Re: drivers/mtd/devices/at91_dataflash26.c

Re: drivers/mtd/devices/at91_dataflash26.c

[David Brownell]

On Friday 27 April 2007, David Brownell wrote:
> On Friday 27 April 2007, Hans-Jürgen Koch wrote:
> 
> > > 	 the m25p80 driver
> > > was pretty close to working with those Atmel devices too ...
> > > most of the commands are identical.
> > 
> > Thanks for that hint, I'll have a look at the data sheet.

Any progress on that?  I looked again, and yes most commands
are identical.

 - That driver relies on an AT91-only SPI stack that's not
   going upstream, instead of drivers/spi/atmel_spi.c;

 - Driver can't even be selected on kernel.org kernel (!!);

 - The m25p80 driver will need at most minor tweaks to
   handle those AT26 (and AT25) series chips;

So 340ea370c2ce89d1c15fbf785460f2f74314ce58 should be reverted.

- Dave

Re: drivers/mtd/devices/at91_dataflash26.c

[Thomas Gleixner]

David,

On Thu, 2007-05-10 at 13:07 -0700, David Brownell wrote:
> On Friday 27 April 2007, David Brownell wrote:
> > On Friday 27 April 2007, Hans-Jürgen Koch wrote:
> > 
> > > > 	 the m25p80 driver
> > > > was pretty close to working with those Atmel devices too ...
> > > > most of the commands are identical.
> > > 
> > > Thanks for that hint, I'll have a look at the data sheet.
> 
> Any progress on that?  I looked again, and yes most commands
> are identical.
> 
>  - That driver relies on an AT91-only SPI stack that's not
>    going upstream, instead of drivers/spi/atmel_spi.c;
> 
>  - Driver can't even be selected on kernel.org kernel (!!);
> 
>  - The m25p80 driver will need at most minor tweaks to
>    handle those AT26 (and AT25) series chips;
> 
> So 340ea370c2ce89d1c15fbf785460f2f74314ce58 should be reverted.

We look into this and fix the problem. There is no need to revert it
right now. It does no harm and we remove it with the merge into the
m25p80 driver, ok ?

	tglx

Re: drivers/mtd/devices/at91_dataflash26.c

[David Brownell]

On Thursday 10 May 2007, Thomas Gleixner wrote:
> David,
> 
> On Thu, 2007-05-10 at 13:07 -0700, David Brownell wrote:
> > On Friday 27 April 2007, David Brownell wrote:
> > > On Friday 27 April 2007, Hans-Jürgen Koch wrote:
> > > 
> > > > > 	 the m25p80 driver
> > > > > was pretty close to working with those Atmel devices too ...
> > > > > most of the commands are identical.
> > > > 
> > > > Thanks for that hint, I'll have a look at the data sheet.
> > 
> > Any progress on that?  I looked again, and yes most commands
> > are identical...
> >
> > So 340ea370c2ce89d1c15fbf785460f2f74314ce58 should be reverted.
> 
> We look into this and fix the problem. There is no need to revert it
> right now. It does no harm and we remove it with the merge into the
> m25p80 driver, ok ?

Sounds like a plan ... but I'd still revert ASAP.

Either way, I'd hope it's all resolved before we
get very far into the 2.6.22-rc series.

- Dave

Re: drivers/mtd/devices/at91_dataflash26.c

[Hans-Jürgen Koch]

Am Donnerstag 10 Mai 2007 22:07 schrieb David Brownell:
> On Friday 27 April 2007, David Brownell wrote:
> > On Friday 27 April 2007, Hans-Jürgen Koch wrote:
> > 
> > > > 	 the m25p80 driver
> > > > was pretty close to working with those Atmel devices too ...
> > > > most of the commands are identical.
> > > 
> > > Thanks for that hint, I'll have a look at the data sheet.
> 
> Any progress on that?  I looked again, and yes most commands
> are identical.
> 
>  - That driver relies on an AT91-only SPI stack that's not
>    going upstream, instead of drivers/spi/atmel_spi.c;
> 
>  - Driver can't even be selected on kernel.org kernel (!!);
> 
>  - The m25p80 driver will need at most minor tweaks to
>    handle those AT26 (and AT25) series chips;
> 
> So 340ea370c2ce89d1c15fbf785460f2f74314ce58 should be reverted.
> 
> - Dave
>

Dave,
sorry for not responding earlier, I was busy with lots of other 
stuff. At the moment, I'm working on this again. I had a bit
of trouble getting a recent kernel working on my test system,
but that seems to be OK now.

You'll hear from me soon.

