This is the mail archive of the gdb-patches@sourceware.org mailing list for the GDB project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: AIX 64bit support


>>>>> ">" == swamy sangamesh <swamy.sangamesh@gmail.com> writes:

>> Below patches adds the support for 64-bit gdb for ppc64 running on
>> AIX, plus some bug fixes which are caused
>> by the way gcc and xlc generated binary are read. Please consider the
>> patches if its fine.

Thanks for writing this.

First, you will need to file copyright assignment paperwork with the FSF
before we can accept this patch.  Contact me off-list and I can get you
started on the process.

>> --- ./gdb/aix-thread.c_orig	2012-08-07 17:11:21.270057686 +0530
>> +++ ./gdb/aix-thread.c	2012-08-07 17:11:17.998084929 +0530
[...]
>>  static int
>> +#ifdef BFD64
>> +ptrace64aix (int req, long long id, long long addr, int data, int *buf)

I tend to doubt that BFD64 is the correct check here.
I think you can use --enable-64-bit-bfd even on a 32-bit machine.
However in that situation you would not want to use ptrace64 --
presumably it wouldn't even exist.

Perhaps checking directly for ptrace64 in configure is best.

I agree with the other commenters about the desirability of a wrapper
function, rather than repeated #ifdefs all over.

>> --- ./gdb/inf-ptrace.c_orig	2012-07-25 21:07:04.273078850 +0530
>> +++ ./gdb/inf-ptrace.c	2012-08-02 00:35:56.425443341 +0530

There are many changes here, but I think this code is shared by many
ports.  So, your changes have to take that into consideration.  A
wrapper function may be the way to go.

>> --- ./gdb/xcoffread.c_orig	2012-08-07 17:36:42.378057756 +0530
>> +++ ./gdb/xcoffread.c	2012-08-07 17:36:48.702060320 +0530

I don't know anything about xcoffread.  Could you say how you tested
this?  Did you test it on a 32-bit host as well?

>> --- ./gdb/symtab.c_orig	2012-08-07 17:52:15.181060405 +0530
>> +++ ./gdb/symtab.c	2012-08-07 17:53:04.653058722 +0530

There is a change here, but this is generic code, and I think shouldn't
have an xlc-specific change.  The problem has to be approached some
other way, preferably in the debuginfo reader.

The BFD changes should go to the binutils list.

Tom


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]