On Tuesday, May 31, 2016 07:20 PM, Vladimir Zapolskiy wrote:
On 31.05.2016 12:40, Chris Ruehl wrote:
Hi,
as imx6 using multiple burst to send data to spi slaves and drop the chipselect
between the words by default my sensor NXP MPL115A1 wasn't working.
The hint comes from a discussion in the Freescale forum from 2013 where Jeff
Coffman posted his solution for a 3.x kernel.
I'd pick-up the idea behind and develop something which works "so far" with
4.6 and linux-next
Up front - I'm not happy using the xfer->cs_change to get set the single burst
I more likely want add a new xfer bit which allow to dedicated set a single burst.
To replace todays:
xfer[0].cs_change = 0;
with a
xfer[0].singleburst = 1;
An other issue with is not yet solved; when I have a odd number of bytes (8 bpw)
in a transfer, its result in 3 bytes eaten on the start and 0x00 added on the
tail -
Do you use PIO or DMA transfer mode? The problem description resembles
a known issue within SDMA firmware (referenced as TKT238285 or ERR008517)
and you should not be able to face it, if SPI controller operates in PIO
mode.
Board is a imx6d and its runs in DMA mode. I have the MPL115A working properly
when I keep an eye on even numbers of bytes in a transfer (which is always the
case in a half duplex mode <cmd><00>..)
But any how thanks for this interesting info!
Chris
--
To unsubscribe from this list: send the line "unsubscribe linux-spi" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
