[Qemu-devel] vmmouse

[Qemu-devel] vmmouse

[Rick Vernam]

For the past few days (10-15, longer?) both my windows vms seem to have broken 
mice - the mouse cursor does not move in the guest...
Both guests run windows (2k,xp) and both have vmware mouse driver installed 
(not sure which version, off the top of my head).

I've not followed the list too closely during this time period either - in 
trying to catch up tonight I haven't seen anything on this subject, so I'm 
posting to see if this is a known issue...

I built qemu with DEBUG_VMMOUSE turned on in hw/vmmouse.c.
When I run qemu I get a single vmmouse_init right when qemu starts, but then 
no additional output.

Perhaps this is a separate issue, but when I "usb_add tablet" the mouse cursor 
becomes constrained to the upper left corner of the guests' screen.

Any ideas about what might be going on?
Thanks

Re: [Qemu-devel] vmmouse

[Juergen Lock]

In article <200901282248.26006.rickv@hobi.com> you write:
>For the past few days (10-15, longer?) both my windows vms seem to have broken 
>mice - the mouse cursor does not move in the guest...
>Both guests run windows (2k,xp) and both have vmware mouse driver installed 
>(not sure which version, off the top of my head).
>
>I've not followed the list too closely during this time period either - in 
>trying to catch up tonight I haven't seen anything on this subject, so I'm 
>posting to see if this is a known issue...
>
>I built qemu with DEBUG_VMMOUSE turned on in hw/vmmouse.c.
>When I run qemu I get a single vmmouse_init right when qemu starts, but then 
>no additional output.
>
>Perhaps this is a separate issue, but when I "usb_add tablet" the mouse cursor 
>becomes constrained to the upper left corner of the guests' screen.
>
>Any ideas about what might be going on?
>Thanks
>

`It's dead, Jim!' ;)

 Yeah, I had reported vmmouse brokenness before (for linux guests):
	http://lists.gnu.org/archive/html/qemu-devel/2009-01/msg01061.html

 I guess its good (in a way) you confirm its broken for windows
guests too...

	Juergen

Re: [Qemu-devel] vmmouse

[Rick Vernam]

On Thursday 29 January 2009 2:58:27 pm you wrote:
> In article <200901282248.26006.rickv@hobi.com> you write:
> >For the past few days (10-15, longer?) both my windows vms seem to have
> > broken mice - the mouse cursor does not move in the guest...
> >Both guests run windows (2k,xp) and both have vmware mouse driver
> > installed (not sure which version, off the top of my head).
> >
> >I've not followed the list too closely during this time period either - in
> >trying to catch up tonight I haven't seen anything on this subject, so I'm
> >posting to see if this is a known issue...
> >
> >I built qemu with DEBUG_VMMOUSE turned on in hw/vmmouse.c.
> >When I run qemu I get a single vmmouse_init right when qemu starts, but
> > then no additional output.
> >
> >Perhaps this is a separate issue, but when I "usb_add tablet" the mouse
> > cursor becomes constrained to the upper left corner of the guests'
> > screen.
> >
> >Any ideas about what might be going on?
> >Thanks
>
> `It's dead, Jim!' ;)
>
>  Yeah, I had reported vmmouse brokenness before (for linux guests):
> 	http://lists.gnu.org/archive/html/qemu-devel/2009-01/msg01061.html
>
>  I guess its good (in a way) you confirm its broken for windows
> guests too...
>
> 	Juergen

It seems the brokenness extends beyond the mouse - the arrow keys on the 
keyboard are not behaving correctly either, although the arrows on the keypad 
are working correctly.
Another example: when qemu grabs the input, in the past I've been able to hit 
ctrl-alt-delete to send those keys to the guest...this no longer works.  I 
have to go to the qemu monitor and do a sendkey ctrl-alt-delete

Re: [Qemu-devel] vmmouse

[Anthony Liguori]

