This fixes jerkiness in doom3 and other apps since the kernel change to
throttle less absurdly, which led to a thundering herd of frames.
Because this is a rather minimal fix, there is at least one downside: If
the whole scene completes in one batchbuffer, we'll end up stalling the GPU.
Thanks to Michel Dänzer for suggesting using glFlush to signal frame end
instead of going to all the effort of adding a new DRI2 extension.
Now that a current surface points back to its binding context, and a
current context points back to its binding thread, make sure there is no
dangling pointers. This commit reworks eglMakeCurrent, adds more checks
to avoid stealing context or surfaces from another thread, and correctly
destroys unlinked context and surfaces.
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
IsBound tells if a context or surface is current. What it does not tell
is, to which thread a context is current, or to which context a surface
is current. This commit replaces IsBound by a pointer to the binding
thread or context.
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
We need to properly set up a fake bo for the texture override,
so add a new function to radeon_bo_legacy.c. This could probably
be used on radeon/r200/r300 to unify the bo handling for
texture override.
compiz now works :)
It might be better to add an instruction to normalize the
coordinates for rectanglular textures as there are some limitations
to wrap modes on unnormalized tex coords.
fixes texrect
R300 hardware (but _not_ R500) hardware requires an enabled texture unit
if KIL is used in fragment programs. We now work around the CS checker
correctly when enabling such a fake texture unit.
Signed-off-by: Nicolai Hähnle <nhaehnle@gmail.com>
We used to not always correctly re-emit the Z buffer size in all cases,
in particular the clear path, and invalidated state was not always picked
up correctly.
This fixes a bug where the kernel CS checker correctly complains about
a Z buffer that is too small.
Note that this bug was probably only visible with ridiculously high
framerates, i.e. glxgears.
Signed-off-by: Nicolai Hähnle <nhaehnle@gmail.com>
This almost fixes compressed mipmapped textures on r200, though some small
mip levels are still broken.
Leave r300 compressed texture stride as is though afaik it's different
to pre-radeon-rewrite too. Also do the fixup for rs600 uncompressed row stride
at same place.
Reported to fix corruption while dragging an active window by John Bridgman.
Signed-off-by: Pauli Nieminen <suokkos@gmail.com>
Signed-off-by: Nicolai Hähnle <nhaehnle@gmail.com>
The LOOP/ENDLOOP pair is renamed to BGNFOR/ENDFOR as its behaviour
is similar to a C language for-loop.
The BGNLOOP2/ENDLOOP2 pair is renamed to BGNLOOP/ENDLOOP as now
there is no name collision.
Casting an unsigned int to or from a pointer directly gives warnings on
x86-64. Add wrappers to silence the warnings.
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
The field is not used, and in any case it would be more interesting to
manipulate from *outside* the compiler if we ever wanted to load several
fragment programs at the same time or something.
Signed-off-by: Nicolai Hähnle <nhaehnle@gmail.com>
When a texutre transfer is mapped for writing, mark the texture dirty
when unmapped. This was done in surface creation, and this commit moves
it to happen in texture unmapping.
This fixes subtex test in progs/tests/.
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
