[Qemu-devel] State of TI OMAP board support?

[Qemu-devel] State of TI OMAP board support?

[Paul Sokolovsky]

Hello qemu-devel,

  I wonder, what's the state of TI OMAP patch, previously announced on
the list by Andrzej Zaborowski
http://lists.gnu.org/archive/html/qemu-devel/2006-03/msg00125.html ?
Is it considered for inclusion in the mainline? If not, what precludes
that?

-- 
Best regards,
 Paul                          mailto:pmiscml@gmail.com

Re: [Qemu-devel] State of TI OMAP board support?

[andrzej zaborowski]

Hi,

On 30/07/06, Paul Sokolovsky <pmiscml@gmail.com> wrote:
> Hello qemu-devel,
>
>   I wonder, what's the state of TI OMAP patch, previously announced on
> the list by Andrzej Zaborowski
> http://lists.gnu.org/archive/html/qemu-devel/2006-03/msg00125.html ?
> Is it considered for inclusion in the mainline? If not, what precludes
> that?

I made some progress on the OMAP 310 emulation but I probably won't
finish it till I absolutely need it (which might be soon). I think I
will consider it finished when I can run PalmOS 5.0 on it. Linux works
absolutely fine, but PalmOS boots only halfway through and gets stuck
in some loop (probably not the OMAP emulation fault, probably one of
the peripheral chips, which are also emulated). Running a
closed-source OS is a real stress-test for an emulator.

I don't think it is considered for inclusion. Before it is merged I
would like to separate the generic OMAP code from board-related code
(the board I emulate is the Palm Tunsgten|E handheld), but I imagine
the QEMU maintainers won't like the way code is formatted and probably
many other things, hard to say.

If you want to emulate a different OMAP processor you'll probably want
to add the emulation of the DSP part, which 310 doesn't have and it
may a whole lot of additional work.
Also, my current tree is based on QEMU 0.8.0.

>
> --
> Best regards,
>  Paul                          mailto:pmiscml@gmail.com
>
>
>
> _______________________________________________
> Qemu-devel mailing list
> Qemu-devel@nongnu.org
> http://lists.nongnu.org/mailman/listinfo/qemu-devel
>

Regards,
-- 
balrog 2oo6

Dear Outlook users: Please remove me from your address books
http://www.newsforge.com/article.pl?sid=03/08/21/143258

Re[2]: [Qemu-devel] State of TI OMAP board support?

[Paul Sokolovsky]

Hello Andrzej,

Sunday, July 30, 2006, 4:29:41 PM, you wrote:

> Hi,

> On 30/07/06, Paul Sokolovsky <pmiscml@gmail.com> wrote:
>> Hello qemu-devel,
>>
>>   I wonder, what's the state of TI OMAP patch, previously announced on
>> the list by Andrzej Zaborowski
>> http://lists.gnu.org/archive/html/qemu-devel/2006-03/msg00125.html ?
>> Is it considered for inclusion in the mainline? If not, what precludes
>> that?

> I made some progress on the OMAP 310 emulation but I probably won't
> finish it till I absolutely need it (which might be soon). I think I
> will consider it finished when I can run PalmOS 5.0 on it. Linux works
> absolutely fine, but PalmOS boots only halfway through and gets stuck
> in some loop (probably not the OMAP emulation fault, probably one of
> the peripheral chips, which are also emulated). Running a
> closed-source OS is a real stress-test for an emulator.

  Yes, I may say that the job you've already done is very impressive,
and aims you have are even more ambitious. If you do PalmOS, WinCE
will be next ;-).

> I don't think it is considered for inclusion. Before it is merged I
> would like to separate the generic OMAP code from board-related code
> (the board I emulate is the Palm Tunsgten|E handheld), but I imagine
> the QEMU maintainers won't like the way code is formatted and probably
> many other things, hard to say.

  I really hoped to get response from QEMU author/ARM emulation maintainer
regarding these points. Paul, any comments? Would you be interested to
add more CPUs/boards? Any suggestion regarding this? Maybe if it is
expected that there may be many implementation and they won't fit into
core distribution, to provide API for plugin development?

> If you want to emulate a different OMAP processor you'll probably want
> to add the emulation of the DSP part, which 310 doesn't have and it
> may a whole lot of additional work.

  Actually, I'm also working on porting a PDA to run Linux (iPaq
4150/XScale). So far I don't have strong need to run kernel compiled
specifically for that device in QEMU, though SD support you
implemented seems like very good addition for emulation of any PDA
device. Anyway, I'm currently using Integrator/CP kernel from here:
http://www.o-hand.com/~richard/qemu.html , which boots h4000 userland
(Familiar Linux GPE) just fine. Well, that GPE image doesn't have
suport for mouse, and QEMU cannot emulate touchscreen. So, first thing
I'd be interested from your patch is that "absolute mouse" support.
Its status is also not clear - googling gives lots of references, and
over good period of time, but it seems that patch is still not in the
mainline?

  Well, I guess I need to look closer at you patch and try do my
builds based on it, before asking specific questions.

> Also, my current tree is based on QEMU 0.8.0.

Thanks,

-- 
 Paul                            mailto:pmiscml@gmail.com

Re: [Qemu-devel] State of TI OMAP board support?

