SPIRV-Tools/source/opt
Steven Perron 6c7db9c630
Handle nested breaks from switches. (#2624)
* Handle nested breaks from switches.

There was a recent decision made to allow branches to the merge node of
a switch even if the switch is not the first enclosing construct.  They
can be generated by glslang from break statements in switches.

Dead branch elimination seems to be the only optimization that will
break because of this change, so I will update that optimizations.

The change made are:

- Track switches in structured cfg analysis.
- In Dead branch elimination:
  - Look for nested breaks that will require a switch instruction.
  - Rewrite, but don't delete, switchs that are required even if it
    could be replaced by an unconditional branch.
  - When looking for the first break, consider the merge of a switch
    as well.

See #2612.

* Fix variable names and comments.

* Add tests for the struct cfg analysis and switches.

* Fix typos in comments.
2019-05-27 16:28:14 -04:00
..
aggressive_dead_code_elim_pass.cpp ADCE support for SPIR-V 1.4 entry points (#2561) 2019-05-07 14:52:22 -04:00
aggressive_dead_code_elim_pass.h Make the constant and type manager analyses. (#2250) 2018-12-20 18:00:05 +00:00
basic_block.cpp Change implementation of post order CFG traversal (#2543) 2019-04-29 17:09:20 -04:00
basic_block.h Change implementation of post order CFG traversal (#2543) 2019-04-29 17:09:20 -04:00
block_merge_pass.cpp Extracted block merging functionality into its own utility file (#2325) 2019-01-25 10:57:13 +00:00
block_merge_pass.h Extracted block merging functionality into its own utility file (#2325) 2019-01-25 10:57:13 +00:00
block_merge_util.cpp Don't merge unreachable blocks (#2375) 2019-02-12 09:24:01 -05:00
block_merge_util.h Extracted block merging functionality into its own utility file (#2325) 2019-01-25 10:57:13 +00:00
build_module.cpp Move make_unique to source/util. (#1836) 2018-08-14 12:44:54 -04:00
build_module.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
ccp_pass.cpp Removing decorations when doing constant propagation. (#2444) 2019-03-13 10:40:49 -04:00
ccp_pass.h Make the constant and type manager analyses. (#2250) 2018-12-20 18:00:05 +00:00
cfg_cleanup_pass.cpp Don't inline recursive functions. (#2130) 2018-11-29 14:24:58 -05:00
cfg_cleanup_pass.h Make the constant and type manager analyses. (#2250) 2018-12-20 18:00:05 +00:00
cfg.cpp Change implementation of post order CFG traversal (#2543) 2019-04-29 17:09:20 -04:00
cfg.h Document in the context what happens with id overflow. (#2159) 2018-12-06 09:07:00 -05:00
CMakeLists.txt Add pass to fix some invalid unreachable blocks for WebGPU (#2563) 2019-05-09 12:56:10 -04:00
code_sink.cpp Handle instructions not in blocks in code sinking. (#2308) 2019-01-21 12:09:56 -05:00
code_sink.h Sink (#2284) 2019-01-17 15:56:36 -05:00
combine_access_chains.cpp Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
combine_access_chains.h Make the constant and type manager analyses. (#2250) 2018-12-20 18:00:05 +00:00
common_uniform_elim_pass.cpp Handle variable pointer in some optimizations (#2490) 2019-04-03 12:47:51 -04:00
common_uniform_elim_pass.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
compact_ids_pass.cpp Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
compact_ids_pass.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
composite.cpp Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
composite.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
const_folding_rules.cpp Add folding rules for OpQuantizeToF16 (#2614) 2019-05-21 23:15:01 -07:00
const_folding_rules.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
constants.cpp Using the instruction folder to fold OpSpecConstantOp (#2598) 2019-05-21 12:45:00 -04:00
constants.h Using the instruction folder to fold OpSpecConstantOp (#2598) 2019-05-21 12:45:00 -04:00
copy_prop_arrays.cpp Fix up type mismatches. (#2545) 2019-05-02 09:31:46 -04:00
copy_prop_arrays.h Fix up type mismatches. (#2545) 2019-05-02 09:31:46 -04:00
