qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
* [Qemu-devel] Windows vioinput guest driver tablet input bug
@ 2018-02-25 10:21 Justin Gatzen
  2018-02-25 11:41 ` Justin Gatzen
  0 siblings, 1 reply; 3+ messages in thread
From: Justin Gatzen @ 2018-02-25 10:21 UTC (permalink / raw)
  To: qemu-devel; +Cc: Ladi Prosek

Hi,

The vioinput Windows guest driver does not seem to work with mouse wheel or
side buttons for virtio-tablet-pci devices, while virtio-mouse-pci works as
expected. Linux guest drivers are unaffected. Tablets are being categorized as
mice due to the EXPOSE_ABS_AXES_WITH_BUTTONS_AS_MOUSE compile flag. But when
there are no relative REL_X or REL_Y input axes the driver ignores all other
relative axis entries. Mouse wheel is one such relative axis. I've attached a
proof of concept fix for the guest driver that seems to fix both the mouse
wheel and side buttons but I believe this needs a more robust fix and deeper
look at the HIDMouseProbe function.

Thanks,
Justin


diff --git a/vioinput/sys/HidMouse.c b/vioinput/sys/HidMouse.c
index 69dac235..ea188991 100644
--- a/vioinput/sys/HidMouse.c
+++ b/vioinput/sys/HidMouse.c
@@ -325,7 +325,11 @@ HIDMouseProbe(
     DynamicArrayReserve(&AxisMap, AXIS_MAP_INITIAL_LENGTH * 2 * sizeof(ULONG));

     // Windows won't drive a mouse without at least the X and Y relative axes
-    if (InputCfgDataHasBit(pRelAxes, REL_X) &&
InputCfgDataHasBit(pRelAxes, REL_Y))
+    if (InputCfgDataHasBit(pRelAxes, REL_X) &&
InputCfgDataHasBit(pRelAxes, REL_Y)
+#ifdef EXPOSE_ABS_AXES_WITH_BUTTONS_AS_MOUSE
+        || (pMouseDesc->uNumOfButtons > 0 &&
InputCfgDataHasBit(pAbsAxes, ABS_X) && InputCfgDataHasBit(pAbsAxes,
ABS_Y))
+#endif // EXPOSE_ABS_AXES_WITH_BUTTONS_AS_MOUSE
+        )
     {
         for (i = 0; i < pRelAxes->size; i++)
         {
@@ -391,7 +395,7 @@ HIDMouseProbe(
     }

 #ifdef EXPOSE_ABS_AXES_WITH_BUTTONS_AS_MOUSE
-    if (uNumOfRelAxes == 0 &&
+    if ((!InputCfgDataHasBit(pRelAxes, REL_X) ||
!InputCfgDataHasBit(pRelAxes, REL_Y)) &&
         pMouseDesc->uNumOfButtons > 0 &&
         InputCfgDataHasBit(pAbsAxes, ABS_X) &&
InputCfgDataHasBit(pAbsAxes, ABS_Y))
     {

^ permalink raw reply related	[flat|nested] 3+ messages in thread

* Re: [Qemu-devel] Windows vioinput guest driver tablet input bug
  2018-02-25 10:21 [Qemu-devel] Windows vioinput guest driver tablet input bug Justin Gatzen
@ 2018-02-25 11:41 ` Justin Gatzen
  2018-02-25 15:32   ` Yan Vugenfirer
  0 siblings, 1 reply; 3+ messages in thread
From: Justin Gatzen @ 2018-02-25 11:41 UTC (permalink / raw)
  To: qemu-devel; +Cc: Ladi Prosek

The last email's diff got mangled due to long lines. The original
source has long lines still but it shouldnt get cut off his time.

Justin,


diff --git a/vioinput/sys/HidMouse.c b/vioinput/sys/HidMouse.c
index 69dac235..528bd3d6 100644
--- a/vioinput/sys/HidMouse.c
+++ b/vioinput/sys/HidMouse.c
@@ -325,7 +325,12 @@ HIDMouseProbe(
     DynamicArrayReserve(&AxisMap, AXIS_MAP_INITIAL_LENGTH * 2 * sizeof(ULONG));
 
     // Windows won't drive a mouse without at least the X and Y relative axes
-    if (InputCfgDataHasBit(pRelAxes, REL_X) && InputCfgDataHasBit(pRelAxes, REL_Y))
+    if (InputCfgDataHasBit(pRelAxes, REL_X) && InputCfgDataHasBit(pRelAxes, REL_Y)
+#ifdef EXPOSE_ABS_AXES_WITH_BUTTONS_AS_MOUSE
+        || (pMouseDesc->uNumOfButtons > 0 && InputCfgDataHasBit(pAbsAxes, ABS_X)
+            && InputCfgDataHasBit(pAbsAxes, ABS_Y))
+#endif // EXPOSE_ABS_AXES_WITH_BUTTONS_AS_MOUSE
+        )
     {
         for (i = 0; i < pRelAxes->size; i++)
         {
@@ -391,7 +396,8 @@ HIDMouseProbe(
     }
 
