i2c-algo-ite and i2c-ite planned for removal

i2c-algo-ite and i2c-ite planned for removal

[Jean Delvare]

Hi all,

I noticed today that we have an i2c bus driver named i2c-ite,
supposedly useful on some MIPS systems which have an ITE8172 chip,
which doesn't compile. It is based on an i2c algorithm driver named
i2c-algo-ite, which doesn't compile either, due to some changes made to
the i2c core which weren't properly reflected there. Going back trough
the versions, I found that the bus driver was previously named
i2c-adap-ite, and was introduced in 2.4.10. And I don't think it even
compiled back then either, as it uses a structure (iic_ite) which isn't
defined anywhere.

So basically we have two drivers in the kernel tree for 5 years or so,
which never were usable, and nobody seemed to care. It's about time to
get rid of these 1296 lines of code, don't you think? So, unless someone
volunteers to take care of these drivers, or otherwise has a very
strong reason to object, I'm going to delete them from the tree soon.

Thanks,
-- 
Jean Delvare

Re: i2c-algo-ite and i2c-ite planned for removal

[Pete Popov]

On Thu, 2006-06-15 at 22:57 +0200, Jean Delvare wrote:
> Hi all,
> 
> I noticed today that we have an i2c bus driver named i2c-ite,
> supposedly useful on some MIPS systems which have an ITE8172 chip,
> which doesn't compile. It is based on an i2c algorithm driver named
> i2c-algo-ite, which doesn't compile either, due to some changes made to
> the i2c core which weren't properly reflected there. Going back trough
> the versions, I found that the bus driver was previously named
> i2c-adap-ite, and was introduced in 2.4.10. And I don't think it even
> compiled back then either, as it uses a structure (iic_ite) which isn't
> defined anywhere.
> 
> So basically we have two drivers in the kernel tree for 5 years or so,
> which never were usable, and nobody seemed to care. 

For historical correctness, this driver was once upon a time usable,
though it was a few years ago. It was written by MV for some ref board
that had the ITE chip and it did work. That ref board is no longer
around so it's probably safe to nuke the driver. 

Pete

> It's about time to
> get rid of these 1296 lines of code, don't you think? So, unless someone
> volunteers to take care of these drivers, or otherwise has a very
> strong reason to object, I'm going to delete them from the tree soon.
> 
> Thanks,

Re: i2c-algo-ite and i2c-ite planned for removal

[Jean Delvare]

Hi Pete,

> > So basically we have two drivers in the kernel tree for 5 years or so,
> > which never were usable, and nobody seemed to care. 
> 
> For historical correctness, this driver was once upon a time usable,
> though it was a few years ago. It was written by MV for some ref board
> that had the ITE chip and it did work. That ref board is no longer
> around so it's probably safe to nuke the driver. 

In which kernel version? In every version I checked (2.4.12, 2.4.30,
2.6.0 and 2.6.16) it wouldn't compile due to struct iic_ite being used
but never defined (and possibly other errors, but I can't test-compile
the driver.)

-- 
Jean Delvare

Re: i2c-algo-ite and i2c-ite planned for removal

[Pete Popov]

On Fri, 2006-06-16 at 22:29 +0200, Jean Delvare wrote:
> Hi Pete,
> 
> > > So basically we have two drivers in the kernel tree for 5 years or so,
> > > which never were usable, and nobody seemed to care. 
> > 
> > For historical correctness, this driver was once upon a time usable,
> > though it was a few years ago. It was written by MV for some ref board
> > that had the ITE chip and it did work. That ref board is no longer
> > around so it's probably safe to nuke the driver. 
> 
> In which kernel version? In every version I checked (2.4.12, 2.4.30,
> 2.6.0 and 2.6.16) it wouldn't compile due to struct iic_ite being used
> but never defined (and possibly other errors, but I can't test-compile
> the driver.)

Honestly, I don't remember. I think it was one of the very first 2.6
kernels because when MV first released a 2.6 product, 2.6 was still
'experimental'. It's quite possible of course that the driver was never
properly merged upstream in the community tree(s). But I do know that it
worked in the internal MV tree and an effort was made to get the driver
accepted upstream.

