Use the READ and WRITE macros to wrap memory accesses that could be in video
memory. Add MEMCPY_WRAPPED and MEMSET_WRAPPED macros to wrap memcpy and
memset, respectively.
Pass the size of the data pointed to by src or dst as an argument to
wfb{Read,Write}Memory. This allows one set of macros to be used with any size
pointer. Assumes that sizeof(FbBits) >= sizeof(FbStip).
Create fbPrepareAccess macros to call into the driver to set up the
wfbReadMemory and wfbWriteWemory pointers. Call these from fbGetDrawable and
fbGetStipDrawable.
Add the READ and WRITE macros, which expand to simple memory accesses for fb,
and calls through the function pointers for wfb.
Add fbFinishAccess macro to give the driver an opportunity to clean up. Add
calls to this in the appropriate places.
For now, just #define all of the exported symbols in wfbrename.h. Later,
we should add FBPREFIX() around the exported symbols and use -fvisiblity=hidden
to hide the rest.
Builds fb/* twice, defining FB_ACCESS_WRAPPER for libwfb.la. Define a macro,
FBPREFIX(X) which expands to fbX for libfb.la and wfbX for libwfb.la. Use the
macro on [w]fbModuleData so the new module loads.
Get rid of almost all uses of these definitions. They're still defined for
delinquent out-of-tree drivers, and also for the Mesa build. As well as
for miinitext.c. But largely gone.
Instead, stick the NULL return default case afterwards, so that the compiler can
warn us when we've got unimplemented cases. Removes some unimplemented and
unused 8bpp, depth 4 picture format names.
transforms so we can avoid doing the expensive
64-bit math. Unroll the bilinear interpolation loops for an extra boost. I
tested this with the up/downscaling cairo-benchmarks with Xvfb and saw
a 12% +/- 4% decrease in time taken to run them.
value is being added to multiple pixels. This improves the speed of
rendering wide trapezoids. I tested this with a small set of xlibs
cairo-benchmarks with Xvfb and saw a 4% decrease in time taken to run
them.
- Merge various fb/ bits of COMPOSITE support from xserver, which weren't
necessary before due to cw hiding the issues. Fixes offset calculations
for a number of operations, and may pull some fixes that cairo has
wanted for XAA as well.
- Add a new call, miDisableCompositeWrapper(), which a DDX can call to keep
cw from getting initialized from the damage code. While it would be
cleaner to have each DDX initialize it if it needs it, we don't have
control over all of them (e.g. nvidia).
- Use the miDisableCompositeWrapper() to keep cw from getting set up for
screens using EXA, because EXA is already aware of composite. Avoiding
cw improved performance 0-35% on operations tested by ajax in x11perf.
Make the fbCompose paths use the existing inline functions instead of
macros. Various other cleanups.
Remove macro definitions, move typedef to fbmmx.c
Add XSERV_t, TRANS_SERVER, TRANS_REOPEN to quash warnings.
Add #include <dix-config.h> or <xorg-config.h>, as appropriate, to all
source files in the xserver/xorg tree, predicated on defines of
HAVE_{DIX,XORG}_CONFIG_H. Change all Xfont includes to
<X11/fonts/foo.h>.
Changed the semantics of the Convolution filter a bit. It now doesn't try
to normalize the filter values but leaves this to the client. This
gives more reasonable behaviour in the limit where the filter
parameters sum up to 0.
change "foo.h" to <X11/foo.h> for core headers, e.g. X.h, Xpoll.h;
change "foo.h", "extensions/foo.h" and "X11/foo.h" to
<X11/extensions/foo.h> for extension headers, e.g. Xv.h;
change "foo.[ch]" to <X11/Xtrans/foo.[ch]> for Xtrans files.
Add MMX implementation of non-repeating source IN repeating mask, aka
"translucent window".
Add MMX implementation of CopyArea.
Use MMX implementation of CopyArea.
Use the new implementations.
Clean-ups and support for AMD64. Bug 1067. Patch by Nicholas Miell
(nmiell@comcast.net)
Add support for AMD64
Many cleanups using <mmintrin.h> instead of __builin_ia32_*, and intrinsics
instead of inline assembly. Also unconditionally use pshufw on AMD64.
s/USE_GCC34_MMX/USE_MMX/g
compiled in or not.
Removed stale make variable.
Updated xaaWrapper.c: Under certain circumstances the accel path was not
used when possible. Removed some debugging stuff and stale code that
had been commented out.
Check if maps are really installed before attempting to list them.
Wrap InstallColormap so that the DDX doesn't see colormaps from our ARGB
visual (avoids lovely green tint to screen). Also, set visual->nplanes
of ARGB visual to all used (including alpha) planes so DIX can set
pixel values correctly.
Translate automatic update regions correctly to account for borders
When nplanes == 32 (ARGB visuals), mask in all ones for alpha values to
allocated pixel values.
Remove redundant fbAddTraps declaration
Fix fbCopyWindow to work on non-screen pixmaps (not needed yet)
Replace broken clipping code with that from modular tree.
Respect subWindowMode.
Xnest was half-using midispcur and doing a bad job of it. Replace all of
that code with mipointer which does a lot of the work.
Support DDXen which don't provide GetWindowPixmap, or which return NULL for
the root pixmap.
the other render symbols.
Initialize screen_x and screen_y to 0. This fixes a problem with the new
Damage based sprite, where intersection test against the root pixmap
would fail because the two fields were not initialized (damage.c,
166-170).
framebuffer formats except cfb and the overlay modes should work, and
r128 and radeon need to be loaded from the ati driver (both issues to
be fixed soon). Tested on i740, s3virge, mach64, tdfx, vesa, and vga
drivers. elfloader users shouldn't be affected.
Call MMX solid fill routine when available.
Call MMX operations when available.
New HasGcc34 macro
New file with many operations implemented with MMX intrinsics, conditional
on having GCC 3.4 on i386.
xc/programs/Xserver/fb/fboverlay.c
xc/programs/Xserver/fb/fbscreen.c
- Change #ifdef for checking for old format miScreenInit to
FB_OLD_MISCREENINIT for easier portability to xservers with updated
screen structs but old function prototypes. Make it automatically
defined if FB_OLD_SCREEN is defined.
- Add _LP64 to list of #ifdefs for 64-bit platforms to support
64-bit Solaris.
- fbCombineSaturate was pointed at fbCombineDisjointOver, instead of
fbCombineDisjointOverReverse as it should. Instead, point
fbCombineDisjointOverReverse at fbCombineSaturate (which is likely to
be faster).
- fix previously-unused fbCombineSaturate implementation.
- fbCombineMaskAlphaC was just a copy of fbCombineMaskValueC. Make it do
what it's supposed to (return a cs.alpha).
- fbCombineAtopC didn't invert the source alpha value.
- fix copy'n'paste errors in fbCombine(Dis/Con)jointGeneralC, also source
alpha wasn't treated in a component fashion.
- fbCompositeSrc_8888* didn't handle when the source lacks an alpha
channel. Rather than adding that and possilby slowing down the (normal)
alpha case, don't let x8r8g8b8/x8b8g8r8 Pictures be used in
fbCompositeSrc_8888* because Over with one of these is just Src.