This is the mail archive of the
gdb-patches@sourceware.cygnus.com
mailing list for the GDB project.
Re: RFA: bfd/rs6000-core.c 64-bit support
- To: Nick Duffek <nsd at redhat dot com>, gdb-patches at sourceware dot cygnus dot com
- Subject: Re: RFA: bfd/rs6000-core.c 64-bit support
- From: Kevin Buettner <kevinb at cygnus dot com>
- Date: Tue, 13 Jun 2000 12:47:25 -0700
- References: <200006131706.NAA19532@nog.bosbc.com>
On Jun 13, 1:06pm, Nick Duffek wrote:
> I sent this for approval to binutils, whence I was redirected here on the
> theory that rs6000-core.c is really owned by GDB.
>
> The appended patch adds 64-bit support to bfd/rs6000-core.c. In addition,
> it adds support for pre-4.3 core dumps on AIX 4.3+, which can generate
> such dumps with the appropriate SMIT setting.
>
> It's one of several BFD patches that accompany the AIX mega-patch I
> published here a few days ago.
>
> Okay to apply?
Nick,
Have you tried your patch on a machine where CORE_VERSION_1 is still
defined?
I am not certain, but it appears to me that you've removed some code
which'll allow rs6000-core.c to build on these older versions of AIX
(pre 4.1?).
E.g, look at the following:
- (bfd_size_type) coredata.CORE_DATA_SIZE_FIELD,
+ (bfd_size_type) core.old.c_u.u_dsize,
As I understand it, CORE_DATA_SIZE_FIELD used to be defined to be
either c_u.U_dsize or c_u.u_dsize depending upon whether CORE_VERSION_1
was defined or not. Now, it seems that you expect that c_u.u_dsize
will work on the CORE_VERSION_1 boxes.
I notice, however, that you've been careful to handle some of the
other fields with similar issues. So either the above was an oversight
or else u_dsize really worked on the CORE_VERSION_1 boxes. In any event,
I'd feel better about the whole patch if you could do a build and make
sure core file support still works on one of the older (pre 4.1) AIX
boxes.
Kevin