XQuartz: Don't enable rootless accelerated functionality... crashy...

This commit is contained in:
Jeremy Huddleston 2008-04-03 16:29:43 -07:00
parent ed15556a9f
commit cdb4c291d8
4 changed files with 8 additions and 11 deletions

View File

@ -351,7 +351,9 @@ xprSetupScreen(int index, ScreenPtr pScreen)
{ {
// Initialize accelerated rootless drawing // Initialize accelerated rootless drawing
// Note that this must be done before DamageSetup(). // Note that this must be done before DamageSetup().
RootlessAccelInit(pScreen);
// These are crashing ugly... better to be stable and not crash for now.
//RootlessAccelInit(pScreen);
#ifdef DAMAGE #ifdef DAMAGE
// The Damage extension needs to wrap underneath the // The Damage extension needs to wrap underneath the

View File

@ -76,10 +76,6 @@ rootlessConfig.h to specify compile time options for its platform.
The following compile-time options are defined in The following compile-time options are defined in
rootlessConfig.h: rootlessConfig.h:
o ROOTLESS_ACCEL: If true, use the optional rootless acceleration
functions where possible to a accelerate X11 drawing primitives.
If false, all drawing will be done with fb.
o ROOTLESS_GLOBAL_COORDS: This option controls the way that frame o ROOTLESS_GLOBAL_COORDS: This option controls the way that frame
coordinates are passed to the rootless implementation. If false, coordinates are passed to the rootless implementation. If false,
the coordinates are passed per screen relative to the origin of the coordinates are passed per screen relative to the origin of

View File

@ -36,12 +36,12 @@
#ifdef __APPLE__ #ifdef __APPLE__
# define ROOTLESS_ACCEL TRUE
# define ROOTLESS_GLOBAL_COORDS TRUE # define ROOTLESS_GLOBAL_COORDS TRUE
# define ROOTLESS_PROTECT_ALPHA TRUE # define ROOTLESS_PROTECT_ALPHA TRUE
# define ROOTLESS_REDISPLAY_DELAY 10 # define ROOTLESS_REDISPLAY_DELAY 10
# define ROOTLESS_RESIZE_GRAVITY TRUE # define ROOTLESS_RESIZE_GRAVITY TRUE
# undef ROOTLESS_TRACK_DAMAGE # undef ROOTLESS_TRACK_DAMAGE
/*# define ROOTLESSDEBUG*/
/* Bit mask for alpha channel with a particular number of bits per /* Bit mask for alpha channel with a particular number of bits per
pixel. Note that we only care for 32bpp data. Mac OS X uses planar pixel. Note that we only care for 32bpp data. Mac OS X uses planar
@ -52,7 +52,6 @@
#if defined(__CYGWIN__) || defined(WIN32) #if defined(__CYGWIN__) || defined(WIN32)
# define ROOTLESS_ACCEL YES
# define ROOTLESS_GLOBAL_COORDS TRUE # define ROOTLESS_GLOBAL_COORDS TRUE
# define ROOTLESS_PROTECT_ALPHA NO # define ROOTLESS_PROTECT_ALPHA NO
# define ROOTLESS_REDISPLAY_DELAY 10 # define ROOTLESS_REDISPLAY_DELAY 10

View File

@ -118,7 +118,7 @@ static GCOps rootlessGCOps = {
/* /*
There are two issues we must contend with when drawing. These are There are two issues we must contend with when drawing. These are
controlled with ROOTLESS_PROTECT_ALPHA and ROOTLESS_ACCEL. controlled with ROOTLESS_PROTECT_ALPHA and RootlessAccelInit().
If ROOTLESS_PROTECT_ALPHA is set, we have to make sure that the alpha If ROOTLESS_PROTECT_ALPHA is set, we have to make sure that the alpha
channel of the on screen windows is always opaque. fb makes this harder channel of the on screen windows is always opaque. fb makes this harder
@ -141,9 +141,9 @@ static GCOps rootlessGCOps = {
from another window since its alpha channel must also be opaque. from another window since its alpha channel must also be opaque.
The other issue to consider is that the rootless implementation may The other issue to consider is that the rootless implementation may
provide accelerated drawing functions if ROOTLESS_ACCEL is set. For some provide accelerated drawing functions if RootlessAccelInit() is called.For
drawing primitives we swap in rootless acceleration functions, which use some drawing primitives we swap in rootless acceleration functions, which
the accelerated drawing functions where possible. use the accelerated drawing functions where possible.
Where both alpha protection and acceleration is used, it is even a bigger Where both alpha protection and acceleration is used, it is even a bigger
win to relax the planemask to all ones because most accelerated drawing win to relax the planemask to all ones because most accelerated drawing