dead_branch_elim_pass.cpp Handle nested breaks from switches. (#2624) 2019-05-27 16:28:14 -04:00
dead_branch_elim_pass.h Handle nested breaks from switches. (#2624) 2019-05-27 16:28:14 -04:00
dead_insert_elim_pass.cpp Don't inline recursive functions. (#2130) 2018-11-29 14:24:58 -05:00
dead_insert_elim_pass.h Make the constant and type manager analyses. (#2250) 2018-12-20 18:00:05 +00:00
dead_variable_elimination.cpp Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
dead_variable_elimination.h Make the constant and type manager analyses. (#2250) 2018-12-20 18:00:05 +00:00
decompose_initialized_variables_pass.cpp Add support for Private & Output to initializer decompose flag (#2537) 2019-04-25 16:24:32 -04:00
decompose_initialized_variables_pass.h Implement WebGPU->Vulkan initializer conversion for 'Function' variables (#2513) 2019-04-16 14:31:36 -04:00
decoration_manager.cpp Invalidate the decoration manager at the start of ADCE. (#2189) 2018-12-10 13:24:33 -05:00
decoration_manager.h Invalidate the decoration manager at the start of ADCE. (#2189) 2018-12-10 13:24:33 -05:00
def_use_manager.cpp Add base and core bindless validation instrumentation classes (#2014) 2018-11-08 13:54:54 -05:00
def_use_manager.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
dominator_analysis.cpp Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
dominator_analysis.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
dominator_tree.cpp Fix readbility/braces warnings (#1804) 2018-08-07 09:09:47 -04:00
dominator_tree.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
eliminate_dead_constant_pass.cpp Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
eliminate_dead_constant_pass.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
eliminate_dead_functions_pass.cpp Fixes #2358. Added to the reducer the ability to remove a function t… (#2361) 2019-02-08 16:20:29 +00:00
eliminate_dead_functions_pass.h Make the constant and type manager analyses. (#2250) 2018-12-20 18:00:05 +00:00
eliminate_dead_functions_util.cpp Fixes #2358. Added to the reducer the ability to remove a function t… (#2361) 2019-02-08 16:20:29 +00:00
eliminate_dead_functions_util.h Fixes #2358. Added to the reducer the ability to remove a function t… (#2361) 2019-02-08 16:20:29 +00:00
eliminate_dead_members_pass.cpp Don't change type of input and output var in dead member elim (#2412) 2019-02-20 18:59:41 -05:00
eliminate_dead_members_pass.h Don't change type of input and output var in dead member elim (#2412) 2019-02-20 18:59:41 -05:00
feature_manager.cpp Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
feature_manager.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
fix_storage_class.cpp Fix up type mismatches. (#2545) 2019-05-02 09:31:46 -04:00
fix_storage_class.h Fix up type mismatches. (#2545) 2019-05-02 09:31:46 -04:00
flatten_decoration_pass.cpp Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
flatten_decoration_pass.h Add testing framework for tools. 2018-08-17 15:03:14 -04:00
fold_spec_constant_op_and_composite_pass.cpp Add folding rules for OpQuantizeToF16 (#2614) 2019-05-21 23:15:01 -07:00
fold_spec_constant_op_and_composite_pass.h Using the instruction folder to fold OpSpecConstantOp (#2598) 2019-05-21 12:45:00 -04:00
fold.cpp Fix overflow when negating INT_MIN. (#2293) 2019-01-17 17:01:55 -05:00
fold.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
folding_rules.cpp Remove duplicates from list of interface IDs in OpEntryPoint instruction (#2449) 2019-03-13 15:46:31 -04:00
folding_rules.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
freeze_spec_constant_value_pass.cpp Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
freeze_spec_constant_value_pass.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
function.cpp Add pass to fix some invalid unreachable blocks for WebGPU (#2563) 2019-05-09 12:56:10 -04:00
function.h Add pass to fix some invalid unreachable blocks for WebGPU (#2563) 2019-05-09 12:56:10 -04:00
