Age | Commit message (Collapse) | Author |
|
|
|
max_index could have been less than min_index, which later caused integer
underflow followed by a segfault in memcpy.
|
|
AA lines drawn as textured quads look a little better with this change.
Conformance/piglit tests still pass.
|
|
Spotted by Jakob Bornecrantz.
|
|
|
|
This can avoid DMAing the whole buffer when just a fraction was changed.
|
|
|
|
|
|
|
|
|
|
This code has originally matured in r300g and was ported to r600g several
times. It was obvious it's a code duplication.
See also comments in the header file.
|
|
The type of u_get_transfer_vtbl of the usage argument in u_transfer.h is
unsigned and not enum pipe_transfer_usage. This patch changes the type
of usage to unsigned to match the prototype in the header file.
|
|
Before, the set_sampler_views() and restore_sampler_views() functions
used MAX2(old,new) to tell the driver how many samplers or sampler
views to set. This could result in cases such as:
pipe->set_fragment_sampler_views(pipe, 4, views={foo, bar, NULL, NULL})
Many/most gallium drivers would take this as-is and set
ctx->num_sampler_views=4 and ctx->sampler_views={foo, bar, NULL, NULL, ...}.
Later, loops over ctx->num_sampler_views would have to check for null
pointers. Worse, the number of sampler views and number of sampler CSOs
could get out of sync:
ctx->num_samplers = 2
ctx->samplers = {foo, bar, ...}
ctx->num_sampler_views = 4
ctx->sampler_views={Foo, Bar, NULL, NULL, ...}
So loops over the num_samplers could run into null sampler_views pointers
or vice versa.
This fixes a failed assertion in the SVGA driver when running the Mesa
engine demo in AA line mode (and possibly other cases).
It looks like all gallium drivers are careful to unreference views
and null-out sampler CSO pointers for the units beyond what's set
with the pipe::bind_x_sampler_states() and pipe::set_x_sampler_views()
functions.
I'll update the gallium docs to explain this as well.
|
|
|
|
This consolidates the code duplicated between the fragment sampler
and vertex sampler functions. Plus, it'll make adding support for
geometry shader samplers trivial.
|
|
To better distinguish from vertex sampler fields.
|
|
Before we were looping to nr_samplers, which is the number of fragment
samplers, not vertex samplers.
NOTE: This is a candidate for the 7.9 and 7.10 branches.
|
|
|
|
User buffers may be the fastest way to upload data.
|
|
|
|
So that 'foo' can be found in: OPTION=prefixfoosuffix,foo
Also allow that debug options can be separated by a non-alphanumeric characters
instead of just commas.
|
|
This drops the memblock manager for ZMASK. Instead, only one zbuffer can be
compressed at a time. Note that this does not necessarily have to be slower.
When there is a large number of zbuffers, compression might be used more often
than it was before. It's also easier to debug.
How it works:
1) 'clear' turns the compression on.
2) If some other zbuffer is set or the currently-bound zbuffer is used
for texturing, the driver decompresses it and then turns the compression off.
Notes:
- The ZMASK clear has been refactored, so that only one packet3 is used to clear
ZMASK.
- The 8x8 compression mode is disabled. I couldn't make it work without issues.
- Also removed driver-specific stuff from u_blitter.
Driver status:
- RV530 and R580 appear to just work (finally).
- RV570 should work, but there may be an issue that we don't correctly
calculate the number of dwords to clear, resulting in a partially
uninitialized zbuffer.
- RS690 misrenders as if no ZMASK clear happened. No idea what's going on.
- RV350 may even hardlock. This issue was already present and this patch doesn't
fix it.
I think we are still missing some hardware info we need to make the zbuffer
compression work fully.
Note that there is also an issue with HiZ, resulting in a sort of blocky
zigzagged corruption around some objects.
|
|
Fixes http://bugs.freedesktop.org/show_bug.cgi?id=33555
|
|
Let's assume there are two options with names such that one is a substring
of another. Previously, if we only specified the longer one as a debug option,
the shorter one would be considered specified as well (because of strstr).
This commit fixes it by checking that each option is surrounded by commas.
(a regexp would be nicer, but this is not a performance critical code)
|
|
|
|
The module uses the 3D engine, so it can blit non-compatible formats.
|
|
|
|
largely a merge of the previously discussed origin/gallium-resource-sampling
but updated.
the idea is to allow arbitrary binding of resources, the way opencl, new gl
versions and dx10+ require, i.e.
DCL RES[0], 2D, FLOAT
LOAD DST[0], SRC[0], RES[0]
SAMPLE DST[0], SRC[0], RES[0], SAMP[0]
|
|
Contrary what the name may suggest, LLVM's opaque types are used for
recursive types -- types whose definition refers itself -- so opaque
types correspond to pre-declaring a structure in C. E.g.:
struct node;
struct link {
....
struct node *next;
};
struct node {
struct link link;
}
Void pointers are also disallowed by LLVM. So the suggested way of creating
what's commonly referred as "opaque pointers" is using byte pointer (i.e.,
uint8_t * ).
|
|
Signed-off-by: Brian Paul <brianp@vmware.com>
|
|
|
|
|
|
|
|
Implement it in terms of pipe_get_tile_rgba_format()
|
|
The new function, pipe_get_tile_rgba_format(), no longer takes a
swizzle (we weren't actually using it anywhere). Rename it to
indicate that the format is passed explicitly.
|
|
|
|
|
|
Conflicts:
src/gallium/auxiliary/draw/draw_llvm.c
src/gallium/drivers/llvmpipe/lp_state_fs.c
src/glsl/ir_set_program_inouts.cpp
src/mesa/tnl/t_vb_program.c
|
|
When use_spoken is true, istart (the first vertex of this segment) is
replaced by i0 (the spoken vertex of the fan). There are still icount
vertices.
Thanks to Brian Paul for spotting this.
|
|
|
|
|
|
If you want to enable noop set GALLIUM_NOOP=1 as an env variable.
You need first to enable noop wrapping for your driver see change
to src/gallium/targets/dri-r600/ in this commit as an example.
Signed-off-by: Jerome Glisse <jglisse@redhat.com>
|
|
|
|
I also specified the array sizes in the header so that one can use
the Elements macro on it.
|
|
So that a state tracker can unreference them after set_vertex_buffers.
|
|
- Added a parameter to specify a minimum offset that should be returned.
r300g needs this to better implement user buffer uploads. This weird
requirement comes from the fact that the Radeon DRM doesn't support negative
offsets.
- Added a parameter to notify a driver that the upload flush occured.
A driver may skip buffer validation if there was no flush, resulting
in a better performance.
- Added a new upload function that returns a pointer to the upload buffer
directly, so that the buffer can be filled e.g. by the translate module.
|
|
The map/unmap overhead can be significant even though there is no waiting on busy
buffers. There is simply a huge number of uploads.
This is a performance optimization for Torcs, a car racing game.
|
|
|
|
|
|
|