PATCH: Fix gdb compilation on Tru64 UNIX

PATCH: Fix gdb compilation on Tru64 UNIX

[Rainer Orth]

Current gdb mainline doesn't even compile on Tru64 UNIX V5.1B with gcc
4.4.2:

cc1: warnings being treated as errors
/vol/src/gnu/gdb/gdb/gdb/solib-osf.c: In function 'osf_current_sos':
/vol/src/gnu/gdb/gdb/gdb/solib-osf.c:539: error: 'tail' may be used uninitialized in this function
make[2]: *** [solib-osf.o] Error 1

This can be easily fixed by initializing tail to NULL.

cc1: warnings being treated as errors
/vol/src/gnu/gdb/gdb/gdb/tui/tui-io.c: In function 'tui_getc':
/vol/src/gnu/gdb/gdb/gdb/tui/tui-io.c:683: error: implicit declaration of function 'napms'
make[2]: *** [tui-io.o] Error 1

napms() is declared only if _XOPEN_SOURCE_EXTENDED is defined, which
requires _XOPEN_SOURCE >= 420.

The following patch fixes both issues and allows the compilation to
conclude.

Unfortunately, I still cannot debug gnat1 generated from GCC mainline:

Reading symbols from /vol/gcc/obj/gcc-4.5.0-20100111/5.1b-gcc/gcc/gnat1...Error reading symbol table: Memory exhausted

I'm investigating this separately.

Ok for mainline?

	Rainer

-- 
-----------------------------------------------------------------------------
Rainer Orth, Center for Biotechnology, Bielefeld University


2010-01-18  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>

	* gdb_curses.h [__osf__] (_XOPEN_SOURCE, _XOPEN_SOURCE_EXTENDED):
	Define.

	* solib-osf.c (osf_current_sos): Initialize tail.

Index: solib-osf.c
===================================================================
RCS file: /cvs/src/src/gdb/solib-osf.c,v
retrieving revision 1.28
diff -u -p -r1.28 solib-osf.c
--- solib-osf.c	8 Jan 2010 22:52:03 -0000	1.28
+++ solib-osf.c	18 Jan 2010 16:33:46 -0000
@@ -536,7 +536,7 @@ close_map (struct read_map_ctxt *ctxt)
 static struct so_list *
 osf_current_sos (void)
 {
-  struct so_list *head = NULL, *tail, *newtail, so;
+  struct so_list *head = NULL, *tail = NULL, *newtail, so;
   struct read_map_ctxt ctxt;
   int skipped_main;
 
Index: gdb_curses.h
===================================================================
RCS file: /cvs/src/src/gdb/gdb_curses.h,v
retrieving revision 1.12
diff -u -p -r1.12 gdb_curses.h
--- gdb_curses.h	1 Jan 2010 07:31:32 -0000	1.12
+++ gdb_curses.h	18 Jan 2010 16:33:46 -0000
@@ -21,6 +21,14 @@
 #ifndef GDB_CURSES_H
 #define GDB_CURSES_H 1
 
+/* On Tru64 UNIX, the napms() prototype is only visible with
+   _XOPEN_SOURCE_EXTENDED.  */
+#ifdef __osf__
+#undef _XOPEN_SOURCE
+#define _XOPEN_SOURCE 500
+#define _XOPEN_SOURCE_EXTENDED
+#endif
+
 #if defined (HAVE_NCURSES_NCURSES_H)
 #include <ncurses/ncurses.h>
 #elif defined (HAVE_NCURSES_H)

Re: PATCH: Fix gdb compilation on Tru64 UNIX

[Tom Tromey]

>>>>> "Rainer" == Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE> writes:

Rainer> napms() is declared only if _XOPEN_SOURCE_EXTENDED is defined, which
Rainer> requires _XOPEN_SOURCE >= 420.

Rainer> 2010-01-18  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
Rainer> 	* gdb_curses.h [__osf__] (_XOPEN_SOURCE, _XOPEN_SOURCE_EXTENDED)
Rainer> 	Define.
Rainer> 	* solib-osf.c (osf_current_sos): Initialize tail.

The solib-osf.c change is ok.

I am not as sure about the gdb_curses.h change.  I don't think we have
many host tests like that.  Perhaps some kind of configury would be
better.

Tom

Re: PATCH: Fix gdb compilation on Tru64 UNIX

[Rainer Orth]

Tom Tromey <tromey@redhat.com> writes:

>>>>>> "Rainer" == Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE> writes:
>
> Rainer> napms() is declared only if _XOPEN_SOURCE_EXTENDED is defined, which
> Rainer> requires _XOPEN_SOURCE >= 420.
>
> Rainer> 2010-01-18  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
> Rainer> 	* gdb_curses.h [__osf__] (_XOPEN_SOURCE, _XOPEN_SOURCE_EXTENDED)
> Rainer> 	Define.
> Rainer> 	* solib-osf.c (osf_current_sos): Initialize tail.
>
> The solib-osf.c change is ok.
>
> I am not as sure about the gdb_curses.h change.  I don't think we have
> many host tests like that.  Perhaps some kind of configury would be
> better.

True.  While researching where/how best to handle this in configure.ac,
I found that TUI wasn't supposed to be built on Tru64 UNIX at all.
Unfortunately, the test is completely broken and can never have worked:

* The test checks $host_os, but tries to match it against the full
  target triplet, which fails since host_os is only the third component.

* Both if conditions failed to invoke the test command, but instead
  tried to execute the value of $enable_tui as a command.

After fixing those issues, TUI is disabled on alpha-dec-osf5.1b (tested
by removing the gdb subdir and running make again) and gdb builds
without problems.

Ok?
	Rainer

-- 
-----------------------------------------------------------------------------
Rainer Orth, Center for Biotechnology, Bielefeld University


2010-01-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>

	* configure.ac: Only use host_os part when disabling TUI on osf.
	Use test to check variables, prefix strings with x.
	* configure: Regenerate.

Index: gdb/configure.ac
===================================================================
RCS file: /cvs/src/src/gdb/configure.ac,v
retrieving revision 1.112
diff -u -p -r1.112 configure.ac
--- gdb/configure.ac	15 Jan 2010 00:34:37 -0000	1.112
+++ gdb/configure.ac	21 Jan 2010 22:44:01 -0000
@@ -447,11 +447,11 @@ AM_ICONV
 # broken on alpha-osf.
 
 case $host_os in
-  alpha*-*-osf* )
-    if "$enable_tui" = "yes"; then
+  osf* )
+    if test x"$enable_tui" = xyes; then
       AC_MSG_ERROR([Building GDB with TUI mode is not supported on this host])
     fi
-    if "$enable_tui" = "auto"; then
+    if test x"$enable_tui" = xauto; then
       enable_tui=no
     fi
     ;;

Re: PATCH: Fix gdb compilation on Tru64 UNIX

[Tom Tromey]

>>>>> "Rainer" == Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE> writes:

Rainer> 2010-01-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
Rainer> 	* configure.ac: Only use host_os part when disabling TUI on osf.
Rainer> 	Use test to check variables, prefix strings with x.
Rainer> 	* configure: Regenerate.

This is ok.  Thanks for finding this.

Tom