Hi Pacho,
On Fri, Sep 21, 2012, Pacho Ramos wrote:
quoted
quoted
Searching, I found this thread that pointed to the culprit, but I
haven't found what finally occurred with it, if patch was reverted or a
different fix was pulled in:
http://www.spinics.net/lists/linux-bluetooth/msg26442.html
This is a different issue but the cause seems to be similar. You don't
need to patch bluetoothd though but just disable the mgmt part by
passing -P mgmtops to bluetoothd. For whatever reason the connection
state isn't cleaned up with mgmt (which shouldn't be dependent on mgmt
to begin with) and the input_device_set_channel function returns
-EALREADY when accepting a connection.
But it sounds like a workaround as mgmt is still broken as compared with
bluetoothd previous above change was committed, no? Thanks for the
info :)
Yes, it's still a workaround and the real cause needs to be figured out
and fixed (either with more detailed logs or by the developers
reproducing this themselves).
Johan