Re: [PATCH 2/5] altera_uart: Add support for polling mode(IRQ-less)

From: Tobias Klauser
Date: Tue Sep 28 2010 - 11:45:46 EST


On 2010-09-28 at 15:35:26 +0200, Anton Vorontsov <cbouatmailru@xxxxxxxxx> wrote:
> Some Altera UART implementations doesn't route the IRQ line, so we have
> to work in polling mode.
>
> Signed-off-by: Anton Vorontsov <cbouatmailru@xxxxxxxxx>
> ---
> drivers/serial/altera_uart.c | 38 +++++++++++++++++++++++++++++---------
> 1 files changed, 29 insertions(+), 9 deletions(-)
>
> diff --git a/drivers/serial/altera_uart.c b/drivers/serial/altera_uart.c
> index f8d8a00..2603f59 100644
> --- a/drivers/serial/altera_uart.c
> +++ b/drivers/serial/altera_uart.c
> @@ -15,6 +15,7 @@
>
> #include <linux/kernel.h>
> #include <linux/init.h>
> +#include <linux/timer.h>
> #include <linux/interrupt.h>
> #include <linux/module.h>
> #include <linux/console.h>
> @@ -76,6 +77,7 @@
> */
> struct altera_uart {
> struct uart_port port;
> + struct timer_list tmr;
> unsigned int sigs; /* Local copy of line sigs */
> unsigned short imr; /* Local IMR mirror */
> };
> @@ -168,6 +170,7 @@ static void altera_uart_set_termios(struct uart_port *port,
> tty_termios_encode_baud_rate(termios, baud, baud);
>
> spin_lock_irqsave(&port->lock, flags);
> + uart_update_timeout(port, termios->c_cflag, baud);
> writel(baudclk, port->membase + ALTERA_UART_DIVISOR_REG);
> spin_unlock_irqrestore(&port->lock, flags);
> }
> @@ -268,6 +271,15 @@ static irqreturn_t altera_uart_interrupt(int irq, void *data)
> return IRQ_RETVAL(isr);
> }
>
> +static void altera_uart_timer(unsigned long data)
> +{
> + struct uart_port *port = (void *)data;
> + struct altera_uart *pp = container_of(port, struct altera_uart, port);
> +
> + altera_uart_interrupt(0, port);
> + mod_timer(&pp->tmr, jiffies + uart_poll_timeout(port));
> +}
> +
> static void altera_uart_config_port(struct uart_port *port, int flags)
> {
> port->type = PORT_ALTERA_UART;
> @@ -284,14 +296,6 @@ static int altera_uart_startup(struct uart_port *port)
> unsigned long flags;
> int ret;
>
> - ret = request_irq(port->irq, altera_uart_interrupt, IRQF_DISABLED,
> - DRV_NAME, port);
> - if (ret) {
> - pr_err(DRV_NAME ": unable to attach Altera UART %d "
> - "interrupt vector=%d\n", port->line, port->irq);
> - return ret;
> - }
> -
> spin_lock_irqsave(&port->lock, flags);
>
> /* Enable RX interrupts now */
> @@ -300,6 +304,19 @@ static int altera_uart_startup(struct uart_port *port)
>
> spin_unlock_irqrestore(&port->lock, flags);

Shouldn't we only enable the RX interrupt if we actually use interrupts?
Also I would retain the order here (first register the interrupt handler
and then enable the interrupts).

Cheers,
Tobias
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/