Rick Vernam wrote:
> On Thursday 29 January 2009 2:58:27 pm you wrote:
>   
>> In article <200901282248.26006.rickv@hobi.com> you write:
>>     
>>> For the past few days (10-15, longer?) both my windows vms seem to have
>>> broken mice - the mouse cursor does not move in the guest...
>>> Both guests run windows (2k,xp) and both have vmware mouse driver
>>> installed (not sure which version, off the top of my head).
>>>
>>> I've not followed the list too closely during this time period either - in
>>> trying to catch up tonight I haven't seen anything on this subject, so I'm
>>> posting to see if this is a known issue...
>>>
>>> I built qemu with DEBUG_VMMOUSE turned on in hw/vmmouse.c.
>>> When I run qemu I get a single vmmouse_init right when qemu starts, but
>>> then no additional output.
>>>
>>> Perhaps this is a separate issue, but when I "usb_add tablet" the mouse
>>> cursor becomes constrained to the upper left corner of the guests'
>>> screen.
>>>
>>> Any ideas about what might be going on?
>>> Thanks
>>>       
>> `It's dead, Jim!' ;)
>>
>>  Yeah, I had reported vmmouse brokenness before (for linux guests):
>> 	http://lists.gnu.org/archive/html/qemu-devel/2009-01/msg01061.html
>>
>>  I guess its good (in a way) you confirm its broken for windows
>> guests too...
>>
>> 	Juergen
>>     
>
> It seems the brokenness extends beyond the mouse - the arrow keys on the 
> keyboard are not behaving correctly either, although the arrows on the keypad 
> are working correctly.
>   

Arrow keys are a separate problem.  I wasn't aware that vmmouse was 
having problems.  Does anyone know what changeset broke it?

Regards,

Anthony Liguori

> Another example: when qemu grabs the input, in the past I've been able to hit 
> ctrl-alt-delete to send those keys to the guest...this no longer works.  I 
> have to go to the qemu monitor and do a sendkey ctrl-alt-delete
>
>
>
>   

Re: [Qemu-devel] vmmouse

[François Revol]

> >Perhaps this is a separate issue, but when I "usb_add tablet" the
> mouse cursor
> >becomes constrained to the upper left corner of the guests' screen.

Oddly it's the same symptoms as in Haiku with the wacom emulation...

In the wacom code coordinates are scaled to [0:INT16_MAX] which is
wrong, the PenPartner model it's emulating has max X of like 5040, and
max Y of 3780.

Though the tablet emulation in usb-hid.c does publish an HID report
desc with X and Y max as 0x7fff, so it should work, unless the windows
hid driver is actually ignoring those maximum values and just using
them unscaled, in which case it'd actually stay near 0,0 as it does...

You can try applying my wacom patch:
http://article.gmane.org/gmane.comp.emulators.qemu/36877/match=wacom

and use -usb wacom

and see if it works for you. Not sure if windows has a wacom driver by
default though...

Btw, I still didn't have any comment on that wacom diff...

François.

Re: [Qemu-devel] vmmouse

[andrzej zaborowski]

2009/1/29 Anthony Liguori <anthony@codemonkey.ws>:
> Rick Vernam wrote:
>>
>> On Thursday 29 January 2009 2:58:27 pm you wrote:
>>
>>>
>>> In article <200901282248.26006.rickv@hobi.com> you write:
>>>
>>>>
>>>> For the past few days (10-15, longer?) both my windows vms seem to have
>>>> broken mice - the mouse cursor does not move in the guest...
>>>> Both guests run windows (2k,xp) and both have vmware mouse driver
>>>> installed (not sure which version, off the top of my head).
>>>>
>>>> I've not followed the list too closely during this time period either -
>>>> in
>>>> trying to catch up tonight I haven't seen anything on this subject, so
>>>> I'm
>>>> posting to see if this is a known issue...
>>>>
>>>> I built qemu with DEBUG_VMMOUSE turned on in hw/vmmouse.c.
>>>> When I run qemu I get a single vmmouse_init right when qemu starts, but
>>>> then no additional output.
>>>>
>>>> Perhaps this is a separate issue, but when I "usb_add tablet" the mouse
>>>> cursor becomes constrained to the upper left corner of the guests'
>>>> screen.
>>>>
>>>> Any ideas about what might be going on?
>>>> Thanks
>>>>
>>>
>>> `It's dead, Jim!' ;)
>>>
>>>  Yeah, I had reported vmmouse brokenness before (for linux guests):
>>>        http://lists.gnu.org/archive/html/qemu-devel/2009-01/msg01061.html
>>>
>>>  I guess its good (in a way) you confirm its broken for windows
>>> guests too...
>>>
>>>        Juergen
>>>
>>
>> It seems the brokenness extends beyond the mouse - the arrow keys on the
>> keyboard are not behaving correctly either, although the arrows on the
>> keypad are working correctly.
>>
>
> Arrow keys are a separate problem.  I wasn't aware that vmmouse was having
> problems.  Does anyone know what changeset broke it?

