Kenneth Graunke [Wed, 26 May 2010 23:41:47 +0000 (16:41 -0700)]
Add mappings between ir_texture_opcode and strings.
Ian Romanick [Thu, 29 Apr 2010 01:42:36 +0000 (18:42 -0700)]
Define IR instruction for texture look-ups
Ian Romanick [Wed, 9 Jun 2010 18:00:00 +0000 (11:00 -0700)]
Use array delete operator to delete an array
This was detected by valgrind. I think GCC still does the right
thing, but the C++ spec allows the compiler to do something
stupid... like crash or only delete the first entry in the array.
Ian Romanick [Tue, 8 Jun 2010 02:10:33 +0000 (19:10 -0700)]
Bump GL_MAX_TEXTURE_COORDS from 2 to 4
Every platform that supports GLSL sets GL_MAX_TEXTURE_COORDS to at
least 4, so hard-code 4 for now.
This causes the following tests to pass:
glslparsertest/glsl2/norsetto-bumptbn_sh_fp.vert
glslparsertest/glsl2/xreal-lighting-d-omni.vert
glslparsertest/glsl2/xreal-lighting-db-omni.vert
glslparsertest/glsl2/xreal-lighting-dbs-omni.vert
Ian Romanick [Tue, 8 Jun 2010 02:02:44 +0000 (19:02 -0700)]
Only allow global precision qualifier for int and float
This causes the following tests to pass:
glslparsertest/glsl2/precision-03.vert
Ian Romanick [Tue, 8 Jun 2010 01:55:41 +0000 (18:55 -0700)]
Fix parsing of precision qualifiers
This causes the following tests to pass:
glslparsertest/glsl2/precision-02.vert
glslparsertest/glsl2/precision-04.vert
glslparsertest/glsl2/precision-06.vert
This causes the following test to fail. This shader was previously
failing to compile, but it was failing for the wrong reasons.
glslparsertest/glsl2/precision-03.vert
Ian Romanick [Tue, 8 Jun 2010 01:53:06 +0000 (18:53 -0700)]
Don't process empty shaders
Some valid shaders, such as 'precision highp float;', evaluate to
empty sets of instructions. This causes some of the optimization
stages to enter infinite loops. Instead, don't bother processing the
empty ones.
Ian Romanick [Tue, 8 Jun 2010 01:49:48 +0000 (18:49 -0700)]
Generate an error on empty declaration lists
This causes an error for code such as 'float;'
Kenneth Graunke [Fri, 4 Jun 2010 19:56:15 +0000 (12:56 -0700)]
Check variable mode when comparing qualifier lists.
Fixes function-05.vert.
Ian Romanick [Wed, 2 Jun 2010 20:50:27 +0000 (13:50 -0700)]
ir_dereference_array always operates on an r-value
ir_dereference_array::array is always an r-value. If the dereference
is of a varaible, that r-value will be an ir_dereference_variable.
This simplifies the code a bit.
Ian Romanick [Wed, 2 Jun 2010 20:39:45 +0000 (13:39 -0700)]
There is no class ir_label, so there's no need for ir_instruction::as_label
Ian Romanick [Wed, 2 Jun 2010 20:13:40 +0000 (13:13 -0700)]
Remove some cruft from the Makefile
This was affecting the build, but the files don't actually exist.
Eric Anholt [Wed, 19 May 2010 21:09:04 +0000 (14:09 -0700)]
Add builtin gl_LightSource[].
Fixes gst-gl-bumper.vert parsing.
Eric Anholt [Wed, 19 May 2010 21:01:33 +0000 (14:01 -0700)]
builtins: Add support for reflect().
Fixes glsl-orangebook-ch06.frag parsing.
Eric Anholt [Wed, 19 May 2010 20:57:01 +0000 (13:57 -0700)]
builtins: Add ftransform().
Fixes glsl-orangebook-ch06-bump.vert.
Eric Anholt [Wed, 19 May 2010 20:38:15 +0000 (13:38 -0700)]
Allow arrays of floats as varyings.
The comment just above the code said arrays were OK, then it didn't
handle arrays. Whoops. Partially fixes CorrectUnsizedArray.frat.
Eric Anholt [Wed, 19 May 2010 20:30:14 +0000 (13:30 -0700)]
gl_Normal is a vec3 not a vec4.
Fixes CorrectSwizzle1.vert.
Eric Anholt [Wed, 19 May 2010 20:05:01 +0000 (13:05 -0700)]
builtins: Add the mix(gentype, gentype, float) variant.
