2016-09-30 13:54 GMT+02:00 Alexandre Belloni
[off-list ref]:

On 30/09/2016 at 13:45:47 +0200, Richard Genoud wrote :

quoted

2016-09-30 13:16 GMT+02:00 Alexandre Belloni
[off-list ref]:

quoted

On 30/09/2016 at 13:04:28 +0200, Richard Genoud wrote :

quoted

Anyway, the problematics setups are all the setups with USMODE_HWHS
enabled on platform without Fifos or PDC,
i.e. all platforms but sama5d2 (Cyrille, correct me if I'm wrong).

This is a wrong assumption, at91rm9200 to at91sam9g45 all have a pdc.
Please, don't break those platforms.

The only affected platforms are sam9x5, sama5d3 and sama5d4.

Have you tested them ?

And why are you saying that rm9200 and g45 will be broken with this patch ?
If they have a pdc, they will fall in the case:
(atmel_use_pdc_rx(port) || atmel_use_fifo(port))
and thus use:
mode |= ATMEL_US_USMODE_HWHS;

won't they ?

Well, this patch wouldn't have my ack if it was breaking them.

It's not.
Read the code.

However, I'm still not sure about 3/3.

As I said 2 weeks from now ( https://lkml.org/lkml/2016/9/14/263 ) the
only case where
we would want to drop the CRTSCTS flag is when there's no pin muxed
for CTS/RTS nor GPIOs.
Fell free to give a patch for that.

To be clear : this patch (2/3) unbreaks ALL platforms with GPIOs as CTS/RTS
and 3/3 unbreaks sam9x5 / sama5d3 / sama5d4 platform with CTS/RTS as !GPIOs.