[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[bluetooth-dev] Ericsson RF Band Seems to Fail




Hello all:

I am using the November 2000 bt stack on a NET+ARM (40Mhz) processor.  I
keep running into problems with my Ericsson development boards from Comtec
Sigma (http://www.comtec.sigma.se/).

A successfull HCI connection is made and reported up to the application
layer (/proc/bt_status reports one connection, with the correct BD_ADDR, as
ACTIVE on both machines...).  But it would seem that the RF communcation
unceremoniously dies:

*	Subsequent L2CAP communication fails by timing out
*	An INQUIRY command from BTD returns AMAZINGLY fast, without any
reported BDs (INQUIRY worked before we established the connection)
*	We have used a serial line analyzer to inspect the
computer<->bluetooth module communication, all seems to be as it should be:
*	The server machine (I am using vocabularly from
l2cap.c:process_event()) waits for an connect_ind() packet
*	The client machine keeps sending connect_req()'s, and in fact the
bluetooth module still communicates HCI NCP data correctly, without any
apparent malfunction
Resetting the bluetooth module simply means we have to reestablish the
communication, and the vicious cycle starts again...

I have tested the software (previously) between two desktop machines, and it
has worked OK
The platform where these errors are occurring is a NET+ARM 40MHz embedded
development board (32MRAM, Linux 2.0.38, RS232 interface to bluetooth
modules)

I have turned the necessary debugging messges in l2cap.c and hci.c.  The HCI
interface does propagate two suspicious messages:  either that the
REMOTE_NAME query has timed out (LMP timeout) or that there has been an LMP
Page Timeout (I do not know what this means...).  My understanding is that
the LMP layer actually resides on the ericsson chipset as firmware; does
anyone one know if:
*	There is a way to alter the timeout metric via HCI
*	If the ericsson chipsets decide to ignore BDs that fail a
REMOTE_NAME query or other LMP message reponse (this would be long shot, but
it would account for all the symptoms)

The biggest difference between the two platforms is easily speed (40MHz vs
733MHz!); does anyone have experience with the Axis stack on slower
processors?

Any suggestions would be appreciated...

As a final word (if you get this far...):  kudos to the developers
responsible for the linux bluetooth stack.  For all the sifting I have done
through the code, at best I have found small copy&paste typo errors in debug
messages; although it is a little less reliable on desktops, it certainly is
adequate for getting development work done.


/*  EMAIL_BEGIN_SIGNATURE ***********************************/
Keith E. Hellman                             khellman@xxxxxxx.com
Software Engineer                    Voice: 303.530.8288 x3106
Penguin Development Team                   



-
To unsubscribe from this list: send the line "unsubscribe bluetooth-dev" in
the body of a message to majordomo@xxxxxxx.com