Re: Reason ofono flakeyness

[Pavel Machek <pavel@ucw.cz>]

[-- Attachment #1: Type: text/plain, Size: 1784 bytes --]

Hi!

> > > OK I think I found the reason fo ofono doing a variable
> > > number of commands on start-up with sudo test/enable-modem.
> > > 
> > > There seems to be some kind of issue with handling of
> > > commands and results in ofono. See the logs below for
> > > test/enable-modem. The AT+SCRN=0 is sent and then AT+CFUN=1
> > > without waiting for results from the previous command :)
> > > 
> > > Is there maybe a missing sync somewhere after writes?
> > 
> > No idea. I just copied the code from other drivers and it seemed to
> > work for me.
> 
> Yeah presumably that works for serial ports.

Still not sure what is going on there. Part of the problem seems to be
ofono expects serial protocol, and will happily split & batch things
together, but motmdm is really packet-based.

Few questions:

drivers/mfd/motorola-mdm.c:

        /* Prepend Motorola custom packet numbering */
        snprintf(cmd, cmdlen, "U%04i%s\r", mot_dlci->id, buf);

We already have \r or \n from the user. Will this add extra one?

I guess this will not work with embedded zero bytes...

Plus it seems kernel auto-splits stream from user into packets...

motmdm_cdev_write:

static ssize_t motmdm_cdev_write(struct file *file, const char __user
*buf,

...should it work byte-by-byte? This way if userland sends
AT+FOO=1\rAT+BAR=2\r, it will be sent as a single packet, which is
probably wrong thing to do.

...should it also treat ^Z as an end of packet? In case of SMS
sending, AT+GCGMS= should go in one packet and hex data in another
one, if I understand things correctly....

Best regards,
									Pavel

-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 181 bytes --]