[patch] objfile_relocate: Call breakpoint_re_set on-demand

[patch] objfile_relocate: Call breakpoint_re_set on-demand

[Jan Kratochvil]

Hi,

this is just a performance optimization to avoid some needless calls.

I created the patch as a safety against regressions on non-PIE (normal)
executables by the PIE patch if breakpoint_re_set would hurt some time.

But it is true there can be a risk of regression for current (rare) users of
objfile_relocate (IMO incorrectly) expecting breakpoint_re_set gets called.
Briefly checked the callers without seeing such expectations.

No regressions on {x86_64,x86_64-m32,i686}-fedora12-linux-gnu.


Thanks,
Jan


2010-01-11  Jan Kratochvil  <jan.kratochvil@redhat.com>

	* objfiles.c (objfile_relocate1): Change the return type to int.
	Describe the new return value.  Return non-zero if data changed.
	(objfile_relocate): New variable changed.  Set it.  Call
	breakpoint_re_set depending on CHANGED.

--- a/gdb/objfiles.c
+++ b/gdb/objfiles.c
@@ -693,9 +693,10 @@ free_all_objfiles (void)
 }
 \f
 /* Relocate OBJFILE to NEW_OFFSETS.  There should be OBJFILE->NUM_SECTIONS
-   entries in new_offsets.  SEPARATE_DEBUG_OBJFILE is not touched here.  */
+   entries in new_offsets.  SEPARATE_DEBUG_OBJFILE is not touched here.
+   Return non-zero iff any change happened.  */
 
-static void
+static int
 objfile_relocate1 (struct objfile *objfile, struct section_offsets *new_offsets)
 {
   struct obj_section *s;
@@ -714,7 +715,7 @@ objfile_relocate1 (struct objfile *objfile, struct section_offsets *new_offsets)
 	  something_changed = 1;
       }
     if (!something_changed)
-      return;
+      return 0;
   }
 
   /* OK, get all the symtabs.  */
@@ -850,6 +851,9 @@ objfile_relocate1 (struct objfile *objfile, struct section_offsets *new_offsets)
       exec_set_section_address (bfd_get_filename (objfile->obfd), idx,
 				obj_section_addr (s));
     }
+
+  /* Data changed.  */
+  return 1;
 }
 
 /* Relocate OBJFILE to NEW_OFFSETS.  There should be OBJFILE->NUM_SECTIONS
@@ -865,8 +869,9 @@ void
 objfile_relocate (struct objfile *objfile, struct section_offsets *new_offsets)
 {
   struct objfile *debug_objfile;
+  int changed = 0;
 
-  objfile_relocate1 (objfile, new_offsets);
+  changed |= objfile_relocate1 (objfile, new_offsets);
 
   for (debug_objfile = objfile->separate_debug_objfile;
        debug_objfile;
@@ -894,13 +899,14 @@ objfile_relocate (struct objfile *objfile, struct section_offsets *new_offsets)
 					     debug_objfile->num_sections,
 					     objfile_addrs);
 
-      objfile_relocate1 (debug_objfile, new_debug_offsets);
+      changed |= objfile_relocate1 (debug_objfile, new_debug_offsets);
 
       do_cleanups (my_cleanups);
     }
 
   /* Relocate breakpoints as necessary, after things are relocated. */
-  breakpoint_re_set ();
+  if (changed)
+    breakpoint_re_set ();
 }
 \f
 /* Return non-zero if OBJFILE has partial symbols.  */

Re: [patch] objfile_relocate: Call breakpoint_re_set on-demand

[Tom Tromey]

>>>>> "Jan" == Jan Kratochvil <jan.kratochvil@redhat.com> writes:

Jan> this is just a performance optimization to avoid some needless calls.

It seems reasonable enough to me.

Jan> 2010-01-11  Jan Kratochvil  <jan.kratochvil@redhat.com>
Jan> 	* objfiles.c (objfile_relocate1): Change the return type to int.
Jan> 	Describe the new return value.  Return non-zero if data changed.
Jan> 	(objfile_relocate): New variable changed.  Set it.  Call
Jan> 	breakpoint_re_set depending on CHANGED.

Ok.

Tom

Re: [patch] objfile_relocate: Call breakpoint_re_set on-demand

[Jan Kratochvil]

On Mon, 11 Jan 2010 18:58:20 +0100, Tom Tromey wrote:
> >>>>> "Jan" == Jan Kratochvil <jan.kratochvil@redhat.com> writes:
> Jan> 2010-01-11  Jan Kratochvil  <jan.kratochvil@redhat.com>
> Jan> 	* objfiles.c (objfile_relocate1): Change the return type to int.
> Jan> 	Describe the new return value.  Return non-zero if data changed.
> Jan> 	(objfile_relocate): New variable changed.  Set it.  Call
> Jan> 	breakpoint_re_set depending on CHANGED.
> 
> Ok.

Checked-in:
	http://sourceware.org/ml/gdb-cvs/2010-01/msg00095.html


Thanks,
Jan