This reverts commit bd25e23bf3.
Apart from introducing a lot of hex magic numbers and being highly impenetable code,
it causes lots of lockups on an average piglit run that always runs without lockups.
Always run piglit before/after doing big things like this.
this adds handling for some more CF instructions and conditions
also adds parameter for stack size emission
These seem to pass on VS with the stack size hack but not on FS,
TODO: fix FS + stack size calcs
this makes op2 emission smaller, since it skips instructions
that don't write to the dst. not sure if this could have unwanted
side effects but try it and see.
Our channel-expressions and vector-splitting changes now happen into a
private copy of the IR that we maintain for ourselves. Uniform
assignment still happens by the core, so we continue using Mesa IR
generation not just for swrast fallbacks but also for uniform values
(since there's no storage for their contents other than
shader_program->FragmentProgram->Parameters->ParameterValues). And
most importantly, at the moment no actual codegen is hooked up other
than emitting our favorite color to the framebuffer.
Combined with the previous pass, this lets other optimization passes
do their work thanks to ir_tree_grafting. Still have regression in
instruction count with INTEL_NEW_FS, but register count is even
better.
This is a step towards implementing a GLSL IR backend for the 965
fragment shader. Because it has downsides with the current codegen,
it is hidden under the environment variable INTEL_NEW_FS.
This results in an increase in instruction count at the moment (1444
-> 1752 for glsl-fs-raytrace, 345 -> 359 on my demo), because dot
products are turned into a series of multiplies and adds instead of a
custom expansion of MULs and MACs, and by not splitting the variable
types up we don't get tree grafting and thus there are extra moves of
temporary storage. However, register count drops for the non-GLSL
path (64 -> 56 on my demo shader) because the register allocator sees
all the sub-operations.
Per the GLSL 1.20 specification (presumably a clarification of 1.10).
Also, when creating user functions, make a new ir_function that shadows the
built-in ir_function, rather than adding new signatures. User functions
are supposed to hide built-ins, not overload them.
Fixes piglit tests redeclaration-{04, 12, 14}.vert.
Moving the check for an earlier variable declaration helps cleanly
separate out the re-declaration vs. new declaration code a bit. With
that in place, conflicts between variable names and structure types or
function names aren't caught by the earlier "redeclaration" error
message, so check the return type on glsl_symbol_table::add_variable
and issue an error there. If one occurs, don't emit the initializer.
Fixes redeclaration-01.vert and redeclaration-09.vert.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
As of 1.20, variable names, function names, and structure type names all
share a single namespace, and should conflict with one another in the
same scope, or hide each other in nested scopes.
However, in 1.10, variables and functions can share the same name in the
same scope. Structure types, however, conflict with/hide both.
Fixes piglit tests redeclaration-06.vert, redeclaration-11.vert,
redeclaration-19.vert, and struct-05.vert.
Intel sometimes uses packed depth/stencil buffers even when only a depth
buffer or only a stencil buffer was requested. Common code currently
uses the _BaseFormat field to determine whether a depth/stencil wrapper
is necessary. But unless the user explicitly requested a packed
depth/stencil buffer, the _BaseFormat field does not encode this
information, and the required wrappers are not created.
The problem was introduced by commit 45e76d2665 ("mesa: remove a
bunch of gl_renderbuffer fields"), which killed off the _ActualFormat
field upon which the decision to create a wrapper used to be made. This
patch changes the logic to use the Format field instead, which is more
like the old code.
Fixes fdo bug 27590.
Signed-off-by: Nick Bowler <nbowler@draconx.ca>
Signed-off-by: Brian Paul <brianp@vmware.com>
when we dont know max_index we cannot calculate vb size from count
anymore - just use the bo size.
Also added an assert to remind that we dont handle GL_INT GL_DOUBLE
upload when we dont' know max_index - will fix later
Keith prefers a clean separation between graw applications and
implementations, where apps do not link libgallium.a but instead
get all functionality they need via graw interface.
Although this is not incompatible with late loading of graw drivers, it
it would make it very hard to maintain, as wrappers for every utility
symbol exposed in graw would have to be written or generated somehow.
stObj->pt should be set in st_bind_surface, just as in st_TexImage. On
the other hand, st_TexImage should unreference stObj->pt. It also needs
to initialize the texture image again as _mesa_clear_texture_object
clears the image.