I hit the issue with the generic tablet.  Hopefully fixed it in r6476.

Cheers

Re: [Qemu-devel] vmmouse

[François Revol]

> > Arrow keys are a separate problem.  I wasn't aware that vmmouse was
> > having
> > problems.  Does anyone know what changeset broke it?
>
> I hit the issue with the generic tablet.  Hopefully fixed it in
> r6476.

Eh, nice catch!

Care to look at my wacom diff ?

Hopefully Haiku will soon include the existing vmmouse driver we have
so it won't be needed anymore, but's more correct, and can always help
someone else.

François.

Re: [Qemu-devel] vmmouse

[Rick Vernam]

On Thursday 29 January 2009 5:20:20 pm andrzej zaborowski wrote:
> 2009/1/29 Anthony Liguori <anthony@codemonkey.ws>:
> > Rick Vernam wrote:
> >> On Thursday 29 January 2009 2:58:27 pm you wrote:
> >>> In article <200901282248.26006.rickv@hobi.com> you write:
> >>>> For the past few days (10-15, longer?) both my windows vms seem to
> >>>> have broken mice - the mouse cursor does not move in the guest...
> >>>> Both guests run windows (2k,xp) and both have vmware mouse driver
> >>>> installed (not sure which version, off the top of my head).
> >>>>
> >>>> I've not followed the list too closely during this time period either
> >>>> - in
> >>>> trying to catch up tonight I haven't seen anything on this subject, so
> >>>> I'm
> >>>> posting to see if this is a known issue...
> >>>>
> >>>> I built qemu with DEBUG_VMMOUSE turned on in hw/vmmouse.c.
> >>>> When I run qemu I get a single vmmouse_init right when qemu starts,
> >>>> but then no additional output.
> >>>>
> >>>> Perhaps this is a separate issue, but when I "usb_add tablet" the
> >>>> mouse cursor becomes constrained to the upper left corner of the
> >>>> guests' screen.
> >>>>
> >>>> Any ideas about what might be going on?
> >>>> Thanks
> >>>
> >>> `It's dead, Jim!' ;)
> >>>
> >>>  Yeah, I had reported vmmouse brokenness before (for linux guests):
> >>>       
> >>> http://lists.gnu.org/archive/html/qemu-devel/2009-01/msg01061.html
> >>>
> >>>  I guess its good (in a way) you confirm its broken for windows
> >>> guests too...
> >>>
> >>>        Juergen
> >>
> >> It seems the brokenness extends beyond the mouse - the arrow keys on the
> >> keyboard are not behaving correctly either, although the arrows on the
> >> keypad are working correctly.
> >
> > Arrow keys are a separate problem.  I wasn't aware that vmmouse was
> > having problems.  Does anyone know what changeset broke it?
>
> I hit the issue with the generic tablet.  Hopefully fixed it in r6476.
tablet is better.  thanks.

>
> Cheers

Re: [Qemu-devel] vmmouse

[Rick Vernam]

[-- Attachment #1: Type: text/plain, Size: 2375 bytes --]

On Wednesday 28 January 2009 10:48:25 pm Rick Vernam wrote:
> For the past few days (10-15, longer?) both my windows vms seem to have
> broken mice - the mouse cursor does not move in the guest...
> Both guests run windows (2k,xp) and both have vmware mouse driver installed
> (not sure which version, off the top of my head).
>
> I've not followed the list too closely during this time period either - in
> trying to catch up tonight I haven't seen anything on this subject, so I'm
> posting to see if this is a known issue...
>
> I built qemu with DEBUG_VMMOUSE turned on in hw/vmmouse.c.
> When I run qemu I get a single vmmouse_init right when qemu starts, but
> then no additional output.

(btw, Revision: 6476)

On  Win XP guest, in addition to the broken vmmouse, windows perpetually tells 
me that a new video controller has been detected.  I assume it's talking about 
the vmware-vga device, so I've tried installing vmware's video drivers from 
workstation 5.0.1, 5.5.9 & 6.0.5.  (this also started happening at the same 
time as the vmmouse dying, I just didn't mention it before).


vmware-vga is also broken with a w2k guest (see attached screenshot).  
Although, it has adjusted the sdl window to the correct resolution.