The broken-in-mesa Regnum Online shader now parses, except for its
preprocessor usage.
Eric Anholt [Wed, 19 May 2010 17:38:37 +0000 (10:38 -0700)]
Handle GLSL 1.20 implicit type conversions.
We were nicely constructing a new expression for the implicit type
conversion, but then checking that the previous types matched instead
of the new expression's type. Fixes errors in Regnum Online shaders.
Eric Anholt [Wed, 19 May 2010 19:44:23 +0000 (12:44 -0700)]
builtins: Add atan().
Eric Anholt [Wed, 19 May 2010 19:19:54 +0000 (12:19 -0700)]
builtins: Add asin().
Eric Anholt [Wed, 12 May 2010 21:42:21 +0000 (14:42 -0700)]
ir_constant_expression: Handle several floating point unops.
Cleans up a bunch of pointless operations in a GStreamer fragment shader.
Eric Anholt [Wed, 12 May 2010 19:10:41 +0000 (12:10 -0700)]
ir_constant_variable: New pass to mark constant-assigned variables constant.
This removes a bunch of gratuitous moving around of constant values
from constructors. Makes a shader ir I was looking at for structure
handling almost readable.
Eric Anholt [Wed, 12 May 2010 19:15:40 +0000 (12:15 -0700)]
ir_constant_folding: Look at instructions in functions.
This was broken in the ir_label -> ir_function rework.
Eric Anholt [Tue, 11 May 2010 19:34:21 +0000 (12:34 -0700)]
ir_swizzle_swizzle: Reduce swizzle chains to a single swizzle.
Eric Anholt [Tue, 11 May 2010 18:31:09 +0000 (11:31 -0700)]
ir_vec_index_to_swizzle: Pass to convert indexing of vectors to swizzles.
This should remove the burden of handling constant vector indexing
well from backend codegen, and could help with swizzle optimizations.
Eric Anholt [Thu, 6 May 2010 21:48:20 +0000 (14:48 -0700)]
ir_expression_flattening: Handle flattening values out of swizzles, too.
Fixes an uninlined normalize() in CorrectSwizzle2.vert.
Eric Anholt [Tue, 1 Jun 2010 22:06:11 +0000 (15:06 -0700)]
ir_expression_flattening: Fix breakage from hierarchichal visitor.
Similar to other situations where the visitor pattern doesn't fit, in
this case we need the pointer to the base instruction in the
instruction stream for where to insert any new instructions we
generate (not the instruction in the tree we're looking at). By
removing the code for setting the base_ir, flattened expressions would
end up, for example, before the function definition where they had appeared.
Ian Romanick [Thu, 27 May 2010 01:58:27 +0000 (18:58 -0700)]
Reimplement ir_function_inlining_visitor using ir_hierarchical_vistor
Ian Romanick [Thu, 27 May 2010 00:04:19 +0000 (17:04 -0700)]
Reimplement ir_copy_propagation_visitor using ir_hierarchical_vistor
Ian Romanick [Wed, 26 May 2010 23:07:49 +0000 (16:07 -0700)]
Reimplement ir_expression_flattening_visitor using ir_hierarchical_vistor
Kenneth Graunke [Wed, 26 May 2010 22:20:59 +0000 (15:20 -0700)]
ir_reader: Read record_refs.
Also changes the print visitor to not emit extraneous parenthesis.
Kenneth Graunke [Wed, 26 May 2010 20:03:14 +0000 (13:03 -0700)]
ir_reader: Fix reading of array deferences and correct error messages.
Previously, the syntax was (array_ref <variable name> <index>), but the
subject is now a general rvalue (not a name). In particular, it might
be a (var_ref ...).
Also, remove "expected ... or (swiz)" from error messages; swiz is not
allowed inside a var_ref.
Ian Romanick [Wed, 26 May 2010 22:08:11 +0000 (15:08 -0700)]
Fix setting the maximum accessed array element
Array dereferences now point to variable dereferences instead of
pointing directly to variables. This necessitated some changes to the
way the variable is accessed when setting the maximum index array element.
