On startup, scale_x and scale_y were set to 1 that didn't reflect the
real situation of the scan-out, resulting in incorrect cursor
coordinates to be sent when moving the mouse pointer. Simply updating
the scales before rendering the image fixes this issue.
Cc: hikalium <hikalium@hikalium.com>
Cc: Alexander Orzechowski <orzechowski.alexander@gmail.com>
Cc: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Weifeng Liu <weifeng.liu.z@gmail.com>
---
ui/gtk-gl-area.c | 6 ++++++
1 file changed, 6 insertions(+)
diff --git a/ui/gtk-gl-area.c b/ui/gtk-gl-area.c
index 2c9a0db425..01235f876a 100644
--- a/ui/gtk-gl-area.c
+++ b/ui/gtk-gl-area.c
@@ -42,6 +42,7 @@ void gd_gl_area_draw(VirtualConsole *vc)
#ifdef CONFIG_GBM
QemuDmaBuf *dmabuf = vc->gfx.guest_fb.dmabuf;
#endif
+ int fbw, fbh;
int ww, wh, ws, y1, y2;
if (!vc->gfx.gls) {
@@ -53,6 +54,11 @@ void gd_gl_area_draw(VirtualConsole *vc)
ww = gtk_widget_get_allocated_width(vc->gfx.drawing_area) * ws;
wh = gtk_widget_get_allocated_height(vc->gfx.drawing_area) * ws;
+ fbw = surface_width(vc->gfx.ds);
+ fbh = surface_height(vc->gfx.ds);
+ vc->gfx.scale_x = (double)ww / fbw / ws;
+ vc->gfx.scale_y = (double)wh / fbh / ws;
+
if (vc->gfx.scanout_mode) {
if (!vc->gfx.guest_fb.framebuffer) {
return;
--
2.49.0Hi
On Sat, Apr 19, 2025 at 1:51 PM Weifeng Liu <weifeng.liu.z@gmail.com> wrote:
>
> On startup, scale_x and scale_y were set to 1 that didn't reflect the
> real situation of the scan-out, resulting in incorrect cursor
> coordinates to be sent when moving the mouse pointer. Simply updating
> the scales before rendering the image fixes this issue.
>
> Cc: hikalium <hikalium@hikalium.com>
> Cc: Alexander Orzechowski <orzechowski.alexander@gmail.com>
> Cc: Gerd Hoffmann <kraxel@redhat.com>
> Signed-off-by: Weifeng Liu <weifeng.liu.z@gmail.com>
Tested-by: Marc-André Lureau <marcandre.lureau@redhat.com>
(Hopefully someone has enough motivation to actually understand this
change better - otherwise I'll simply queue it?)
> ---
> ui/gtk-gl-area.c | 6 ++++++
> 1 file changed, 6 insertions(+)
>
> diff --git a/ui/gtk-gl-area.c b/ui/gtk-gl-area.c
> index 2c9a0db425..01235f876a 100644
> --- a/ui/gtk-gl-area.c
> +++ b/ui/gtk-gl-area.c
> @@ -42,6 +42,7 @@ void gd_gl_area_draw(VirtualConsole *vc)
> #ifdef CONFIG_GBM
> QemuDmaBuf *dmabuf = vc->gfx.guest_fb.dmabuf;
> #endif
> + int fbw, fbh;
> int ww, wh, ws, y1, y2;
>
> if (!vc->gfx.gls) {
> @@ -53,6 +54,11 @@ void gd_gl_area_draw(VirtualConsole *vc)
> ww = gtk_widget_get_allocated_width(vc->gfx.drawing_area) * ws;
> wh = gtk_widget_get_allocated_height(vc->gfx.drawing_area) * ws;
>
> + fbw = surface_width(vc->gfx.ds);
> + fbh = surface_height(vc->gfx.ds);
> + vc->gfx.scale_x = (double)ww / fbw / ws;
> + vc->gfx.scale_y = (double)wh / fbh / ws;
> +
> if (vc->gfx.scanout_mode) {
> if (!vc->gfx.guest_fb.framebuffer) {
> return;
> --
> 2.49.0
>
>
--
Marc-André Lureau
Sorry, just realized that I forgot to cc Marc-André :D.
