[Bluez-devel] RFCOMM server problem

[Bluez-devel] RFCOMM server problem

[Marcel Holtmann]

Hi Max,

if I set up a RFCOMM server on a 2.4.20-mh7 (plus all new patches) and
try to connect from another machine which runs 2.4.20-mh6, the
communications stops after the initial SABM for dlci 0. The server
didn't send the UA for this dlci. The connection establishment fails
because of the timeout. If I don't start the server and try to establish
the connection I see the UA for dlci 0. If I change the direction and
put the server on the machine with the older kernel all is working fine.

Any idea what is going wrong?

Regards

Marcel




-------------------------------------------------------
Enterprise Linux Forum Conference & Expo, June 4-6, 2003, Santa Clara
The only event dedicated to issues related to Linux enterprise solutions
www.enterpriselinuxforum.com

_______________________________________________
Bluez-devel mailing list
Bluez-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bluez-devel

Re: [Bluez-devel] RFCOMM server problem

[Max Krasnyansky]

At 05:05 PM 5/13/2003, Marcel Holtmann wrote:
>Hi Max,
>
>if I set up a RFCOMM server on a 2.4.20-mh7 (plus all new patches) and
>try to connect from another machine which runs 2.4.20-mh6, the
>communications stops after the initial SABM for dlci 0. The server
>didn't send the UA for this dlci. The connection establishment fails
>because of the timeout. If I don't start the server and try to establish
>the connection I see the UA for dlci 0. If I change the direction and
>put the server on the machine with the older kernel all is working fine.
>
>Any idea what is going wrong?
Sounds weird. MSC,RLS,etc changes couldn't affect that. On my machines
everything works as expected. I'll test again.
Make sure you recompiled everything in RFCOMM module. We changed dlc struct 
a bit may be something like rfcomm/sock.c wasn't recompiled and still uses 
old struct. I can't think of anything else at this point.

Max

Re: [Bluez-devel] RFCOMM server problem

[Marcel Holtmann]

Hi Max,

> >if I set up a RFCOMM server on a 2.4.20-mh7 (plus all new patches) and
> >try to connect from another machine which runs 2.4.20-mh6, the
> >communications stops after the initial SABM for dlci 0. The server
> >didn't send the UA for this dlci. The connection establishment fails
> >because of the timeout. If I don't start the server and try to establish
> >the connection I see the UA for dlci 0. If I change the direction and
> >put the server on the machine with the older kernel all is working fine.
> >
> >Any idea what is going wrong?
> Sounds weird. MSC,RLS,etc changes couldn't affect that. On my machines
> everything works as expected. I'll test again.
> Make sure you recompiled everything in RFCOMM module. We changed dlc struct 
> a bit may be something like rfcomm/sock.c wasn't recompiled and still uses 
> old struct. I can't think of anything else at this point.

I was thinking the same, but after some more testing it gets much more
weird. First the kernels I have used are both clean builds from scratch
and packed up in a Debian package, so it is not a compile problem. I
tested this with two of my machines, that have mostly the same kernel
now:

	Sony C1VE	2.4.20-mh7 (plus bt-2.4)
	Pentium 4	2.4.20-mh7 (plus bt-2.4 except the LM_RELIABLE)

The problem only happens if I want to connect from my C1VE to the P4.
The other direction is working fine. I ran hcidump on both sides and
also on the P4 side I see the SABM package, but the rfcomm.o module is
not reacting. For me it looks like an initial processing of incoming
packets is missing. Did you have any other idea?

Regards

Marcel




-------------------------------------------------------
Enterprise Linux Forum Conference & Expo, June 4-6, 2003, Santa Clara
The only event dedicated to issues related to Linux enterprise solutions
www.enterpriselinuxforum.com

_______________________________________________
Bluez-devel mailing list
Bluez-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bluez-devel

Re: [Bluez-devel] RFCOMM server problem

[Max Krasnyansky]

At 03:21 PM 5/14/2003, Marcel Holtmann wrote:
>        Sony C1VE       2.4.20-mh7 (plus bt-2.4)
>        Pentium 4       2.4.20-mh7 (plus bt-2.4 except the LM_RELIABLE)
>
>The problem only happens if I want to connect from my C1VE to the P4.
>The other direction is working fine. I ran hcidump on both sides and
>also on the P4 side I see the SABM package, but the rfcomm.o module is
>not reacting. For me it looks like an initial processing of incoming
>packets is missing. Did you have any other idea?
I cannot reproduce that. I just tried to connect from my Dual PPro, which 
is still running 2.4.21-pre4 with no updates to BT stuff, to my laptop
with very latest bt-2.4. 
I get normal trace
1053045305.509342 > RFCOMM(s): SABM: cr 1 dlci 0 pf 1 ilen 0 fcs 0x1c 
1053045305.509438 < RFCOMM(s): UA: cr 1 dlci 0 pf 1 ilen 0 fcs 0xd7 
1053045305.540498 > RFCOMM(s): PN CMD: cr 1 dlci 0 pf 0 ilen 10 fcs 0x70 mcc_len 8
  dlci 20 frame_type 0 credit_flow 15 pri 0 ack_timer 0 frame_size 1019 max_retrans 0 credits 7
1053045305.540619 < RFCOMM(s): PN RSP: cr 0 dlci 0 pf 0 ilen 10 fcs 0xaa mcc_len 8
  dlci 20 frame_type 0 credit_flow 14 pri 0 ack_timer 0 frame_size 1019 max_retrans 0 credits 7
1053045305.549164 > RFCOMM(s): SABM: cr 1 dlci 20 pf 1 ilen 0 fcs 0xfd 
1053045305.549206 < RFCOMM(s): UA: cr 1 dlci 20 pf 1 ilen 0 fcs 0x36 
1053045305.549248 < RFCOMM(s): MSC CMD: cr 0 dlci 0 pf 0 ilen 4 fcs 0xaa mcc_len 2
  dlci 20 fc 0 rtc 1 rtr 1 ic 0 dv 1 b1 1 b2 1 b3 0 len 0
1053045305.579298 > RFCOMM(s): MSC CMD: cr 1 dlci 0 pf 0 ilen 4 fcs 0x70 mcc_len 2
  dlci 20 fc 0 rtc 1 rtr 1 ic 0 dv 1 b1 1 b2 1 b3 0 len 0
1053045305.579387 < RFCOMM(s): MSC RSP: cr 0 dlci 0 pf 0 ilen 4 fcs 0xaa mcc_len 2
  dlci 20 fc 0 rtc 1 rtr 1 ic 0 dv 1 b1 1 b2 1 b3 0 len 0
1053045305.579660 > RFCOMM(d): UIH: cr 1 dlci 20 pf 1 ilen 0 fcs 0x2d 
1053045305.589091 > RFCOMM(s): MSC RSP: cr 1 dlci 0 pf 0 ilen 4 fcs 0x70 mcc_len 2
  dlci 20 fc 0 rtc 1 rtr 1 ic 0 dv 1 b1 1 b2 1 b3 0 len 0
1053045305.589117 < RFCOMM(d): UIH: cr 0 dlci 20 pf 1 ilen 0 fcs 0xf7 

