[PATCH] bfd/rs6000-core.c

[PATCH] bfd/rs6000-core.c

[David Edelsohn]

	The new AIX 4.3+ core support assumes a field in the core_dumpx
structure defined in /usr/include/sys/core.h that only exists in AIX 4.3.3
and not earlier releases.  This prevents binutils from building on AIX
4.3.2.  Albert and I have worked out the appended patch.

David


	* rs6000-core.c: Use autoconf variable HAVE_ST_C_IMPL to guard use
	of c_impl field.
	* configure.in: Auto-configure HAVE_ST_C_IMPL.
	* configure.in: Likewise.

--- bfd/configure.in.orig	Fri Jun 22 13:09:23 2001
+++ bfd/configure.in	Fri Jun 22 13:18:11 2001
@@ -301,6 +301,15 @@
 changequote([,])dnl
         COREFILE=rs6000-core.lo
 	COREFLAG="$COREFLAG -DAIX_CORE_DUMPX_CORE"
+
+	# Not all versions of AIX with -DAIX_CORE_DUMPX_CORE
+	# have c_impl as a member of struct core_dumpx
+	AC_MSG_CHECKING([for c_impl in struct core_dumpx])
+	AC_TRY_COMPILE([#include <core.h>],
+	  [struct core_dumpx c; c.c_impl = 0;],
+	  [AC_DEFINE(HAVE_ST_C_IMPL, 1,
+	    [Define if struct core_dumpx has member c_impl])
+	  AC_MSG_RESULT(yes)],[AC_MSG_RESULT(no)])
 	;;
   rs6000-*-aix4*)	COREFILE=rs6000-core.lo ;;
   rs6000-*-*)		COREFILE=rs6000-core.lo ;;
--- bfd/config.in.orig	Fri Jun 22 13:17:50 2001
+++ bfd/config.in	Fri Jun 22 13:18:31 2001
@@ -208,6 +208,9 @@
 /* Define if getenv is not declared in system header files. */
 #undef NEED_DECLARATION_GETENV
 
+/* Define if struct core_dumpx has member c_impl */
+#undef HAVE_ST_C_IMPL
+
 /* Define if <sys/procfs.h> has prstatus_t. */
 #undef HAVE_PRSTATUS_T
 
--- bfd/rs6000-core.c.orig	Fri Jun 22 13:11:09 2001
+++ bfd/rs6000-core.c	Fri Jun 22 13:11:26 2001
@@ -182,7 +182,7 @@
 
 /* Return the c_impl field from struct core_dumpx C.  */
 
-#ifdef AIX_CORE_DUMPX_CORE
+#ifdef HAVE_ST_C_IMPL
 # define CNEW_IMPL(c)	(c).c_impl
 #else
 # define CNEW_IMPL(c)	0

Re: [PATCH] bfd/rs6000-core.c

[Nick Duffek]

On 24-Jun-2001, David Edelsohn wrote:

>	* rs6000-core.c: Use autoconf variable HAVE_ST_C_IMPL to guard use
>	of c_impl field.
>	* configure.in: Auto-configure HAVE_ST_C_IMPL.
>	* configure.in: Likewise.

The patch looks good to me.  Approved, unless Kevin has any objections.

I suggest also getting the binutils mailing list to rubber-stamp it before
committing it.  There was talk about officially giving GDB ownership of
GDB-only files in bfd, but I don't think that's happened.

Nick

Re: [PATCH] bfd/rs6000-core.c

[Kevin Buettner]

On Jun 25,  2:20pm, Nick Duffek wrote:

> On 24-Jun-2001, David Edelsohn wrote:
> 
> >	* rs6000-core.c: Use autoconf variable HAVE_ST_C_IMPL to guard use
> >	of c_impl field.
> >	* configure.in: Auto-configure HAVE_ST_C_IMPL.
> >	* configure.in: Likewise.
> 
> The patch looks good to me.  Approved, unless Kevin has any objections.

I've just looked it over and it looks good to me too.

> I suggest also getting the binutils mailing list to rubber-stamp it before
> committing it.  There was talk about officially giving GDB ownership of
> GDB-only files in bfd, but I don't think that's happened.

I agree that it ought to be sent to the binutils list prior to committing
it.

Kevin

Re: [PATCH] bfd/rs6000-core.c

[David Edelsohn]

>>>>> Kevin Buettner writes:

>> I suggest also getting the binutils mailing list to rubber-stamp it before
>> committing it.  There was talk about officially giving GDB ownership of
>> GDB-only files in bfd, but I don't think that's happened.

Kevin> I agree that it ought to be sent to the binutils list prior to committing
Kevin> it.

	Note that the binutils people said that the earlier patch should
be approved on the gdb-patches mailinglist.  The two groups need to make
up their mind.

David