Re: [PATCH RFC 0/11] usbserial: Serial Core port.

[Pete Zaitcev <zaitcev@redhat.com>]

On Sun, 4 Jun 2006 20:12:23 -0300, "Luiz Fernando N. Capitulino" <lcapitulino@mandriva.com.br> wrote:

> | I understand. My intent was different, however. One of the bigger sticking
> | points for usb-serial was its interaction with line disciplines, which are
> | notorious for looping back and requesting writes from callbacks
> | (e.g. h_hdlc.c). They are also sensitive to drivers lying about the
> | amount of free space in their FIFOs. This is something you never test
> | when driving a serial port from an application, no matter how cleverly
> | written.

>   In all the tests the modem was configured to answer the calls, and the
> cell phone was configured to dial to the modem (my home's number).

This is exactly backwards, and so it tests different code paths.
The line discipline is involved into driving a cooked mode port,
e.g. the one where getty is.

Running uploads and downloads with things like xmodem is a good
test of hardware flow control, so someone will have to do it too.

>  Unfortunatally this is a very expensive test environment, and I can't use
> it for development. The best one would be to have a USB<->DB9 cable..

PL-2303 already has a DB-9, you actually you need a DB-9-to-DB-9
Null Modem (cross-over) cable.

Anyway, I do not expect pl2303 failing this test, mind. It's more
of a problem for simpler devices.

-- Pete