Bug 1300250

Can you check if this is a regression?

(In reply to Fabiano Fidêncio from comment #2)
> Can you check if this is a regression?

Hi Fidencio,

Because I also can't see mouse of vnc+qxl guest in virt-manager, the bug maybe not caused by virt-viewer, I try to downgrade version of qxl to
xorg-x11-drv-qxl-0.1.1-16.el6.x86_64, the problem still exists, so this bug maybe a new one, thanks

VNC pkg:gtk-vnc-0.3.10-3.el6.x86_64

(In reply to Fabiano Fidêncio from comment #2)
> Can you check if this is a regression?

Hi Fidencio,

Because I also can't see mouse of vnc+qxl guest in virt-manager, the bug maybe not caused by virt-viewer, I try to downgrade version of qxl to
xorg-x11-drv-qxl-0.1.1-16.el6.x86_64, the problem still exists, so this bug maybe a new one, thanks

VNC pkg:gtk-vnc-0.3.10-3.el6.x86_64

(In reply to mxie from comment #4)
> (In reply to Fabiano Fidêncio from comment #2)
> > Can you check if this is a regression?
> 
> Hi Fidencio,
> 
> Because I also can't see mouse of vnc+qxl guest in virt-manager, the bug
> maybe not caused by virt-viewer, I try to downgrade version of qxl to
> xorg-x11-drv-qxl-0.1.1-16.el6.x86_64, the problem still exists, so this bug
> maybe a new one, thanks
> 
> VNC pkg:gtk-vnc-0.3.10-3.el6.x86_64

I'm moving the bug to gtk-vnc for now.

Just to be sure ...

You have this problem when connecting to a RHEL-6 guest, right? If yes, is this RHEL-6 guest a 6.8 or a released 6.7? What's the version of the xorg-x11-drv-qxl on the GUEST? When you downgraded the xorg-x11-drv-qxl, you did in the GUEST, right?

Does this problem occur when connecting to another guests? Like, Fedora or RHEL-7?

Just for the record ...

I am not able to reproduce the problem at all on RHEL-6.6 guest (it means xorg-x11-drv-qxl-0.1.1-15). I've upgrade to -16 and -17 and still cannot reproduce the problem.

I have another guest here, a fully updated RHEL-6.8 (it means xorg-x11-drv-qxl-0.1.1-18) and I can reproduce the issue on this guest.

Hi Fidencio,
 
> You have this problem when connecting to a RHEL-6 guest, right? If yes, is
> this RHEL-6 guest a 6.8 or a released 6.7? What's the version of the
> xorg-x11-drv-qxl on the GUEST? 

The bug can be reproduced on rhel6.7 released guest(also can be reproduced after updating kernel to 2.6.32-615.el6.x86_64  )
qxl version in guest :xorg-x11-drv-qxl-0.1.1-17.el6.x86_64



>When you downgraded the xorg-x11-drv-qxl, you
> did in the GUEST, right?

I downgraded xorg-x11-drv-qxl in host, I'm sorry that I didn't take the focus on guest's qxl package at that time

> Does this problem occur when connecting to another guests? Like, Fedora or
> RHEL-7?

The bug can be reproduced on rhel7.2 guest. its qxl package version is xorg-x11-drv-qxl-0.1.1-18.el7.x86_64


> I am not able to reproduce the problem at all on RHEL-6.6 guest (it means 
> xorg-x11-drv-qxl-0.1.1-15). I've upgrade to -16 and -17 and still cannot > reproduce the problem.

It is very strange that I still can reproduce the problem at RHEL-6.6 guest with xorg-x11-drv-qxl-0.1.1-15

Okay. Using a Fedora23 as host I do have the cursor problem using a updated RHEL-6.8 guest.

From  my investigations so far the mouse shows up when booting up the machine and then disappear in case of reconnection (or connection from another client).
The cursor comes back though when a cursor redraw is forced (or when opening an app, or when switching to the console and coming back to the graphical mode),

Assigning the bug to xorg-x11-drv-qxl for further investigation.

