serial8250: too much work for irq73

serial8250: too much work for irq73

[Robert Wörle]

Hi List

I am running into issues , when using ttyS1 on a heavy loaded 
application. It trys to send and recieve around 60bytes every 100ms with 
57600baud.  There is a PIC Microcontroller attached on the otherside 
which immidiatly returns the data when requested.

After some hours , the kernel starts to flood with the above messages 
and the system gets into an unusable state.

i am running 2.6.35.9 ( released from isee.com ) on their Igep Module ( 
arch igep0030).

What would you guys do ? Anything i can provide you to help here ? Its 
really a major issue for us and showstopping an important release.

Thanks in advance for any help or suggestion

Regards Rob

Re: serial8250: too much work for irq73

[Mathias Nyman]

ext Robert Wörle wrote:
> Hi List
> 
> I am running into issues , when using ttyS1 on a heavy loaded
> application. It trys to send and recieve around 60bytes every 100ms with
> 57600baud.  There is a PIC Microcontroller attached on the otherside
> which immidiatly returns the data when requested.
> 
> After some hours , the kernel starts to flood with the above messages
> and the system gets into an unusable state.
> 
> i am running 2.6.35.9 ( released from isee.com ) on their Igep Module (
> arch igep0030).
> 
> What would you guys do ? Anything i can provide you to help here ? Its
> really a major issue for us and showstopping an important release.
> 
> Thanks in advance for any help or suggestion
> 

Not sure if this is the same cause, but here's some code that fixed a similar issue:

Serial: 8250.c: Clear FIFOs before enabling them

There's a short time when FIFO's are disabled and interrupts enabled
between startup and set_termios.
If rx FIFO is not cleared before enabled, and there are lots of serial I/O,
an interrupt indicating FIFO data error can occur.
(LSR register bit 7 set, no other error bits set).

The current interrupt handle is unable to clear this interrupt and will continue looping with a
flood of "too much work" messages.

This is seen on an omap board with ST16654 type uart if a key is kept pressed in serial console while
booting.

Signed-off-by: Mathias Nyman <mathias.nyman@nokia.com>
---
 drivers/serial/8250.c |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)

diff --git a/drivers/serial/8250.c b/drivers/serial/8250.c
index 5ed1b82..d3add25 100644
--- a/drivers/serial/8250.c
+++ b/drivers/serial/8250.c
@@ -2403,6 +2403,8 @@ serial8250_set_termios(struct uart_port *port, struct ktermios *termios,
                if (fcr & UART_FCR_ENABLE_FIFO) {
                        /* emulated UARTs (Lucent Venus 167x) need two steps */
                        serial_outp(up, UART_FCR, UART_FCR_ENABLE_FIFO);
+                       serial_outp(up, UART_FCR, UART_FCR_ENABLE_FIFO |
+                                   UART_FCR_CLEAR_RCVR | UART_FCR_CLEAR_XMIT);
                }
                serial_outp(up, UART_FCR, fcr);         /* set fcr */
        }
-- 





--
To unsubscribe from this list: send the line "unsubscribe linux-omap" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html