 #ifdef EXPOSE_ABS_AXES_WITH_BUTTONS_AS_MOUSE
-    if (uNumOfRelAxes == 0 &&
+    if ((!InputCfgDataHasBit(pRelAxes, REL_X) ||
+         !InputCfgDataHasBit(pRelAxes, REL_Y)) &&
         pMouseDesc->uNumOfButtons > 0 &&
         InputCfgDataHasBit(pAbsAxes, ABS_X) && InputCfgDataHasBit(pAbsAxes, ABS_Y))
     {


On Sun, Feb 25, 2018 at 05:21:29AM -0500, Justin Gatzen wrote:
> Hi,
> 
> The vioinput Windows guest driver does not seem to work with mouse wheel or
> side buttons for virtio-tablet-pci devices, while virtio-mouse-pci works as
> expected. Linux guest drivers are unaffected. Tablets are being categorized as
> mice due to the EXPOSE_ABS_AXES_WITH_BUTTONS_AS_MOUSE compile flag. But when
> there are no relative REL_X or REL_Y input axes the driver ignores all other
> relative axis entries. Mouse wheel is one such relative axis. I've attached a
> proof of concept fix for the guest driver that seems to fix both the mouse
> wheel and side buttons but I believe this needs a more robust fix and deeper
> look at the HIDMouseProbe function.
> 
> Thanks,
> Justin
> 
> 
> diff --git a/vioinput/sys/HidMouse.c b/vioinput/sys/HidMouse.c
> index 69dac235..ea188991 100644
> --- a/vioinput/sys/HidMouse.c
> +++ b/vioinput/sys/HidMouse.c
> @@ -325,7 +325,11 @@ HIDMouseProbe(
>      DynamicArrayReserve(&AxisMap, AXIS_MAP_INITIAL_LENGTH * 2 * sizeof(ULONG));
> 
>      // Windows won't drive a mouse without at least the X and Y relative axes
> -    if (InputCfgDataHasBit(pRelAxes, REL_X) &&
> InputCfgDataHasBit(pRelAxes, REL_Y))
> +    if (InputCfgDataHasBit(pRelAxes, REL_X) &&
> InputCfgDataHasBit(pRelAxes, REL_Y)
> +#ifdef EXPOSE_ABS_AXES_WITH_BUTTONS_AS_MOUSE
> +        || (pMouseDesc->uNumOfButtons > 0 &&
> InputCfgDataHasBit(pAbsAxes, ABS_X) && InputCfgDataHasBit(pAbsAxes,
> ABS_Y))
> +#endif // EXPOSE_ABS_AXES_WITH_BUTTONS_AS_MOUSE
> +        )
>      {
>          for (i = 0; i < pRelAxes->size; i++)
>          {
> @@ -391,7 +395,7 @@ HIDMouseProbe(
>      }
> 
>  #ifdef EXPOSE_ABS_AXES_WITH_BUTTONS_AS_MOUSE
> -    if (uNumOfRelAxes == 0 &&
> +    if ((!InputCfgDataHasBit(pRelAxes, REL_X) ||
> !InputCfgDataHasBit(pRelAxes, REL_Y)) &&
>          pMouseDesc->uNumOfButtons > 0 &&
>          InputCfgDataHasBit(pAbsAxes, ABS_X) &&
> InputCfgDataHasBit(pAbsAxes, ABS_Y))
>      {

^ permalink raw reply related	[flat|nested] 3+ messages in thread

* Re: [Qemu-devel] Windows vioinput guest driver tablet input bug
  2018-02-25 11:41 ` Justin Gatzen
@ 2018-02-25 15:32   ` Yan Vugenfirer
  0 siblings, 0 replies; 3+ messages in thread
From: Yan Vugenfirer @ 2018-02-25 15:32 UTC (permalink / raw)
  To: Justin Gatzen; +Cc: qemu-devel, Ladi Prosek

Hello Justin,

Thanks a lot for the patch proposal.
I opened the issue on virtio-win GitHub page to track the issue: https://github.com/virtio-win/kvm-guest-drivers-windows/issues/249

Best regards,
Yan.

