Re: GDB record patch 0.1.3.1 for GDB-6.8 release

[Tea <teawater at gmail dot com>]

On Wed, May 21, 2008 at 1:40 AM, Michael Snyder <msnyder@specifix.com> wrote:
> On Tue, 2008-05-20 at 11:55 +0800, Tea wrote:
>> Agree with Daniel.
>> I think current way is better than include Linux kernel header files.
>>
>> To compile GDB with Linux KERNEL? I think it will make compile GDB not
>> very easy. If the Linux Kernel change. The user need re-compile the
>> GDB?
>> Or add Linux Kernel files to GDB directory? It looks not very
>> professional. And maybe will cause some copyright problem.
>>
>> And this types size will not change very continually. The GLIBC use it
>> too. The kernel more like add new types than change the old types.
>
> There is another possibility.  You might build something
> resembling gdbserver, and put the record/playback functionality
> in there.  Have gdb talk to it by the remote serial protocol
> (over a socket, of course).
>
> There is an old branch in the cvs repository where we have
> stashed away a version of gdb that has some extra "reverse"
> syntax in the remote protocol.
>
> This would have the advantage of a clear separation layer,
> so that gdb would not have to be "poluted" by foreign header
> files with different copyrights and licensing, plus keeping
> a big chunk of complex code separate from gdb itself.
>
> Plus, you could now reverse-debug from one host to another,
> even across different architectures.
>
> Just an idea...
>
> Michael
>
>
>

I will try to make all the part of record to be the part of GDB.
BTW, I received the ASSIGNMENT - GNU GDB from FSF in Monday and I had
signed it and send it back.

Make simulator reverse-run with itself is a cool idea. It will make
speed higher than before.

Thanks,
teawater