Fix for the continuing issues with pointer lock operations in HiDPI Chrome.
While the recent fixes to the pointer lock code have helped, we still have problems when we move the cursor near
the bounds of the content window. The pointer lock code in RWHVA detects that and forces a move to the center of
the window via a SetCursorPos operation. The expectation here is that we would see a mouse move message with the
location specified by SetCursorPos which would ignore.
With fractional scale factors like 1.25/1.5, etc it turns out that the conversions of the mouse messages from pixel
to DPI and vice versa causes us to end up with messages which are off by 1 or 2 px which causes us to miss the move
to center event in the locked mode. The renderer sees the message and bounces the cursor back to the center leading
to erratic behavior.
Proposed fix is to add an allowance of 2 for the mouse event x and y coordinates if we are waiting for the move to
center event.
BUG=411634
TEST=As described in the bug. Please try a clockwise and anticlockwise mouse rotation with device scales set to 1.25/
1.5, etc.
Review URL: https://codereview.chromium.org/1005213003
Cr-Commit-Position: refs/heads/master@{#321450}
1 file changed
