[PATCH 2/2] hid: Don't sign-extend 16-bit values.

[Kai Krakow]

FWIW, this only shows on the horizontal axis, not the vertical. This makes
it even stranger...

Am So., 10. Feb. 2019, 22:04 hat Aric Stewart <aric at codeweavers.com>
geschrieben:

> I am noticing this as well with one of my games and a ps4 pad. I have
> brought it up to Zeb but have not had a chance to dig.
>
> And my mirror problem did go away with this patch reverted.
>
> -aric
>
> > On Feb 10, 2019, at 1:10 PM, Kai Krakow <kai at kaishome.de> wrote:
> >
> > Hello again!
> >
> > I'm not sure if this commit is the problem, with it reverted I still
> > see strange effects: Moving the left stick only half-way right still
> > works correctly, moving it further turns it mirrored into left. Have
> > there been any other changes related to this? It worked fine in 4.0,
> > and the only other related change I could find here in my system was
> > updating from libSDL2-2.0.8 to 2.0.9 but I AFAIR this version still
> > worked with 4.0 correctly.
> >
> > Thanks,
> > Kai
> >
> >> Am So., 10. Feb. 2019 um 19:02 Uhr schrieb Kai Krakow <kai at kaishome.de
> >:
> >>
> >> Hello!
> >>
> >> This change seems to introduce odd behavior in at least some games:
> >> Gamepad axis are mirrored to one side only now. This happens in Shadow
> >> of War (found by me) and Rocket League (reported by a user). I expect
> >> other games to fail in a similar way but didn't test those yet.
> >>
> >> I guess the clue is in "_may_ report a logical range"...
> >>
> >> I'm using the xpadneo driver with an Xbox One S wireless controller,
> >> and it reports -32768..32767 according to the driver (which follows
> >> the Linux spec on joysticks for this). This used to work perfectly
> >> before this change.
> >>
> >> Regards,
> >> Kai
> >>
> >>> Am Mi., 6. Feb. 2019 um 19:42 Uhr schrieb Aric Stewart <
> aric at codeweavers.com>:
> >>>
> >>> Signed-off-by: Aric Stewart <aric at codeweavers.com>
> >>>
> >>>> On 2/5/19 1:09 PM, Zebediah Figura wrote:
> >>>> From: Zebediah Figura <zfigura at codeweavers.com>
> >>>>
> >>>> Some controllers (including, with the previous patch, any reported
> >>>> through SDL) may report a logical range of [0,65535], which takes up
> >>>> 16 bits but should not be sign-extended.
> >>>>
> >>>> Signed-off-by: Zebediah Figura <zfigura at codeweavers.com>
> >>>> ---
> >>>>  dlls/hid/hidp.c | 4 ----
> >>>>  1 file changed, 4 deletions(-)
> >>>>
> >>>> diff --git a/dlls/hid/hidp.c b/dlls/hid/hidp.c
> >>>> index f9978038e3..15d827edf1 100644
> >>>> --- a/dlls/hid/hidp.c
> >>>> +++ b/dlls/hid/hidp.c
> >>>> @@ -277,8 +277,6 @@ NTSTATUS WINAPI
> HidP_GetScaledUsageValue(HIDP_REPORT_TYPE ReportType, USAGE Usag
> >>>>                               element->valueStartBit,
> element->bitCount, &rawValue);
> >>>>          if (rc != HIDP_STATUS_SUCCESS)
> >>>>              return rc;
> >>>> -        if (element->caps.value.BitSize == 16)
> >>>> -            rawValue = (short)rawValue;
> >>>>          *UsageValue = rawValue;
> >>>>      }
> >>>>
> >>>> @@ -925,8 +923,6 @@ NTSTATUS WINAPI HidP_GetData(HIDP_REPORT_TYPE
> ReportType, HIDP_DATA *DataList, U
> >>>>                                       element->valueStartBit,
> element->bitCount, &v);
> >>>>                  if (rc != HIDP_STATUS_SUCCESS)
> >>>>                      return rc;
> >>>> -                if (element->caps.value.BitSize == 16)
> >>>> -                    v = (short)v;
> >>>>                  DataList[uCount].DataIndex =
> element->caps.value.u.NotRange.DataIndex;
> >>>>                  DataList[uCount].u.RawValue = v;
> >>>>              }
> >>>>
> >>>
> >>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.winehq.org/pipermail/wine-devel/attachments/20190210/2561b5c9/attachment.html>