It seems I was mistaken before when I said vmmouse didn't work in w2k guest - 
it does works fine with w2k guest.
I get many debug_vmmouse messages like so:
vmmouse_get_status()
set_data = {0, 0, 28, 5658, 80471854, 818700d8}
vmmouse_mouse_event(0, 0, 0, 0)
get_data = {564d5868, 0, 80420028, 817a5658, 817a05e8, 0}
vmmouse_get_status()
set_data = {4, 0, 80420028, 817a5658, 817a05e8, 0}
get_data = {564d5868, 4, 80420027, 817a5658, 817a05e8, 0}
vmmouse_data(4)
set_data = {0, 0, 0, 0, 817a05e8, 0}
get_data = {564d5868, 0, 28, 5658, 80471854, 818700d8}
vmmouse_get_status()
set_data = {0, 0, 28, 5658, 80471854, 818700d8}
get_data = {564d5868, 0, 28, 5658, 80471854, 818700d8}
vmmouse_get_status()
set_data = {0, 0, 28, 5658, 80471854, 818700d8}
get_data = {564d5868, 0, 28, 5658, 80471854, 818700d8}
vmmouse_get_status()
set_data = {0, 0, 28, 5658, 80471854, 818700d8}
get_data = {564d5868, 0, 28, 5658, 80471854, 818700d8}
vmmouse_get_status()
set_data = {0, 0, 28, 5658, 80471854, 818700d8}

It seems to vmware stuff is affected universally.  I'm going to try to narrow 
down an SVN commit tomorrow, if I can get some time...

Thanks.

[-- Attachment #2: qemu-w2k.png --]
[-- Type: image/png, Size: 21299 bytes --]

Re: [Qemu-devel] vmmouse

[Juergen Lock]

On Fri, Jan 30, 2009 at 12:20:20AM +0100, andrzej zaborowski wrote:
> 2009/1/29 Anthony Liguori <anthony@codemonkey.ws>:
> > Rick Vernam wrote:
> >>
> >> On Thursday 29 January 2009 2:58:27 pm you wrote:
> >>
> >>>
> >>> In article <200901282248.26006.rickv@hobi.com> you write:
> >>>
> >>>>
> >>>> For the past few days (10-15, longer?) both my windows vms seem to have
> >>>> broken mice - the mouse cursor does not move in the guest...
> >>>> Both guests run windows (2k,xp) and both have vmware mouse driver
> >>>> installed (not sure which version, off the top of my head).
> >>>>
> >>>> I've not followed the list too closely during this time period either -
> >>>> in
> >>>> trying to catch up tonight I haven't seen anything on this subject, so
> >>>> I'm
> >>>> posting to see if this is a known issue...
> >>>>
> >>>> I built qemu with DEBUG_VMMOUSE turned on in hw/vmmouse.c.
> >>>> When I run qemu I get a single vmmouse_init right when qemu starts, but
> >>>> then no additional output.
> >>>>
> >>>> Perhaps this is a separate issue, but when I "usb_add tablet" the mouse
> >>>> cursor becomes constrained to the upper left corner of the guests'
> >>>> screen.
> >>>>
> >>>> Any ideas about what might be going on?
> >>>> Thanks
> >>>>
> >>>
> >>> `It's dead, Jim!' ;)
> >>>
> >>>  Yeah, I had reported vmmouse brokenness before (for linux guests):
> >>>        http://lists.gnu.org/archive/html/qemu-devel/2009-01/msg01061.html
> >>>
> >>>  I guess its good (in a way) you confirm its broken for windows
> >>> guests too...
> >>>
> >>>        Juergen
> >>>
> >>
> >> It seems the brokenness extends beyond the mouse - the arrow keys on the
> >> keyboard are not behaving correctly either, although the arrows on the
> >> keypad are working correctly.
> >>
> >
> > Arrow keys are a separate problem.  I wasn't aware that vmmouse was having
> > problems.  Does anyone know what changeset broke it?
> 
Hum.  Should I Cc you (Anthony) next time I report a regression?

> I hit the issue with the generic tablet.  Hopefully fixed it in r6476.

 Yup, vmmouse is back to working order now; -vga vmware is still
broken tho.

 Oh and I just tried -usbdevice tablet with my usual sidux iso guest,
and unlike with vmmouse, the guest cursor wasnt warped to the right
position when the host cursor entered the qemu window...  does that
work for anywone else?

 Cheers,
	Juergen