Re: Removing USB keyboard after filesystems synced causes panic with destroyed mutex twa(4)?

On Wed, Mar 10, 2010 at 9:58 PM, Garrett Cooper <yanefbsd_at_gmail.com> wrote:
> On Wed, Mar 10, 2010 at 2:07 PM, Tom Couch <tom.couch.storage_at_gmail.com> wrote:
>> Hi FreeBSD-current,
>>     My name is Tom Couch,
>> I am part of the 3ware driver team recently acquired by LSI.
>> I believe Giovanni's patch, below, is the correct fix for this bug.
>>
>> I am available to maintain the twa driver, now that I am on this list.
>> Let me know how I can help,
>>
>> Tom
>>
>>
>> On Wed, Mar 10, 2010 at 1:56 PM, Giovanni Trematerra <
>> giovanni.trematerra_at_gmail.com> wrote:
>>
>>> On Sun, Mar 7, 2010 at 11:24 AM, Garrett Cooper <yanefbsd_at_gmail.com>
>>> wrote:
>>> > On Sun, Mar 7, 2010 at 2:07 AM, Garrett Cooper <yanefbsd_at_gmail.com>
>>> wrote:
>>> >> Hi Alexander and Hans,
>>> >>    I recently did the following which generated a panic on a
>>> >> 9-CURRENT kernel compiled on the 26th:
>>> >>
>>> >> 1. Executed reboot
>>> >> 2. Removed keyboard.
>>> >> 3. Some time after `All buffers synced\nUptime: ...' was displayed,
>>> >> the keyboard was registered disconnected.
>>> >> 4. The interrupt was delivered to my twa(4) enabled card and the
>>> >> kernel panicked, like so:
>>> >>
>>> >> ugen2.2: <Mitsumi Electric> at usbus2 (disconnected)
>>> >> uhub8: at uhub2, port 1, addr 2 (disconnected)
>>> >> ugen2.3: <Mitsumi Electric> at usbus2 (disconnected)
>>> >> ukbd0: at uhub8, port 3, addr 3 (disconnected)
>>> >> uhid0: at uhub8, port 3, addr 3 (disconnected)
>>> >> panic: mtx_lock_spin() of destroyed mutex _at_
>>> /usr/src/sys/dev/twa/tw_cl_intr.c:88
>>> >>
>>> >> cpuid = 1
>>> >> KDB: enter: panic
>>> >> [thread pid 12 tid 100025 ]
>>> >> Stopped at         kdb_enter+0x3d: movq     $0,0x40289c(%rip)
>>> >> db>
>>> >>
>>> >>    I wish I could provide you with more details, but unfortunately I
>>> >> the USB bus isn't registering the fact that I'm reattaching the
>>> >> keyboard right now and the box won't reboot automatically :( (didn't
>>> >> set the right sysctl beforehand to panic automatically). I'll try and
>>> >> reproduce the issue again, but I was just wondering whether or not you
>>> >> guys had seen this problem before.
>>> >
>>> >    Phew... it's reproducible with that kernel. Here's what I did
>>> > exactly (because my original directions were incorrect):
>>> >    1. Hit power button (for S5).
>>> >    2. Disconnect keyboard RIGHT as `Uptime: ...' is displayed.
>>> >    Kernel panicked on my system again. Now to figure out if it still
>>> > exists with a kernel compiled today, and also how to debug it if it
>>> > still does exist :/...
>>> > Thanks,
>>> > -Garrett
>>>
>>> Hi Garrett,
>>> Could you please try the patch below and report back?
>>>
>>> Thank you
>>>
>>> diff -r cab6489de66d sys/dev/twa/tw_cl_intr.c
>>> --- a/sys/dev/twa/tw_cl_intr.c        Wed Mar 03 04:51:13 2010 -0500
>>> +++ b/sys/dev/twa/tw_cl_intr.c        Wed Mar 10 06:29:05 2010 -0500
>>> _at__at_ -75,9 +75,12 _at__at_ tw_cl_interrupt(struct tw_cl_ctlr_handle
>>>      if (ctlr == NULL)
>>>               goto out;
>>>
>>> -     /* If we get an interrupt while resetting, it is a shared
>>> -        one for another device, so just bail */
>>> -     if (ctlr->state & TW_CLI_CTLR_STATE_RESET_IN_PROGRESS)
>>> +     /*
>>> +      *  If we get an interrupt while resetting or shutting down,
>>> +      *  it is a shared one for another device, so just bail
>>> +      */
>>> +     if (ctlr->state & TW_CLI_CTLR_STATE_RESET_IN_PROGRESS ||
>>> +                     (ctrl->state & TW_CLI_CTLR_STATE_ACTIVE) == 0)
>>>              goto out;
>>>
>>>      /*

Apart from the typo above (s/ctrl/ctlr/), things work appropriately
now at reboot. The only problem is that bootup is really wonky now,
because the RAID had a LOT of issues attaching to cam(4) (failed in
2/3 cold boot attempts); an additional branch condition may need to be
added to the above if-statement if this change didn't take that into
account. However, if the old behavior was incorrect and the new
behavior is correct, s.t. the RAID controller demonstrating bus
detection timeout issue that is occurring with a lot of USB devices
and some RAID controllers today, this could be extremely problematic.

So, while it looks better than before at reboot, it's not ready yet
for prime time; I'd rather that the bug was filed with the patch you
provided after the typo fixed, with the caveat mentioned and NOT
committed, because the adverse affect(s) seem a bit more annoying than
the previous panic issue described.

> I'll give the patch a try sometime before the weekend; I have a
> critical deadline that I need to work through and the machine can't be
> taken offline until then.

Thanks :),
-Garrett