Working on the x86_64-windows native port...

Working on the x86_64-windows native port...

[Joel Brobecker]

[ChrisF, I noticed you prefer if we don't Cc you in addition to reaching
through the mailing-list. I work a little differently, so hopefully
you'll see this message]

Hello everyone,

Just a quick message to say that I started looking at porting GDB to
x86_64-windows. Currently, if I use a 32bit compiler and configure
GDB as a i686-pc-mingw32, I get a fine debugger.

Things get a little more interesting when I use a 64bit compiler
and configure GDB with x86_64-pc-mingw32 [1]. Basically, the root
of the issues I am seeing comes from the fact that win32-nat has
some ia32-specific code, in particular in terms of the registers
(see the CONTEXT structure for instance).

My intentions are to split off the 32bit-specific parts to
i386-windows-nat.c, and write the equivalent x86_64-windows-nat.c.
I hope that's OK with everyone (Chris?).

-- 
Joel

[1]: Currently, config.guess (IIRC) doesn't like mingw64. For now,
     I'll keep mingw32 as meaning either 32bit or 64bit. I'll probably
     fix this too, but later.

Re: Working on the x86_64-windows native port...

[Christopher Faylor]

On Thu, Nov 13, 2008 at 02:44:31PM -0800, Joel Brobecker wrote:
>My intentions are to split off the 32bit-specific parts to
>i386-windows-nat.c, and write the equivalent x86_64-windows-nat.c.
>I hope that's OK with everyone (Chris?).

I'd like to see the patches before they are applied but I have no
objections to this in principle.

cgf

Re: Working on the x86_64-windows native port...

[Joel Brobecker]

On Thu, Nov 13, 2008 at 05:59:43PM -0500, Christopher Faylor wrote:
> I'd like to see the patches before they are applied but I have no
> objections to this in principle.

Of course! I wouldn't touch a windows-related file without review.

On Fri, Nov 14, 2008 at 12:07:30AM +0100, Mark Kettenis wrote:
> Please call files amd64-xxx.c instead of x86_64-xxx.c.

oh yes, right. Will do. Thanks.

On Fri, Nov 14, 2008 at 10:13:49AM +0100, Kai Tietz wrote:
> This is really good. I begun with a port of gdb (now about a half a year 
> ago). If you like I can provide you my work on it.

Sure! I'm traveling again next week, and I have had a very busy schedule
so any help is very welcome. Thanks!

> The mingw64 should be an alias to x86_64-pc-mingw32, nothing more, nothing 
> less. There were several discussion on that on gcc, and the conclusion was 
> that the target triplet should be in future on gcc (i?86|x86_64)-pc-mingw.

Makes sense. I think I'll adjust GDB to accept mingw*.

Thanks to everyone who responded,
-- 
Joel

Re: Working on the x86_64-windows native port...

[Kai Tietz]

Hello Joel,

maybe you are interested in a experimental port of gdb, I did some
time ago on old code base (see
http://sourceforge.net/project/showfiles.php?group_id=202880&package_id=259447
for the source tar ball).

Cheers,
Kai

PS: My papers with FSF for gdb are already done,
-- 
|  (\_/) This is Bunny. Copy and paste
| (='.'=) Bunny into your signature to help
| (")_(") him gain world domination

Re: Working on the x86_64-windows native port...

[Joel Brobecker]

> maybe you are interested in a experimental port of gdb, I did some
> time ago on old code base (see
> http://sourceforge.net/project/showfiles.php?group_id=202880&package_id=259447
> for the source tar ball).

Sure! This may prove to be a convenient source of information when
I start working on the port (scheduled this week, if all goes well).

Thanks a lot :)

-- 
Joel

Re: Working on the x86_64-windows native port...

[Mark Kettenis]

> Date: Thu, 13 Nov 2008 14:44:31 -0800
> From: Joel Brobecker <brobecker@adacore.com>
> 
> My intentions are to split off the 32bit-specific parts to
> i386-windows-nat.c, and write the equivalent x86_64-windows-nat.c.
> I hope that's OK with everyone (Chris?).

Please call files amd64-xxx.c instead of x86_64-xxx.c.

Re: Working on the x86_64-windows native port...

[Kai Tietz]

Hi Joel,

> [ChrisF, I noticed you prefer if we don't Cc you in addition to reaching
> through the mailing-list. I work a little differently, so hopefully
> you'll see this message]
> 
> Hello everyone,
> 
> Just a quick message to say that I started looking at porting GDB to
> x86_64-windows. Currently, if I use a 32bit compiler and configure
> GDB as a i686-pc-mingw32, I get a fine debugger.

This is really good. I begun with a port of gdb (now about a half a year 
ago). If you like I can provide you my work on it. On the other hand, if 
you need some support for the x86_64 mingw target in crt, gcc, or 
binutils, please let me know.

> [1]: Currently, config.guess (IIRC) doesn't like mingw64. For now,
>      I'll keep mingw32 as meaning either 32bit or 64bit. I'll probably
>      fix this too, but later.
The mingw64 should be an alias to x86_64-pc-mingw32, nothing more, nothing 
less. There were several discussion on that on gcc, and the conclusion was 
that the target triplet should be in future on gcc (i?86|x86_64)-pc-mingw.

Best regards,
 Kai Tietz

|  (\_/)  This is Bunny. Copy and paste Bunny
| (='.'=) into your signature to help him gain
| (")_(") world domination.