Ian Romanick [Wed, 26 May 2010 18:43:40 +0000 (11:43 -0700)]
ir_dereference::mode is no longer used, kill with fire
Ian Romanick [Wed, 26 May 2010 18:32:52 +0000 (11:32 -0700)]
Refactor whole-variable assigment checking into member function
Ian Romanick [Wed, 19 May 2010 11:52:29 +0000 (13:52 +0200)]
Refactor ir_dereference data fields to subclasses
Ian Romanick [Wed, 19 May 2010 11:47:39 +0000 (13:47 +0200)]
Replace open coded deref navigation with hierarchical visitors
Ian Romanick [Wed, 19 May 2010 11:20:12 +0000 (13:20 +0200)]
Refactor ir_dereference support for ir_visitor
Move the accept method for visitors from ir_dereference to the derived
classes.
Ian Romanick [Wed, 19 May 2010 10:02:19 +0000 (12:02 +0200)]
Refactor ir_dereference support for ir_hierarchical_visitor
Move the accept method for hierarchical visitors from ir_dereference
to the derived classes. This was mostly straight-forward, but I
suspect that ir_dead_code_local may be broken now.
Ian Romanick [Wed, 19 May 2010 09:37:35 +0000 (11:37 +0200)]
Begin refactoring ir_dereference
Create separate subclasses of ir_dereference for variable, array, and
record dereferences. As a side effect, array and record dereferences
no longer point to ir_variable objects directly. Instead they each
point to an ir_dereference_variable object.
This is the first of several steps in the refactoring process. The
intention is that ir_dereference will eventually become an abstract
base class.
Ian Romanick [Tue, 18 May 2010 11:53:20 +0000 (13:53 +0200)]
Use ir_rvalue::variable_referenced instead of open coding it
Ian Romanick [Fri, 14 May 2010 21:32:17 +0000 (14:32 -0700)]
ir_visit_tree is no longer used, remove ir_visit_tree.{cpp,h}
Ian Romanick [Fri, 14 May 2010 21:31:25 +0000 (14:31 -0700)]
Reimplement kill_for_derefs using ir_hierarchical_vistor
The output of all test cases was verified to be the same using diff.
Ian Romanick [Fri, 14 May 2010 21:11:06 +0000 (14:11 -0700)]
Reimplement has_call_callback using ir_hierarchical_vistor
This has the added advantage that it will stop traversing the tree as
soon as the first call is found.
The output of all test cases was verified to be the same using diff.
Ian Romanick [Fri, 14 May 2010 20:35:27 +0000 (13:35 -0700)]
Reimplement ir_if_simplicifation_visitor using ir_hierarchical_vistor
The output of all test cases was verified to be the same using diff.
Ian Romanick [Fri, 14 May 2010 20:34:43 +0000 (13:34 -0700)]
Make visit_list_elements safe against node removals
Ian Romanick [Fri, 14 May 2010 19:53:33 +0000 (12:53 -0700)]
Reimplement ir_function_can_inline_visitor using ir_hierarchical_vistor
The output of all test cases was verified to be the same using diff.
Ian Romanick [Fri, 14 May 2010 19:41:22 +0000 (12:41 -0700)]
Reimplement ir_dead_code_visitor using ir_hierarchical_vistor
The output of all test cases was verified to be the same using diff.
Ian Romanick [Fri, 14 May 2010 19:39:23 +0000 (12:39 -0700)]
Add ir_hierarchical_visitor base class and associated infrastructure
This type of visitor should eventually replace all or almost all
current uses of ir_visitor.
Ian Romanick [Tue, 11 May 2010 00:44:57 +0000 (17:44 -0700)]
Replace find_dead_code with visit_exec_list
find_dead_code appears to be an open-coded version of visit_exec_list
that was implemented first.
Ian Romanick [Sat, 15 May 2010 00:36:00 +0000 (17:36 -0700)]
Use ir_rvalue::variable_referenced instead of open-coding it
Ian Romanick [Sat, 15 May 2010 00:35:42 +0000 (17:35 -0700)]
Add ir_rvalue::variable_referenced
Kenneth Graunke [Fri, 30 Apr 2010 22:20:06 +0000 (15:20 -0700)]
Refresh autogenerated builtin_function.cpp.
Kenneth Graunke [Tue, 4 May 2010 05:12:35 +0000 (22:12 -0700)]
Implement "tan" builtin.
Kenneth Graunke [Tue, 4 May 2010 05:11:17 +0000 (22:11 -0700)]
Implement "sin" and "cos" builtins via new expression operators.
Kenneth Graunke [Tue, 4 May 2010 04:29:01 +0000 (21:29 -0700)]
Implement "cross" builtin.
Kenneth Graunke [Tue, 4 May 2010 03:59:09 +0000 (20:59 -0700)]
Implement "fract" builtin.