> On 25 Feb 2018, at 13:41, Justin Gatzen <justin.gatzen@gmail.com> wrote:
> 
> The last email's diff got mangled due to long lines. The original
> source has long lines still but it shouldnt get cut off his time.
> 
> Justin,
> 
> 
> diff --git a/vioinput/sys/HidMouse.c b/vioinput/sys/HidMouse.c
> index 69dac235..528bd3d6 100644
> --- a/vioinput/sys/HidMouse.c
> +++ b/vioinput/sys/HidMouse.c
> @@ -325,7 +325,12 @@ HIDMouseProbe(
>     DynamicArrayReserve(&AxisMap, AXIS_MAP_INITIAL_LENGTH * 2 * sizeof(ULONG));
> 
>     // Windows won't drive a mouse without at least the X and Y relative axes
> -    if (InputCfgDataHasBit(pRelAxes, REL_X) && InputCfgDataHasBit(pRelAxes, REL_Y))
> +    if (InputCfgDataHasBit(pRelAxes, REL_X) && InputCfgDataHasBit(pRelAxes, REL_Y)
> +#ifdef EXPOSE_ABS_AXES_WITH_BUTTONS_AS_MOUSE
> +        || (pMouseDesc->uNumOfButtons > 0 && InputCfgDataHasBit(pAbsAxes, ABS_X)
> +            && InputCfgDataHasBit(pAbsAxes, ABS_Y))
> +#endif // EXPOSE_ABS_AXES_WITH_BUTTONS_AS_MOUSE
> +        )
>     {
>         for (i = 0; i < pRelAxes->size; i++)
>         {
> @@ -391,7 +396,8 @@ HIDMouseProbe(
>     }
> 
> #ifdef EXPOSE_ABS_AXES_WITH_BUTTONS_AS_MOUSE
> -    if (uNumOfRelAxes == 0 &&
> +    if ((!InputCfgDataHasBit(pRelAxes, REL_X) ||
> +         !InputCfgDataHasBit(pRelAxes, REL_Y)) &&
>         pMouseDesc->uNumOfButtons > 0 &&
>         InputCfgDataHasBit(pAbsAxes, ABS_X) && InputCfgDataHasBit(pAbsAxes, ABS_Y))
>     {
> 
> 
> On Sun, Feb 25, 2018 at 05:21:29AM -0500, Justin Gatzen wrote:
>> Hi,
>> 
>> The vioinput Windows guest driver does not seem to work with mouse wheel or
>> side buttons for virtio-tablet-pci devices, while virtio-mouse-pci works as
>> expected. Linux guest drivers are unaffected. Tablets are being categorized as
>> mice due to the EXPOSE_ABS_AXES_WITH_BUTTONS_AS_MOUSE compile flag. But when
>> there are no relative REL_X or REL_Y input axes the driver ignores all other
>> relative axis entries. Mouse wheel is one such relative axis. I've attached a
>> proof of concept fix for the guest driver that seems to fix both the mouse
>> wheel and side buttons but I believe this needs a more robust fix and deeper
>> look at the HIDMouseProbe function.
>> 
>> Thanks,
>> Justin
>> 
>> 
>> diff --git a/vioinput/sys/HidMouse.c b/vioinput/sys/HidMouse.c
>> index 69dac235..ea188991 100644
>> --- a/vioinput/sys/HidMouse.c
>> +++ b/vioinput/sys/HidMouse.c
>> @@ -325,7 +325,11 @@ HIDMouseProbe(
>>     DynamicArrayReserve(&AxisMap, AXIS_MAP_INITIAL_LENGTH * 2 * sizeof(ULONG));
>> 
>>     // Windows won't drive a mouse without at least the X and Y relative axes
>> -    if (InputCfgDataHasBit(pRelAxes, REL_X) &&
>> InputCfgDataHasBit(pRelAxes, REL_Y))
>> +    if (InputCfgDataHasBit(pRelAxes, REL_X) &&
>> InputCfgDataHasBit(pRelAxes, REL_Y)
>> +#ifdef EXPOSE_ABS_AXES_WITH_BUTTONS_AS_MOUSE
>> +        || (pMouseDesc->uNumOfButtons > 0 &&
>> InputCfgDataHasBit(pAbsAxes, ABS_X) && InputCfgDataHasBit(pAbsAxes,
>> ABS_Y))
>> +#endif // EXPOSE_ABS_AXES_WITH_BUTTONS_AS_MOUSE
>> +        )
>>     {
>>         for (i = 0; i < pRelAxes->size; i++)
>>         {
>> @@ -391,7 +395,7 @@ HIDMouseProbe(
>>     }
>> 
>> #ifdef EXPOSE_ABS_AXES_WITH_BUTTONS_AS_MOUSE
>> -    if (uNumOfRelAxes == 0 &&
>> +    if ((!InputCfgDataHasBit(pRelAxes, REL_X) ||
>> !InputCfgDataHasBit(pRelAxes, REL_Y)) &&
>>         pMouseDesc->uNumOfButtons > 0 &&
>>         InputCfgDataHasBit(pAbsAxes, ABS_X) &&
>> InputCfgDataHasBit(pAbsAxes, ABS_Y))
>>     {
> 

^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2018-02-25 15:32 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2018-02-25 10:21 [Qemu-devel] Windows vioinput guest driver tablet input bug Justin Gatzen
2018-02-25 11:41 ` Justin Gatzen
2018-02-25 15:32   ` Yan Vugenfirer

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).