| Age | Commit message (Collapse) | Author |
|---|
|
|
|
Relative addressing of constant buffers can't work properly through the
kcache, since you can only address within the currently locked kcache window.
Instead, this patch binds the constant buffer as a shader resource, and then
explicitly fetches the constant using a vertex fetch with fetch type
VTX_FETCH_NO_INDEX_OFFSET from the shader. There's probably still some room
for improvement, doing the fetch right before the instruction that needs the
value may not be quite optimal for example.
|
|
|
|
|
|
This allows them to be used for VS or PS buffer resources as well.
|
|
These are practically unused, only the vs_resource array is being abused for
fetch shader resources.
|
|
|
|
r600_shader_from_tgsi().
|
|
The r600_bc_alu_src structure is used in two different ways, as a vector and
for the individual channels of that same vector. This is somewhat fragile,
and probably confusing.
|
|
This is much easier to work with, and allows use to get rid of some of the
literal handling hacks.
|
|
|
|
|
|
|
|
MESA_drm_image isnt limited to drm platform,
others can enable the extension too.
|
|
|
|
|
|
|
|
|
|
|
|
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.
|
|
|
|
We don't have to unmap and recreate the upload buffer when a flush occurs.
This should also prevent buffer allocations from failing.
|
|
|
|
This uses the surface format to set the CB up not the underlying texture
format, since these can and do differ.
Fixes piglit fbo-srgb.
|
|
Since we never bind the actual DB to the CB/texture only the flushed one
we don't need to track the tile type at the moment.
|
|
|
|
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.
|
|
this fixes the mipmap tests with tiling forced on.
Signed-off-by: Dave Airlie <airlied@redhat.com>
|
|
Since these textures still have the depth bit set.
Signed-off-by: Dave Airlie <airlied@redhat.com>
|
|
|
|
For mipmap generation we need all levels in the flushing texture.
Signed-off-by: Dave Airlie <airlied@redhat.com>
|
|
|
|
This just adds a flag to create the texture without doing any
flushing to it. Flushing occurs in the draw function. This avoids
unnecessary flushes when we end up rebinding a CB/DB/texture due
to the blitter just restoring state.
Signed-off-by: Dave Airlie <airlied@redhat.com>
|
|
This checks the color buffer bindings to make sure there is something
to flush.
Signed-off-by: Dave Airlie <airlied@redhat.com>
|
|
If we are going to blit from a depth texture we need to flush
it before we blit from it.
Signed-off-by: Dave Airlie <airlied@redhat.com>
|
|
|
|
|
|
Plus other assorted clarifications.
|
|
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.
|
|
|
|
|
|
|
|
|
|
Leak was introduced when fixing strict aliasing violation in this code:
the reference counting was preserved, but the destructor call on zero
reference count was not.
|
|
See comment for more details.
|
|
Exactly one half would be the ideal, but this is a soft limit, and one
more byte over brings us to synchronous behavior.
Flushing when the referred GMR exceeds one third of the aperture gives us
statistically better performance.
|
|
|
|
Only svga_context_flush should do it, to ensure upload commands are not
submitted to hardware in an inconsistent state.
|
|
|
|
|