Kenneth Graunke [Tue, 4 May 2010 03:05:57 +0000 (20:05 -0700)]
Implement "sign" builtin via a new expression operator.
Kenneth Graunke [Tue, 4 May 2010 02:10:04 +0000 (19:10 -0700)]
Implement "smoothstep" builtin.
Kenneth Graunke [Fri, 30 Apr 2010 22:18:07 +0000 (15:18 -0700)]
Implement "step" builtin.
Kenneth Graunke [Mon, 3 May 2010 21:39:03 +0000 (14:39 -0700)]
Integrate generate_builtins.pl into the build process.
make will now regenerate builtin_function.cpp whenever you
change/add/remove files in the builtins/* folders.
Kenneth Graunke [Thu, 29 Apr 2010 06:01:33 +0000 (23:01 -0700)]
Replace old builtin_function.cpp with new autogenerated one.
Kenneth Graunke [Thu, 29 Apr 2010 06:00:43 +0000 (23:00 -0700)]
Add a perl script to generate builtin_function.cpp.
Usage: ./builtins/tools/generate_builtins.pl > builtin_function.cpp
Kenneth Graunke [Mon, 3 May 2010 09:40:10 +0000 (02:40 -0700)]
Fix bogus expression typing in various builtins.
Kenneth Graunke [Thu, 29 Apr 2010 05:57:47 +0000 (22:57 -0700)]
Initial commit of IR for builtins.
These were all generated by Eric's existing builtin_functions.cpp; I split
the uvec* signatures out of 110 into the 130 folder.
Kenneth Graunke [Thu, 29 Apr 2010 06:17:58 +0000 (23:17 -0700)]
ir_reader: Set function signatures as defined.
Eric Anholt [Thu, 6 May 2010 19:57:23 +0000 (12:57 -0700)]
FS gl_FragCoord and and gl_FrontFacing are FS ins, not outs.
Ian Romanick [Fri, 14 May 2010 23:06:41 +0000 (16:06 -0700)]
Fix function call parameter printer to omit extraneous leading comma
The output of all test cases was verified to be the same using diff.
Ian Romanick [Tue, 11 May 2010 00:42:05 +0000 (17:42 -0700)]
Replace many uses of foreach_list with foreach_list_typed
Ian Romanick [Tue, 11 May 2010 00:40:41 +0000 (17:40 -0700)]
exec_list: Add foreach_list_typed and foreach_list_typed_const
These variations are parameterized by the type of the nodes in the
list. This enables skipping the explicit usage of exec_node_data in
the loop body.
Ian Romanick [Mon, 10 May 2010 18:17:53 +0000 (11:17 -0700)]
Convert ast_node use of simple_node to exec_list and exec_node
Ian Romanick [Mon, 10 May 2010 18:17:23 +0000 (11:17 -0700)]
exec_list: Add simpler exec_list for-each macros
Ian Romanick [Mon, 10 May 2010 18:16:24 +0000 (11:16 -0700)]
exec_list: Add macros to get ptr to structure containing a node
This has some ugly hackery to work-around C++ fail. I have
emperically determined that it works in all the cases that matter.
Ian Romanick [Mon, 10 May 2010 18:10:26 +0000 (11:10 -0700)]
Loop bodies, then-statements, and else-statements are not lists
The statement making up a loop body, a then-statement, or an
else-statement are single nodes. If the statement is a block, the
single node will be an ast_compound_statement. There is no need to
loop at the top level when processing these statements.
Ian Romanick [Mon, 10 May 2010 17:56:03 +0000 (10:56 -0700)]
Remove unnecessary include of simple_list.h
Ian Romanick [Mon, 10 May 2010 17:47:14 +0000 (10:47 -0700)]
Store AST function call parameters in expressions
Previously the list of function call parameters was stored as a
circular list in ast_expression::subexpressions[1]. They are now
stored as a regular list in ast_expression::expressions.
Eric Anholt [Wed, 5 May 2010 18:45:30 +0000 (11:45 -0700)]
Move optimization pass prototypes to a single header.
Eric Anholt [Wed, 5 May 2010 17:37:25 +0000 (10:37 -0700)]
ir_dead_code_local: Remove redundant assignments within basic blocks.
This cleans up a bunch of junk code in some of the GLSL parser tests,
and could potentially help real-world too (particularly after copy
propagation has happened).
