On Sunday 27 Apr 2014 08:27:31 William Kenworthy wrote:
> On 04/27/14 11:14, William Kenworthy wrote:
> > On 04/27/14 02:33, Mick wrote:
> >> On Thursday 24 Apr 2014 02:11:57 William Kenworthy wrote:
> >>> I was able to get it working manually - gentoo's init scripts are out
> >>> of date with bluez 5, blutoothctl is broken (or probably just poorly
> >>> documented which equates to the same thing if the command doesn't
> >>> work) .
> >>> 
> >>> In bluetoothctl:
> >>> power on
> >>> scan on
> >>> agent on
> >>> default-agent
> >>> pair <dev_id>
> >>> trust <dev_id>
> >>> exit
> >>> 
> >>> In a shell:
> >>> rfcomm bind rfcomm0 <dev_id>
> >>> 
> >>> do serial port stuff with /dev/rfcomm0
> >>> 
> >>> rfcomm unbind rfcomm0
> >>> 
> >>> bluetoothctl connect command does not work - connects and immediately
> >>> disconnects with an error
> >>> gentoo's rfcomm initscript has removed the -f flag which bluez 5 does
> >>> not have, but it also looks like the bind all in the 5.17 ebuild is
> >>> also not supported by late bluez5 so it immediately exits and no
> >>> rfcomm device is created.
> >>> 
> >>> Ive adapted my python script to the changes now - but the pairing does
> >>> not survive restarting bluetooth so I'll need an expect script to set
> >>> it up each bluetooth re-init as it looks like there are no scripting
> >>> hooks in bluetoothctl.
> >>> 
> >>> BillK
> >> 
> >> Thanks BillK, your suggestions above helped somewhat, because I was able
> >> to connect with my phone, but it didn't get me far enough.  I was not
> >> able to connect with rfcomm to my mobile.  When I ran 'pon
> >> <connection_name>' pppd started, but I got errors like:
> >> 
> >> Apr 26 18:15:12 dell_xps chat[29579]:  -- write failed: Transport
> >> endpoint is not connected
> >> Apr 26 18:15:12 dell_xps chat[29579]: Failed
> >> 
> >> 
> >> This was despite the fact that I had created manually the rfcomm0 device
> >> and binded it to the bdaddr of my phone as you suggested.
> >> 
> >> Googling for this error revealed that this is because the rfcomm code
> >> has
> >> 
> >> changed - but there is a patch which may fix things:
> >>   http://comments.gmane.org/gmane.linux.bluez.kernel/42303
> >> 
> >> I ran out of time and did not try 'rfcomm connect' instead of 'rfcomm
> >> bind' to see if it makes a difference in my case.
> >> 
> >> FYI, I'm on net-wireless/bluez-5.15 and kernel 3.12.13-gentoo.
> > 
> > I just upgraded to 3.12.13 and it stopped working with the same error
> > you have.
> > 
> > I did see some other messages saying that certain kernel versions are
> > broken but I'll now need to look into that now.
> > 
> > BillK
> 
> I used the patch from the reply above - worked!
> 
> It did take a couple of goes but after restarting the bluetooth
> initscript before using bluetoothctl via expect (took 3 goes before I
> got a clean run from "expect" - timing might need adjusting?)

This is getting worse!  O_O

I am on net-wireless/bluez-5.18 and gentoo-sources-3.12.20 without the patch.

Now I have no rfcomm service at all listed under rc-update and if I try to 
start /etc/init.d/bluetooth I get:

# /etc/init.d/bluetooth restart
 * ERROR: bluetooth needs service(s) rfcomm


Creating /dev/rfcomm0 with:

# rfcomm bind rfcomm0 hci0

does not change the error.  Starting KDE's BlueDevil shows no adaptor found.  
Indeed, hci0 is not configured.  O_o

So, running:

# hciconfig hci0 up

allows me to list it:

$ hcitool dev 
Devices:
        hci0    90:4C:E5:FA:F2:A8

but NOT under bluetoothctl!

[bluetooth]# power on
No default controller available
[bluetooth]# show
No default controller available
[bluetooth]# list
[bluetooth]#
[bluetooth]# devices
[bluetooth]# 

NOTE: using hcitool I can scan my mobile phone, but without rfcomm I can't use 
it.


Hmm ... am I alone in this quest?

-- 
Regards,
Mick

Attachment: signature.asc
Description: This is a digitally signed message part.

Reply via email to