The problem is that RedWindow::show calls the XLib MoveWindow function
on the window after it has been mapped, moving it to the location in
_show_pos. This is seen by the window manager as the application saying
I know exactly where I want my window to be placed, don't do placing for
me. Which causes the client window to always be shown at pos 0x0, even
though that may not be the best location.
What this patch does is:
1) It makes RedWindow::show not call MoveWindow when a window is
first created normally and then shown
2) It makes RedWindow::show still call MoveWindow when:
-when the window has been shown before, and was hidden for some
reason (controller interface), and is now being re-shown
so that it ends up being re-shown at its old position
-when the window is a fullscreen window (screen.cpp always
calls move on the window before showing it to set its position)
-when the user switch from windowed mode -> fullscreen ->
windowed mode again, to make sure that the windowed mode window
is shown in the same position as before switching to fullscreen
mode
drawable_count was becoming negative. It tracks the number of
items in the worker->current_list ring. It was decremented correctly,
but incremented only in several cases. The cases it wasn't incremented
where:
red_current_add_equal found an equivalent drawable
by moving the increment to where the item is added to current_list, in
__current_add_drawable, the accounting remains correct.
This has no affect other then correct accounting, as drawable_count isn't
used for anything.
While we are at it: There is no reason for chardev
support to stay in the experimental area, so move it out.
qemu should not need the "spice-experimental.h" file.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
(cherry picked from commit 7bbc2ba090)
Implement server-side support for switch-host client migration. Client
side support is present already in the tree.
Setting the migration information is done using the existing
spice_server_migrate_info() function. A new
spice_server_migrate_switch() function has been added which triggers
sending out the switch-host message.
Seamless migration functions are left there for now.
spice_server_migrate_start() has been chamnged to just fail
unconditionally though as seamless migration is broken anyway.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
(cherry picked from commit 4b1ea4e102)
We should pass up errors instead of aborting. Do that at least
for bind() failures which actually happen in real live due to the
tcp port being busy.
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
(cherry picked from commit aafd8eea74)
If connect_unsecure failed due to protocol version mismatch, don't try to
connect_secure with the same version, but retry (connect_secure or
connect_unsecure) with older version. catch (...) is handled by caller
at RedChannel::run().
This solves the following bug: when "new" Spice client (protocol version 2)
with given secure_port connects to "old" server which is not using the same
secure_port (or not using a secure_port at all), the client exits immediately.
In this scenario, the client first tries to use Spice protocol version 2 to
connect the unsecure port, and altough this fails due to version mismatch, it
tries to connect to the secure port with the same protocol version 2, which is
a wrong behavior, fails due to socket error 10061 (WSAECONNREFUSED -
Connection refused) and handled mistakenly by immediate exit, instead of
retrying with protocol version 1.
We erronously ignored data from guest on the serial channel if no client is
connected. This leads to an assert when the guest writes a second time, since
there is still data unconsumed by us (the host).
Fix by reading data anyway, and discarding it after parsing (and reading) whole
messages from the guest.
Net affect is that any messages the agent sends while no client is connected
get discarded, but only full messages are discarded.
This fixes an abort if booting a winxp guest with vdagent without a connected
client.
updates taken from spice vga mode updates, i.e. non cacheable, glz compressed
(depends on whatever settings you apply to the server) opaque draw operations.
+ completed the SpiceCoreInterface implementation (timers)
v1->v2:
removed test_util.c (Hans)
replaced mallocz with calloc (Hans)
Currently when going / starting fullscreen if the guest resolution for one of
the monitors is higher then that monitor on the client can handle, we show a
white screen. Leaving the user stuck (unless they know the fullscreen key
switch combi) with a white screen when starting the client fullscreen from
the XPI.
This patch changes the client to fall back to windowed mode in this case
instead.
When XIM + ibus is in use XIM creates an invisible window for its own
purposes, we sometimes get a _GTK_LOAD_ICONTHEMES ClientMessage event on
this window. Since this window was not explicitly created by spicec, it
does not have a Window Context (with the event handling function for the
window in question) set. This would cause spicec to throw an unhandled
exception and exit.
This patch replaces the exception throwing with silently ignoring
ClientMessage events on Windows without a Context and logging a warning
for other event types.
Currently modifier keys (ctrl, alt) can get stuck when using the x11 client.
To reproduce under gnome:
-focus the client window without causing it to grab the keyborad (click on
the title bar not the window)
-press crlt + alt + right arrow to switch virtual desktop
-press crlt + alt + left arrow to switch back
-notice ctrl + alt are stuck pressed
What is happening here is:
-We get a focus out event, caused by the hotkey combi key grab, focus event
notify mode == NotifyGrab, and release all keys -> good
-We get another focus out event, as we really loose the focus.
notify mode == NotifyWhileGrabbed, which we ignore as we already lost
focus before
-We get a focus in event, as the focus is returning to us, but we don't
really have the focus yet, as the hotkey combi key grab is still active
(ie ctrl + alt are still pressed).
We now sync the vm's modifier key state with the current X-server state,
telling the vm ctrl + alt are pressed. Note we do this by directly reading
the X-server keyboard status, we are not getting any key press events from the
X-server -> bad
-We get another focus in event, as we really get the focus back,
notify mode == NotifyUngrab. We ignore this one as already have gained the
focus before. If we were to sync the vm modifier state here, all would be
well we would no longer see the modifier keys pressed, or if we would we
would get a release event when they get released (testing has shown both).
The solution here is to ignore the first focus in event, and do the modifier
sync on the second focus in event, or more in general to ignore focus events
where notify mode == NotifyWhileGrabbed.
Currently when compiled with the gui enabled if you specify a host to connect
to on the cmdline the gui flashes by (show_gui gets called, then the connect
handler calls hide_gui as soon as the connection is made).
This patch removes this ugly flashing by of the gui.
Currently we are calling show_info_layer from hide_gui in application.cpp, but
there are 2 cases where this does not happen:
1) When compiled without gui support hide_gui is a complete nop, so we never
show the info layer when compiled without gui support
2) When run with --controller we never show the gui, and hide_gui
checks if there is a gui to hide as the first thing and if not returns
resulting in show_info_layer not being called, and thus the info layer
not showing when launched from the xpi
This patch fixes both by adding a call to show_info_layer from
on_visibility_start note that on_visibility_start also calls hide_gui,
so in some cases show_info_layer may be called twice, this is not a
problem as show_info_layer is protected against this.
