Strange regression in hid_llogitech_dj (was: Re: Linux 5.2-rc4)

Strange regression in hid_llogitech_dj (was: Re: Linux 5.2-rc4)

[Rafael J. Wysocki]

On Sunday, June 9, 2019 5:46:48 AM CEST Linus Torvalds wrote:
> No, I'm not confused, and I haven't lost track of what day it is, I do
> actually know that it's still Saturday here, not Sunday, and I'm just
> doing rc4 a bit early because I'll be on an airplane during my normal
> release time. And while I've done releases on airports and airplanes
> before, I looked at my empty queue of pull requests and went "let's
> just do it now".
> 
> We've had a fairly calm release so far, and on the whole that seems to
> hold. rc4 isn't smaller than rc3 was (it's a bit bigger), but rc3 was
> fairly small, so the size increase isn't all that worrisome. I do hope
> that we'll start actually shrinking now, though.

I noticed that the cordless mouse used by me with one of the machines here
stopped to work in 5.2-rc (up to and including the -rc4).

Bisection turned up commit 74808f9115ce ("HID: logitech-dj: add support for non
unifying receivers").

Of course, that commit does not revert cleanly from 5.2-rc4, but I have reverted
the changes made by it in hid/hid-ids.h and I took the version of hid/hid-logitech-dj.c
from commit b6aeeddef68d ("HID: logitech-dj: add logi_dj_recv_queue_unknown_work
helper"), which is the parent of commit 74808f9115ce, and that made the mouse
work again for me.

Here's the output of "dmesg | grep -i logitech" from 5.2-rc4 with the above changes:

[    4.288905] usb 1-2: Manufacturer: Logitech
[    5.444621] input: Logitech USB Receiver as /devices/pci0000:00/0000:00:14.0/usb1/1-2/1-2:1.0/0003:046D:C52F.0002/input/input23
[    5.446960] hid-generic 0003:046D:C52F.0002: input,hidraw1: USB HID v1.11 Mouse [Logitech USB Receiver] on usb-0000:00:14.0-2/input0
[    5.451265] input: Logitech USB Receiver Consumer Control as /devices/pci0000:00/0000:00:14.0/usb1/1-2/1-2:1.1/0003:046D:C52F.0003/input/input24
[    5.507545] hid-generic 0003:046D:C52F.0003: input,hiddev96,hidraw2: USB HID v1.11 Device [Logitech USB Receiver] on usb-0000:00:14.0-2/input1

Please let me know what you need to diagnose this.

Thanks,
Rafael

Re: Strange regression in hid_llogitech_dj (was: Re: Linux 5.2-rc4)

[Jiri Kosina]

On Tue, 11 Jun 2019, Rafael J. Wysocki wrote:

> I noticed that the cordless mouse used by me with one of the machines here
> stopped to work in 5.2-rc (up to and including the -rc4).
> 
> Bisection turned up commit 74808f9115ce ("HID: logitech-dj: add support for non
> unifying receivers").
> 
> Of course, that commit does not revert cleanly from 5.2-rc4, but I have reverted
> the changes made by it in hid/hid-ids.h and I took the version of hid/hid-logitech-dj.c
> from commit b6aeeddef68d ("HID: logitech-dj: add logi_dj_recv_queue_unknown_work
> helper"), which is the parent of commit 74808f9115ce, and that made the mouse
> work again for me.
> 
> Here's the output of "dmesg | grep -i logitech" from 5.2-rc4 with the above changes:
> 
> [    4.288905] usb 1-2: Manufacturer: Logitech
> [    5.444621] input: Logitech USB Receiver as /devices/pci0000:00/0000:00:14.0/usb1/1-2/1-2:1.0/0003:046D:C52F.0002/input/input23
> [    5.446960] hid-generic 0003:046D:C52F.0002: input,hidraw1: USB HID v1.11 Mouse [Logitech USB Receiver] on usb-0000:00:14.0-2/input0
> [    5.451265] input: Logitech USB Receiver Consumer Control as /devices/pci0000:00/0000:00:14.0/usb1/1-2/1-2:1.1/0003:046D:C52F.0003/input/input24
> [    5.507545] hid-generic 0003:046D:C52F.0003: input,hiddev96,hidraw2: USB HID v1.11 Device [Logitech USB Receiver] on usb-0000:00:14.0-2/input1

Hi Rafael,

0x046d/0xc52f is known to have issues in 5.2-rcX. There is a patch queued 
[1] that is believed to fix all this; my plan is to send it to Linus in 
the coming 1-2 days. If you could report whether it fixes the issues 
you've been seeing yourself as well, it'd be helpful.