Thanks,
Hans

Re: drivers/mtd/devices/at91_dataflash26.c

[David Brownell]

On Thursday 10 May 2007, Hans-Jürgen Koch wrote:
> Am Donnerstag 10 Mai 2007 22:07 schrieb David Brownell:
> > On Friday 27 April 2007, David Brownell wrote:
> > > On Friday 27 April 2007, Hans-Jürgen Koch wrote:
> > > 
> > > > > 	 the m25p80 driver
> > > > > was pretty close to working with those Atmel devices too ...
> > > > > most of the commands are identical.
> > > > 
> > > > Thanks for that hint, I'll have a look at the data sheet.
> > 
> > Any progress on that?  I looked again, and yes most commands
> > are identical.

Still true.  :)

For that and a boatload of other SPI flash chips.  I'm
wondering just what that JEDEC standard covers; either
it's got useful command coverage, or vendors of SPI flash
noticed that converging command sets is a Good Thing.


> >  - That driver relies on an AT91-only SPI stack that's not
> >    going upstream, instead of drivers/spi/atmel_spi.c;
> > 
> >  - Driver can't even be selected on kernel.org kernel (!!);
> > 
> >  - The m25p80 driver will need at most minor tweaks to
> >    handle those AT26 (and AT25) series chips;
> > 
> > So 340ea370c2ce89d1c15fbf785460f2f74314ce58 should be reverted.

Also still true...


> > 
> > - Dave
> >
> 
> Dave,
> sorry for not responding earlier, I was busy with lots of other 
> stuff. At the moment, I'm working on this again. I had a bit
> of trouble getting a recent kernel working on my test system,
> but that seems to be OK now.
> 
> You'll hear from me soon.

Ping?  Have a look at

    http://lists.infradead.org/pipermail/linux-mtd/2007-June/018706.html

which depends on the resent mutex conversion patch in the
"previous message" link of that patch.

That should highlight more reasons why at91_dataflash should
be reverted ASAP.  I'd expect you would be able to just drop
that in and have it work ... unless you're seeing that nCS0
issue built out of at91rm9200 errata, in which case there's
an experiment I can forward to you.  :)

- Dave

Re: drivers/mtd/devices/at91_dataflash26.c

[Hans-Jürgen Koch]

Am Montag 25 Juni 2007 04:51 schrieb David Brownell:
> On Thursday 10 May 2007, Hans-Jürgen Koch wrote:
> > Am Donnerstag 10 Mai 2007 22:07 schrieb David Brownell:
> > > On Friday 27 April 2007, David Brownell wrote:
> > > > On Friday 27 April 2007, Hans-Jürgen Koch wrote:
> > > > 
> > > > > > 	 the m25p80 driver
> > > > > > was pretty close to working with those Atmel devices too ...
> > > > > > most of the commands are identical.
> > > > > 
> > > > > Thanks for that hint, I'll have a look at the data sheet.
> > > 
> > > Any progress on that?  I looked again, and yes most commands
> > > are identical.
> 
> Still true.  :)

I came to the same conclusion.

> 
> For that and a boatload of other SPI flash chips.  I'm
> wondering just what that JEDEC standard covers; either
> it's got useful command coverage, or vendors of SPI flash
> noticed that converging command sets is a Good Thing.
> 
> 
> > >  - That driver relies on an AT91-only SPI stack that's not
> > >    going upstream, instead of drivers/spi/atmel_spi.c;
> > > 
> > >  - Driver can't even be selected on kernel.org kernel (!!);
> > > 
> > >  - The m25p80 driver will need at most minor tweaks to
> > >    handle those AT26 (and AT25) series chips;
> > > 
> > > So 340ea370c2ce89d1c15fbf785460f2f74314ce58 should be reverted.
> 
> Also still true...

I agree.

> 
> 
> > > 
> > > - Dave
> > >
> > 
> > Dave,
> > sorry for not responding earlier, I was busy with lots of other 
> > stuff. At the moment, I'm working on this again. I had a bit
> > of trouble getting a recent kernel working on my test system,
> > but that seems to be OK now.
> > 
> > You'll hear from me soon.
> 
> Ping?  Have a look at
> 
>     http://lists.infradead.org/pipermail/linux-mtd/2007-June/018706.html

Great! Didn't notice that one.

> 
> which depends on the resent mutex conversion patch in the
> "previous message" link of that patch.
> 
> That should highlight more reasons why at91_dataflash should
> be reverted ASAP.  I'd expect you would be able to just drop
> that in and have it work ... 

Looks like that.

> unless you're seeing that nCS0 
> issue built out of at91rm9200 errata, in which case there's
> an experiment I can forward to you.  :)

I'll look at that ASAP. I had tons of other work on my desk and last time
I tried, SPI was broken on my device. It would be OK for me if you threw out
that at91_dataflash/at91_dataflash26 stuff now.

Thanks,
Hans