Re: [Gumstix-users] Recent pxa2xx-spi updates, was: Gumsitx policy on contributions to the Linux kernel tree

[Ned F. <nfo...@wh...>]

Harry J Mason wrote:
> On Tue, 26 Feb 2008, Ned Forrester wrote:
> 
>> Unfortunately for me, the pxa2xx-spi driver already has a significant update
>> in 2.6.25, so I will have to wait a while longer to base my version on the
>> mainline version.
> 
> Do you know if any of the 2.6.25 fixes would affect a DMA underrun I have
> been experiencing (interfacing to an MMC card)?

No.  The large recent changes are structural changes that adapt the
driver to changes going on elsewhere in the kernel.  There are no recent
functional changes other than a small bug fix relating to the sequence
of setting modes and chip select.

How do you suppose you are generating a TX FIFO underrun when using DMA?
 The entire transmit buffer must exist in memory before starting the
DMA, so this could only happen if the DMA does not service the TX FIFO
in time.  I expect that you may not have set the burst size, and
thresholds optimally.  If you post your clock rate, burst, threshold and
other parameters, I will see if I see anything wrong.

I assume you are using the version of pxa2xx_spi.c that is distributed
with 2.6.20 or later.  Earlier versions certainly might have this problem.

-- 
Ned Forrester                                       nfo...@wh...
Oceanographic Systems Lab                                  508-289-2226
Applied Ocean Physics and Engineering Dept.
Woods Hole Oceanographic Institution          Woods Hole, MA 02543, USA
http://www.whoi.edu/sbl/liteSite.do?litesiteid=7212
http://www.whoi.edu/hpb/Site.do?id=1532
http://www.whoi.edu/page.do?pid=10079