Comment 31 for bug 1072082

Hi Benjamin

Following the install, the workaround succeeded on 10/10 power cycle
tests to be operational on the login screen.

It looks like you are on the right track. I await further news with
interest. I won't sign off the bug till you are happy.

Regards Bob Bowles

On 14/02/13 15:05, Benjamin Tissoires wrote:
> On Thu, Feb 14, 2013 at 2:36 PM, Bob Bowles <email address hidden> wrote:
>> Hi Benjamin
>>
>> Thanks for sticking at this.
>>
>> I did as requested, and the pull, make, rmmod, insmod, dmesg, and the
>> kern.log tail are logged in log3.txt (attached). At this point, as before,
>> both mouse and keyboard were still inoperative, but as you can see, the
>> error messages have been replaced with 'success' messages in the logs (?!).
>>
>> I did the replug, and both mouse and keyboard worked right away. I have
>> included the dmesg for that in log3a.txt.
> Ok. So I understood the problem:
> - during the boot, the distribution's module hid-logitech-dj trigger a
> command to switch the receiver to the "dj" mode.
> - however, this command is sent during a phase where the USB3 hw or
> driver is not cooperative.
> - this blocks the receiver in a way it can not receive nor emit any data.
> - if you rmmod/modprobe the driver, it won't work until either the
> receiver or the usb subsystem reset the device and the communication
> in some way.
>
> This explains why the 'couple of replug' is working: the usb
> communication is aborted, and the receiver accepts to receive/emit new
> data, so it can be switched to the dj mode.
>
>> This seems to be progress of sorts, though at the moment it does not seem to
>> improve over the 3.2.0 version of the driver, which typically works after a
>> couple of replugs. The intermittent nature of this bug makes testing quite
>> tricky. Will a power cycle wipe out the changes, or are they permanent (or
>> permanent enough) to test through a power cycle?
> A power cycle won't be of any help here. If you want to actually reset
> the receiver, just unplug it and replug it. It will go back to its
> original state.
>
> If you want to directly use the new driver at boot, you will have to
> follow this (I just updated the repo):
> $ git pull
> $ sudo make install
>
> Then reboot. Normally, the mouse should work directly.
>
> However, you must understand that I disabled the enhanced mode of the
> logitech receiver, meaning that it should not be used as a permanent
> fix (please do not close the bug: I won't put this workaround
> upstream).
> So I will come back to you for other tests when I have something.
>
> Cheers,
> Benjamin
>
>> On 14/02/13 10:46, Benjamin Tissoires wrote:
>>> On Thu, Feb 14, 2013 at 10:38 AM, Benjamin Tissoires
>>> <email address hidden> wrote:
>>>> Hi Bob,
>>>>
>>>> mmm... ok, let's try something else.
>>>>
>>>> Please "git pull" again, and insmod the new compiled module. The
>>>> receiver should be in the very same state than when it was during
>>>> grub.
>>>>
>>>> Oh, and I hope you did not experienced the crash during resume with
>>>> the previous version. My bad, sorry if it happened.
>>> And if it's still not working, please unplug and replug the receiver.
>>> It will reset the usb communications that must block the incoming
>>> data.
>>>
>>> Cheers,
>>> Benjamin
>>>
>>>> On Wed, Feb 13, 2013 at 11:09 PM, Bob Bowles <email address hidden>
>>>> wrote:
>>>>> Hi Benjamin
>>>>>
>>>>> The good news is the log has some more information in it this time, the
>>>>> bad
>>>>> news is it is still failing. The error message looks much like it did
>>>>> when
>>>>> it failed before.
>>>>>
>>>>> Please find the log attached.
>>>>>
>>>>>
>>>>> On 13/02/13 20:53, Benjamin Tissoires wrote:
>>>>>> Hi Bob,
>>>>>>
>>>>>> ok, I understood why you are not receiving any events. I made a
>>>>>> temporary fix which should allow you to have the mouse/keyboard
>>>>>> working.
>>>>>>
>>>>>> $ cd hid-logitech-dj/
>>>>>> $ git pull
>>>>>> $ make
>>>>>> $ sudo rmmod hid_logitech_dj ; sudo insmod ./hid-logitech-dj.ko
>>>>>>