Ben Hutchings <b...@decadent.org.uk> writes:
> On Wed, 2016-04-06 at 16:51 +0200, Bjørn Mork wrote:
>> Antonis Christofides <anth...@itia.ntua.gr> writes:
>>
>> >
>> > Apr 5 13:32:47 thames kernel: [86003.993994] BUG: unable to handle kernel
>> > paging request at ffffeba40a800380
>> > Apr 5 13:32:47 thames kernel: [86003.994005] IP: [] kfree+0x76/0x220
>> ..
>> >
>> > Apr 5 13:32:47 thames kernel: [86003.994679] Call Trace:
>> > Apr 5 13:32:47 thames kernel: [86003.994689] [] ?
>> > usb_release_bos_descriptor+0x1d/0x40 [usbcore]
>> I believe this bug is fixed by commit e5bdfd50d6f7 ("Revert "usb: hub:
>> do not clear BOS field during reset device"") in mainline. Don't know
>> the stable status of that. I don't see it in v3.16.x yet. I assume Ben
>> keeps track of it :)
>
> The crash after disconnection certainly seems like it could be fixed by
> that. But the rapid disconnections (at 13:31:32 and 13:32:47) appear
> to be a different bug.
Yes. My guess would be a voltage dip since it happens when spinning up
the disk. There is little the host can do about USB disconnects anyway.
That's a device hardware or firmware issue. It won't help tracking it as
a kernel bug.
I would start by adding more power, using an external power supply if
supported or a split USB cable drawing power from 2 ports.
Bjørn
