[Qemu-devel] [PATCH 0/2] sdl2: Fix relative mouse mode

[Qemu-devel] [PATCH 0/2] sdl2: Fix relative mouse mode

[Cole Robinson]

sdl.c received a fix for relative mouse mode, but it wasn't copied to
sdl2.c.

Patch #2 uses new SDL2 API to make relative mode behave similar to sdl.c,
but I'm not fully confident that it's safe. See patch comments.

Also, relative mode is a mess in gtk as well. I started fixing these only
to find that patches were already posted:

https://lists.gnu.org/archive/html/qemu-devel/2014-02/msg02278.html

Cole Robinson (2):
  input: sdl2: Fix guest_cursor logic
  input: sdl2: Fix relative mode to match SDL1 behavior

 ui/sdl2.c | 21 +++++++++++++--------
 1 file changed, 13 insertions(+), 8 deletions(-)

-- 
1.9.0

[Qemu-devel] [PATCH 1/2] input: sdl2: Fix guest_cursor logic

[Cole Robinson]

Unbreaks relative mouse mode with sdl2, just like was done with sdl.c
in c3aa84b6.

Signed-off-by: Cole Robinson <crobinso@redhat.com>
---
 ui/sdl2.c | 18 +++++++++++-------
 1 file changed, 11 insertions(+), 7 deletions(-)

diff --git a/ui/sdl2.c b/ui/sdl2.c
index f1532e9..e4cb9fb 100644
--- a/ui/sdl2.c
+++ b/ui/sdl2.c
@@ -403,13 +403,17 @@ static void sdl_send_mouse_event(struct sdl2_state *scon, int dx, int dy,
         }
         qemu_input_queue_abs(scon->dcl.con, INPUT_AXIS_X, off_x + x, max_w);
         qemu_input_queue_abs(scon->dcl.con, INPUT_AXIS_Y, off_y + y, max_h);
-    } else if (guest_cursor) {
-        x -= guest_x;
-        y -= guest_y;
-        guest_x += x;
-        guest_y += y;
-        qemu_input_queue_rel(scon->dcl.con, INPUT_AXIS_X, x);
-        qemu_input_queue_rel(scon->dcl.con, INPUT_AXIS_Y, y);
+    } else {
+        if (guest_cursor) {
+            x -= guest_x;
+            y -= guest_y;
+            guest_x += x;
+            guest_y += y;
+            dx = x;
+            dy = y;
+        }
+        qemu_input_queue_rel(scon->dcl.con, INPUT_AXIS_X, dx);
+        qemu_input_queue_rel(scon->dcl.con, INPUT_AXIS_Y, dy);
     }
     qemu_input_event_sync();
 }
-- 
1.9.0

[Qemu-devel] [RFC PATCH 2/2] input: sdl2: Fix relative mode to match SDL1 behavior

[Cole Robinson]

Right now relative mode accelerates too fast, and has the 'invisible wall'
problem. SDL2 added an explicit API to handle this use case, so let's use
it.

Signed-off-by: Cole Robinson <crobinso@redhat.com>
---
While this fixes things for me, I'm not positive it doesn't have other
side effects. I didn't test transition from relative mode to absolute
mode and back, which might hit issues.

 ui/sdl2.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/ui/sdl2.c b/ui/sdl2.c
index e4cb9fb..7506e2e 100644
--- a/ui/sdl2.c
+++ b/ui/sdl2.c
@@ -278,7 +278,7 @@ static void sdl_hide_cursor(void)
         SDL_ShowCursor(1);
         SDL_SetCursor(sdl_cursor_hidden);
     } else {
-        SDL_ShowCursor(0);
+        SDL_SetRelativeMouseMode(SDL_TRUE);
     }
 }
 
@@ -289,6 +289,7 @@ static void sdl_show_cursor(void)
     }
 
     if (!qemu_input_is_absolute()) {
+        SDL_SetRelativeMouseMode(SDL_FALSE);
         SDL_ShowCursor(1);
         if (guest_cursor &&
             (gui_grab || qemu_input_is_absolute() || absolute_enabled)) {
-- 
1.9.0