generate_webgpu_initializers_pass.cpp Implement WebGPU->Vulkan initializer conversion for 'Function' variables (#2513) 2019-04-16 14:31:36 -04:00
generate_webgpu_initializers_pass.h Implement WebGPU->Vulkan initializer conversion for 'Function' variables (#2513) 2019-04-16 14:31:36 -04:00
if_conversion.cpp Fix restrictions in if conversion (#1998) 2018-10-19 15:16:46 -04:00
if_conversion.h Make the constant and type manager analyses. (#2250) 2018-12-20 18:00:05 +00:00
inline_exhaustive_pass.cpp Handle id overflow in inlining. (#2196) 2018-12-18 19:34:03 +00:00
inline_exhaustive_pass.h Handle id overflow in inlining. (#2196) 2018-12-18 19:34:03 +00:00
inline_opaque_pass.cpp Handle id overflow in inlining. (#2196) 2018-12-18 19:34:03 +00:00
inline_opaque_pass.h Handle id overflow in inlining. (#2196) 2018-12-18 19:34:03 +00:00
inline_pass.cpp Handle id overflow in inlining. (#2196) 2018-12-18 19:34:03 +00:00
inline_pass.h Handle id overflow in inlining. (#2196) 2018-12-18 19:34:03 +00:00
inst_bindless_check_pass.cpp Bindless Validation: Instrument descriptor-based loads and stores (#2583) 2019-05-15 19:43:23 -04:00
inst_bindless_check_pass.h Bindless Validation: Instrument descriptor-based loads and stores (#2583) 2019-05-15 19:43:23 -04:00
instruction_list.cpp Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
instruction_list.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
instruction.cpp Do not fold floating point if float controls used (#2569) 2019-05-10 11:03:22 -04:00
instruction.h New spirv-reduce reduction pass: operand to dominating id. (#2099) 2018-11-26 17:06:21 -05:00
instrument_pass.cpp Update instrumentation passes to handle 1.4 interfaces (#2573) 2019-05-10 11:08:28 -04:00
instrument_pass.h Bindless Validation: Descriptor Initialization Check (#2419) 2019-03-19 09:53:43 -04:00
ir_builder.h Upgrade modf and frexp (#2266) 2019-01-07 12:36:38 -05:00
ir_context.cpp Fix Constants Analyses bug inserted by #2302 (#2306) 2019-01-21 12:34:12 -05:00
ir_context.h Handle id overflow in sroa better. (#2582) 2019-05-15 09:29:28 -04:00
ir_loader.cpp spirv-opt: Behave a bit better in the face of unknown instructions (#2487) 2019-04-05 13:36:42 -04:00
ir_loader.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
iterator.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
legalize_vector_shuffle_pass.cpp Implement WebGPU->Vulkan initializer conversion for 'Function' variables (#2513) 2019-04-16 14:31:36 -04:00
legalize_vector_shuffle_pass.h Implement WebGPU->Vulkan initializer conversion for 'Function' variables (#2513) 2019-04-16 14:31:36 -04:00
licm_pass.cpp In LICM don't place code between merge instruction and branch. (#2252) 2018-12-20 18:33:52 -05:00
licm_pass.h In LICM don't place code between merge instruction and branch. (#2252) 2018-12-20 18:33:52 -05:00
local_access_chain_convert_pass.cpp Handle variable pointer in some optimizations (#2490) 2019-04-03 12:47:51 -04:00
local_access_chain_convert_pass.h Make the constant and type manager analyses. (#2250) 2018-12-20 18:00:05 +00:00
local_redundancy_elimination.cpp Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
local_redundancy_elimination.h Make the constant and type manager analyses. (#2250) 2018-12-20 18:00:05 +00:00
local_single_block_elim_pass.cpp Handle variable pointer in some optimizations (#2490) 2019-04-03 12:47:51 -04:00
local_single_block_elim_pass.h Make the constant and type manager analyses. (#2250) 2018-12-20 18:00:05 +00:00
local_single_store_elim_pass.cpp Handle variable pointer in some optimizations (#2490) 2019-04-03 12:47:51 -04:00
local_single_store_elim_pass.h Make the constant and type manager analyses. (#2250) 2018-12-20 18:00:05 +00:00
local_ssa_elim_pass.cpp Handle variable pointer in some optimizations (#2490) 2019-04-03 12:47:51 -04:00
local_ssa_elim_pass.h Make the constant and type manager analyses. (#2250) 2018-12-20 18:00:05 +00:00
log.h Rewrite include guards (#1793) 2018-08-03 08:05:33 -04:00
