sf_mask is the same as sf except that it must compute both src and mask uvw
cooefficients, which are conveniently adjacent in the same registers, and so
need only an extended execution width
This involves correctly computing u/v locations based on x/y vectors and
line constants computed in new sf program.
Also, use fewer instructions to make this go a bit faster (2X for 500x500
composite).
If the legacy bit is set, use both the BLC_PWM_CTL and LBB regs to control the
backlight, rather than just LBB. Looks like more platforms want that than what
the current code does. Note that kernel provided interfaces will always be
used if available, so this shouldn't affect users with /sys/class/backlight
interfaces at all.
Fixes#14721.
o Check for RANDR_GET_CRTC_INTERFACE before defining functions that
are used only if it is defined.
o Declare a variable before code, and rename it from ret to xvmc_status
to better describe it.
o if 0 some static functions not used.
o Don't declare some unused variables.
o Declare as static some functions that are used only in the file defining it.
o Add a default/fallback return True to the Bool function
src/xvmc/intel_batchbuffer.c:intelInitBatchBuffer().
o Ansify src/xvmc/xf86dri.c.
o Add missing prototype to src/xvmc/xf86dri.h and follow pattern of other
headers by adding "extern" before function prototype.
The new chips no longer automatically flush the rendering cache, so if we
don't flush the RC at blockhandler, the last rendering done may not
appear on the screen. This was particularly noticable with a bare Xorg with
some missing root weave, and terminals where the last character wouldn't
appear until the cursor blinked. A flush in the DRI blockhandler path had
hidden this issue for most people.
While I still like the idea, the mprotect calls themselves are failing on
Linux and causing more trouble than they're worth.
This reverts commit a1612b7728.
Conflicts:
src/i830_driver.c
src/i830_memory.c
Besides our driver having fallen through to the GM965 path for
RENCLK_GATE_D1, the BIOS was turning some of these on. It may be relevant
for previous platforms as well to zero out the fields that should be zero
in the other registers.
The homogeneous coordinate computation in the core server cannot be used for
many legal matrices as it overflows. Just use floats in the driver; faster
and avoids troubles.
When compositing with bilinear filter, don't push the dst coordinates around
as that makes the output blurry when pixels are aligned.
Projective transforms require un-normalized texture coordinates and the use
of the texldp instruction. The coordinates are passed as x/y/z/w (the z is
unused, but there isn't a vertext format for just x/y/w).
When we have unset channels (color for a8, or alpha for x8*), force them
to the appropriate value in the texture combiner rather than relying on getting
the desired results from teture mapping. Performance is the same on 865, so
multiple paths aren't necessary and we can use the less friendly texture
formats everywhere.
This fixed the crash appearing for Xv buffer memory, which appears
allocation and free happening in stolen memory. We should always
try to bind for stolen memory for correct protect setup, otherwise
we might get fault that stolen memory protect reverted to non-r/w
after previous free and unbind.
Credit goes to Lukas Hejtmanek to identify the real problem of Xv
crash!
This lets us get better sanity asserts, and avoid mysterious braces when you
do two BEGIN_LP_RING()s in a single function. Potential minor performance
loss isn't too exciting, as ring access is about to become a compat path
anyway. This change also removes the requirement for ring emits to be aligned
to dwords.
Keith Packard
Eric Anholt
Jesse Barnes
Matthias Hopf
Zhenyu Wang
Paulo Cesar Pereira de Andrade
Julien Cristau
Maxim Levitsky
Joerg Sonnenberger
Jesse Barnes
Jesse Barnes