Eric Anholt [Wed, 5 May 2010 18:07:21 +0000 (11:07 -0700)]
ir_visit_tree: Make sure we visit dereference targets, too.
Found this with the local dead code pass, which never saw variable
dereferences occurring.
Eric Anholt [Wed, 5 May 2010 16:38:09 +0000 (09:38 -0700)]
ir_copy_propagation: Fix up the doxygen about the file.
Eric Anholt [Wed, 5 May 2010 16:31:53 +0000 (09:31 -0700)]
ir_copy_propagation: Return true if we optimized out any assignments.
This may trigger other optimization phases to make more progress themselves.
Eric Anholt [Wed, 5 May 2010 16:26:46 +0000 (09:26 -0700)]
ir_copy_propagation: Handle swizzles and array derefs on LHS of assign.
This improves the ACP to not get cleared when more complicated assignments
occur, cleaning up more redundant copies in programs.
Eric Anholt [Tue, 4 May 2010 20:04:40 +0000 (13:04 -0700)]
ir_copy_propagation: New pass to rewrite dereferences to avoid copies.
This is pretty basic. Right now it only handles pure assignments --
same type on each side, no swizzling, and only within basic blocks.
Eric Anholt [Tue, 4 May 2010 00:08:01 +0000 (17:08 -0700)]
ir_to_mesa.cpp: Fix missing types on some ir_swizzles.
Debugging this took forever as I only looked at constructors in ir.cpp
to find who wasn't setting up ->type. I dislike hiding code (as
opposed to prototypes and definitions) in C++ header files, but in
this case I have only myself to blame.
Eric Anholt [Mon, 3 May 2010 18:41:00 +0000 (11:41 -0700)]
Remove the pedantic C junk.
Mesa doesn't use pedantic ANSI C89, so I have no idea why we would.
Eric Anholt [Fri, 30 Apr 2010 01:00:33 +0000 (18:00 -0700)]
Store warnings and errors in a parser state infolog.
Cleans up compile warning about unused state in _mesa_glsl_warning. We
would want infolog handling roughly like this anyway.
Eric Anholt [Fri, 30 Apr 2010 00:57:28 +0000 (17:57 -0700)]
Quiet warnings about ir_shader not being handled in places it's not needed.
Eric Anholt [Fri, 30 Apr 2010 00:54:52 +0000 (17:54 -0700)]
Use the AM_SILENT_RULES when available.
Eric Anholt [Fri, 30 Apr 2010 00:53:26 +0000 (17:53 -0700)]
glsl_lexer: Quiet warning about unused unput().
Ian Romanick [Mon, 3 May 2010 17:15:02 +0000 (10:15 -0700)]
Merge branch 'fixes'
Kenneth Graunke [Thu, 29 Apr 2010 06:18:25 +0000 (23:18 -0700)]
Use %p rather than %08x when printing pointers to fix compile.
Kenneth Graunke [Sat, 1 May 2010 07:31:35 +0000 (00:31 -0700)]
Fix ir_return cloning to actually use the cloned subexpression.
This caused a nasty bug where the function inliner would create new
variables for each of the formal parameters, but the body would still
reference the old copies.
This was highly visible since the dead code eliminator (rightly) removed
the new declarations, leading to printed IR that referenced non-existent
variable names.
Kenneth Graunke [Sat, 1 May 2010 06:38:50 +0000 (23:38 -0700)]
Fix incorrect comments in function inliner.
Eric Anholt [Thu, 29 Apr 2010 20:54:19 +0000 (13:54 -0700)]
ir_dead_code: Fix segfault on handling a return statement with no value.
Ian Romanick [Wed, 28 Apr 2010 20:14:53 +0000 (13:14 -0700)]
Track and print user defined structure types
Ian Romanick [Wed, 28 Apr 2010 20:04:15 +0000 (13:04 -0700)]
IR print visitor: Move logic for printing the whole program to _mesa_print_ir
Ian Romanick [Wed, 28 Apr 2010 19:01:40 +0000 (12:01 -0700)]
IR print visitor: Just print the name of structures
Treat structure types like other non-array types. We'll have to print
the structure defintion elsewhere.
Eric Anholt [Mon, 26 Apr 2010 22:02:40 +0000 (15:02 -0700)]
ir_function_inlining: Handle inlining of structure dereferences.
Eric Anholt [Mon, 26 Apr 2010 22:01:50 +0000 (15:01 -0700)]
Correctly handle remapping of array dereferences if ->var is a variable.