Re: [patch 05/14] mfd: PCAP2 driver

[David Brownell <david-b-yBeKhBN/0LDR7s880joybQ@public.gmane.org>]

On Saturday 22 November 2008, Daniel Ribeiro wrote:
> >  - You should make ezx_pcap_write() and ezx_pcap_read() switch
> >    over to spi_write_then_read(), to ensure it's never doing
> >    DMA to/from the stack.  I see a byte-order dependency too...
> 
> I tried spi_write_then_read before, but it didn't work. I supposed it
> was because it was doing 2 transfers as the second transfer rx_buf
> always came zeroed.

You may have been using it wrong.  Also, verify against
current kernels ... there have been bugfixes to pxa2xx_spi.


> I see that commit 
> f9b90e39cbc5c4d6ef60022fd1f25d541df0aad1 changed it to do a single
> transfer, so i will try it again.

Unless it needs the full duplex capability of SPI, then
you should be able to make write_then_read work ... on
the other hand, if you do need full duplex, then you
should address the DMA.  (And byteswapping, either way.)


> >  - If you're going to mark the probe() as __devinit, then mark
> >    the remove() as __devexit and use __devexit_p() in the driver
> >    struct.
> 
> Ok. Shouldn't i use __init instead?

I wouldn't.  There are ways that the probe() could be called
more than once -- e.g. after rebinding either of the relevant
SPI drivers through sysfs, or after rmmod/modprobe of the SPI
master controller driver -- and using an __init annotation
would make that point to memory that's unlikely to still hold
that code.

- Dave

-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/