_64_regis
*** 116,121 ****
--- 116,124 ----
  int x86_64_num_regs = X86_64_NUM_REGS;
  int x86_64_num_gregs = X86_64_NUM_GREGS;
  
+ /* Did we already print a note about frame pointer?  */
+ int omit_fp_note_printed = 0;
+ 
  /* Number of bytes of storage in the actual machine representation for
     register REGNO.  */
  int
*************** x86_64_skip_prologue (CORE_ADDR pc)
*** 811,820 ****
  
    read_memory (pc, (char *) prolog_buf, PROLOG_BUFSIZE);
  
!   /* First check, whether pc points to pushq %rbp, movq %rsp,%rbp.  */
!   for (i = 0; i < PROLOG_BUFSIZE; i++)
!     if (prolog_expect[i] != prolog_buf[i])
        return pc;
  
    v_function = find_pc_function (pc);
    v_sal = find_pc_line (pc, 0);
--- 814,836 ----
  
    read_memory (pc, (char *) prolog_buf, PROLOG_BUFSIZE);
  
!   /* First check, whether pc points to pushq %rbp. If not, 
!    * print a recommendation to enable frame pointer.  */
!   if (prolog_expect[0] != prolog_buf[0])
!     {
!       if (!omit_fp_note_printed)
! 	{
! 	  printf_filtered
! 	    ("NOTE: This function doesn't seem to have a valid prologue.\n"
! 	     "      Try to add -fno-omit-frame-pointer tou your gcc's CFLAGS.\n");
! 	  omit_fp_note_printed++;
! 	}
        return pc;
+     }
+   /* Valid prolog continues with movq %rsp,%rbp.  */
+   for (i = 1; i < PROLOG_BUFSIZE; i++)
+     if (prolog_expect[i] != prolog_buf[i])
+       return pc + 1;		/* First instruction after pushq %rbp.  */
  
    v_function = find_pc_function (pc);
    v_sal = find_pc_line (pc, 0);

Re: [RFA] Simple patch for x86-64 (can go to 5.2?)

[Andrew Cagney]

> This patch is a really simple one without any side-effects. The problem is, that gcc for x86-64 by default has -fomit-frame-pointer switched on when optimalization is requested. People don't know it and are complaining that (diplomaticaly said) "gdb doesn't work" (depending on their attitude :-). The only thing this patch does is, that it prints a note, when x86_64_skip_prologue doesn't see "push %rbp" as the very first instruction of a given function. The note is printed only once. That's all. I'd like to see it in 5.2 to teach people that they should explicitly use -fno-omit-frame-pointer when compiling a code for debugging. Can I commit? I'll remove this note as soon as I fix dwarf2 reader to handle even optimized code.
> 
> 2002-04-26  Michal Ludvig  <mludvig@suse.cz>
>      * x86-64-tdep.c (x86_64_skip_prologue): Print note when debugging
>      code without frame pointers.

Sigh (you've figured out the one thing that will make me delay a release :-)

Yes, I've committed it (can you please create a bug report explaining 
the warning message).

Andrew