[9fans] eia overruns at 115k baud...

[9fans] eia overruns at 115k baud...

[steve.simon]

Hi,

This is not really surprising but I get serial port overruns
on /dev/eia1 when reading at 115k baud (450Mhz PII).

Unfortunately I need this rate to talk to the bootstrap the
embedded systems we use here, and big xilinxes have BIG image files.

The serial interface does not use any handshaking, though it is a stop and
wait protocol and the packest are only 512 bytes, so that size of RX buffer
should be sufficent.

Just before I start hacking, anyone know of any inherent limitations
in the which might scupper me? I guess I need 1ms interrupt latency
(16 byte FIFOs), is this too much to expect from a multitasking kernel?

-Steve

Re: [9fans] eia overruns at 115k baud...

[David Presotto]

[-- Attachment #1: Type: text/plain, Size: 123 bytes --]

Shouldn't be, I use 115k serial ports all the time.  Sure the fifo's
are on?  Sure they are at 16?  Look at the uart code.

[-- Attachment #2: Type: message/rfc822, Size: 2295 bytes --]

From: steve.simon@snellwilcox.com
To: 9fans@cse.psu.edu
Subject: [9fans] eia overruns at 115k baud...
Date: Thu, 18 Sep 2003 11:05:44 +0100
Message-ID: <948281b86b66dede835fe0b44ba0c92b@yourdomain.dom>

Hi,

This is not really surprising but I get serial port overruns
on /dev/eia1 when reading at 115k baud (450Mhz PII).

Unfortunately I need this rate to talk to the bootstrap the
embedded systems we use here, and big xilinxes have BIG image files.

The serial interface does not use any handshaking, though it is a stop and
wait protocol and the packest are only 512 bytes, so that size of RX buffer
should be sufficent.

Just before I start hacking, anyone know of any inherent limitations
in the which might scupper me? I guess I need 1ms interrupt latency
(16 byte FIFOs), is this too much to expect from a multitasking kernel?

-Steve

Re: [9fans] eia overruns at 115k baud...

[Scott Schwartz]

| Shouldn't be, I use 115k serial ports all the time.  Sure the fifo's
| are on?  Sure they are at 16?  Look at the uart code.

We talked about this a few months ago... at that time the uart was
definately getting overruns, but we decided it was ok because the network
layer retried the few lost packets automatically.

Re: [9fans] eia overruns at 115k baud...

[jmk]

On Thu Sep 18 11:26:34 EDT 2003, schwartz@bio.cse.psu.edu wrote:
> | Shouldn't be, I use 115k serial ports all the time.  Sure the fifo's
> | are on?  Sure they are at 16?  Look at the uart code.
>
> We talked about this a few months ago... at that time the uart was
> definately getting overruns, but we decided it was ok because the network
> layer retried the few lost packets automatically.

I don't remember that being the conclusion at all.