Re: [RFC] Entry point update with "run" command

[Jan Kratochvil <jan.kratochvil@redhat.com>]

[-- Attachment #1: Type: text/plain, Size: 429 bytes --]

On Thu, 24 Apr 2008 17:58:14 +0200, Luis Machado wrote:
> Hi folks,
> 
> It seems GDB doesn't really update the entry point of an object file if
> we "reload" the modified binary through the "run" command.

There is a problem whole EXEC_BFD becomes stale - using the more general
attached patch.

Still I agree it is a hack as besides whole reload-on-change is a hack also GDB
still misses some real OO framework.


Regards,
Jan

[-- Attachment #2: gdb-6.7-reread-exec_bfd.patch --]
[-- Type: text/plain, Size: 1799 bytes --]

2007-10-29  Jan Kratochvil  <jan.kratochvil@redhat.com>

	* symfile.c (reread_symbols): Reread also EXEC_BFD if changed.

2008-04-11  Jan Kratochvil  <jan.kratochvil@redhat.com>

	* symfile.c (reread_symbols): Reload whole EXEC_BFD by the EXEC module
	as its in-place patching did cause regressions.

Testcase: Regressed by the gdb-6.7 version of `gdb-6.3-pie-20050110.patch':
 Running ../../../gdb/testsuite/gdb.base/reread.exp ...
 PASS: gdb.base/reread.exp: breakpoint foo in first file (PRMS 13484)
 PASS: gdb.base/reread.exp: run to foo() (PRMS 13484)
-PASS: gdb.base/reread.exp: run to foo() second time
+FAIL: gdb.base/reread.exp: run to foo() second time
 PASS: gdb.base/reread.exp: second pass: breakpoint foo in first file
-PASS: gdb.base/reread.exp: second pass: run to foo()
-PASS: gdb.base/reread.exp: second pass: continue to completion
-PASS: gdb.base/reread.exp: second pass: run to foo() second time
+FAIL: gdb.base/reread.exp: second pass: run to foo()
+FAIL: gdb.base/reread.exp: second pass: continue to completion
+FAIL: gdb.base/reread.exp: second pass: run to foo() second time

--- gdb-6.7-orig/gdb/symfile.c	2007-10-29 01:04:35.000000000 +0100
+++ gdb-6.7-patched/gdb/symfile.c	2007-10-29 01:03:13.000000000 +0100
@@ -2810,6 +2810,12 @@ reread_symbols (void)
 	      /* We need to do this whenever any symbols go away.  */
 	      make_cleanup (clear_symtab_users_cleanup, 0 /*ignore*/);
 
+	      if (exec_bfd != NULL && strcmp (bfd_get_filename (objfile->obfd),
+					      bfd_get_filename (exec_bfd)) == 0)
+		{
+		  exec_ops.to_open (bfd_get_filename (objfile->obfd), 0);
+		}
+
 	      /* Clean up any state BFD has sitting around.  We don't need
 	         to close the descriptor but BFD lacks a way of closing the
 	         BFD without closing the descriptor.  */