i2c: removal of s6000 driver

i2c: removal of s6000 driver

[Wolfram Sang]

[-- Attachment #1: Type: text/plain, Size: 581 bytes --]

Hi,

I recently discovered that the I2C core of the S6000 is basically a
Designware IP core. For the latter we have a frequently used and updated
i2c driver while the S6000 driver seems to bitrot. My wish is to remove
the s6000 driver in favor of the DesignWare one. My question is if there
is someone who could test such a change on real hardware?
Grepping through sources, it looks to me like the only candidate in
mainline is the s6105, yet its defconfig doesn't even have I2C
activated? Can we remove this driver right away? Am I missing something?

Kind regards,

   Wolfram


[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: i2c: removal of s6000 driver

[Oskar Schirmer]

Hi Wolfram,

On Wed, May 21, 2014 at 13:30:29 +0200, Wolfram Sang wrote:
> I recently discovered that the I2C core of the S6000 is basically a
> Designware IP core. For the latter we have a frequently used and updated
> i2c driver while the S6000 driver seems to bitrot. My wish is to remove
> the s6000 driver in favor of the DesignWare one. My question is if there
> is someone who could test such a change on real hardware?
> Grepping through sources, it looks to me like the only candidate in
> mainline is the s6105, yet its defconfig doesn't even have I2C
> activated? Can we remove this driver right away? Am I missing something?

Removing drivers without a replacement tested and running
doesn't make much sense, except one chooses to drop support
for the whole S6 thing right away. Which I don't support.

best regards,
  Oskar

Re: i2c: removal of s6000 driver

[Max Filippov]

On Fri, May 23, 2014 at 5:17 PM, Oskar Schirmer <oskar-fYPSZ7JpQqsAvxtiuMwx3w@public.gmane.org> wrote:
> Removing drivers without a replacement tested and running
> doesn't make much sense, except one chooses to drop support
> for the whole S6 thing right away. Which I don't support.

Oscar, do you mean that you don't support the platform or that
you don't support its removal?

I'm pretty sure it's broken now, since there's no hardware to test on
and even no compiler for it (IIRC a recent attempt to get a toolchain
overlay for s6000 failed, see
http://lists.linux-xtensa.org/pipermail/linux-xtensa/Week-of-Mon-20130617/001099.html
).

-- 
Thanks.
-- Max

Re: i2c: removal of s6000 driver

[czankel]

Hi Oskar,

I might have asked you before, but do you still have the overlay for the 
S6000 somewhere laying around?

Thanks,
-Chris

On 5/23/14, 6:17 AM, Oskar Schirmer wrote:
> Hi Wolfram,
>
> On Wed, May 21, 2014 at 13:30:29 +0200, Wolfram Sang wrote:
>> I recently discovered that the I2C core of the S6000 is basically a
>> Designware IP core. For the latter we have a frequently used and updated
>> i2c driver while the S6000 driver seems to bitrot. My wish is to remove
>> the s6000 driver in favor of the DesignWare one. My question is if there
>> is someone who could test such a change on real hardware?
>> Grepping through sources, it looks to me like the only candidate in
>> mainline is the s6105, yet its defconfig doesn't even have I2C
>> activated? Can we remove this driver right away? Am I missing something?
> Removing drivers without a replacement tested and running
> doesn't make much sense, except one chooses to drop support
> for the whole S6 thing right away. Which I don't support.
>
> best regards,
>    Oskar

Re: i2c: removal of s6000 driver

[Oskar Schirmer]

On Fri, May 23, 2014 at 10:33:06 -0700, czankel wrote:
> I might have asked you before, but do you still have the overlay for the 
> S6000 somewhere laying around?

No, but maybe Daniel has?

  Oskar

Re: i2c: removal of s6000 driver

[Daniel Glöckner]

On Sat, May 24, 2014 at 03:47:27PM +0000, Oskar Schirmer wrote:
> On Fri, May 23, 2014 at 10:33:06 -0700, czankel wrote:
> > I might have asked you before, but do you still have the overlay for the 
> > S6000 somewhere laying around?
> 
> No, but maybe Daniel has?

At work we still have the overlay as part of the board support package
we created for the s6105. But I'd prefer to talk to someone at Stretch
first before handing out copies. Unfortunately I'm on vacation right
now and will be without internet access from tomorrow to Sunday.

  Daniel

Re: i2c: removal of s6000 driver

[Wolfram Sang]

[-- Attachment #1: Type: text/plain, Size: 700 bytes --]

On Wed, May 28, 2014 at 12:35:57AM +0200, Daniel Glöckner wrote:
> On Sat, May 24, 2014 at 03:47:27PM +0000, Oskar Schirmer wrote:
> > On Fri, May 23, 2014 at 10:33:06 -0700, czankel wrote:
> > > I might have asked you before, but do you still have the overlay for the 
> > > S6000 somewhere laying around?
> > 
> > No, but maybe Daniel has?
> 
> At work we still have the overlay as part of the board support package
> we created for the s6105. But I'd prefer to talk to someone at Stretch
> first before handing out copies. Unfortunately I'm on vacation right
> now and will be without internet access from tomorrow to Sunday.

Ping. Is that the only known platform for the s6000?


