Re: btusb not working in 2.6.38.4

[Andrej Podzimek <andrej@podzimek.org>]

[-- Attachment #1: Type: text/plain, Size: 3379 bytes --]

> Hi Andrej,
>
> * Andrej Podzimek<andrej@podzimek.org>  [2011-04-28 01:22:02 +0200]:
>
>> Hello,
>>
>> after a recent kernel update on Arch Linux (2.6.38.4), I can see the following problem in dmesg:
>>
>> ------------[ cut here ]------------
>> WARNING: at lib/kobject.c:595 kobject_put+0x40/0x50()
>> Hardware name: To Be Filled By O.E.M.
>> kobject: '(null)' (ffff88023f60d5a0): is not initialized, yet kobject_put() is being called.
>> Modules linked in: btusb(+) bluetooth rfkill usbhid hid button i2c_i801 i2c_core r8169 mii iTCO_wdt iTCO_vendor_support psmouse sg serio_raw evdev pcspkr ip_tables ip6_tables x_tables tpm_tis tpm tpm_bios w83627ehf hwmon_vid coretemp ext4 mbcache jbd2 crc16 dm_mod usb_storage raid456 async_raid6_recov async_pq raid6_pq async_xor xor async_memcpy async_tx md_mod uhci_hcd xhci_hcd ehci_hcd usbcore sd_mod crc_t10dif ahci libahci libata scsi_mod cpufreq_ondemand acpi_cpufreq freq_table processor mperf i7core_edac edac_core
>> Pid: 1820, comm: modprobe Not tainted 2.6.38-ARCH #1
>> Call Trace:
>>    [<ffffffff8103a86b>] ? warn_slowpath_common+0x7b/0xc0
>>    [<ffffffff8103a965>] ? warn_slowpath_fmt+0x45/0x50
>>    [<ffffffff81248d3f>] ? skb_dequeue+0x5f/0x80
>>    [<ffffffff8116b6f0>] ? kobject_put+0x40/0x50
>>    [<ffffffffa025f85c>] ? btusb_probe+0x57c/0x5e0 [btusb]
>>    [<ffffffffa00b85d5>] ? usb_probe_interface+0xe5/0x1e0 [usbcore]
>>    [<ffffffff81215aaf>] ? driver_probe_device+0x6f/0x190
>>    [<ffffffff81215c63>] ? __driver_attach+0x93/0xa0
>>    [<ffffffff81215bd0>] ? __driver_attach+0x0/0xa0
>>    [<ffffffff81214b23>] ? bus_for_each_dev+0x53/0x80
>>    [<ffffffff812153f8>] ? bus_add_driver+0x188/0x260
>>    [<ffffffff81215e3a>] ? driver_register+0x6a/0x130
>>    [<ffffffffa00b748a>] ? usb_register_driver+0x8a/0x180 [usbcore]
>>    [<ffffffffa011c000>] ? btusb_init+0x0/0x1000 [btusb]
>>    [<ffffffff810001dc>] ? do_one_initcall+0x3c/0x170
>>    [<ffffffff8106db3a>] ? sys_init_module+0xda/0x230
>>    [<ffffffff8100243b>] ? system_call_fastpath+0x16/0x1b
>> ---[ end trace 9c4648061ddb0b67 ]---
>>
>> Presumably, Bluetooth does not seem to work at all (i.e., hciconfig -a does not output anything).
>>
>> The output from lsusb -v relevant to the specific device is attached.
>>
>> The device worked 100% flawlessly with 2.6.38.3. I had an a2dp headset and a mouse paired with it, both of which worked perfectly fine in parallel.
>
> The only Bluetooth change in 2.6.38.4 is: https://lkml.org/lkml/2011/4/19/406
> However I don't think this patch is causing this issue. Do you care to patch -R it
> and test?
> If the problem is somewhere else we have to go with a bisect, there are only
> 70 patches, so it shouldn't take to long.
>
> Greg, any idea on that? Are there similar reports of warnings for 2.6.38.4?

Hello,

I'm sorry for this false positive. It was just a typical PEBKAC. The machine runs a WiFi access point and requires ath9k from compat-wireless (since the mainline ath9k can do no better than 2 Mb/s in AP mode). I installed compat-wireless as usual, but forgot to run scripts/driver-select. Consequently, all drivers (including Bluetooth, namely btusb) got replaced. So the backtrace was caused by using btusb from compat-wireless (April 26 2011) with kernel 2.6.38.4. The problem does *not* occur in a standard vanilla kernel.

Andrej


[-- Attachment #2: S/MIME Cryptographic Signature --]
[-- Type: application/pkcs7-signature, Size: 5804 bytes --]