[PATCH v2] ARM: dts: imx6dl: disable dma support for spi on i.mx6dl

Alexander Holler holler at ahsoftware.de
Tue Sep 16 02:41:55 PDT 2014


Am 16.09.2014 05:52, schrieb Robin Gong:
> On Mon, Sep 15, 2014 at 01:50:02PM +0200, Alexander Holler wrote:
>> Am 10.09.2014 07:30, schrieb Robin Gong:
>>> There is one weird data in rxfifo after one full rx/tx transfer
>>> done sometimes. It looks a design issue and hard to workaround
>>> totally, so disable dma functhion here. And will re-enable it
>>> once the root cause found.
>>
>> Hmm, I experience problems with DMA too but on uart3. I'm using the same
>> workaround for the uart (I've just commented out the dma entries in the DT).
>> The problem manifests itself here such, that brcm_patchram_plus
>> hangs while uploading the firmware to a BCM4330 connected at uart3
>> (reproducible).
>>
>> So maybe there is a bug in the DMA-engine which not only effects
>> SPI. Or both drivers contain the same error in handling DMA (maybe
>> through c&p).
>> But that's just specualtion from me, I haven't looked further into
>> that problem.
>>
>> Regards,
>>
>> Alexander Holler
> Thanks for your information share. But my issue should be caused by hardware,
> since everything is ok if it runs on other i.mx6 chip. Is your board also based
> on i.mx6 chip? If yes, hope you can raise your issue in freescale community or
> contact with Andy whose mail address added in CC list fugang.duan at freescale.cm.
> We have fix some bugs in UART DMA case.

It's an i.mx6q (Wandboard quad c1) where I have this problem with 
mainline and much older (but heavily patched freescale 3.10.x based) 
kernels. A quick web-search suggests that this problem exists since a 
long time (noticed mainly by people which try to use BT as this seems to 
be the major use case for high speed serial communication).

Regards,

Alexander Holler



More information about the linux-arm-kernel mailing list