Hi!
> > On 27.10.22 00:42, Alexander Zhang wrote:
> > > On 8/11/22 8:27 AM, Benjamin Tissoires wrote:
> > > > On Thu, Aug 4, 2022 at 8:00 PM José Expósito
> > > > <jose.exposito89@xxxxxxxxx> wrote:
> > > > > On Thu, Aug 04, 2022 at 05:18:32PM +0200, Torge Matthies wrote:
> > > > > > Commit f7d8e387d9ae ("HID: uclogic: Switch to Digitizer usage for
> > > > > > styluses") broke input from my XP-Pen Star G640. This is because the
> > > > > > "Digitizer" usage is not recognized as a valid usage for input devices.
> > > > > >
> > > > > > This patch changes the IS_INPUT_APPLICATION macro so that the
> > > > > > "Digitizer"
> > > > > > (HID_DG_DIGITIZER) usage is recognized as an input device usage.
> > > > > >
> > > > > > Fixes: f7d8e387d9ae ("HID: uclogic: Switch to Digitizer usage for
> > > > > > styluses")
> > > > > > Signed-off-by: Torge Matthies <openglfreak@xxxxxxxxxxxxxx>
> > > > > > ---
> > > > > > This patch could be risky, because any digitizer devices that were
> > > > > > previously not treated as input devices are now used for input.
> > > > > > Alternatively the linked commit could be reverted, but that would
> > > > > > re-introduce the problem detailed in its commit message.
> > > > > >
> > > > > > include/linux/hid.h | 2 +-
> > > > > > 1 file changed, 1 insertion(+), 1 deletion(-)
> > > > >
> > > > > I hesitated about this when I sent the patch you mentioned. In the end,
> > > > > I didn't include any fix because the digitizer use was tested for 2
> > > > > years in DIGImend, so I (wrongly) assumed that it was safe enough.
> > > > >
> > > > > However, my initial thought was to add in uclogic_probe():
> > > > >
> > > > > hdev->quirks |= HID_QUIRK_HIDINPUT_FORCE;
> > > > > + hdev->quirks |= HID_QUIRK_HIDINPUT_FORCE;
> > > > >
> > > > > Let's see if we can hear more opinions, but if you are worried about
> > > > > affecting other drivers, that could be a good solution.
> > > >
> > > > Sadly, my automated regression tests are broken for a while and I
> > > > haven't checked if that patch is introducing errors in hid-multitouch.
> > > >
> > > > FWIW, this part has always been painful because some tablets were not
> > > > using the correct usages. And so that's why we are ending up in that
> > > > weird situation.
> > > >
> > > > Anyway, just to mention that any code that touches this part should be
> > > > tested against the hid regression tests suite[0], because that's the
> > > > only way to find out if the change is affecting other devices.
Since it seems like this patch is kind of stuck, I sent you a different
patch [1] that aims to fix the same issue but using a more conservative
approach.
Could you test it and confirm that it fixes your problem, please?
Hopefully, we'd manage to get it merged as the risk of doing so would
be smaller.
Jose
[1] https://lore.kernel.org/linux-input/20221110174056.393697-1-jose.exposito89@xxxxxxxxx/T/