loop_dependence_helpers.cpp Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
loop_dependence.cpp Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
loop_dependence.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
loop_descriptor.cpp Fix memory leak in unrolling. (#2301) 2019-01-17 16:02:43 -05:00
loop_descriptor.h Fix memory leak in unrolling. (#2301) 2019-01-17 16:02:43 -05:00
loop_fission.cpp Document in the context what happens with id overflow. (#2159) 2018-12-06 09:07:00 -05:00
loop_fission.h Add testing framework for tools. 2018-08-17 15:03:14 -04:00
loop_fusion_pass.cpp Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
loop_fusion_pass.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
loop_fusion.cpp Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
loop_fusion.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
loop_peeling.cpp Document in the context what happens with id overflow. (#2159) 2018-12-06 09:07:00 -05:00
loop_peeling.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
loop_unroller.cpp Fix memory leak in unrolling. (#2301) 2019-01-17 16:02:43 -05:00
loop_unroller.h Make the constant and type manager analyses. (#2250) 2018-12-20 18:00:05 +00:00
loop_unswitch_pass.cpp Don't fold specialized branches in loop unswitch (#2245) 2018-12-19 04:40:30 +00:00
loop_unswitch_pass.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
loop_utils.cpp Document in the context what happens with id overflow. (#2159) 2018-12-06 09:07:00 -05:00
loop_utils.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
mem_pass.cpp Consider atomics that load when analyzing live stores in ADCE (#1956) (#1958) 2018-10-12 08:46:35 -04:00
mem_pass.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
merge_return_pass.cpp Fix merge return in the face of breaks (#2466) 2019-04-02 10:05:28 -04:00
merge_return_pass.h Fix merge return in the face of breaks (#2466) 2019-04-02 10:05:28 -04:00
module.cpp Add option for the max id bound. (#1870) 2018-09-10 11:49:41 -04:00
module.h Add option for the max id bound. (#1870) 2018-09-10 11:49:41 -04:00
null_pass.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
optimizer.cpp Add in individual flags for Vulkan <-> WebGPU passes (#2615) 2019-05-22 10:06:53 -07:00
pass_manager.cpp Add option to validate after each pass (#2462) 2019-03-26 14:38:59 -04:00
pass_manager.h Add option to validate after each pass (#2462) 2019-03-26 14:38:59 -04:00
pass.cpp Fix clang-tidy warning about definition/declaration mismatch. (#2571) 2019-05-09 00:15:08 -04:00
pass.h Fix up type mismatches. (#2545) 2019-05-02 09:31:46 -04:00
passes.h Add pass to fix some invalid unreachable blocks for WebGPU (#2563) 2019-05-09 12:56:10 -04:00
pch_source_opt.cpp Enable precompiled headers for spirv-tools(-shared) and some unit tests (#2026) 2018-11-06 09:26:23 -05:00
pch_source_opt.h Enable precompiled headers for spirv-tools(-shared) and some unit tests (#2026) 2018-11-06 09:26:23 -05:00
private_to_local_pass.cpp Update priv-to-local for SPIR-V 1.4 (#2567) 2019-05-08 12:38:49 -04:00
private_to_local_pass.h Make the constant and type manager analyses. (#2250) 2018-12-20 18:00:05 +00:00
process_lines_pass.cpp Add passes to propagate and eliminate redundant line instructions (#2027). (#2039) 2018-11-15 14:06:17 -05:00
process_lines_pass.h Make the constant and type manager analyses. (#2250) 2018-12-20 18:00:05 +00:00
propagator.cpp Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
propagator.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
reduce_load_size.cpp Have replace load size handle extact with no index. (#2261) 2019-01-03 13:02:10 -05:00
reduce_load_size.h Make the constant and type manager analyses. (#2250) 2018-12-20 18:00:05 +00:00
redundancy_elimination.cpp Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
redundancy_elimination.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
reflect.h Add validation for SPV_NV_cooperative_matrix (#2404) 2019-02-25 17:43:11 -05:00
register_pressure.cpp Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