(In reply to mxie from comment #8)
> Hi Fidencio,
>  
> > You have this problem when connecting to a RHEL-6 guest, right? If yes, is
> > this RHEL-6 guest a 6.8 or a released 6.7? What's the version of the
> > xorg-x11-drv-qxl on the GUEST? 
> 
> The bug can be reproduced on rhel6.7 released guest(also can be reproduced
> after updating kernel to 2.6.32-615.el6.x86_64  )
> qxl version in guest :xorg-x11-drv-qxl-0.1.1-17.el6.x86_64
> 
> 
> 
> >When you downgraded the xorg-x11-drv-qxl, you
> > did in the GUEST, right?
> 
> I downgraded xorg-x11-drv-qxl in host, I'm sorry that I didn't take the
> focus on guest's qxl package at that time
> 
> > Does this problem occur when connecting to another guests? Like, Fedora or
> > RHEL-7?
> 
> The bug can be reproduced on rhel7.2 guest. its qxl package version is
> xorg-x11-drv-qxl-0.1.1-18.el7.x86_64
> 
> 
> > I am not able to reproduce the problem at all on RHEL-6.6 guest (it means 
> > xorg-x11-drv-qxl-0.1.1-15). I've upgrade to -16 and -17 and still cannot > reproduce the problem.
> 
> It is very strange that I still can reproduce the problem at RHEL-6.6 guest
> with xorg-x11-drv-qxl-0.1.1-15

Not sure we should support vnc + qxl. It has never worked before (see BZ 707055)!

Hi，David
As rjones said, qxl + vnc is supported.please see  
https://bugzilla.redhat.com/show_bug.cgi?id=1312254#c5

(In reply to mxie from comment #11)
> Hi，David
> As rjones said, qxl + vnc is supported.please see  
> https://bugzilla.redhat.com/show_bug.cgi?id=1312254#c5

Not sure it is/was supported, as the mouse problem was reported a while ago (https://bugzilla.redhat.com/show_bug.cgi?id=707055), and was not fixed.

Regardless, spice team will investigate, but it is too late for 6.8

*** Bug 707055 has been marked as a duplicate of this bug. ***

The problem was that mouse cursor was sent only on change but not when a new
client is connecting. This small patch fix the issue (the if is to avoid core
if qxl did not set the mouse like during boot):


diff -ru qemu-2.4.1.orig/ui/vnc.c qemu-2.4.1.my/ui/vnc.c
--- qemu-2.4.1.orig/ui/vnc.c	2016-03-02 13:41:50.066454216 +0000
+++ qemu-2.4.1.my/ui/vnc.c	2016-03-02 14:16:13.182081871 +0000
@@ -2124,6 +2124,8 @@
             break;
         case VNC_ENCODING_RICH_CURSOR:
             vs->features |= VNC_FEATURE_RICH_CURSOR_MASK;
+            if (vs->vd->cursor)
+                vnc_cursor_define(vs);
             break;
         case VNC_ENCODING_EXT_KEY_EVENT:
             send_ext_key_event_ack(vs);

Patch posted upstream at http://lists.nongnu.org/archive/html/qemu-devel/2016-03/msg00507.html

Reassignment as fix is for Qemu component.

This bug was accidentally moved from POST to MODIFIED via an error in automation, please see mmccune with any questions

(In reply to Frediano Ziglio from comment #15)
> Patch posted upstream at
> http://lists.nongnu.org/archive/html/qemu-devel/2016-03/msg00507.html

commit 91ec41dc3f9cf28f93af0630b20ba8e7dcefa16b

(In reply to Gerd Hoffmann from comment #19)
> (In reply to Frediano Ziglio from comment #15)
> > Patch posted upstream at
> > http://lists.nongnu.org/archive/html/qemu-devel/2016-03/msg00507.html
> 
> commit 91ec41dc3f9cf28f93af0630b20ba8e7dcefa16b

Ahem, well, turned out to not be that simple.  It's not that VNC_ENCODING_RICH_CURSOR support needs a bugfix in RHEL-6.  RHEL-6 has no rich cursor support in the first place (which is required to show the qxl cursor).

Given the lifecycle RHEL-6 is in and the fact that only the qxl/spice combination is supported I think this qualifies as WONTFIX.  Ademar?