[Qemu-devel] [PATCH] fix usb-hid SET_IDLE behaviour

[Qemu-devel] [PATCH] fix usb-hid SET_IDLE behaviour

[Stefano Stabellini]

Hi all,
the usb-hid spec states that the SET_IDLE request has a 16bit value,
where the upper byte specifies the idle rate (currently unimplemented,
we handle only the 0 case, meaning infinite duration) and the lower
byte specifies the report id (0 means all reports).
In our code we do idle = value, while it should be idle = "upper byte",
especially if the guest issues a GET_IDLE, we should return only the
idle rate while we are returning only the report id.
In practice it doesn't make much difference because I have only seen
SET_VALUE with both bytes set to 0 so far, but still it is wrong.

Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>

---

diff --git a/hw/usb-hid.c b/hw/usb-hid.c
index 972543f..76fdce6 100644
--- a/hw/usb-hid.c
+++ b/hw/usb-hid.c
@@ -65,7 +65,7 @@ typedef struct USBHIDState {
     };
     int kind;
     int protocol;
-    int idle;
+    uint8_t idle;
     int changed;
     void *datain_opaque;
     void (*datain)(void *);
@@ -794,7 +794,7 @@ static int usb_hid_handle_control(USBDevice *dev, int request, int value,
         data[0] = s->idle;
         break;
     case SET_IDLE:
-        s->idle = value;
+        s->idle = (uint8_t) (value >> 8);
         ret = 0;
         break;
     default:

Re: [Qemu-devel] [PATCH] fix usb-hid SET_IDLE behaviour

[Ian Jackson]

Stefano Stabellini writes ("[Qemu-devel] [PATCH] fix usb-hid SET_IDLE behaviour"):
> the usb-hid spec states that the SET_IDLE request has a 16bit value,
> where the upper byte specifies the idle rate (currently unimplemented,
> we handle only the 0 case, meaning infinite duration) and the lower
> byte specifies the report id (0 means all reports).
> In our code we do idle = value, while it should be idle = "upper byte",
> especially if the guest issues a GET_IDLE, we should return only the
> idle rate while we are returning only the report id.
> In practice it doesn't make much difference because I have only seen
> SET_VALUE with both bytes set to 0 so far, but still it is wrong.
> 
> Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>

Stefano's patch looks good to me.

Reviewed-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>

Ian.

Re: [Qemu-devel] [PATCH] fix usb-hid SET_IDLE behaviour

[Alexander Graf]

Ian Jackson wrote:
> Stefano Stabellini writes ("[Qemu-devel] [PATCH] fix usb-hid SET_IDLE behaviour"):
>   
>> the usb-hid spec states that the SET_IDLE request has a 16bit value,
>> where the upper byte specifies the idle rate (currently unimplemented,
>> we handle only the 0 case, meaning infinite duration) and the lower
>> byte specifies the report id (0 means all reports).
>> In our code we do idle = value, while it should be idle = "upper byte",
>> especially if the guest issues a GET_IDLE, we should return only the
>> idle rate while we are returning only the report id.
>> In practice it doesn't make much difference because I have only seen
>> SET_VALUE with both bytes set to 0 so far, but still it is wrong.
>>
>> Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
>>     
>
> Stefano's patch looks good to me.
>
> Reviewed-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
>   

Shouldn't that be Ian rather than Stefano? :-)

Alex

Re: [Qemu-devel] [PATCH] fix usb-hid SET_IDLE behaviour

[Ian Jackson]

Alexander Graf writes ("Re: [Qemu-devel] [PATCH] fix usb-hid SET_IDLE behaviour"):
> Ian Jackson wrote:
> > Reviewed-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
> 
> Shouldn't that be Ian rather than Stefano? :-)

Sorry, I'm an idiot :-).

Ian.

Re: [Qemu-devel] [PATCH] fix usb-hid SET_IDLE behaviour

[Anthony Liguori]

Stefano Stabellini wrote:
> Hi all,
> the usb-hid spec states that the SET_IDLE request has a 16bit value,
> where the upper byte specifies the idle rate (currently unimplemented,
> we handle only the 0 case, meaning infinite duration) and the lower
> byte specifies the report id (0 means all reports).
> In our code we do idle = value, while it should be idle = "upper byte",
> especially if the guest issues a GET_IDLE, we should return only the
> idle rate while we are returning only the report id.
> In practice it doesn't make much difference because I have only seen
> SET_VALUE with both bytes set to 0 so far, but still it is wrong.
>
> Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
>   

Applied. Thanks.

Regards,

Anthony Liguori