Bug 33482 - [Regression] Bluetooth pairing does not work anymore in 2.6.39-rc3 (works in 2.6.38.3)
Summary: [Regression] Bluetooth pairing does not work anymore in 2.6.39-rc3 (works in ...
Status: CLOSED INVALID
Alias: None
Product: Drivers
Classification: Unclassified
Component: Bluetooth (show other bugs)
Hardware: All Linux
: P1 normal
Assignee: drivers_bluetooth@kernel-bugs.osdl.org
URL:
Keywords:
Depends on:
Blocks: 32012
  Show dependency tree
 
Reported: 2011-04-17 18:39 UTC by Maciej Rutecki
Modified: 2020-06-17 18:03 UTC (History)
8 users (show)

See Also:
Kernel Version: 2.6.39-rc3
Subsystem:
Regression: Yes
Bisected commit-id:


Attachments

Description Maciej Rutecki 2011-04-17 18:39:32 UTC
Subject    : [Regression] Bluetooth pairing does not work anymore in 2.6.39-rc3 (works in 2.6.38.3)
Submitter  : Gottfried Haider <gottfried.haider@gmail.com>
Date       : 2011-04-16 16:38
Message-ID : BANLkTi=LvyZ+7BHfVL849pztfvsYaVM4SQ@mail.gmail.com
References : http://marc.info/?l=linux-kernel&m=130297197128328&w=2

This entry is being used for tracking a regression from 2.6.38. Please don't
close it until the problem is fixed in the mainline.
Comment 1 gottfried.haider 2011-04-21 15:09:24 UTC
This is probably no real regression.

The Bluetooth KConfig changes in 2.6.39 make it easy to exclude BT_L2CAP from the config when you had it build as a module before and don't know that it's required for pairing devices etc.

Once I figured this out and re-enabled BT_L2CAP everything was working again like before. Shall we close the bug?
Comment 2 Eduard Bloch 2011-04-22 13:40:17 UTC
Sorry? I think I can reproduce this or similar problem and your explanation does not make it go away.

Here, the device cannot be brought up anymore:

$ hciconfig 
hci0:	Type: BR/EDR  Bus: USB
	BD Address: 00:02:72:01:0E:2D  ACL MTU: 192:8  SCO MTU: 64:8
	DOWN 
	RX bytes:712 acl:0 sco:0 events:24 errors:0
	TX bytes:108 acl:0 sco:0 commands:24 errors:0
$ hciconfig hci0 up
Can't init device hci0: Invalid argument (22)

The major difference I could detect so far is that L2CAP and BNEP were modules before and now they are forcibly built-in, and the init order has changed:

[    8.432938] Bluetooth: Core ver 2.16
[    8.432992] NET: Registered protocol family 31
[    8.433036] Bluetooth: HCI device and connection manager initialized
[    8.433630] Bluetooth: HCI socket layer initialized
[    8.433667] Bluetooth: L2CAP socket layer initialized
[    8.433747] Bluetooth: SCO socket layer initialized
[    8.457962] Bluetooth: Generic Bluetooth USB driver ver 0.6

[   26.335569] Bluetooth: L2CAP ver 2.15
[   26.339905] Bluetooth: L2CAP socket layer initialized
[   26.492119] Bluetooth: BNEP (Ethernet Emulation) ver 1.3
[   26.496587] Bluetooth: BNEP filters: protocol multicast
[   26.509742] Bluetooth: RFCOMM TTY layer initialized
[   26.514265] Bluetooth: RFCOMM socket layer initialized
[   26.518825] Bluetooth: RFCOMM ver 1.11
[   26.849430] Bluetooth: SCO (Voice Link) ver 0.6
[   26.853984] Bluetooth: SCO socket layer initialized
Comment 3 gottfried.haider 2011-04-22 15:04:30 UTC
I think you are seeing a different bug.

"hciconfig hci0 up" worked for me at all times, it was only pairing ("bluez-simple-agent hci0 $MAC") that produced the following error message:

"Creating device failed: org.bluez.Error.Failed: Input/output error"

Any my problem was only related to BT_L2CAP not being available - whereas it judging from your dmesg.
Comment 4 Rafael J. Wysocki 2011-05-02 21:14:06 UTC
On Monday, May 02, 2011, Gottfried Haider wrote:
> 
> > Bug-Entry       : http://bugzilla.kernel.org/show_bug.cgi?id=33482
> > Subject         : [Regression] Bluetooth pairing does not work anymore in
> 2.6.39-rc3 (works in 2.6.38.3)
> > Submitter       : Gottfried Haider <gottfried.haider@gmail.com>
> > Date            : 2011-04-16 16:38 (15 days old)
> 
> This can be removed from the regression list as it turned out to be
> only a mis-config-uration (oldconfig listed BT_L2CAP as a new config
> option, defaulting to false, when it had been build as a module
> before).

Note You need to log in before you can comment on or make changes to this bug.