Thanks.

[1] https://git.kernel.org/pub/scm/linux/kernel/git/hid/hid.git/commit/?h=for-5.2/fixes&id=3ed224e273ac5880eeab4c3043a6b06b0478dd56

-- 
Jiri Kosina
SUSE Labs

Re: Strange regression in hid_llogitech_dj (was: Re: Linux 5.2-rc4)

[Rafael J. Wysocki]

On Wednesday, June 12, 2019 12:02:21 AM CEST Jiri Kosina wrote:
> On Tue, 11 Jun 2019, Rafael J. Wysocki wrote:
> 
> > I noticed that the cordless mouse used by me with one of the machines here
> > stopped to work in 5.2-rc (up to and including the -rc4).
> > 
> > Bisection turned up commit 74808f9115ce ("HID: logitech-dj: add support for non
> > unifying receivers").
> > 
> > Of course, that commit does not revert cleanly from 5.2-rc4, but I have reverted
> > the changes made by it in hid/hid-ids.h and I took the version of hid/hid-logitech-dj.c
> > from commit b6aeeddef68d ("HID: logitech-dj: add logi_dj_recv_queue_unknown_work
> > helper"), which is the parent of commit 74808f9115ce, and that made the mouse
> > work again for me.
> > 
> > Here's the output of "dmesg | grep -i logitech" from 5.2-rc4 with the above changes:
> > 
> > [    4.288905] usb 1-2: Manufacturer: Logitech
> > [    5.444621] input: Logitech USB Receiver as /devices/pci0000:00/0000:00:14.0/usb1/1-2/1-2:1.0/0003:046D:C52F.0002/input/input23
> > [    5.446960] hid-generic 0003:046D:C52F.0002: input,hidraw1: USB HID v1.11 Mouse [Logitech USB Receiver] on usb-0000:00:14.0-2/input0
> > [    5.451265] input: Logitech USB Receiver Consumer Control as /devices/pci0000:00/0000:00:14.0/usb1/1-2/1-2:1.1/0003:046D:C52F.0003/input/input24
> > [    5.507545] hid-generic 0003:046D:C52F.0003: input,hiddev96,hidraw2: USB HID v1.11 Device [Logitech USB Receiver] on usb-0000:00:14.0-2/input1
> 
> Hi Rafael,
> 
> 0x046d/0xc52f is known to have issues in 5.2-rcX. There is a patch queued 
> [1] that is believed to fix all this; my plan is to send it to Linus in 
> the coming 1-2 days. If you could report whether it fixes the issues 
> you've been seeing yourself as well, it'd be helpful.
> 
> Thanks.
> 
> [1] https://git.kernel.org/pub/scm/linux/kernel/git/hid/hid.git/commit/?h=for-5.2/fixes&id=3ed224e273ac5880eeab4c3043a6b06b0478dd56

It kind of helps, but there is a catch.

hid-logitech-dj is not loaded after a fresh boot, so I need to modprobe it manually and that
appears to be blocking (apparently indefinitely) until terminated with ^C.  But then it turns
out that hid-logitech-dj is there in the list of modules and it is in use (by usbhid) and the
mouse works.

I guess I need to update the mkinitrd configuration, but even so that is not exactly
straightforward IMO. :-)

Cheers!

Re: Strange regression in hid_llogitech_dj (was: Re: Linux 5.2-rc4)

[Jiri Kosina]

On Wed, 12 Jun 2019, Rafael J. Wysocki wrote:

> It kind of helps, but there is a catch.
> 
> hid-logitech-dj is not loaded after a fresh boot, so I need to modprobe it manually and that
> appears to be blocking (apparently indefinitely) until terminated with ^C.  But then it turns
> out that hid-logitech-dj is there in the list of modules and it is in use (by usbhid) and the
> mouse works.