[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 819 bytes --]

Re: i2c: removal of s6000 driver

[Daniel Glöckner]

On Tue, Jun 10, 2014 at 12:06:35AM +0200, Wolfram Sang wrote:
> On Wed, May 28, 2014 at 12:35:57AM +0200, Daniel Glöckner wrote:
> > At work we still have the overlay as part of the board support package
> > we created for the s6105. But I'd prefer to talk to someone at Stretch
> > first before handing out copies. Unfortunately I'm on vacation right
> > now and will be without internet access from tomorrow to Sunday.
>
> Ping. Is that the only known platform for the s6000?

The only platform we know of.

These chips were showcased as H.264 encoders, so it is more likely
to find them inside cameras, DVRs, and studio equipment. But there
is no H.264 IP block inside the chips and their instruction set is
generic (actually runtime reconfigurable), so they are not restricted
to video processing.

  Daniel

Re: i2c: removal of s6000 driver

[Daniel Glöckner]

On Tue, Jun 10, 2014 at 12:06:35AM +0200, Wolfram Sang wrote:
> On Wed, May 28, 2014 at 12:35:57AM +0200, Daniel Glöckner wrote:
> > On Sat, May 24, 2014 at 03:47:27PM +0000, Oskar Schirmer wrote:
> > > On Fri, May 23, 2014 at 10:33:06 -0700, czankel wrote:
> > > > I might have asked you before, but do you still have the overlay for
> > > > the S6000 somewhere laying around?
> > > 
> > > No, but maybe Daniel has?
> > 
> > At work we still have the overlay as part of the board support package
> > we created for the s6105. But I'd prefer to talk to someone at Stretch
> > first before handing out copies. Unfortunately I'm on vacation right
> > now and will be without internet access from tomorrow to Sunday.
> 
> Ping. Is that the only known platform for the s6000?

After exchanging some mails with Stretch (now Exar), it is still uncertain
if the Linux port ever made it into a product. As the chip is out of
production, there won't be any new s6000 devices. Google turns up three
press releases of companies using the s6000 in DVR devices, but these
appear to have the chips on add-in cards.

The good news is that Exar doesn't mind me sharing the overlay.
You can download it from this page: http://www.emlix.com/download/

I removed some files from the tarball:

create-config-pkgs
	had a restrictive copyright header requiring written permission from
	Tensilica. It's a simple shell script grouping the other files for
	building various toolchain packages.

src/gdb/reg-xtensa.dat and src/binutils/xtensa-config.sh
	had no license header and Exar could not easily determine the original
	licensing conditions. The first file basically contains the size
	and name of the first 176 registers in the "rmap" array. The other
	file just defines MAXPAGESIZE to a huge value. Well, it's nommu.

gnu-xcalibur/
	Appeared to contain old copies of GNU libstdc++, libio, and libiberty.
	Most files were four years older than the Xtensa core headers.
	I doubt anyone would want to use these today.

Marc, according to the tarball someone with your name at Tensilica was
involved in the creation of the overlay. Maybe you remember if any of these
files is needed?

Trying to build CONFIG_XTENSA_PLATFORM_S6105 in mainline will fail.
Mainline never accepted all of our patches and this apparently led to
two arrays being mixed when 759c67e87bce was merged.

Best regards,

  Daniel

-- 
Dipl.-Math. Daniel Glöckner, emlix GmbH, http://www.emlix.com
Fon +49 551 30664-0, Fax +49 551 30664-11,
Bertha-von-Suttner-Straße 9, 37085 Göttingen, Germany
Sitz der Gesellschaft: Göttingen, Amtsgericht Göttingen HR B 3160
Geschäftsführung: Dr. Uwe Kracke, Ust-IdNr.: DE 205 198 055

emlix - your embedded linux partner

Re: i2c: removal of s6000 driver

[Wolfram Sang]

[-- Attachment #1: Type: text/plain, Size: 837 bytes --]

Hi Daniel,

> After exchanging some mails with Stretch (now Exar), it is still uncertain
> if the Linux port ever made it into a product.

Thanks for keeping at it!

> The good news is that Exar doesn't mind me sharing the overlay.
> You can download it from this page: http://www.emlix.com/download/

That is nice.

> Trying to build CONFIG_XTENSA_PLATFORM_S6105 in mainline will fail.
> Mainline never accepted all of our patches and this apparently led to
> two arrays being mixed when 759c67e87bce was merged.

That not ;) Anyhow, my latest pull-request includes the removal of the
s6000 I2C driver. There is the designware driver if anyone feels like
it. It doesn't really look like there will be one, though. Sounds like
this platform is a good candidate for removal, but I will leave this up
to you guys...

Regards,

   Wolfram


[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 819 bytes --]

Re: i2c: removal of s6000 driver

[Wolfram Sang]

[-- Attachment #1: Type: text/plain, Size: 960 bytes --]


> > I recently discovered that the I2C core of the S6000 is basically a
> > Designware IP core. For the latter we have a frequently used and updated
> > i2c driver while the S6000 driver seems to bitrot. My wish is to remove
> > the s6000 driver in favor of the DesignWare one. My question is if there
> > is someone who could test such a change on real hardware?
> > Grepping through sources, it looks to me like the only candidate in
> > mainline is the s6105, yet its defconfig doesn't even have I2C
> > activated? Can we remove this driver right away? Am I missing something?
> 
> Removing drivers without a replacement tested and running
> doesn't make much sense, except one chooses to drop support
> for the whole S6 thing right away. Which I don't support.

Huh? I explicitly asked for help to test the replacement driver in the
first place. And was wondering why it is not in a defconfig. You know,
it could have been broken, too.


[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]