[ corrected Nicolas and Cyrille addresses ]
2018-02-22 5:50 GMT+01:00 Sergey Organov <sorganov@xxxxxxxxx>:
> Hi Richard,
>
> Richard Genoud <richard.genoud@xxxxxxxxx> writes:
>> Hi Sergey,
>>
>> In the commit 9bcffe7575b7 ("tty/serial: at91: fix hardware handshake
>> on Atmel platforms"),
>> it's said that the RTS/CTS handshake *driven by the controller* is
>> broken on sama5d3 platforms.
>>
>> But, it works perfectly fine if it is driven by Linux.
>
> Yes, I thought I made it clear I only consider hardware handshake case.
>
> As a side-note, in my experience software handshake rarely works
> perfectly in practice, due to its inherent limitations. It works
> expected, yes, but "perfectly fine" sounds too optimistic to me.
:) I have to agree with you.
Having RTS/CTS controlled by hardware with FIFOs is the only way to
have a reliable communication at high speed (say 3Mbit/s for example)
>
>> That said, I've effectively seen the RTS breakage on sama5d3. (it's
>> not random, it's 100% reproductible).
>
> That contradicts my experience though. I believe I just effectively
> commented-out the
>
> if (atmel_use_dma_rx(port) &&
> !mctrl_gpio_to_gpiod(atmel_port->gpios, UART_GPIO_CTS)) {
I guess you mean:
if (atmel_use_fifo(port) &&
!mctrl_gpio_to_gpiod(atmel_port->gpios, UART_GPIO_CTS)) {
>
> part of the code to always use exactly hardware handshake, and it does
> work.
>
> Looks like I need to further research the issue then and make sure I do see
> what I think I see.
Yes, please. I would be quite surprised.
I've just tried it again on a sam9g35 (it's the same usart as the sama5d3).
In the driver:
- if (atmel_use_fifo(port) &&
- !mctrl_gpio_to_gpiod(atmel_port->gpios, UART_GPIO_CTS)) {
+ if (!mctrl_gpio_to_gpiod(atmel_port->gpios, UART_GPIO_CTS)) {
Used a null modem cable between ttyS1/ttyS2.
for i in 1 2 ; do stty -F /dev/ttyS$i -crtscts ; done
one one terminal:
cat /dev/ttyS2
on the other :
echo "hello" > /dev/ttyS1
=> it works
for i in 1 2 ; do stty -F /dev/ttyS$i crtscts ; done
one one terminal:
cat /dev/ttyS2
on the other :
echo "hello" > /dev/ttyS1
=> nothing is received/send
>
>> I did not try to run atmel's bare metal code, but IMHO, the
>> usart_hw_handshaking example should fail.
>
> I didn't either, but I did run Linux driver with hardware handshake
> re-enabled as I've said above, and it seems to be working on my
> sama5d3-based board.
>
> -- Sergey
Richard.
--
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
