Difference between revisions of "Category:How to - bluetooth"
m |
m |
||
| Line 312: | Line 312: | ||
Tomas Targownik | Tomas Targownik | ||
| − | |||
| − | |||
| − | |||
| − | |||
| − | |||
[[Category:How_to_-_general]] | [[Category:How_to_-_general]] | ||
Revision as of 13:51, 15 October 2009
Contents
Essentials
Install the basic bluetooth stuff with
ipkg install bluez-utils
First off, I had to move the "sleep 1" from after "sdptool add ..." to before the sdptool command. The startup script seemed to be failing to run to completion before I did that.
Networking: pan access point
ipkg install dnsmasq
/etc/dnsmasq.conf needs to contain (among other things it comes with):
interface=bnep0 dhcp-range=10.0.1.10,10.0.1.200,2h
/etc/network/interfaces:
allow-hotplug bnep0
iface bnep0 inet static
address 10.0.1.1
netmask 255.255.255.0
network 10.0.1.0
/etc/default/bluetooth:
PAND_ENABLE=true PAND_OPTIONS="--listen --role NAP"
/etc/network/if-up.d/btap
#!/bin/sh iptables -A POSTROUTING -t nat -j MASQUERADE -s 10.0.1.0/24 echo 1 > /proc/sys/net/ipv4/ip_forward iptables -P FORWARD ACCEPT
You may need to re-run pand or re-run the post-up script.
Networking: pan client
/etc/network/interfaces:
allow-hotplug bnep0 iface bnep0 inet dhcp
/etc/default/bluetooth:
PAND_ENABLE=true PAND_OPTIONS="--role PANU --search"
For some reason pan can't find my access point so I have to hardcode it
PAND_OPTIONS="-c 00:0B:5D:44:33:22"
Dialup networking: server
ipkg install pppd
ipkg install (some kernel modules iirc)
I'm not sure if both the bridging and masquerading scripts are required. The pan0 device would need to be created in advance (with brctl iirc)
/etc/default/bluetooth:
DUND_ENABLE=true DUND_OPTIONS="--master --channel 3 --auth --encrypt --secure --pppd /bin/mypppd"
/bin/mypppd:
#!/bin/sh LEASES=/var/run/mypppd.leases grep -q $DUN_BDADDR $LEASES || echo $DUN_BDADDR >>$LEASES LOCAL=192.168.32.10 REMOTE=192.168.42.`grep -n $DUN_BDADDR $LEASES | cut -d : -f1` pppd $LOCAL:$REMOTE $*
/etc/ppp/peers/dun:
460800 crtscts debug local noipdefault passive connect "/etc/ppp/peers/at-command.pl" noauth nodefaultroute noipx noaccomp nobsdcomp nodeflate ms-dns 198.60.22.2
/etc/ppp/peers/at-command.pl:
#!/usr/bin/perl
$/ = "\r";
while(<STDIN>)
{
$_ = uc;
if(/^ATD/) {
print "CONNECT\r\n";
exit(0);
} elsif(/^AT/) {
print "OK\r\n";
}
}
/etc/ppp/ip-up.d/09nat
iptables -F -t nat
echo "1" > /proc/sys/net/ipv4/ip_forward
echo "1" > /proc/sys/net/ipv4/ip_dynaddr
iptables -t nat -A POSTROUTING -s 192.168.42.0/24 -j MASQUERADE
iptables -t nat -A POSTROUTING -s 10.0.1.0/24 -j MASQUERADE
iptables -P FORWARD ACCEPT
/etc/ppp/ip-up.d/10bridge
#!/bin/sh brctl addif pan0 $2
Dialup networking: client
GPS repeater
Install gpsd and configure it for your gps connection.
In /etc/init.d/bluetooth, change
$RFCOMM_EXEC -r watch 0 1 /sbin/getty -w -L rfcomm0 115200 vt100 &
to
$RFCOMM_EXEC -r watch 0 1 sh -c "gpspipe -r >/dev/rfcomm0" &
Unfortunately, advertising your gps uses the same sdp record as advertising your bluetooth console, so this knocks out your console. You can bring it back on an unadvertised channel:
$RFCOMM_EXEC -r watch 1 2 /sbin/getty -w -L rfcomm1 115200 vt100 &
but then you'll have to be able to manually choose channel 2. That's not a problem from linux, but problematic for others.
Keyboard or mouse
Put your keyboard or mouse in pairing mode and run:
hidd --search
The keyboard should connect and on subsequent runs it will reconnect automatically if you boot up with this in /etc/default/bluetooth
HIDD_ENABLE=true HIDD_OPTIONS="--master --server"
Voice audio
Currently the way to use voice over bluetooth requires a vx board and a USB bluetooth adapter with a CSR chip in it.
High-fidelity audio
You can use A2DP with either the internal bluetooth adapter (model 08 or newer) or any brand of USB adapter if you have a vx board.
Install the packages you'll need:
ipkg install bluez-utils bluez-utils-alsa alsa-utils-aplay madplay
Edit /etc/bluetooth/audio.service to enable autostart:
[Bluetooth Service] Identifier=audio Name=Audio service Description=Bluetooth Audio service Autostart=true
Reboot now if you had to change this.