Pete

Re: i2c-algo-ite and i2c-ite planned for removal

[Jean Delvare]

Hi Pete,

> > > For historical correctness, this driver was once upon a time usable,
> > > though it was a few years ago. It was written by MV for some ref board
> > > that had the ITE chip and it did work. That ref board is no longer
> > > around so it's probably safe to nuke the driver. 
> > 
> > In which kernel version? In every version I checked (2.4.12, 2.4.30,
> > 2.6.0 and 2.6.16) it wouldn't compile due to struct iic_ite being used
> > but never defined (and possibly other errors, but I can't test-compile
> > the driver.)
> 
> Honestly, I don't remember. I think it was one of the very first 2.6
> kernels because when MV first released a 2.6 product, 2.6 was still
> 'experimental'. It's quite possible of course that the driver was never
> properly merged upstream in the community tree(s). But I do know that it
> worked in the internal MV tree and an effort was made to get the driver
> accepted upstream.

I couldn't find any evidence of this effort. Whatever, past is past, if
someone fixes the i2c-ite and i2c-algo-ite drivers soon, fine with me,
if not, the drivers will be deleted (which doesn't mean they can't be
resurrected later if there is interest and someone takes over
maintenance.) I'm setting the deadline to September 2006.

-- 
Jean Delvare

Re: i2c-algo-ite and i2c-ite planned for removal

[Ralf Baechle]

On Fri, Jun 16, 2006 at 12:23:17AM +0300, Pete Popov wrote:

> For historical correctness, this driver was once upon a time usable,
> though it was a few years ago. It was written by MV for some ref board
> that had the ITE chip and it did work. That ref board is no longer
> around so it's probably safe to nuke the driver. 

Not quite true; the board support for that board is still around and it's
on my to-be-nuked list for after 2.6.18.

  Ralf

Re: i2c-algo-ite and i2c-ite planned for removal

[Ralf Baechle]

On Wed, Jun 21, 2006 at 11:08:34AM +0530, ankur maheshwari wrote:

> I have used once i2c-adap-ite and i2c-algo-ite for ite-8712 chip and it
> worked fine for me in MV 2.4.25.

The fact that is used to work in some vendor kernel is irrelevant.  And
2.4 hardly indicates anyway.

> Its been an year ago, I asked on same forum if some one has used it
> before but I didn't got any reply.

You see how amazingly popular the board was.  On April 2 just after
2.6.16 was out I announced my intension to remove the board in

  http://www.linux-mips.org/cgi-bin/mesg.cgi?a=linux-mips&i=20060402194822.GA26358%40linux-mips.org

Nobody raised hand to take care of the maintenance of any of these boards,
thus http://www.linux-mips.org/wiki/ITE-8172G also marks the board as
to be deleted.

> It's just an info on ite-chip works, to remove it from kernel tree .....
> decision is up to you : ).

There is more that just that wrong with the board support.  And if the
fact that it is was broken does not even result bug reports this is
another indicator the board a board should go.

The usual reminder: patches to fix the issues will be accepted.

  Ralf

RE: i2c-algo-ite and i2c-ite planned for removal

[Shane McDonald]

Hi:

  PMC-Sierra has an eval board (the "Xiao Hu") that uses these files.  We haven't (yet) pushed out patches to add support for this board, but we hope to at some time.  The kernel we use on the board is based on 2.6.14, so I could generate a patch so that these files will compile against 2.6.14.  I'm guessing the patch will also apply to 2.6.16.20, since these files haven't changed for so long.

  I'll put together the patch and make sure it's still correct against 2.6.16.20, then submit it.

Shane

> -----Original Message-----
> From: linux-mips-bounce@linux-mips.org 
> [mailto:linux-mips-bounce@linux-mips.org] On Behalf Of Jean Delvare
> Sent: Thursday, June 15, 2006 2:57 PM
> To: linux-mips@linux-mips.org; LKML
> Subject: i2c-algo-ite and i2c-ite planned for removal
> 
> Hi all,
> 
> I noticed today that we have an i2c bus driver named i2c-ite,
> supposedly useful on some MIPS systems which have an ITE8172 chip,
> which doesn't compile. It is based on an i2c algorithm driver named
> i2c-algo-ite, which doesn't compile either, due to some 
> changes made to
> the i2c core which weren't properly reflected there. Going back trough
> the versions, I found that the bus driver was previously named
> i2c-adap-ite, and was introduced in 2.4.10. And I don't think it even
> compiled back then either, as it uses a structure (iic_ite) 
> which isn't
> defined anywhere.
> 
> So basically we have two drivers in the kernel tree for 5 years or so,
> which never were usable, and nobody seemed to care. It's about time to
> get rid of these 1296 lines of code, don't you think? So, 
> unless someone
> volunteers to take care of these drivers, or otherwise has a very
> strong reason to object, I'm going to delete them from the tree soon.
> 
> Thanks,
> -- 
> Jean Delvare
>