[demo-merge] [patch] Regression fix for vla [Re: Branch mergingfor demo]

[Jan Kratochvil <jan dot kratochvil at redhat dot com>]

On Thu, 18 Dec 2008 17:16:47 +0100, Rick Moseley wrote:
> OK guys, here is the branches I have merged so far using the  
> "archer-tromey-python" as the base branch:
[...]
> archer-jkratochvil-vla branch

The merge had a regression against the archer-jkratochvil-vla branch:
FAIL: gdb.fortran/dynamic.exp: p varx unallocated
FAIL: gdb.fortran/dynamic.exp: p varv unassociated
FAIL: gdb.fortran/dynamic.exp: p varv deassociated
FAIL: gdb.fortran/dynamic.exp: p varx deallocated
FAIL: gdb.fortran/dynamic.exp: p vary
FAIL: gdb.fortran/dynamic.exp: p varw
FAIL: gdb.fortran/dynamic.exp: p varw filled

Attached a fix, please check it in incl. the demo machine update.

Regression-tested archer-rmoseley-demo-merge on x86_64-unknown-linux-gnu.
While it is a change in a generic code I doubt it can affect the other merged
branches in any way.


Thanks,
Jan

Author: Jan Kratochvil <jan.kratochvil@redhat.com>
Date:   Sun Jan 18 18:17:37 2009 +0100

    Fix a archer-jankratochvil-vla -> archer-rmoseley-demo-merge merge regression.
    
    I do not find any conflict with the other merged patches at that place.

diff --git a/gdb/valops.c b/gdb/valops.c
index c3935ef..5c2af5e 100644
--- a/gdb/valops.c
+++ b/gdb/valops.c
@@ -677,10 +677,8 @@ value_fetch_lazy (struct value *val)
 {
   if (VALUE_LVAL (val) == lval_memory)
     {
-      CORE_ADDR addr = value_address (val);
-      int length = TYPE_LENGTH (check_typedef (value_enclosing_type (val)));
+      CORE_ADDR addr = VALUE_ADDRESS (val);
 
-      addr = VALUE_ADDRESS (val);
       if (object_address_get_data (value_type (val), &addr))
 	{
 	  struct type *type = value_enclosing_type (val);