[PATCH] xf86/cursor: Check hw cursor after xf86SeCursor() call, too

Takashi Iwai tiwai at suse.de
Thu Jul 17 12:04:19 PDT 2014

At Thu, 17 Jul 2014 11:09:19 -0700,
Keith Packard wrote:
> Takashi Iwai <tiwai at suse.de> writes:
> > When a video driver changed the availability of hw cursor inside the
> > xf86SetCursor() call, f86CursorSetCursor() still assumes the hw
> > cursor and clears SWCursor flag of the screen unconditionally.  This
> > results in the inconsistent state, typically the invisible cursor
> > pointer.  For example, the problem is seen with modesetting driver
> > with Cirrus or MGA KMS where the hw cursor isn't always available.
> >
> > This patch adds the check of availability of hw cursor again after the
> > call of xf86SetCursor(), and proceeds only when it passes the test
> > after the call.
> This has been fixed with Michael Thayer's patch which has xf86SetCursor
> return whether the hw cursor was enabled successfully.

OK, thanks!  I myself forgot about this patch completely :)


More information about the xorg-devel mailing list