What happened here is that when fixing bug #5734 we cahnged the
glyph image cache lookup to never return a NULL pointer, but to
instead return a 0x0 image surface. We didn't also update code
that was looking for a NULL pointer here, and as such we were
calling XRenderAddGlyphs with a 0x0 image which apparently
tickles a bug in some X servers.
If the BitBlt in get_image fails, we pretty much can't do anything -- so
fill the destination with white and hope for the best. This enables
somewhat accurate printing of complex operations. Also, check the
destination device caps before calling AlphaBlend; return UNSUPPORTED if
the destination DC can't do AlphaBlend.
(cherry picked from 9831de538e347a624af5b0ca38242b198b64bd45 commit)
(cherry picked from 1a14419126 commit)
If the DC is a display DC, inspect its depth and set out local format
appropriately. If it's not a display DC, assume RGB24.
(cherry picked from 6dd0a70d271f93df95f4bcaff5073b9bf90cecb6 commit)
(cherry picked from 2d784815ff commit)
Fixes up src coords and width/height before calling AlphaBlend/BitBlt; it's
an error to try to use a region that extents outside of the source surface
as a source DC.
Doesn't repair the extra region relative to the operator -- e.g. regions
outside of an ARGB source surface with SOURCE operator should be cleared
to fully transparent black in the destination.
(cherry picked from bc19c5b64b0e38e9d20045907d7b47d79f6afc60 commit)
(cherry picked from 68ed40b6da commit)
- Save extents at surface creation and always return these
- Simplify clipping paths
(cherry picked from a67722b3d7280e6c2375778c2d46556cca261bfc commit)
(cherry picked from 7ce09732a978749a4da814feb77f50161dc42c91 commit)
(cherry picked from e0f4eecb91 commit)
(cherry picked from d0cc56b9a24fa59febc15ac5de073da0e785e1cc commit)
(cherry picked from a6a054abe45287eb950c294e20366594313138f5 commit)
(cherry picked from f6b4fabaf5 commit)
Use & not && for bitwise AND.
src/fbpict.c src/fbmmx.c: Take 'dest' as an argument, if pict and dest don't match for RGB, BGR, swap result.
(cherry picked from a1bca924b9 commit)
Surely, the ! instead of ~ here was just a typo. I'd feel better if I
had a failing test case for this though, (time to break out gcov
again).
Fixes bug #6185:
Strange pointer operations in fbCompositeTrans_0888xnx0888 (pixman/src/fbpict.c)
https://bugs.freedesktop.org/show_bug.cgi?id=6185
(cherry picked from 5d89faf6b3 commit)
Change a little the logic in hinting to follow the same
logic as Xft and doesn't try to mix FT_LOAD_TARGET which should not be
mixed (based on David Turner comments)
