Hello.

On Wed, 2009-03-25 at 17:26, Marcel Holtmann wrote:
> 
> > > I have a system here where the wifi card, AR6000 SDIO, which gets probed 
> > > during
> > > kernel bootup and shutdown from a resource handling system during from 
> > > the init
> > > system. This seems to kill connmand and leaves back only wpa-supplicant 
> > > activaed
> > > via dbus service activation. Loging connmand debug output to a file gives 
> > > me no
> > > real clue here:
> > > 
> > > connmand[1476]: detect.c:detect_newlink() type 1 index 2
> > > connmand[1476]: rtnl.c:rtnl_message() buf 0x3ca0b7c len 248
> > > connmand[1476]: rtnl.c:rtnl_message() NEWLINK len 248 type 16 flags 
> > > 0x0000 seq 0
> > > connmand[1476]: rtnl.c:rtnl_newlink() ifi_type 1 ifi_index 2 ifi_flags 
> > > 0x11003
> > > ifi_change 0x0000
> > > connmand[1476]: detect.c:detect_newlink() type 1 index 2
> > > connmand[1476]: supplicant.c:add_interface_reply() task 0x2fa30
> > > 
> > > (If the full log is needed I can send it over)
> > 
> > This part is stil valid. Connman dies and I have to restart it manually.
> > 
> > > On the other hand connman survives if I make sure that wpa-supplicant get 
> > > not
> > > started via service activation. I can start it later, activate the wifi 
> > > device
> > > and all works fine.
> > > 
> > > I wonder if this is a known bug? Testing was done with connman 0.12 and 
> > > 0.14 and
> > > wpa-supplicant 0.6.3 and 0.6.9
> > 
> > I was now able to deactivate and activate the wifi chip with connman still 
> > alive
> > and able to connect to a network. So this part was more a heisenbug.
> 
> I don't understand what you are doing and what breaks. Can you please
> describe it again so that I can follow.

Sure, I'm less confused myself now so it should work better. ;)

The problem I have is that during system startup connman gets started via an
init-script, but dies later. When the system is up and I can log in only
wpa-supplicant is still there. (Started via debus service actication).

To get a clue I modified the init-script to let connman print out its debug
output into a file. Sadly I could not spot any problem there. The lines quoted
above are the last lines from the log. I can upload the complete log if that
helps.

Once I'm logged in I can start connman again and use it. From that point on I
can use it normally. I tested this with 0.12 and 0.14 (0.13 had the segfault on
startup). 0.15 testing is on my list, but I could not spot anything in the logs
that may fix this. Hope this was less confusing this time.

regards
Stefan Schmidt
_______________________________________________
connman mailing list
[email protected]
https://lists.moblin.org/mailman/listinfo/connman

Reply via email to