[RFC]first set of 3 patches to build GDB with xlc on AIX 5.3

[RFC]first set of 3 patches to build GDB with xlc on AIX 5.3

[Manoj Iyer]

Here are 3 patches that I used to (partially) build GDB on AIX 5.3 using
the XLC compiler. Please review and approve for commit. Is there a
maintainer for GDB on AIX? I would love to volunteer as one in the absence
of a full time maintainer for GDB for AIX. Ok, here are is the first set
of patches, I will post a few more tomorrow...

Patch #1
--------
This patch fixes what appears to be a typo in the hearder file, where for
ALPHA unsigned32 and unsigned64 gets multiply defined in sim-types.h. This
is caught by the xlc compiler, and it exits with an error saying that
these vars are already defined.

--- ./old/src/sim/common/sim-types.h	2002-11-22 19:12:05.000000000 -0600
+++ ./new/src/sim/common/sim-types.h	2005-05-31 05:46:01.000000000 -0500
@@ -112,11 +112,11 @@ typedef struct { signed64 a[2]; } signed
 typedef signed char signed8;
 typedef signed short signed16;
 #if defined (__ALPHA__)
-typedef signed int unsigned32;
-typedef signed long unsigned64;
+typedef signed int signed32;
+typedef signed long signed64;
 #else
-typedef signed long unsigned32;
-typedef signed long long unsigned64;
+typedef signed long signed32;
+typedef signed long long signed64;
 #endif

 typedef unsigned char unsigned8;

Patch #2
--------
When compiling/linking tmp-ld-insn, the linker exits with the error
message, cc -g -O -I. -I. -I./../../include -I../../bfd -I./../../bfd
-I../../gdb -I./../../gdb  -I./../../gdb/config   -o tmp-ld-insn -DMAIN
./ld-insn.c ld-cache.o misc.o lf.o table.o ld-decode.o filter_host.o
filter.o
cc: 1501-228 input file ld-cache.o not found
cc: 1501-228 input file ld-decode.o not found
Rearranging the build order fixes this build break. Here is the patch.

diff -Naurp ./old/src/sim/ppc/Makefile.in new/src/sim/ppc/Makefile.in
--- ./old/src/sim/ppc/Makefile.in	2005-01-03 09:34:19.000000000 -0600
+++ new/src/sim/ppc/Makefile.in	2005-05-31 05:14:10.000000000 -0500
@@ -683,7 +683,7 @@ tmp-dgen: dgen ppc-spr-table $(srcdir)/.
 	$(SHELL) $(srcdir)/../../move-if-change tmp-spreg.c spreg.c
 	touch tmp-dgen

-tmp-igen: igen $(srcdir)/ppc-instructions $(srcdir)/altivec.igen $(srcdir)/e500.igen $(IGEN_OPCODE_RULES) $(srcdir)/../../move-if-change tmp-ld-decode tmp-ld-cache tmp-ld-insn tmp-filter
+tmp-igen: igen $(srcdir)/ppc-instructions $(srcdir)/altivec.igen $(srcdir)/e500.igen $(IGEN_OPCODE_RULES) $(srcdir)/../../move-if-change tmp-ld-insn tmp-ld-decode tmp-ld-cache tmp-filter
 	./igen	$(IGEN_FLAGS) \
 		-o $(srcdir)/$(IGEN_OPCODE_RULES) \
 		-I $(srcdir) -i $(srcdir)/ppc-instructions \

Patch #3
--------
The psim.c file is missing a header file libiberty.h, this causes the
linker to report an unresolved symbol namely ".alloca".
cc -g   -o psim main.o libsim.a ../../bfd/libbfd.a
../../libiberty/libiberty.a
ld: 0711-317 ERROR: Undefined symbol: .alloca
ld: 0711-345 Use the -bloadmap or -bnoquiet option to obtain more
information.

The patch below adds the headerfile to psim.c file.

--- old/src/sim/ppc/psim.c	2003-06-22 11:48:12.000000000 -0500
+++ new/src/sim/ppc/psim.c	2005-05-31 08:31:45.000000000 -0500
@@ -49,6 +49,7 @@


 #include "bfd.h"
+#include "libiberty.h"


 /* system structure, actual size of processor array determined at


=========== END PATCHES =====================

Please review, ok to commit???

-----
manjo
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+ Cogito ergo sum +
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

Re: [RFC]first set of 3 patches to build GDB with xlc on AIX 5.3

[Daniel Jacobowitz]

On Wed, Apr 13, 2005 at 08:57:55PM -0500, Manoj Iyer wrote:
> 
> Here are 3 patches that I used to (partially) build GDB on AIX 5.3 using
> the XLC compiler. Please review and approve for commit. Is there a
> maintainer for GDB on AIX? I would love to volunteer as one in the absence
> of a full time maintainer for GDB for AIX. Ok, here are is the first set
> of patches, I will post a few more tomorrow...

Please, post patches one at a time with changelogs if you want them to
be reviewed.  #1 and #3 look plausible.  #2 is definitely wrong; there
are missing dependencies somewhere.  The order of things in a
dependency line is not significant to make except accidentally.


-- 
Daniel Jacobowitz
CodeSourcery, LLC