Re: XHCI is slow during boot (bios/efi) and leaves many dmesg messages

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Hi,

Maël Lavault <mael.lavault@xxxxxxxxxxxxxxxxxxx> writes:
> Le 28 avr. 2017 15:57, "Maël Lavault" <mael.lavault@xxxxxxxxxxxxxxxxxxx
>> a écrit :
> On Fri, 2017-04-28 at 16:58 +0300, Mathias Nyman wrote:
>> On 21.04.2017 11:08, Maël Lavault wrote:
>> > On Tue, 2017-04-18 at 16:58 +0200, Maël Lavault wrote:
>> > > Hi,
>> > >
>> > > I can't find how to reply to an old thread with majordomo, sorry
>> > > for
>> > > the inconvenience it might cause.
>> > >
>> > > I'm reposting an issue [0] that has been inactive for a few month
>> > > but
>> > > still present in kernel 4.10.10 on a Macbook pro 12,1.
>> > >
>> > > I can provide more informations if needed but the issue is
>> > > explained
>> > > in
>> > > details in the bugzilla issue.
>> > >
>> > > Thanks.
>> > >
>> > > ---------------
>> > > Hi list,
>> > >
>> > > I have a Apple Inc. MacBookPro11,1 (with the most recent 'bios':
>> > > BIOS
>> > > MBP111.88Z.0138.B16.1509081438 09/08/2015).
>> > >
>> > > At the beginning, USB worked normally. After a while (and after
>> > > newer
>> > > kernel versions released by debian?) things started to act
>> > > strangely.
>> > > For
>> > > one, the bios/efi boot takes a very long time (probably due to
>> > > the
>> > > same
>> > > reason I describe later) just to get to the bootloader/grub.
>> > > Likley
>> > > resetting and probing for USB ports/mass storage. When grub
>> > > finally
>> > > pops
>> > > up, I can use the (internal USB based keyboard) normally to
>> > > select a
>> > > grub
>> > > entry etc.
>> > >
>> > > Booting the kernel then works reasonably fine, until it loads the
>> > > xhci
>> > > module.
>> > > It spews some messages in dmesg (taking some 15 seconds) and only
>> > > then,
>> > > the
>> > > keyboard starts to work again.
>> > >
>> > > The log is filled with messages like:
>> > > [    7.248479] xhci_hcd 0000:00:14.0: Command completion event
>> > > does
>> > > not
>> > > match command
>> > > [    7.248495] xhci_hcd 0000:00:14.0: Timeout while waiting for
>> > > setup
>> > > device command
>> > > [   12.256347] xhci_hcd 0000:00:14.0: Timeout while waiting for
>> > > setup
>> > > device command
>> > > [   12.256363] usb 1-2: hub failed to enable device, error -62
>> > > [   17.264166] xhci_hcd 0000:00:14.0: Timeout while waiting for
>> > > setup
>> > > device command
>> > > (followed by USB hub/device enumeration)
>> > >
>> > > I've tried several combinations and quirks, updating to the
>> > > latest rc
>> > > kernels since 3.16 (am on 4.5.0 right now) and it only seems to
>> > > get
>> > > worse.
>> > >
>> > > Last year, on the 3.x series of kernels occasionally after a
>> > > reboot
>> > > the
>> > > 'bios' would go through quickly and fine and also no problems
>> > > loading
>> > > the
>> > > module and logging in. But now it always fails.
>> > >
>> > > Additionally it (may or may not) seems to cause the internal usb
>> > > card
>> > > reader to not even show up almost all of the time, though under
>> > > OSX
>> > > it
>> > > works fine. There is/was a known issue with this cardreader where
>> > > it
>> > > would
>> > > disappear after a suspend.
>> > >
>> > > Adding various seemingly related intel usb3 quirks I had no
>> > > change,
>> > > as
>> > > I
>> > > think all of them are already applied to this chipset.
>> > >
>> > > I'm guessing that somehow the usb chipset has some configuration
>> > > option
>> > > miss-set (which persists over reboots/power down) and the driver
>> > > doesn't
>> > > quite understand it.
>> > >
>> > > Unfortunately it seems that this chipset does not work in pure
>> > > USB2.0
>> > > (ehci) mode and needs the xhci module to work at all, so even
>> > > falling
>> > > to
>> > > USB2 is no option. Also disconnecting all USB perhipials is
>> > > nearly
>> > > impossible as the touchpad, bluetooth cardreader and keyboard are
>> > > internally all wired to USB.
>> > >
>> > > I'm attaching 3 dmesg logs with various kernels and levels of
>> > > debugging
>> > > information. I tried to google for errors from these logs, but to
>> > > no
>> > > avail.
>> > >
>> > > I have attached some log files on the bugzilla issue tracker [1]
>> > > (they
>> > > are
>> > > to big for the ML I think).
>> > >
>> > > [1] https://bugzilla.kernel.org/show_bug.cgi?id=115741
>> > >
>> > >
>> > > Olliver
>> > >
>> > >
>> > >
>> > > [0] http://thread.gmane.org/gmane.linux.usb.general/139697
>> >
>> > I updated to fedora 26 with kernel 4.11.0-rc7 and the issue is
>> > still
>> > present.
>>
>> 4.11-rc7 should have xhci tracepoints installed.
>> Could you take some traces at boot with most recent kernel you got.
>>
>> xhci traces for boot can be enabled by adding "trace_event=xhci-hcd"
>> to
>> your kernel cmdline
>>
>> Traces should then appear in /sys/kernel/debug/tracing/trace
>>
>> -Mathias
>
> Thanks ! I'll try to do that, but it's utterly difficult since each
> keypress takes between 5 and 10 seconds due to this bug ...
>
> So I tried it and I cannot access the trace. Are the trace persisted
> across reboot ? Since all 4.11 kernel freeze on the login page I have
> no other choice but rebooting on a old 4.10 kernel to look at the
> trace.

You could SSH from a different machine, no? tracepoints don't persist
across reboots, AFAICT.

-- 
balbi
--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html



[Index of Archives]     [Linux Media]     [Linux Input]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]     [Old Linux USB Devel Archive]

  Powered by Linux