Re: [PATCH 08/11] dmaengine: cppi41: Implement the glue for da8xx

[Alexandre Bailon <abailon-rdvid1DuHRBWk0Htik3J/w@public.gmane.org>]

On 01/10/2017 06:53 PM, Sergei Shtylyov wrote:
> On 01/09/2017 07:06 PM, Alexandre Bailon wrote:
> 
>> The da8xx has a cppi41 dma controller.
> 
>    It's called CPPI 4.1. :-)
> 
>> This is add the glue layer required to make it work on da8xx,
>> as well some changes in driver (e.g to manage clock).
>>
>> Signed-off-by: Alexandre Bailon <abailon-rdvid1DuHRBWk0Htik3J/w@public.gmane.org>
>> ---
>>  drivers/dma/cppi41.c | 95
>> ++++++++++++++++++++++++++++++++++++++++++++++++++++
>>  1 file changed, 95 insertions(+)
>>
>> diff --git a/drivers/dma/cppi41.c b/drivers/dma/cppi41.c
>> index 939398e..4318e53 100644
>> --- a/drivers/dma/cppi41.c
>> +++ b/drivers/dma/cppi41.c
> [...]
>> @@ -86,10 +87,19 @@
>>
>>  #define USBSS_IRQ_PD_COMP    (1 <<  2)
>>
>> +/* USB DA8XX */
>> +#define DA8XX_INTR_SRC_MASKED    0x38
>> +#define DA8XX_END_OF_INTR    0x3c
>> +
>> +#define DA8XX_QMGR_PENDING_MASK    (0xf << 24)
>> +
>> +
>> +
> 
>    One empty line is enough.
> 
>>  /* Packet Descriptor */
>>  #define PD2_ZERO_LENGTH        (1 << 19)
>>
>>  #define AM335X_CPPI41        0
>> +#define DA8XX_CPPI41        1
>>
>>  struct cppi41_channel {
>>      struct dma_chan chan;
> [...]
>> @@ -366,6 +393,26 @@ static irqreturn_t am335x_cppi41_irq(int irq,
>> void *data)
>>      return cppi41_irq(cdd);
>>  }
>>
>> +static irqreturn_t da8xx_cppi41_irq(int irq, void *data)
>> +{
>> +    struct cppi41_dd *cdd = data;
>> +    u32 status;
>> +    u32 usbss_status;
>> +
>> +    status = cppi_readl(cdd->qmgr_mem + QMGR_PEND(0));
>> +    if (status & DA8XX_QMGR_PENDING_MASK)
>> +        cppi41_irq(cdd);
>> +    else
>> +        return IRQ_NONE;
> 
>    Seems correct...
> 
>> +
>> +    /* Re-assert IRQ if there no usb core interrupts pending */
>> +    usbss_status = cppi_readl(cdd->usbss_mem + DA8XX_INTR_SRC_MASKED);
>> +    if (!usbss_status)
>> +        cppi_writel(0, cdd->usbss_mem + DA8XX_END_OF_INTR);
> 
>    I don't understand this...
Well, it might not be necessary anymore.
I had an issue with teardown. After a teardown, USB were not working
anymore.
It was because an interrupt was fired on teardown completion  but
because the completion queue was empty, interrupt was ignored by the
interrupt handler. And in USB driver, because the interrupt was not
fired by USB core, then interrupt was ignored and never re-asserted.

But I guess the change I made in patch 11 should prevent this issue.
> 
>> +
>> +    return IRQ_HANDLED;
>> +}
>> +
>>  static dma_cookie_t cppi41_tx_submit(struct dma_async_tx_descriptor *tx)
>>  {
>>      dma_cookie_t cookie;
> [...]
> 
> MBR, Sergei
> 

--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html