Bug 209137 - USB is not working since update from 5.0 to 5.3.0-26. And with 5.4 still not working.
Summary: USB is not working since update from 5.0 to 5.3.0-26. And with 5.4 still not ...
Status: NEW
Alias: None
Product: Drivers
Classification: Unclassified
Component: USB (show other bugs)
Hardware: All Linux
: P1 blocking
Assignee: Default virtual assignee for Drivers/USB
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2020-09-02 21:25 UTC by Philipp
Modified: 2020-10-09 08:37 UTC (History)
2 users (show)

See Also:
Kernel Version: 5.4
Subsystem:
Regression: No
Bisected commit-id:


Attachments
journalctl -b -1 (68.00 KB, text/plain)
2020-09-02 21:25 UTC, Philipp
Details
journalctl -xe (153.11 KB, text/plain)
2020-09-11 09:06 UTC, Philipp
Details
dmesg (63.10 KB, text/plain)
2020-09-11 09:07 UTC, Philipp
Details
'dmesg' for 5.0 for comparison (103.45 KB, text/plain)
2020-10-07 07:29 UTC, Philipp
Details
lsusb -v (86.41 KB, text/plain)
2020-10-07 20:37 UTC, Philipp
Details
report descriptor of a working C52B receiver (3.85 KB, text/plain)
2020-10-09 08:37 UTC, Benjamin Tissoires
Details

Description Philipp 2020-09-02 21:25:46 UTC
Created attachment 292309 [details]
journalctl -b -1

USB is not working since update from 5.0 to 5.3.0-26. And with 5.4 still not working.

It looks like everything is frozen, but also ALT+PRINT REISUB is not working, so I the new kernel versions have problems with the USB system. If I choose in GRUB an older kernel version (e.g. 5.0.0) everything is fine.
Same problem with kernel 5.3 and 5.4.
Comment 1 nvaert1986 2020-09-07 07:12:45 UTC
(In reply to Philipp from comment #0)
> Created attachment 292309 [details]
> journalctl -b -1
> 
> USB is not working since update from 5.0 to 5.3.0-26. And with 5.4 still not
> working.
> 
> It looks like everything is frozen, but also ALT+PRINT REISUB is not
> working, so I the new kernel versions have problems with the USB system. If
> I choose in GRUB an older kernel version (e.g. 5.0.0) everything is fine.
> Same problem with kernel 5.3 and 5.4.

You kernel seems to detect various USB ports and hubs, so the kernel configuration seems to be fine and stuff seems to get detected. 

Can you provide a 'journalctl -xe' or a 'dmesg' after booting (using a PS/2 keyboard for example or via SSH)? That way we can see whether one of the the USB kernel modules crashes. From what I can see you're using a wireless keyboard. Have you tried using a wired keyboard just to be sure and test whether that's working properly?
Comment 2 Philipp 2020-09-11 09:06:54 UTC
Created attachment 292465 [details]
journalctl -xe

journalctl -xe via ssh
Comment 3 Philipp 2020-09-11 09:07:30 UTC
Created attachment 292467 [details]
dmesg

'dmesg' via ssh
Comment 4 Philipp 2020-09-11 09:10:20 UTC
I just connect via ssh and did 'journalctl -xe' and 'dmesg'.

>Have you tried using a wired keyboard just to be sure and test whether that's
>working properly?<
I don't have a wired keyboard any more. But the keyboard is working properly, because it is working right now with the old kernel 5.0 (on the same pc) and also using my windows laptop.
Comment 5 Philipp 2020-09-28 09:09:10 UTC
If I connect a wired mouse the mouse is working, so USB is not the problem.
The new kernel just does not accept the logitech wireless mouse and keyboard (K520 and M310) but the old does. Hopefully the bug can be fixed soon.
Comment 6 Alan Stern 2020-09-28 18:22:28 UTC
You should be more careful about what you write.  In the original bug description you said that USB isn't working, but in comment #5 you said that USB isn't the problem.  Try to be more consistent.

Your dmesg log from the 5.4 kernel does show some errors from the HID driver.  You should collect a similar dmesg log under a 5.0 kernel for comparison.

Another thing that would help is the output from running as superuser "lsusb -v" for the Logitech receiver.  Under either 5.0 or 5.4, the output should be the same.
Comment 7 Philipp 2020-09-28 20:30:45 UTC
I didn't had a wired mouse as I wrote report the bug first (see date and conversation). Therefor as time went on knowledge changes. On 28th of Sept I connected a wired mouse the first time, because I didn't had a wired mouse before. Thus there is no problem with consistency. I'm happy of any help that I can get with this problem but I'm not a child which should be monished and you should be a bit more careful with your conclusions about "consistency".

..I will do dmesg log soon. Thx!
Comment 8 Philipp 2020-10-07 07:29:53 UTC
Created attachment 292879 [details]
'dmesg' for 5.0 for comparison

I just uploeded 'dmesg' for 5.0.
'dmesg' for 5.0 doesn't not show the HID driver error as 5.4 did.
Comment 9 Alan Stern 2020-10-07 20:29:52 UTC
As mentioned in comment #6, can you also provide the output from "lsusb -v" (run as root) for the Logitech receiver?  It should be the same for 5.0 and 5.4, so only one of them is needed.
Comment 10 Philipp 2020-10-07 20:37:12 UTC
Created attachment 292897 [details]
lsusb -v

Oh, sorry for that. Just forgot it. Here is the 'lsusb -v'.
Comment 11 Alan Stern 2020-10-08 14:35:56 UTC
Did you run lsusb as root?  I ask because the report descriptors are missing from the output, and this is typically what happens when lsusb is run as a normal user, not root.

Also, the dmesg logs indicate that you weren't running the -stable kernel versions.  It's possible that this problem has already been fixed in those versions.  For the same reason, you might try running a 5.8 kernel.

If none of those work, you can always try running git-bisect to find the particular commit which caused the problem to appear somewhere between 5.0 and 5.4.
Comment 12 Philipp 2020-10-08 16:44:02 UTC
Yes. I run lsusb as root. I run ´sudo lsusb -v > lsusb.txt´ and uploaded lsusb.txt

I`m using Ubuntu and kernel 5.8 was not indicated via automatical update. Never installed an other kernel by myself especially because my pc is my working pc and therefor I`m usually not doing experiments. Just tested other live CDs like the newest fedora or the latest debian, to see, if there is the same problem.

I ordered a wired keyboard and will bypass this bug.
Comment 13 Benjamin Tissoires 2020-10-09 08:37:53 UTC
Created attachment 292905 [details]
report descriptor of a working C52B receiver

Regarding the `lsub -v` not working, could you:
- boot on the working kernel
- use hid-recorder (as root) from https://gitlab.freedesktop.org/libevdev/hid-tools/ to record all the "Logitech USB Receiver" that shows up in the list

For a point of comparison, I attached the report descriptors of the working C52B I have here (only the wireless interface).

If there is no differences between your dumps and mine, then there is something wrong happening in the usbhid driver or lower, and we will need more logs at that point.

FTR, this Logitech receiver I have here works fine with 5.8.4 on Fedora 32...

Note You need to log in before you can comment on or make changes to this bug.