FYI: fix small regression with gcc trunk

FYI: fix small regression with gcc trunk

[Tom Tromey]

I'm checking this in.

I recently ran the gdb test suite against GCC svn trunk.  This revealed
a number of regressions, mostly little test case bugs.

This patch fixes one regression that appears in gdb itself.  GCC changed
how it emits TLS references, to fix a bug.  This caused "info address"
output to change, causing a gdb test suite regression.

Built and regtested on x86-64 (compile farm).
I also tested the appropriate test case locally with both the new and
old GCC.

Tom

2010-07-26  Tom Tromey  <tromey@redhat.com>

	* dwarf2loc.c (locexpr_describe_location_piece): Also recognize
	TLS with DW_OP_const4u or DW_OP_const8u.

Index: dwarf2loc.c
===================================================================
RCS file: /cvs/src/src/gdb/dwarf2loc.c,v
retrieving revision 1.95
diff -u -r1.95 dwarf2loc.c
--- dwarf2loc.c	13 Jul 2010 15:09:03 -0000	1.95
+++ dwarf2loc.c	26 Jul 2010 20:22:24 -0000
@@ -2016,15 +2016,19 @@
 
      DW_AT_location    : 10 byte block: 3 4 0 0 0 0 0 0 0 e0
                         (DW_OP_addr: 4; DW_OP_GNU_push_tls_address)
-     
+
      0x3 is the encoding for DW_OP_addr, which has an operand as long
      as the size of an address on the target machine (here is 8
-     bytes).  0xe0 is the encoding for DW_OP_GNU_push_tls_address.
-     The operand represents the offset at which the variable is within
-     the thread local storage.  */
+     bytes).  Note that more recent version of GCC emit DW_OP_const4u
+     or DW_OP_const8u, depending on address size, rather than
+     DW_OP_addr.  0xe0 is the encoding for
+     DW_OP_GNU_push_tls_address. The operand represents the offset at
+     which the variable is within the thread local storage.  */
 
   else if (data + 1 + addr_size < end
-	   && data[0] == DW_OP_addr
+	   && (data[0] == DW_OP_addr
+	       || (addr_size == 4 && data[0] == DW_OP_const4u)
+	       || (addr_size == 8 && data[0] == DW_OP_const8u))
 	   && data[1 + addr_size] == DW_OP_GNU_push_tls_address
 	   && piece_end_p (data + 2 + addr_size, end))
     {

Re: FYI: fix small regression with gcc trunk

[Jan Kratochvil]

On Mon, 26 Jul 2010 22:24:45 +0200, Tom Tromey wrote:
> I'm checking this in.

Thanks, would it be for 7.2?


Regards,
Jan

Re: FYI: fix small regression with gcc trunk

[Joel Brobecker]

> > I'm checking this in.
> 
> Thanks, would it be for 7.2?

This patch seems very reasonable for 7.2.

-- 
Joel

Re: FYI: fix small regression with gcc trunk

[Tom Tromey]

Jan> Thanks, would it be for 7.2?

Joel> This patch seems very reasonable for 7.2.

I will commit it to the branch soon.

Tom