My bad, I should've asked you to test with the complete 'for-5.2/fixes' 
branch which contains two reverts [1] [2] that should fix this issue as 
well.

[1] https://git.kernel.org/pub/scm/linux/kernel/git/hid/hid.git/commit/?h=for-5.2/fixes&id=e0b7f9bc0246bc642d1de2ff3ff133730584c956
[2] https://git.kernel.org/pub/scm/linux/kernel/git/hid/hid.git/commit/?h=for-5.2/fixes&id=f9482dabfd1686987cc6044e06ae0e4c05915518

-- 
Jiri Kosina
SUSE Labs

Re: Strange regression in hid_llogitech_dj (was: Re: Linux 5.2-rc4)

[Rafael J. Wysocki]

On Wednesday, June 12, 2019 10:31:45 AM CEST Jiri Kosina wrote:
> On Wed, 12 Jun 2019, Rafael J. Wysocki wrote:
> 
> > It kind of helps, but there is a catch.
> > 
> > hid-logitech-dj is not loaded after a fresh boot, so I need to modprobe it manually and that
> > appears to be blocking (apparently indefinitely) until terminated with ^C.  But then it turns
> > out that hid-logitech-dj is there in the list of modules and it is in use (by usbhid) and the
> > mouse works.
> 
> My bad, I should've asked you to test with the complete 'for-5.2/fixes' 
> branch which contains two reverts [1] [2] that should fix this issue as 
> well.
> 
> [1] https://git.kernel.org/pub/scm/linux/kernel/git/hid/hid.git/commit/?h=for-5.2/fixes&id=e0b7f9bc0246bc642d1de2ff3ff133730584c956
> [2] https://git.kernel.org/pub/scm/linux/kernel/git/hid/hid.git/commit/?h=for-5.2/fixes&id=f9482dabfd1686987cc6044e06ae0e4c05915518

Yes, with the two reverts applied in addition to the fix, it all works as expected.

Thanks!

Re: Strange regression in hid_llogitech_dj (was: Re: Linux 5.2-rc4)

[Jiri Kosina]

On Wed, 12 Jun 2019, Rafael J. Wysocki wrote:

> > > hid-logitech-dj is not loaded after a fresh boot, so I need to modprobe it manually and that
> > > appears to be blocking (apparently indefinitely) until terminated with ^C.  But then it turns
> > > out that hid-logitech-dj is there in the list of modules and it is in use (by usbhid) and the
> > > mouse works.
> > 
> > My bad, I should've asked you to test with the complete 'for-5.2/fixes' 
> > branch which contains two reverts [1] [2] that should fix this issue as 
> > well.
> > 
> > [1] https://git.kernel.org/pub/scm/linux/kernel/git/hid/hid.git/commit/?h=for-5.2/fixes&id=e0b7f9bc0246bc642d1de2ff3ff133730584c956
> > [2] https://git.kernel.org/pub/scm/linux/kernel/git/hid/hid.git/commit/?h=for-5.2/fixes&id=f9482dabfd1686987cc6044e06ae0e4c05915518
> 
> Yes, with the two reverts applied in addition to the fix, it all works as expected.

Rafael, thanks a lot for testing. I am sending the pile to Linus today or 
tomorrow latest.

-- 
Jiri Kosina
SUSE Labs

[Regression] 5.3-rc1: hid_llogitech_dj does not work

[Rafael J. Wysocki]

Hi Hans,

This is similar to a problem I reported some time ago:

https://lore.kernel.org/lkml/2268131.Lc39eCoc3j@kreacher/

and the device affected by it is the same.

The symptom is simply that the mouse just doesn't work (no reaction).  If I do
"rmmod hid_llogitech_dj", it says "Killed", but the module does go away and
the mouse starts to work (through the generic code I suppose), but then
the machine hangs on attempts to suspend (nasty).

