sdio-linux-devel

[Sdio-linux-devel] sdio-linux and recent kernels

[Werner A. <we...@op...>]

Hi all,

I'm in the process of adding drivers for the Samsung 2410 and 2440
MCUs for the OpenMoko project. We're tracking the latest mainstream
kernel, currently we're at 2.6.21.3.

For now, I went back to an older kernel for sdio-linux, but I
wonder if you are already working on patches for more recent
kernels. If not, I'll look into this when done with the drivers.

I'm also curious about plans for merging sdio-linux into the
mainstream kernel. I see mainly two areas where there could be
issues: 1) functionality overlaps with the existing SD/MMC
driver, and 2) the coding style is quite different from what's
required for mainline. Any thoughts on this ?

- Werner

-- 
  _________________________________________________________________________
 / Werner Almesberger, Buenos Aires, Argentina     we...@al... /
/_http://www.almesberger.net/____________________________________________/

Re: [Sdio-linux-devel] sdio-linux and recent kernels

[Paul L. <pl...@at...>]

MontaVista is working on various kernel versions. They may be able to=20
give some additional insight. They are also working on supporting a=20
merged version. I do not have any details on it.

Regards

Paul

-----Original Message-----
From: Werner Almesberger [mailto:we...@op...]=20
Sent: Tuesday, June 12, 2007 3:09 AM
To: sdi...@li...
Subject: [Sdio-linux-devel] sdio-linux and recent kernels

Hi all,

I'm in the process of adding drivers for the Samsung 2410 and 2440 MCUs=20
for the OpenMoko project. We're tracking the latest mainstream kernel,=20
currently we're at 2.6.21.3.

For now, I went back to an older kernel for sdio-linux, but I wonder if=20
you are already working on patches for more recent kernels. If not, I'll=20
look into this when done with the drivers.

I'm also curious about plans for merging sdio-linux into the mainstream=20
kernel. I see mainly two areas where there could be
issues: 1) functionality overlaps with the existing SD/MMC driver, and=20
2) the coding style is quite different from what's required for=20
mainline. Any thoughts on this ?

- Werner

--
 =20
________________________________________________________________________
_
 / Werner Almesberger, Buenos Aires, Argentina    =20
we...@al... /
/_http://www.almesberger.net/___________________________________________
_/

------------------------------------------------------------------------
-
This SF.net email is sponsored by DB2 Express Download DB2 Express C -=20
the FREE version of DB2 express and take control of your XML. No limits.=20
Just data. Click to get it now.
http://sourceforge.net/powerbar/db2/
_______________________________________________
Sdio-linux-devel mailing list
Sdi...@li...
https://lists.sourceforge.net/lists/listinfo/sdio-linux-devel

Re: [Sdio-linux-devel] sdio-linux and recent kernels

[Nicolas P. <ni...@ca...>]

On Tue, 12 Jun 2007, Werner Almesberger wrote:

> Hi all,
> 
> I'm in the process of adding drivers for the Samsung 2410 and 2440
> MCUs for the OpenMoko project. We're tracking the latest mainstream
> kernel, currently we're at 2.6.21.3.
> 
> For now, I went back to an older kernel for sdio-linux, but I
> wonder if you are already working on patches for more recent
> kernels. If not, I'll look into this when done with the drivers.
> 
> I'm also curious about plans for merging sdio-linux into the
> mainstream kernel. I see mainly two areas where there could be
> issues: 1) functionality overlaps with the existing SD/MMC
> driver, and 2) the coding style is quite different from what's
> required for mainline. Any thoughts on this ?

I'm currently working closely with Pierre Ossman to add SDIO support 
into mainline.  Because of the issues you noted above, the mainline SDIO 
support will probably be a complete rewrite.  This is why I'm currently 
directing my current efforts towards this rewrite for the latest 
mainstream kernel instead of bringing the sdio-linux codebase forward.

While the sdio-linux code base has its value as an already functional 
SDIO stack, it has some structural and major style issues that makes it 
impossible to ever be considered for mainline.  There is another SDIO 
stack contender from Marvell that was turned down as well for the same 
reasons.  Therefore I may only use this code as a working reference at 
this point.

