1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
|
graphic: 0.978
other: 0.952
device: 0.946
socket: 0.945
assembly: 0.940
semantic: 0.937
instruction: 0.929
boot: 0.927
vnc: 0.917
network: 0.905
mistranslation: 0.894
KVM: 0.887
Unable to release mouse in SDL2 mode
Starting with commit 8f4ea9cd0b770dbe496d9d24f0ef8813fdbfe0d0 "sdl: prefer sdl2 over sdl1", I can no longer release mouse pointer grab unless I use --with-sdlabi=1.2 configure option.
This easily reproduces in e.g. guest Kubuntu, when I let it start Xorg and then click into the QEMU window. After this the mouse is trapped and no matter how I combine Ctrl+Alt and motion of the cursor, the pointer never goes out from the window. When at the border, QEMU window switches from "Press Ctrl+Alt to exit grab" to "QEMU", i.e. it thinks that it has released the grab. But it hasn't really, so I have to go to VT1 and do "pkill qemu" from there to get my pointer back.
Which version of SDL2 are you using? Could you please try to reproduce it with the latest version to see whether the problem has been fixed there?
I'm using SDL2-2.0.5, which I believe is the latest already.
I am seeing this on my system as well - the exact same symptoms. Has anyone investigated this problem?
I'm also seeing this problem with -vga vmware in case that matters.
Additionally 2ec78706d188df7d3dab43d07b19b05ef7800a44 also broke keyboard with SDL1 so that e.g. in a Windows password prompt backspace and enter just inserts chars instead of doing actions so now the workaround to avoid this bug (--with-sdlabi=1.2) is also unusable.
Can you file a separate bug for the SDL1 backspace problem - it was not intended to cause problems like that.
On Thu, 1 Feb 2018, Daniel Berrange wrote:
> Can you file a separate bug for the SDL1 backspace problem - it was not
> intended to cause problems like that.
Unless you want it tracked in a bug I'm happy with a fix submitted to the
list eventually without being separately tracked on launchpad. I would've
reported this on the mailing list normally but since this is also relevant
here due to SDL2 also not uasable beacause of this bug I've commented
here so people affected by this know about both problems.
workaround: don't click into the guest window before the tablet driver loads.
When qemu mouse mode changes from relative to absolute
we must turn off sdl relative mouse mode too.
Fixes: https://bugs.launchpad.net/qemu/+bug/1703795
Signed-off-by: Gerd Hoffmann <email address hidden>
---
ui/sdl2.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/ui/sdl2.c b/ui/sdl2.c
index 812c315891..858e04d7c0 100644
--- a/ui/sdl2.c
+++ b/ui/sdl2.c
@@ -249,6 +249,7 @@ static void sdl_mouse_mode_change(Notifier *notify, void *data)
if (qemu_input_is_absolute()) {
if (!absolute_enabled) {
absolute_enabled = 1;
+ SDL_SetRelativeMouseMode(SDL_FALSE);
absolute_mouse_grab(&sdl2_console[0]);
}
} else if (absolute_enabled) {
--
2.9.3
I can confirm that the patch from comment #9 appears to fix the original problem.
The patch works for me too. Thanks.
Fix has been included here:
https://git.qemu.org/?p=qemu.git;a=commitdiff;h=8dfa3061ce56d871dc9df
|