Thanks Greg for giving me the push I needed to migrate to 2.6.  I am doing
much better with the airprime driver here.

I have found a number of references to the buffer size patch in the
archives.  Is it the general consensus that this is a bad thing?

I have seen a couple of references to frame size versus buffer size and also
your comments regarding flow control.

I have also seen a couple of discussions elsewhere relating this buffer
overflow to the use of compression.

Is this buffer size issue related to the V4.2 frame size?  I am working with
the KPC650 and Novatel 620 (MSM6500 and MSM5500 respectively) and I see that
they have configurable compression frame size (max_dict?) settings.  Does it
make sense to match the driver buffer size to the v.42 frame size to
optimize performance and/or prevent buffer overflow?

I've been searching the archives for a couple of days now and some of the
related subjects go back a year or more, so please let me know if this horse
has been sufficiently beaten already.

Thanks.

-----Original Message-----
From: Greg KH [mailto:[EMAIL PROTECTED] 
Sent: Thursday, May 04, 2006 4:07 PM
To: Thomas L. Marshall
Cc: [email protected]
Subject: Re: [linux-usb-devel] PC Card cellular modems, EVDO, 1xRTT etc and
usbserial

On Thu, May 04, 2006 at 07:51:32AM -0700, Thomas L. Marshall wrote:
> Sorry if this is a duplicate, I did not see this make it onto the list.
> 
> Greg KH and anyone else that can help,
> 
> I am trying to better understand the details with these cellular data
cards
> and have found a number of resources but still have some questions as
> follows:
> 
> Are the majority of these cellular/evdo data cards (kyocera 650, novatel
> 620, etc) ultimately USB serial devices?  In other words, I believe that
> they contain USB controllers that sit on top of the PCI bus via the PCMCIA
> adapter.

Yes, it seems that they are just dumb USB pipes :)

> Given that, I have seen the general instruction to use modprobe providing
> the vendor and device ids to recognize the devices.  Additionally, I have
> seen two implementations contributed by Greg KH (tiny serial driver from
LJ
> and another possibly more obscure airprime or airframe specific driver
> basically a specification of the tiny driver).  These two examples contain
> the vendor and device ids so that the user is not required to enter them
> directly.

Yes, it's better not to use the generic usb-serial driver for something
that you care about data rates and line speed changes.

> My next question is if this driver can then be linked to a non-loadable
> module kernel and then recognize the device at boot without the need to
use
> modprobe, etc?

The generic driver works just fine built into the kernel, just use the
modprobe options on the command line like the kernel command line
documentation states to do so.

> Secondly, these sample drivers are for 2.6 and I have tried to use the
same
> approach in 2.4.32.

Ah, sorry, you're on your own here.  2.4 is in deep maintenance mode
with no real new changes going into it at all.  I really recommend you
move to a 2.6 based system.

good luck,

greg k-h



-------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
_______________________________________________
[email protected]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel

Reply via email to