Re: [Bug #15127] Bluetooth: sleeping function called from invalid context

[Luiz Augusto von Dentz <luiz.dentz-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>]

Hi,

On Mon, Feb 1, 2010 at 11:14 AM, Marcel Holtmann <marcel-kz+m5ild9QBg9hUCZPvPmw@public.gmane.org> wrote:
> Hi David,
>
>> >> This message has been generated automatically as a part of a report
>> >> of regressions introduced between 2.6.31 and 2.6.32.
>> >>
>> >> The following bug entry is on the current list of known regressions
>> >> introduced between 2.6.31 and 2.6.32. аPlease verify if it still should
>> >> be listed and let me know (either way).
>> >>
>> >>
>> >> Bug-Entry а: http://bugzilla.kernel.org/show_bug.cgi?id=15127
>> >> Subject а а а а а а: Bluetooth: sleeping function called from invalid context
>> >> Submitter а: David John <davidjon-XRr60H37pjdAfugRpC6u6w@public.gmane.org>
>> >> Date а а а а а а а : 2010-01-12 9:19 (20 days old)
>> >> First-Bad-Commit: http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=9e726b17422bade75fba94e625cd35fd1353e682
>> >> References : http://marc.info/?l=linux-kernel&m=126328727021949&w=4
>> >
>> > you have an outdated email from Luiz and I change it to the right one
>> > now.
>> >
>> > I looked with him at the patch and I think this will fix it:
>> >
>> > diff --git a/net/bluetooth/rfcomm/core.c b/net/bluetooth/rfcomm/core.c
>> > index fc5ee32..2b50637 100644
>> > --- a/net/bluetooth/rfcomm/core.c
>> > +++ b/net/bluetooth/rfcomm/core.c
>> > @@ -252,7 +252,6 @@ static void rfcomm_session_timeout(unsigned long
>> > arg)
>> > а а BT_DBG("session %p state %ld", s, s->state);
>> >
>> > а а set_bit(RFCOMM_TIMED_OUT, &s->flags);
>> > - а rfcomm_session_put(s);
>> > а а rfcomm_schedule(RFCOMM_SCHED_TIMEO);
>> > а}
>> >
>> > @@ -1920,6 +1919,7 @@ static inline void rfcomm_process_sessions(void)
>> > а а а а а а if (test_and_clear_bit(RFCOMM_TIMED_OUT, &s->flags)) {
>> > а а а а а а а а а а s->state = BT_DISCONN;
>> > а а а а а а а а а а rfcomm_send_disc(s, 0);
>> > + а а а а а а а а а rfcomm_session_put(s);
>> > а а а а а а а а а а continue;
>> > а а а а а а }
>> >
>> > We need some extra testing on this with the actual hardware we did the
>> > patch for. So this will take at least a few days before we get our hands
>> > on it.
>>
>> FWIW, your patch fixes the issue.
>
> nice. So I can add a tested-by line to the final patch?
>
> Just our of curiosity, which hardware did you test this with. We only
> know about one headset that should cause this issue.
>

Just in case, here is the hcidump of the Nokia HS-12W, the one that
has problem when we connection authorization is denied:

> ACL data: handle 11 flags 0x02 dlen 8
    L2CAP(d): cid 0x0041 len 4 [psm 3]
      RFCOMM(s): SABM: cr 1 dlci 26 pf 1 ilen 0 fcs 0xe7
< ACL data: handle 11 flags 0x02 dlen 12
    L2CAP(s): Disconn req: dcid 0x0042 scid 0x0040
< ACL data: handle 11 flags 0x02 dlen 8
    L2CAP(d): cid 0x0044 len 4 [psm 3]
      RFCOMM(s): DM: cr 1 dlci 26 pf 1 ilen 0 fcs 0xcd
> HCI Event: Number of Completed Packets (0x13) plen 5
> ACL data: handle 11 flags 0x02 dlen 12
    L2CAP(s): Disconn rsp: dcid 0x0042 scid 0x0040
< ACL data: handle 11 flags 0x02 dlen 8
    L2CAP(d): cid 0x0044 len 4 [psm 3]
      RFCOMM(s): DISC: cr 0 dlci 0 pf 1 ilen 0 fcs 0x9c
> ACL data: handle 11 flags 0x02 dlen 8
    L2CAP(d): cid 0x0041 len 4 [psm 3]
      RFCOMM(s): UA: cr 0 dlci 0 pf 1 ilen 0 fcs 0xb6
< ACL data: handle 11 flags 0x02 dlen 12
    L2CAP(s): Disconn req: dcid 0x0044 scid 0x0041
> HCI Event: Number of Completed Packets (0x13) plen 5
> ACL data: handle 11 flags 0x02 dlen 12
    L2CAP(s): Disconn rsp: dcid 0x0044 scid 0x0041
< HCI Command: Disconnect (0x01|0x0006) plen 3
> HCI Event: Command Status (0x0f) plen 4
> HCI Event: Disconn Complete (0x05) plen 4

So this means the patch works. DISC 0 is send from our side (due to
the session timeout) when normally it should be other end that
disconnects right away when we respond with DM.

-- 
Luiz Augusto von Dentz
Engenheiro de Computaчуo