register_pressure.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
remove_duplicates_pass.cpp Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
remove_duplicates_pass.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
replace_invalid_opc.cpp Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
replace_invalid_opc.h Add testing framework for tools. 2018-08-17 15:03:14 -04:00
scalar_analysis_nodes.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
scalar_analysis_simplification.cpp Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
scalar_analysis.cpp Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
scalar_analysis.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
scalar_replacement_pass.cpp Handle id overflow in sroa better. (#2582) 2019-05-15 09:29:28 -04:00
scalar_replacement_pass.h Handle id overflow in sroa better. (#2582) 2019-05-15 09:29:28 -04:00
set_spec_constant_default_value_pass.cpp Move make_unique to source/util. (#1836) 2018-08-14 12:44:54 -04:00
set_spec_constant_default_value_pass.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
simplification_pass.cpp Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
simplification_pass.h Make the constant and type manager analyses. (#2250) 2018-12-20 18:00:05 +00:00
split_invalid_unreachable_pass.cpp Add pass to fix some invalid unreachable blocks for WebGPU (#2563) 2019-05-09 12:56:10 -04:00
split_invalid_unreachable_pass.h Add pass to fix some invalid unreachable blocks for WebGPU (#2563) 2019-05-09 12:56:10 -04:00
ssa_rewrite_pass.cpp SSA-rewriter: make sure phi entries are unique. (#2206) 2018-12-18 18:14:27 +00:00
ssa_rewrite_pass.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
strength_reduction_pass.cpp Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
strength_reduction_pass.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
strip_atomic_counter_memory_pass.cpp Implement WebGPU->Vulkan initializer conversion for 'Function' variables (#2513) 2019-04-16 14:31:36 -04:00
strip_atomic_counter_memory_pass.h Implement WebGPU->Vulkan initializer conversion for 'Function' variables (#2513) 2019-04-16 14:31:36 -04:00
strip_debug_info_pass.cpp Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
strip_debug_info_pass.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
strip_reflect_info_pass.cpp Remove MemberDecorateStringGOOGLE during stript-refect. (#2021) 2018-10-30 16:17:35 -04:00
strip_reflect_info_pass.h Make the constant and type manager analyses. (#2250) 2018-12-20 18:00:05 +00:00
struct_cfg_analysis.cpp Handle nested breaks from switches. (#2624) 2019-05-27 16:28:14 -04:00
struct_cfg_analysis.h Handle nested breaks from switches. (#2624) 2019-05-27 16:28:14 -04:00
tree_iterator.h Rewrite include guards (#1793) 2018-08-03 08:05:33 -04:00
type_manager.cpp Added additional changes for the new AccelerationStructureNV type. (#2218) 2018-12-19 21:42:39 +00:00
type_manager.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
types.cpp Recognize OpTypeAccelerationStructureNV as a type instruction (#2190) 2018-12-11 19:03:55 -05:00
types.h Added additional changes for the new AccelerationStructureNV type. (#2218) 2018-12-19 21:42:39 +00:00
unify_const_pass.cpp Move make_unique to source/util. (#1836) 2018-08-14 12:44:54 -04:00
unify_const_pass.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
upgrade_memory_model.cpp Mem model spv 1.4 (#2565) 2019-05-15 19:06:37 -04:00
upgrade_memory_model.h Mem model spv 1.4 (#2565) 2019-05-15 19:06:37 -04:00
value_number_table.cpp Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
value_number_table.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
vector_dce.cpp Handle extract with no index in VDCE. (#2305) 2019-01-18 15:43:36 -05:00
vector_dce.h Handle extract with no index in VDCE. (#2305) 2019-01-18 15:43:36 -05:00
workaround1209.cpp Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00
workaround1209.h Cleanup includes. (#1795) 2018-08-03 15:06:09 -04:00