Reverting all of the hid_llogitech_dj changes between 5.2 and 5.3-rc1:

dbcbabf7da92 HID: logitech-dj: fix return value of logi_dj_recv_query_hidpp_devices
39d21e7e0043 HID: logitech-dj: make const array template static
423dfbc362b7 HID: logitech-dj: Add usb-id for the 27MHz MX3000 receiver

helps here, but the first two don't really look like they can make any difference,
so I guess I'm an unlucky owner of a MX3000 that doesn't quite work as expected.

Any help will be appreciated. :-)

Cheers,
Rafael

Re: [Regression] 5.3-rc1: hid_llogitech_dj does not work

[Hans de Goede]

Hi Rafael,

On 25-07-19 12:07, Rafael J. Wysocki wrote:
> Hi Hans,
> 
> This is similar to a problem I reported some time ago:
> 
> https://lore.kernel.org/lkml/2268131.Lc39eCoc3j@kreacher/
> 
> and the device affected by it is the same.
> 
> The symptom is simply that the mouse just doesn't work (no reaction).  If I do
> "rmmod hid_llogitech_dj", it says "Killed", but the module does go away and
> the mouse starts to work (through the generic code I suppose), but then
> the machine hangs on attempts to suspend (nasty).
> 
> Reverting all of the hid_llogitech_dj changes between 5.2 and 5.3-rc1:
> 
> dbcbabf7da92 HID: logitech-dj: fix return value of logi_dj_recv_query_hidpp_devices
> 39d21e7e0043 HID: logitech-dj: make const array template static
> 423dfbc362b7 HID: logitech-dj: Add usb-id for the 27MHz MX3000 receiver
> 
> helps here, but the first two don't really look like they can make any difference,
> so I guess I'm an unlucky owner of a MX3000 that doesn't quite work as expected.
> 
> Any help will be appreciated. :-)

Actually we received another bug report about this and the reporter there
has come up with a patch with points to

dbcbabf7da92 HID: logitech-dj: fix return value of logi_dj_recv_query_hidpp_devices

Being the culprit, can you try just reverting that one?

I will take a closer look at this soonish.

Thank & Regards,

Hans

Re: [Regression] 5.3-rc1: hid_llogitech_dj does not work

[Hans de Goede]

Hi Rafael,

On 25-07-19 17:50, Hans de Goede wrote:
> Hi Rafael,
> 
> On 25-07-19 12:07, Rafael J. Wysocki wrote:
>> Hi Hans,
>>
>> This is similar to a problem I reported some time ago:
>>
>> https://lore.kernel.org/lkml/2268131.Lc39eCoc3j@kreacher/
>>
>> and the device affected by it is the same.
>>
>> The symptom is simply that the mouse just doesn't work (no reaction).  If I do
>> "rmmod hid_llogitech_dj", it says "Killed", but the module does go away and
>> the mouse starts to work (through the generic code I suppose), but then
>> the machine hangs on attempts to suspend (nasty).
>>
>> Reverting all of the hid_llogitech_dj changes between 5.2 and 5.3-rc1:
>>
>> dbcbabf7da92 HID: logitech-dj: fix return value of logi_dj_recv_query_hidpp_devices
>> 39d21e7e0043 HID: logitech-dj: make const array template static
>> 423dfbc362b7 HID: logitech-dj: Add usb-id for the 27MHz MX3000 receiver
>>
>> helps here, but the first two don't really look like they can make any difference,
>> so I guess I'm an unlucky owner of a MX3000 that doesn't quite work as expected.
>>
>> Any help will be appreciated. :-)
> 
> Actually we received another bug report about this and the reporter there
> has come up with a patch with points to
> 
> dbcbabf7da92 HID: logitech-dj: fix return value of logi_dj_recv_query_hidpp_devices
> 
> Being the culprit, can you try just reverting that one?
> 
> I will take a closer look at this soonish.

Thank you for reporting this.

After upgrading to 5.3-rc2 I can reproduce this myself and the dbcbabf7da92 commit
indeed is the culprit. I've prepared a fix for this which I'm about to submit upstream.

I've put you in the Cc of the fix.

Regards,

Hans