(first UIH sent before last MSC response came from 2.4.21-pre4 
which doesn't have MSC fixes).

So. I'm absolutely sure that new code is 100% compatible with the old one. I see
no delays in SABM processing.
And I have no idea what's going on with your P4 :).

Max

Re: [Bluez-devel] RFCOMM server problem

[Marcel Holtmann]

Hi Max,

> I cannot reproduce that. I just tried to connect from my Dual PPro, which 
> is still running 2.4.21-pre4 with no updates to BT stuff, to my laptop
> with very latest bt-2.4. 
> I get normal trace
> 1053045305.509342 > RFCOMM(s): SABM: cr 1 dlci 0 pf 1 ilen 0 fcs 0x1c 
> 1053045305.509438 < RFCOMM(s): UA: cr 1 dlci 0 pf 1 ilen 0 fcs 0xd7 
> 1053045305.540498 > RFCOMM(s): PN CMD: cr 1 dlci 0 pf 0 ilen 10 fcs 0x70 mcc_len 8
>   dlci 20 frame_type 0 credit_flow 15 pri 0 ack_timer 0 frame_size 1019 max_retrans 0 credits 7
> 1053045305.540619 < RFCOMM(s): PN RSP: cr 0 dlci 0 pf 0 ilen 10 fcs 0xaa mcc_len 8
>   dlci 20 frame_type 0 credit_flow 14 pri 0 ack_timer 0 frame_size 1019 max_retrans 0 credits 7
> 1053045305.549164 > RFCOMM(s): SABM: cr 1 dlci 20 pf 1 ilen 0 fcs 0xfd 
> 1053045305.549206 < RFCOMM(s): UA: cr 1 dlci 20 pf 1 ilen 0 fcs 0x36 
> 1053045305.549248 < RFCOMM(s): MSC CMD: cr 0 dlci 0 pf 0 ilen 4 fcs 0xaa mcc_len 2
>   dlci 20 fc 0 rtc 1 rtr 1 ic 0 dv 1 b1 1 b2 1 b3 0 len 0
> 1053045305.579298 > RFCOMM(s): MSC CMD: cr 1 dlci 0 pf 0 ilen 4 fcs 0x70 mcc_len 2
>   dlci 20 fc 0 rtc 1 rtr 1 ic 0 dv 1 b1 1 b2 1 b3 0 len 0
> 1053045305.579387 < RFCOMM(s): MSC RSP: cr 0 dlci 0 pf 0 ilen 4 fcs 0xaa mcc_len 2
>   dlci 20 fc 0 rtc 1 rtr 1 ic 0 dv 1 b1 1 b2 1 b3 0 len 0
> 1053045305.579660 > RFCOMM(d): UIH: cr 1 dlci 20 pf 1 ilen 0 fcs 0x2d 
> 1053045305.589091 > RFCOMM(s): MSC RSP: cr 1 dlci 0 pf 0 ilen 4 fcs 0x70 mcc_len 2
>   dlci 20 fc 0 rtc 1 rtr 1 ic 0 dv 1 b1 1 b2 1 b3 0 len 0
> 1053045305.589117 < RFCOMM(d): UIH: cr 0 dlci 20 pf 1 ilen 0 fcs 0xf7 
> 
> (first UIH sent before last MSC response came from 2.4.21-pre4 
> which doesn't have MSC fixes).
> 
> So. I'm absolutely sure that new code is 100% compatible with the old one. I see
> no delays in SABM processing.
> And I have no idea what's going on with your P4 :).

after my latest tests I am pretty sure that this has nothing to do with
our changes in the RFCOMM code. It must be some kind of side effect,
which hasn't been noticed yet.

Did anyone had similar problems?

Regards

Marcel




-------------------------------------------------------
Enterprise Linux Forum Conference & Expo, June 4-6, 2003, Santa Clara
The only event dedicated to issues related to Linux enterprise solutions
www.enterpriselinuxforum.com

_______________________________________________
Bluez-devel mailing list
Bluez-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bluez-devel