[PATCH] ibm ocp gpio

The most common thing I want to use the GPIO pins for is
simple lighting of some LED's. I decided to roll in some timers
into the generic GPIO driver rather than haveing each user of the
GPIO driver be forced to do some form of timers to get a usable
LED visual as well.

I think I got the timer stuff correct but I am not sure.
I didn't bother cleaning it up until I get some feedback on it.

andrew may wrote:

The most common thing I want to use the GPIO pins for is
simple lighting of some LED's. I decided to roll in some timers
into the generic GPIO driver rather than haveing each user of the
GPIO driver be forced to do some form of timers to get a usable
LED visual as well.

I think I got the timer stuff correct but I am not sure.
I didn't bother cleaning it up until I get some feedback on it.

  -----------------------------------------------------------------------------
                    Name: gpio.patch.gz
   gpio.patch.gz    Type: unspecified type (application/octet-stream)
                Encoding: base64

I don't think we should add this It's not really something that needs to be
done in the driver. Also you have not exported this functionality to
userspace.


--
Kenneth Johansson
Ericsson AB                       Tel: +46 8 404 71 83
Borgafjordsgatan 9                Fax: +46 8 404 72 72
164 80 Stockholm                  kenneth.johansson@etx.ericsson.se

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

On Tue, May 21, 2002 at 11:11:33AM +0200, Kenneth Johansson wrote:

andrew may wrote:

quoted

The most common thing I want to use the GPIO pins for is
simple lighting of some LED's. I decided to roll in some timers
into the generic GPIO driver rather than haveing each user of the
GPIO driver be forced to do some form of timers to get a usable
LED visual as well.

I think I got the timer stuff correct but I am not sure.
I didn't bother cleaning it up until I get some feedback on it.

I don't think we should add this It's not really something that needs to be
done in the driver. Also you have not exported this functionality to
userspace.

Well no it doesn't need to be in the driver, but you then are forced with
having every other driver doing there own timer to turn off the bit rather
than having one timer take care of them all. And do you really want to
have you UserSpace app wake up and do an ioctl just to turn off an LED
a quarter second after you turn it on?

As I said I didn't clean it up. That would include not having it exported
as an ioctl or having the symbol export to modules. Or making it a config
option to compile it out if you are worried about the size of the code.
Or removing the extra printk's.

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

On Tue, 2002-05-21 at 18:08, andrew may wrote:

Well no it doesn't need to be in the driver, but you then are forced with
having every other driver doing there own timer to turn off the bit rather
than having one timer take care of them all. And do you really want to

Well I don't think that would be a problem and if we are talking led's
its not going to happen more than a few time a second you have to see
the thing flash. I don't want more code in the kernel than absolutely
necessary all it takes is one bug.

If you really want to make something that can't be done other than in
the driver it's some type of streaming mode and then you have to have
timers to control the speed. Think any card that needs to program sram
based FPGA it would probably go many times faster than todays interface.



--
Kenneth Johansson
Ericsson AB                       Tel: +46 8 404 71 83
Borgafjordsgatan 9                Fax: +46 8 404 72 72
164 80 Stockholm                  kenneth.johansson@etx.ericsson.se


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

On Tue, May 21, 2002 at 06:28:41PM +0200, Kenneth Johansson wrote:

On Tue, 2002-05-21 at 18:08, andrew may wrote:

quoted

Well no it doesn't need to be in the driver, but you then are forced with
having every other driver doing there own timer to turn off the bit rather
than having one timer take care of them all. And do you really want to

Well I don't think that would be a problem and if we are talking led's
its not going to happen more than a few time a second you have to see
the thing flash. I don't want more code in the kernel than absolutely
necessary all it takes is one bug.

If you really want to make something that can't be done other than in
the driver it's some type of streaming mode and then you have to have
timers to control the speed. Think any card that needs to program sram
based FPGA it would probably go many times faster than todays interface.

Look at the code and stop the hand-waving generalities.

The other motto of the kernel is do it once right. If this code is not
there every thing that just wants to turn on a LED for a half second needs
to do their own timer. I have a couple drivers where I use some LED's on the
GPIO pins and I don't want to do a timer in each one of them.

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

On Tue, 2002-05-21 at 20:26, andrew may wrote:

On Tue, May 21, 2002 at 06:28:41PM +0200, Kenneth Johansson wrote:

quoted

On Tue, 2002-05-21 at 18:08, andrew may wrote:

quoted

Well no it doesn't need to be in the driver, but you then are forced with
having every other driver doing there own timer to turn off the bit rather
than having one timer take care of them all. And do you really want to

Well I don't think that would be a problem and if we are talking led's
its not going to happen more than a few time a second you have to see
the thing flash. I don't want more code in the kernel than absolutely
necessary all it takes is one bug.

If you really want to make something that can't be done other than in
the driver it's some type of streaming mode and then you have to have
timers to control the speed. Think any card that needs to program sram
based FPGA it would probably go many times faster than todays interface.

Look at the code and stop the hand-waving generalities.

relax I did not say your code was buggy it was more .. well general
hand-waving :)

The other motto of the kernel is do it once right. If this code is not
there every thing that just wants to turn on a LED for a half second needs
to do their own timer. I have a couple drivers where I use some LED's on the
GPIO pins and I don't want to do a timer in each one of them.

I do not see a problem with everyone doing the timers but it's not me
you have to convince I can't commit the code anyway.

It would be better if you made a more generic interface. perhaps
something that took commands in the form of.

struct gpio_command{
	int action;
	int value;
	int ns_wait;
	}

action would code in what register to use and if you want to set or
clear the bits specified in value.

Now you could do a mode that loop over an array of this type of
commands.

This would have the "side effect" of speeding up large transfers over
SPI bus or similar things and help you with flashing your led.


--
Kenneth Johansson
Ericsson AB                       Tel: +46 8 404 71 83
Borgafjordsgatan 9                Fax: +46 8 404 72 72
164 80 Stockholm                  kenneth.johansson@etx.ericsson.se


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/