Because of that, if you're targeting the latest mainline kernel, I 
strongly suggest that you consider writing a host driver for the 
existing MMC/SD stack in mainline.  The current SDIO effort is based on 
that.

Pierre already pushed some of our SDIO work in his GIT tree at 
git://git2.kernel.org/pub/scm/linux/kernel/git/drzeus/mmc.git.
This is still early WIP but things are moving rather fast.


Nicolas

Re: [Sdio-linux-devel] sdio-linux and recent kernels

[<ra...@us...>]

Hi,
On 6/12/07, Nicolas Pitre <ni...@ca...> wrote:
>
> I'm currently working closely with Pierre Ossman to add SDIO support
> into mainline.  Because of the issues you noted above, the mainline SDIO
> support will probably be a complete rewrite.  This is why I'm currently
> directing my current efforts towards this rewrite for the latest
> mainstream kernel instead of bringing the sdio-linux codebase forward.
>
Does  MMC/SD/SDIO subsystem has a discussion mailing list ?

Best regards,

--=20
Ragner N Magalh=E3es

Instituto Nokia de Tecnologia - INdT
Open Source Mobile Research Center - OSMRC
Linux Kernel Team

E-mail: rag...@in...
            ra...@us...

Re: [Sdio-linux-devel] sdio-linux and recent kernels

[Werner A. <we...@op...>]

Nicolas Pitre wrote:
> I'm currently working closely with Pierre Ossman to add SDIO support 
> into mainline.

Perfect !

> While the sdio-linux code base has its value as an already functional 
> SDIO stack, it has some structural and major style issues that makes it 
> impossible to ever be considered for mainline.

Yup :-(

> Because of that, if you're targeting the latest mainline kernel, I 
> strongly suggest that you consider writing a host driver for the 
> existing MMC/SD stack in mainline.  The current SDIO effort is based on 
> that.

Okay, I'll have a look at it. How far is AR6001 support from working
with the "new" SDIO stack ? We need that rather urgently, at least at
a proof of concept level.

If AR6001 will take some time to work with the "new" stack, the best
approach is probably if we use sdio-linux with an old kernel for our
hardware testing, and then work with the "new" stack in parallel.

Thanks,
- Werner

-- 
  _________________________________________________________________________
 / Werner Almesberger, Buenos Aires, Argentina     we...@al... /
/_http://www.almesberger.net/____________________________________________/

Re: [Sdio-linux-devel] sdio-linux and recent kernels

[Nicolas P. <ni...@ca...>]

On Tue, 12 Jun 2007, Ragner Magalhães wrote:

> Hi,
> On 6/12/07, Nicolas Pitre <ni...@ca...> wrote:
> > 
> > I'm currently working closely with Pierre Ossman to add SDIO support
> > into mainline.  Because of the issues you noted above, the mainline SDIO
> > support will probably be a complete rewrite.  This is why I'm currently
> > directing my current efforts towards this rewrite for the latest
> > mainstream kernel instead of bringing the sdio-linux codebase forward.
> > 
> Does  MMC/SD/SDIO subsystem has a discussion mailing list ?

The MMC/SD discussions typically happened on 
lin...@vg....  Since SDIO is just lifting up, 
discussions about the current stuff happened between Pierre and myself 
directly.  I suspect that things will move to lkml when things start to 
get interesting, unless Pierre can be convinced to set up a dedicated 
list.


Nicolas

Re: [Sdio-linux-devel] sdio-linux and recent kernels

[Nicolas P. <ni...@ca...>]

On Tue, 12 Jun 2007, Werner Almesberger wrote:

> Nicolas Pitre wrote:
> > Because of that, if you're targeting the latest mainline kernel, I 
> > strongly suggest that you consider writing a host driver for the 
> > existing MMC/SD stack in mainline.  The current SDIO effort is based on 
> > that.
> 
> Okay, I'll have a look at it. How far is AR6001 support from working
> with the "new" SDIO stack ? We need that rather urgently, at least at
> a proof of concept level.

It is still a bit far off.

> If AR6001 will take some time to work with the "new" stack, the best
> approach is probably if we use sdio-linux with an old kernel for our
> hardware testing, and then work with the "new" stack in parallel.

Most probably the best course of action if you're in a hurry.


Nicolas