On 04/12/2016 08:32 AM, Sebastian Andrzej Siewior wrote:
> On 04/10/2016 07:14 AM, Peter Hurley wrote:
>> --- a/drivers/tty/serial/8250/8250_dma.c
>> +++ b/drivers/tty/serial/8250/8250_dma.c
>> @@ -201,6 +201,15 @@ int serial8250_request_dma(struct uart_8250_port *p)
>> goto release_rx;
>> }
>>
>> + /* 8250 tx dma requires dmaengine driver to support terminate */
>> + ret = dma_get_slave_caps(dma->txchan, &caps);
>> + if (ret)
>> + goto err;
>> + if (!caps.cmd_terminate) {
>> + ret = -EINVAL;
>> + goto err;
>> + }
>
> So you want to be able to terminate the TX transfer instead of waiting
> for its completion because with flow control you could wait for ever?
Yes.
And I don't want to get stuck not being able to fix things that
should already work in the 8250 base operations without major
surgery occasioned by preexisting dma limitations.
Like I wrote in the cover letter, I intend to add XON/XOFF transmit
fix for 8250 dma, and if the "best" way ends up being pause/terminate,
I'll end up adding pause requirement here as well.
"best" in the sense of clean/minimal/effective intersection.
>> +
>> dmaengine_slave_config(dma->txchan, &dma->txconf);
>>
>> /* RX buffer */
>
> Sebastian
>
--
To unsubscribe from this list: send the line "unsubscribe linux-serial" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