Here is a description of the issue this patch is trying to solve:
--------------------------
How to reproduce the issue
--------------------------
Launch the guest with gtk backed virtual display using command like
IMAGE_FOLDER=$HOME/workspace/images/fedora
DISK_PATH=$IMAGE_FOLDER/rootfs.qcow2
UEFI_CODE_PATH=$IMAGE_FOLDER/OVMF_CODE.fd
UEFI_DATA_PATH=$IMAGE_FOLDER/OVMF_VARS.fd
ISO_PATH=$IMAGE_FOLDER/Fedora-Workstation-Live-42-1.1.x86_64.iso
MEM=4G
$QEMU \
-enable-kvm \
-M q35 \
-object memory-backend-memfd,id=mem1,size=$MEM \
-machine memory-backend=mem1 \
-smp 4 \
-cpu host \
-boot order=c,menu=on,splash-time=3 \
-cdrom $ISO_PATH \
-drive if=pflash,format=raw,readonly=on,file=$UEFI_CODE_PATH \
-drive if=pflash,format=raw,file=$UEFI_DATA_PATH \
-hda $DISK_PATH \
-display gtk,gl=on,show-cursor=on \
-device virtio-vga-gl,venus=false,blob=true,hostmem=$MEM \
-usb -device usb-tablet
When guest GUI is ready, click the guest display and we are likely to
find that the cursor coordinates are incorrectly reported to the guest
with an offset.
------------------
Cause of the issue
------------------
There is a mapping from the gtk glarea widget coordinates to the guest
display coordinates, and the scale factor plays an essential role in
the calculation. However, the scale factor defaults to 1 and is not
updated to reflect the fact on startup, rendering incorrect cursor
coordinates.
Best regards,
Weifeng
On Sat, 2025-04-19 at 17:48 +0800, Weifeng Liu wrote:
> On startup, scale_x and scale_y were set to 1 that didn't reflect the
> real situation of the scan-out, resulting in incorrect cursor
> coordinates to be sent when moving the mouse pointer. Simply updating
> the scales before rendering the image fixes this issue.
>
> Cc: hikalium <hikalium@hikalium.com>
> Cc: Alexander Orzechowski <orzechowski.alexander@gmail.com>
> Cc: Gerd Hoffmann <kraxel@redhat.com>
> Signed-off-by: Weifeng Liu <weifeng.liu.z@gmail.com>
> ---
> ui/gtk-gl-area.c | 6 ++++++
> 1 file changed, 6 insertions(+)
>
> diff --git a/ui/gtk-gl-area.c b/ui/gtk-gl-area.c
> index 2c9a0db425..01235f876a 100644
> --- a/ui/gtk-gl-area.c
> +++ b/ui/gtk-gl-area.c
> @@ -42,6 +42,7 @@ void gd_gl_area_draw(VirtualConsole *vc)
> #ifdef CONFIG_GBM
> QemuDmaBuf *dmabuf = vc->gfx.guest_fb.dmabuf;
> #endif
> + int fbw, fbh;
> int ww, wh, ws, y1, y2;
>
> if (!vc->gfx.gls) {
> @@ -53,6 +54,11 @@ void gd_gl_area_draw(VirtualConsole *vc)
> ww = gtk_widget_get_allocated_width(vc->gfx.drawing_area) * ws;
> wh = gtk_widget_get_allocated_height(vc->gfx.drawing_area) * ws;
>
> + fbw = surface_width(vc->gfx.ds);
> + fbh = surface_height(vc->gfx.ds);
> + vc->gfx.scale_x = (double)ww / fbw / ws;
> + vc->gfx.scale_y = (double)wh / fbh / ws;
> +
> if (vc->gfx.scanout_mode) {
> if (!vc->gfx.guest_fb.framebuffer) {
> return;
© 2016 - 2025 Red Hat, Inc.