lkml.org 
[lkml]   [2003]   [Aug]   [24]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PM] Patrick: which part of "maintainer" and "peer review" needs explaining to you?
Ok, another issue.

UCB1x00 devices. I now have two implementations of drivers for these
devices, one trying to use the device/driver model and failing badly
with only half the functionality supported. The other one is based
around one class, one class device, and several class interfaces.

What we have for the class model is:

MCP driver (on platform bus) ----> UCB1x00 device
UCB1x00 core driver ---> UCB1x00 class device contained within core driver
data structure

Then we have the UCB1x00 touchscreen, audio, random platform specific
GPIO drivers (0 to 13), random ADC drivers (0 to 4) which are class
interfaces. The GPIO signals get used as interrupts, key inputs
(so would be an input device driver), another (slow) control bus,
or random device control.

The advantage with the class implementation is that you can register
as many class interfaces as you like against one class device - so
the UCB1x00 driver itself doesn't have to worry about the platform
specific details. Those can be handled by a separate per-platform
"interface".

The class implementation seems to be better, but I'm hitting a snag -
power management. The touchscreen and audio needs suspend/resume
support to work correctly (so DMA can be shut down, and touchscreen
threads woken up to ensure we aren't missing an interrupt.)

Unfortunately, the class model doesn't provide this functionality.

The device model does, but that then gives us a problem - how to handle
all the platform specific implementation details (ie, gpio and adc
drivers) sanely. We can't create 17 struct devices just to let
platform specific drivers bind to what they need - that'd be insane
in terms of the overhead (struct bus_type + n * struct device
+ m * struct device_driver) but it does get us power management.

And one big UCB1x00 driver would be stuffed full of ifdefs, with
sound, input, and other random subsystem drivers all rolled into one.

--
Russell King (rmk@arm.linux.org.uk) The developer of ARM Linux
http://www.arm.linux.org.uk/personal/aboutme.html

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:48    [from the cache]
©2003-2014 Jasper Spaans. Advertise on this site