You'll likely need the passkey agent running, at least the first time you connect. A few itech headsets need "8888" but most need "0000" for the pin:
passkey-agent --default 0000 &
put your headset in pairing mode. Usually this means starting from off and holding the power button down until it flashes. Then find the headset's address:
hcitool scan
edit your /etc/asound.conf with an adapter for bluetooth with the address you just found. My headset is made by itech, so I name the adapter after it:
pcm.itech {
type bluetooth
device "00:0D:3C:8A:13:06"
}
now try playing something to it:
madplay song.mp3 -r 44100 --output=wave:- | aplay -D itech
you can also live-encode the line in source to a2dp:
arecord -c 2 -f s16 -r 44100 | aplay -D itech
unfortunately for live encoding, the delay is pretty bad and for some reason I can never get stereo audio from the line-in source.
Debugging connection problems
The best way to figure out what is happening is to use hcidump.
ipkg install bluez-hcidump
In a separate window, or in the background, start the dump:
hcidump -XV
And then in another window, try out your bluetooth command that isn't working. hcidump output is very verbose, but you'll probably be able to zero in on the issue by studying it a little.
Bluetooth advice from Australia
kick-start
The Infineon v1.01 modules require a once-only 'kick-start' to become operational at a speed of 921600. We have found the kick-start procedure below must be adhered to exactly, in order for the module to properly invoke the 'infineon manufacturer mode' function (that changes the speed) within hciattach.
The kick-start procedure is as follows-
/etc/init.d/S30bluetooth stop echo \"AF1 out\" > /proc/gpio/GPIO12 echo \"GPIO out clear\" >/proc/gpio/GPIO7 sleep 1 echo \"GPIO out set\" >/proc/gpio/GPIO7 sleep 1 echo \"AF1 in\" > /proc/gpio/GPIO42 echo \"AF2 out\" > /proc/gpio/GPIO43 echo \"AF1 in\" > /proc/gpio/GPIO44 echo \"AF2 out\" > /proc/gpio/GPIO45 /usr/sbin/hciattach -s 115200 ttyS1 gumstix 115200 /etc/init.d/S30bluetooth stop echo \"AF1 out\" > /proc/gpio/GPIO12 echo \"GPIO out clear\" >/proc/gpio/GPIO7 sleep 1 echo \"GPIO out set\" >/proc/gpio/GPIO7 sleep 1 echo \"AF1 in\" > /proc/gpio/GPIO42 echo \"AF2 out\" > /proc/gpio/GPIO43 echo \"AF1 in\" > /proc/gpio/GPIO44 echo \"AF2 out\" > /proc/gpio/GPIO45 /usr/sbin/hciattach -s 115200 ttyS1 gumstix 921600 <-- (Infineon Manufacturer Mode is invoked here, within hciattach) /etc/init.d/S30bluetooth stop echo \"AF1 out\" > /proc/gpio/GPIO12 echo \"GPIO out clear\" >/proc/gpio/GPIO7 sleep 1 echo \"GPIO out set\" >/proc/gpio/GPIO7 sleep 1 echo \"AF1 in\" > /proc/gpio/GPIO42 echo \"AF2 out\" > /proc/gpio/GPIO43 echo \"AF1 in\" > /proc/gpio/GPIO44 echo \"AF2 out\" > /proc/gpio/GPIO45 /usr/sbin/hciattach -s 921600 ttyS1 gumstix 921600 <-- (Will now connect at 921600 using this command only)
BlueZ utilities
The BlueZ utilities version we were using was too old- V2.24, and it did not contain the appropriate 'infineon manufacturer mode' functions required to change the speed of the module. It gave the following output when trying to bring up the module-
Read wrong response size: 4
0x04 0xff 0x01 0x00
We upgraded to BlueZ V3.24 and this solved the above problem, however we had to comment out the following lines in the 'gumstix_reset' function of 'hciattach.c'-
//system("echo clear > /proc/gpio/GPIO12");
//system("echo set > /proc/gpio/GPIO12");
We felt these were more appropriate in S30Bluetooth.
HCIATTACH commands
The Bluetooth appears to be on ttyS1, where previously it was present on ttyS3. Thus the following changes needed to be made to /etc/default/bluetooth-
HCIATTACH_TTY=ttyS1
...and we also had to put in new speed parameters to bring the module up properly after a kick-start-
HCIATTACH_START_SPEED=921600 HCIATTACH_SPEED=921600
Configure extra GPIOs
Lastly, some extra GPIOs needed to be configured as appropriate. We only discovered this when we looked at the Open Embedded version of /etc/init.d/bluetooth and discovered that it had some mysterious extra GPIOs being set, that were not present in the buildroot script or any documentation. These are the following-
echo AF1 out > /proc/gpio/GPIO12 sleep 1 echo GPIO out clear >/proc/gpio/GPIO7 sleep 1 echo GPIO out set >/proc/gpio/GPIO7 sleep 1 echo "AF1 in" > /proc/gpio/GPIO42 #(Not present in buildroot) echo "AF2 out" > /proc/gpio/GPIO43 #(Not present in buildroot) echo "AF1 in" > /proc/gpio/GPIO44 #(Not present in buildroot, not documented) echo "AF2 out" > /proc/gpio/GPIO45 #(Not present in buildroot, not documented)
The last two GPIOs (44/45) do not seem to be documented anywhere?- http://docwiki.gumstix.org/Gumstix_UARTs
All of these changes were required to make the modules work properly.
We have also identified that the new v1.01 infineon module has improved pairing ability over its predecessor. Its predecessor required that a passkey be entered every time the unit connects to a paired device. The new module only requires a passkey once, and subsequent connection attempts are automatically connected, as they should be.
Tomas Targownik
This category currently contains no pages or media.
