Re: [PATCH 00/18] HID: logitech-hidpp: G920 force-feedback fixes, rework, and cleanups

[Gideon Heinz]

Hi Dmitry,

I tested the full 18-patch series on my G920 (046d:c262) as you asked.
Everything on your list works for me, details below.

Setup: Linux 7.3.1-1-cachyos (clang build). The series applies cleanly
to v7.3's hid-logitech-hidpp.c (the pre-image blob of patch 1 matches),
so I built only that file as an out-of-tree module with all 18 patches
applied, against the 7.3.1 headers, and loaded it with insmod. No full
kernel build.

Results:

 - Initialization: "Force feedback support loaded (firmware release
   4)" on every plug-in (7 in total), followed by the input node and
   "HID++ 4.2 device connected". No -110.

 - fftest: all effect types are listed, uploads are OK, and the
   constant effect plays and produces force. After exiting with -1 the
   log stays clean for 2+ minutes. With only your earlier two-line init
   fix, the same exit produced "Failed to send command to device!"
   about 15 s later, so the drain-before-close change fixes that for me.

 - range sysfs: with evtest holding the device open, writing 540 and
   then 900 reads back 540 and 900 and returns instantly. With nothing
   holding the device open, writes stall as you described: I got two
   "Failed to send command to device!" lines 15.36 s apart (22:06:16
   and 22:06:32). I believe they come from two range writes I did with
   the device closed, but I ran those before starting that log, so I
   can't tie them to a timestamp.

 - Input: evtest shows the wheel axis, hat, buttons and the full FF
   capability list, and everything responds. I use other pedals, so the
   G920 pedals are untested.

 - Replug x4: "Unloading HID++ force feedback" then a normal reload
   each time, nothing else in dmesg.

 - Unplug while evtest had the device open: "Unloading HID++ force
   feedback" about 17 ms after "USB disconnect", nothing else in dmesg.

 - Unplug while fftest had the device open (6 effects loaded): 12
   "Failed to send command to device!" lines within 25 us of each
   other, about 0.3 ms after "USB disconnect", i.e. they fail
   immediately without the 3 x 5 s retries. "Unloading HID++ force
   feedback" follows 17 ms later. No warnings, oops, call traces or
   hangs in the full kernel log. After I plugged the wheel back in, FF
   loaded normally.

Not tested: in-game FFB with the series, driver unbind or module
unload with effects loaded (the case you said is still open), and the
G920 pedals.

Happy to test follow-ups, and I can send the full logs if useful.

Tested-by: Gideon Heinz <gideonheinz02@xxxxxxxxx>