Hi Uli,
On Fri, Dec 9, 2016 at 1:36 PM, Ulrich Hecht
<ulrich.hecht+renesas@xxxxxxxxx> wrote:
> To allow operation with a higher RX FIFO interrupt threshold in PIO
> mode, it is necessary to consider the DR bit ("FIFO not full, but no
> data received for 1.5 frames") as an indicator that data can be read.
> Otherwise the driver will let data rot in the FIFO until the threshold
> is reached.
>
> Signed-off-by: Ulrich Hecht <ulrich.hecht+renesas@xxxxxxxxx>
> --- a/drivers/tty/serial/sh-sci.c
> +++ b/drivers/tty/serial/sh-sci.c
> @@ -677,7 +677,7 @@ static int sci_poll_get_char(struct uart_port *port)
> break;
> } while (1);
>
> - if (!(status & SCxSR_RDxF(port)))
> + if (!(status & (SCxSR_RDxF(port) | SCxSR_DR(port))))
> return NO_POLL_CHAR;
>
> c = serial_port_in(port, SCxRDR);
> --- a/drivers/tty/serial/sh-sci.h
> +++ b/drivers/tty/serial/sh-sci.h
> @@ -156,6 +156,7 @@ enum {
> #define SCxSR_FER(port) (((port)->type == PORT_SCI) ? SCI_FER : SCIF_FER)
> #define SCxSR_PER(port) (((port)->type == PORT_SCI) ? SCI_PER : SCIF_PER)
> #define SCxSR_BRK(port) (((port)->type == PORT_SCI) ? 0x00 : SCIF_BRK)
> +#define SCxSR_DR(port) (((port)->type == PORT_SCI) ? 0x00 : SCIF_DR)
Makes sense, as SCIF_RDxF_CLEAR already includes SCIF_DR, and we thus already
clear both RDF and DR.
However, if you would handle this inside the SCxSR_RDxF() macro, your patch
would reduce to a single line:
-#define SCxSR_RDxF(port) (((port)->type == PORT_SCI) ?
SCI_RDRF : SCIF_RDF)
+#define SCxSR_RDxF(port) (((port)->type == PORT_SCI) ?
SCI_RDRF : SCIF_DR | SCIF_RDF)
Gr{oetje,eeting}s,
Geert
--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@xxxxxxxxxxxxxx
In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds
