Hi Weijie,
On 5 November 2018 at 19:48, Weijie Gao weijie.gao@mediatek.com wrote:
On Mon, 2018-11-05 at 16:37 +0800, Ryder Lee wrote:
On Mon, 2018-11-05 at 10:20 +0800, Ryder Lee wrote:
On Sat, 2018-11-03 at 00:09 -0600, Simon Glass wrote:
Hi Ryder,
On 2 November 2018 at 09:15, Ryder Lee ryder.lee@mediatek.com wrote:
This patch adds an extra operation in ns16550.c to suuport MediaTek SoCs as we have a highspeed register which influences the calcualtion of the divisor.
Note that we don't support the baudrate greater than 115200 currently.
Signed-off-by: Ryder Lee ryder.lee@mediatek.com Tested-by: Matthias Brugger matthias.bgg@gmail.com Reviewed-by: Simon Glass sjg@chromium.org
drivers/serial/ns16550.c | 10 ++++++++++ 1 file changed, 10 insertions(+)
Actually it seems to me that we should have a compatible string for this and deal with this at run-time?
Is that easy to do here?
How about this:
uart0: serial@11002000 { compatible = "ns16550a"; .... mediatek,highspeed = <0>; ....
Ryder
Sorry, I didn't get it right. I will add a compatible string in ns16550_serial_ids[] for MTK chips.
Ryder
Hi Simon,
We have tried the compatible string, but it made the ns16550 driver more complicated.
To use the compatible string we have to add a new field in ns16550_platdata, and change the flow of ns16550_serial_probe. Moreover, it's totally useless for debug uart.
At present using a macro is the easiest way here.
It might be easier but it is not correct. We should not have platform-specific #ifdefs in a common driver.
Here's what I think you should do:
1. Create your own separate driver which calls into the code of this one 2. Add a 'quirks' flag word to the platform data and refactor the code to deal with your quirk, or add a separate field like you say 3. For the debug uart, put it in your separate driver, if you cannot easily change _debug_uart_init(). I suppose we could have a CONFIG_DEBUG_UART_QUIRKS flag wod?
Regards, Simon