[Paul Brook]

> > I don't think it is considered for inclusion. Before it is merged I
> > would like to separate the generic OMAP code from board-related code
> > (the board I emulate is the Palm Tunsgten|E handheld), but I imagine
> > the QEMU maintainers won't like the way code is formatted and probably
> > many other things, hard to say.
>
>   I really hoped to get response from QEMU author/ARM emulation maintainer
> regarding these points. Paul, any comments? Would you be interested to
> add more CPUs/boards? Any suggestion regarding this? Maybe if it is
> expected that there may be many implementation and they won't fit into
> core distribution, to provide API for plugin development?

I've no objection to adding new boards/cpus[1].

However, unless I have some personal/commercial interest in that particular 
board it's up to the patch author(s) to get that support into a state where 
I'm happy merging it. If the original author thinks it needs cleanup before 
submission I'm inclined to believe them ;-)

All the normal guidelines for patch submission apply. i.e. follow coding 
conventions, split changes into logically separate patches, don't mix 
cleanups with new features, etc.

Paul

[1] As discussed previously on this list there are unresolved legal issues 
with emulating ARMv6/v7 cpus, however that's not relevant in this case.

Re: Re[2]: [Qemu-devel] State of TI OMAP board support?

[andrzej zaborowski]

On 03/08/06, Paul Sokolovsky <pmiscml@gmail.com> wrote:
> Hello Andrzej,
>
> Sunday, July 30, 2006, 4:29:41 PM, you wrote:
>
> > Hi,
>
> > On 30/07/06, Paul Sokolovsky <pmiscml@gmail.com> wrote:
> >> Hello qemu-devel,
> >>
> >>   I wonder, what's the state of TI OMAP patch, previously announced on
> >> the list by Andrzej Zaborowski
> >> http://lists.gnu.org/archive/html/qemu-devel/2006-03/msg00125.html ?
> >> Is it considered for inclusion in the mainline? If not, what precludes
> >> that?
>
> > I made some progress on the OMAP 310 emulation but I probably won't
> > finish it till I absolutely need it (which might be soon). I think I
> > will consider it finished when I can run PalmOS 5.0 on it. Linux works
> > absolutely fine, but PalmOS boots only halfway through and gets stuck
> > in some loop (probably not the OMAP emulation fault, probably one of
> > the peripheral chips, which are also emulated). Running a
> > closed-source OS is a real stress-test for an emulator.
>
>   Yes, I may say that the job you've already done is very impressive,
> and aims you have are even more ambitious. If you do PalmOS, WinCE
> will be next ;-).
>
> > I don't think it is considered for inclusion. Before it is merged I
> > would like to separate the generic OMAP code from board-related code
> > (the board I emulate is the Palm Tunsgten|E handheld), but I imagine
> > the QEMU maintainers won't like the way code is formatted and probably
> > many other things, hard to say.
>
>   I really hoped to get response from QEMU author/ARM emulation maintainer
> regarding these points. Paul, any comments? Would you be interested to
> add more CPUs/boards? Any suggestion regarding this? Maybe if it is
> expected that there may be many implementation and they won't fit into
> core distribution, to provide API for plugin development?
>
> > If you want to emulate a different OMAP processor you'll probably want
> > to add the emulation of the DSP part, which 310 doesn't have and it
> > may a whole lot of additional work.
>
>   Actually, I'm also working on porting a PDA to run Linux (iPaq
> 4150/XScale). So far I don't have strong need to run kernel compiled

Ah, now I know, I've seen your commits on handhelds.org!
There is a person working on PXA (I think 270) emulation in QEMU, but
I think they're at implementing the CPU now and out of the other
hardware there's only the serial port, but I don't know the details. I
only remember the person's nick on IRC.

> specifically for that device in QEMU, though SD support you
> implemented seems like very good addition for emulation of any PDA
> device. Anyway, I'm currently using Integrator/CP kernel from here:
> http://www.o-hand.com/~richard/qemu.html , which boots h4000 userland
> (Familiar Linux GPE) just fine. Well, that GPE image doesn't have
> suport for mouse, and QEMU cannot emulate touchscreen. So, first thing
> I'd be interested from your patch is that "absolute mouse" support.

The absolute mouse support in my patch was basically stolen from
Anthony Liguori's early Wacom tablet patch. There are links to this
patch in list archives, I'm not sure if it's integrated into QEMU yet,
but there's a pointing device using absolute coordintates known as
usb-tablet which is in the mainline already.

I've been running GPE in QEMU as well, using mostly unmodified fs
images from familiar.handhelds.org and SD emulation which you
mentioned (IIRC I had to disable udev everytime). I took some
screenshots which can be seen at http://zabor.org/balrog/palmte/
(filenames ending in "-emulator").

> Its status is also not clear - googling gives lots of references, and
> over good period of time, but it seems that patch is still not in the
> mainline?
>
>   Well, I guess I need to look closer at you patch and try do my
> builds based on it, before asking specific questions.
>
> > Also, my current tree is based on QEMU 0.8.0.
>
> Thanks,
>
> --
>  Paul                            mailto:pmiscml@gmail.com
>
>

Good luck with your port.
Regards,
Andrzej
-- 
balrog 2oo6

Dear Outlook users: Please remove me from your address books
http://www.newsforge.com/article.pl?sid=03/08/21/143258