Hi
On Wed, Mar 9, 2022 at 12:05 PM Akihiko Odaki <akihiko.odaki@gmail.com
<mailto:akihiko.odaki@gmail.com>> wrote:
On 2022/03/09 17:02, Marc-André Lureau wrote:
> Hi
>
> On Tue, Mar 8, 2022 at 6:43 PM Akihiko Odaki
<akihiko.odaki@gmail.com <mailto:akihiko.odaki@gmail.com>
> <mailto:akihiko.odaki@gmail.com
<mailto:akihiko.odaki@gmail.com>>> wrote:
>
> On 2022/03/08 23:26, Marc-André Lureau wrote:
> > Hi
> >
> > On Mon, Mar 7, 2022 at 4:49 PM Akihiko Odaki
> <akihiko.odaki@gmail.com <mailto:akihiko.odaki@gmail.com>
<mailto:akihiko.odaki@gmail.com <mailto:akihiko.odaki@gmail.com>>>
wrote:
> >>
> >
> > (taking notes, mostly for myself)
> >
> >>> Could you provide a failing test case or a more concrete
> suggestion on
> >>> what to do instead? I am all ears :)
> >>>
> >>> thanks
> >>>
> >>
> >> Let me describe a more concrete case. Think that egl-headless
> and vnc
> >> are enabled. The guest devices are serial virtio-gpu-gl.
vnc selects
> >> serial at first.
> >
> > (I am not sure how you would select serial first, but let's
> assume you did)
> >
> >>
> >> vnc uses register_displaychangelistener and
console_select to switch
> >> consoles.
> >>
> >> displaychangelistener_display_console does the actual
switching, and
> >> calls dpy_gfx_switch and dpy_gfx_update if
con->scanout.kind is
> >> SCANOUT_SURFACE. It calls dpy_gl_scanout_texture or
> >> dpy_gl_scanout_dmabuf if con->scanout.kind is
SCANOUT_TEXTURE or
> >> SCANOUT_DMABUF. It works for a OpenGL display, but it
doesn't vnc in
> >> combination with egl-headless.
> >
> > (hmm, what doesn't work? With this patch, the
DisplaySurface is
> always
> > switched, no matter what con->scanout.kind is)
> >
> >>
> >> virtio-gpu-gl starts scanning out texture. It would
happen in the
> >> following sequence:
> >> 1. virtio-gpu-gl calls qemu_console_resize. con->scanout.kind
> will be
> >> SCANOUT_SURFACE.
> >> 2. qemu_console_resize calls dpy_gfx_switch, delivering
> DisplaySurface
> >> to egl-headless. con->scanout.kind will be SCANOUT_TEXTURE.
> >
> > (qemu_console_resize calls dpy_gfx_replace_surface.
con.scanout.kind
> > is still SCANOUT_SURFACE)
> >
> > (It calls displaychangelistener_gfx_switch() which will call
> > egl_gfx_switch() and update the current surface)
> >
> >> 3. virtio-gpu-gl calls dpy_gl_scanout_texture.
egl-headless starts
> >> rendering the texture to the DisplaySurface.
> >
> > (now con.scanout.kind becomes SCANOUT_TEXTURE)
> >
> >>
> >> In the end, the DisplaySurface will have the image
rendered, and
> >> con->scanout.kind will be SCANOUT_TEXTURE.
> >
> > (so far it works as expected, right?)
> >
> >
> >
> >>
> >> Now vnc switches to virtio-gpu-gl.
> >>
> >> 4. vnc calls console_select
> >> 5. displaychangelistener_display_console finds
con->scanout.kind is
> >> SCANOUT_TEXTURE so it tries to scanout texture, but vnc
is not
> an OpenGL
> >> display. It essentially becomes no-op and the display
would be
> blank.
> >>
> >> This patch will change it to call dpy_gfx_switch but not
to call
> >> dpy_gfx_update.
> >
> > Alright, I think I see what you mean. egl-headless is
associated with
> > a specific con, and thus is not involved during vnc console
> switching.
> >
> > However, dpy_gfx_switch on vnc is not a no-op. It updates the
> surface,
> > resize the client and mark the area dirty. Because the
con.surface is
> > kept updated by egl-headless, the client gets the updated
content.
> > Iow, it still works.
> >
> > If we always called dpy_gfx_update() during
> > displaychangelistener_display_console(), it would mean for the
> > listener to display the surface content, even when the scanout
> kind is
> > set for a texture. Or we need to change the behaviour of
> > dpy_gfx_update() to depend on the current scanout kind.
>
> vnc has to display the surface content so dpy_gfx_update
should be
> called for vnc, but not for OpenGL displays.
>
>
> VNC already marks the whole surface as dirty during
vnc_dpy_switch().
> This is like calling dpy_gfx_update().
Then why does it call dpy_gfx_update for SCANOUT_SURFACE first
place? It
shouldn't call the function if it is not required to display the
surface
content.
This is a pre-existing discrepancy. vnc handles switch with an implicit
update. Imho, we need to improve the code so that updates/display become
explicit.
But this is extra, not necessary to fix the regressions you pointed out.
I'd rather not mix concerns, and proceed step by step.
--
Marc-André Lureau