summaryrefslogtreecommitdiff
path: root/src/passes/OptimizeInstructions.cpp
Commit message (Collapse)AuthorAgeFilesLines
...
* remove unneeded masks using getMaxBitsAlon Zakai (kripken)2017-02-161-3/+16
|
* tiny refactoring in OptimizeInstructions, for clarityAlon Zakai (kripken)2017-02-161-7/+8
|
* fix fuzz testcase, xor maxBits is the max, not the minAlon Zakai (kripken)2017-02-161-2/+2
|
* handle fallthrough values in load_s/u and sign/zero-extend optimizationAlon Zakai (kripken)2017-02-161-6/+30
|
* handle load in getMaxBitsAlon Zakai (kripken)2017-02-161-0/+2
|
* handle tee_local in getMaxBitsAlon Zakai (kripken)2017-02-161-0/+3
|
* optimize out a sign-ext into a store of the same sizeAlon Zakai (kripken)2017-02-161-0/+6
|
* optimize out add/sub of 0Alon Zakai (kripken)2017-02-161-1/+9
|
* optimize sign-extends to a booleanAlon Zakai (kripken)2017-02-161-0/+4
|
* optimize sign-extends to eqzAlon Zakai (kripken)2017-02-161-0/+7
|
* optimize sign-extends to neAlon Zakai (kripken)2017-02-161-2/+2
|
* Optimize "squared" operations (#905)Alon Zakai2017-02-161-23/+86
| | | | * optimize 'almost' sign extends: when we can remove one entirely, then extra shifts can be left behind. with that in place, we can then optimize 'squared' operations like shl on shl, as doing so does not break our sign extend opts
* optimize linear sums (#904)Alon Zakai2017-02-161-0/+110
|
* Optimize sign-extends (#902)Alon Zakai2017-02-161-18/+139
| | | | | | * optimize sign-extend output * optimize sign-extend input
* Hide dead code (#909)Loo Rong Jie2017-02-151-0/+8
|
* Improve handling of implicit traps (#898)Alon Zakai2017-02-061-4/+4
| | | | | | | | * add --ignore-implicit-traps option, and by default do not ignore them, to properly preserve semantics * implicit traps can be reordered, but are side effects and should not be removed * add testing for --ignore-implicit-traps
* Make ast_utils into a library (#892)Derek Schuff2017-01-311-45/+47
| | | | Split ExpressionAnalyzer and ExpressionManipulator into cpp files, and turn their giant template functions into simple functions which take a callback. More organization, fewer mammoth headers, makes the build a few seconds faster, and the binaries a couple MB smaller.
* Fix regression from #850 (#851)Alon Zakai2016-11-301-27/+1
| | | | | | * fix regression from #850 - it is not always safe to fold added offsets into load/store offsets, as the add wraps but offset does not * small refactoring to simplify asm2wasm pass invocation
* Optimize added factors into load/store offsets (#850)Alon Zakai2016-11-291-0/+40
| | | | | * optimize added factors into load/store offsets, removing the add * optimize offset of load/store of a constant into the constant
* optimize booleans through i32.orAlon Zakai2016-11-041-2/+6
|
* recurse in optimizeBoolean (#809)Alon Zakai2016-10-271-0/+10
|
* optimize ne of 0 in a boolean context (#808)Alon Zakai2016-10-271-7/+17
|
* Conditionalize boolean operations based on cost (#805)Alon Zakai2016-10-261-0/+46
| | | When we have expensive | expensive, and both are boolean, then we can execute one of them conditionally if it doesn't have side effects.
* Optimize out bool & 1 (#804)Alon Zakai2016-10-251-0/+10
| | | | | | * asm.js corrections to unit.asm.js test * optimize out bool&1
* Optimize to i64.store[less] (#792)Alon Zakai2016-10-201-4/+10
|
* optimize out i32.and of -1Alon Zakai2016-10-151-4/+9
|
* optimize loads and stores that mask their values, as the load/store ↵Alon Zakai2016-10-151-0/+28
| | | | operation masks itself
* Type check block/loop/if sigs (#717)Alon Zakai2016-09-281-0/+60
| | | | | | * type check using block/loop/if types provided in text and binary formats. * print if and loop sigs which were missing. * remove dsl from OptimizeInstructions as after those changes it needs rethinking.
* refactor wasm.h to remove numericIndex hacks, and move indexing to the parsersAlon Zakai2016-09-211-9/+14
|
* clean up database-ensuring code using the new prepareToRun methodAlon Zakai2016-09-121-9/+6
|
* refactor an optimizeBoolean methodAlon Zakai2016-09-101-20/+24
|
* optimize eqz^2 in select and br_ifAlon Zakai2016-09-101-8/+23
|
* flip select-eqz when possibleAlon Zakai2016-09-071-2/+13
|
* optimize out a set of a get of a globalAlon Zakai2016-09-071-0/+8
|
* optimize loads+shifts into signed loadsAlon Zakai2016-09-071-0/+27
|
* ensure we create the OptimizeInstructions database on demand, avoiding ↵Alon Zakai2016-09-071-3/+17
| | | | global ctors
* rewrite OptimizeInstructions to use a dsl of patternsAlon Zakai2016-06-261-50/+136
|
* Add initialization functions for passes to avoid missing pass registration ↵Jukka Jylänki2016-06-211-1/+3
| | | | due to linker dead code elimination. Fixes #577.
* move function parallelism to pass and pass runner, which allows more ↵Alon Zakai2016-06-031-1/+3
| | | | efficient parallel execution (#564)
* use separate internal opcodes for binary variantsAlon Zakai2016-05-181-10/+32
|
* use separate internal opcodes for unary variantsAlon Zakai2016-05-181-2/+2
|
* create a UnifiedExpressionVisitor for passes that want a single visitor ↵Alon Zakai2016-04-181-1/+1
| | | | function, to avoid confusion with having both visit* and visitExpression in a single pass (#357)
* Function parallelism (#343)Alon Zakai2016-04-151-0/+2
| | | | * allow traversals to mark themselves as function-parallel, in which case we run them using a thread pool. also mark some thread-safety risks (interned strings, arena allocators) with assertions they modify only on the main thread
* dyn_cast => dynCastAlon Zakai2016-04-111-2/+2
|
* De-recurse traversals (#333)Alon Zakai2016-04-111-1/+1
| | | | | | | | | | | | * refactor core walking to not recurse * add a simplify-locals test * reuse parent's non-branchey scan logic in SimpleExecutionWalker, reduce code duplication * update wasm.js * rename things following comments
* fold eqz+comparisonsAlon Zakai2016-04-091-0/+22
|
* optimize if-eqzAlon Zakai2016-04-061-0/+43