[PATCH v2] ARM: pxa: PalmZ72: Add OV9640 camera support

[marek.vasut@gmail.com (Marek Vasut)]

On Saturday 15 January 2011 18:49:16 Bj?rn Forsman wrote:
> 2011/1/15 Russell King - ARM Linux <linux@arm.linux.org.uk>:
> > On Sat, Jan 15, 2011 at 05:37:08PM +0100, Bj?rn Forsman wrote:
> >> Hi,
> >> 
> >> On 15 January 2011 02:17, Marek Vasut <marek.vasut@gmail.com> wrote:
> >> > Rework of patch from 2009:
> >> > PalmZ72: Add support for OV9640 camera sensor
> >> > 
> >> > Signed-off-by: Marek Vasut <marek.vasut@gmail.com>
> >> > ---
> >> > v2: Use gpio_request_array()/gpio_free_array() array
> >> > 
> >> >  arch/arm/mach-pxa/include/mach/palmz72.h |    5 +
> >> >  arch/arm/mach-pxa/palmz72.c              |  127
> >> > ++++++++++++++++++++++++++++++ 2 files changed, 132 insertions(+), 0
> >> > deletions(-)
> >> > 
> >> > diff --git a/arch/arm/mach-pxa/include/mach/palmz72.h
> >> > b/arch/arm/mach-pxa/include/mach/palmz72.h index 2bbcf70..0d4700a
> >> > 100644
> >> > --- a/arch/arm/mach-pxa/include/mach/palmz72.h
> >> > +++ b/arch/arm/mach-pxa/include/mach/palmz72.h
> >> 
> >> [snip]
> >> 
> >> > +static int palmz72_camera_power(struct device *dev, int power)
> >> > +{
> >> > +       gpio_set_value(GPIO_NR_PALMZ72_CAM_PWDN, !power);
> >> > +       mdelay(50);
> >> > +       return 0;
> >> > +}
> >> > +
> >> > +static int palmz72_camera_reset(struct device *dev)
> >> > +{
> >> > +       gpio_set_value(GPIO_NR_PALMZ72_CAM_RESET, 1);
> >> > +       mdelay(50);
> >> > +       gpio_set_value(GPIO_NR_PALMZ72_CAM_RESET, 0);
> >> > +       mdelay(50);
> >> > +       return 0;
> >> > +}
> >> 
> >> (Sorry if these are stupid newbie-questions.)
> >> 
> >> Is the entire kernel blocked during the above mdelay()s or
> >> just some kernel thread? (Or: will this add like 100ms or 150ms
> >> to boot time?)
> >> 
> >> Can mdelay() be replaced with something non-blocking?
> > 
> > mdelay() is a spinning delay - it doesn't voluntarily relinquish the
> > CPU to another task.
> > 
> > If you have preempt enabled and the region is preemptable, you can be
> > preempted in the middle of the delay (in which case it becomes delay
> > time + time away from the thread.)
> > 
> > msleep() can be used for places where you can sleep, but is limited to
> > the jiffy granularity for !HIRES kernels.
> 
> Thanks for explaining!
> 
> Marek, care to try msleep() so we avoid busy-waits?

Let's keep it at mdelay() for now, can we? I'll send a subsequent patch once I 
can test with msleep() ... I don't have the hardware available now.

Cheers

> 
> Best regards,
> Bj?rn Forsman