Dynamic analysis #1
Conversation
| DeducedConstraints deduceConstraints(int argIndex); | ||
|
|
||
| private: | ||
| llvm::SmallVector<llvm::SmallVector<mlir::Value>> expandToDNF(mlir::Value constraint); |
There was a problem hiding this comment.
i think that DNF will be something that will come up a lot, you can wrap the vectors into a class like and put this in a header of their own
class DisjunctiveNormalFormExpression {
public:
<getters>
private:
llvm::SmallVector<llvm::SmallVector<mlir::Value>> Content;
};
to be honest i would not be even opposed to have a operation called DNFExpression and have this analysis as a pass that mutates the ir and turns every precondition into a DNF
| } | ||
|
|
||
|
|
||
| void helper( |
There was a problem hiding this comment.
rename with a meaningful name, furthemore this function can be made static
| */ | ||
| llvm::SmallVector<llvm::SmallVector<mlir::Value>> DynamicArgumentAnalysis::expandToDNF(mlir::Value constraint) { | ||
| llvm::SmallVector<llvm::SmallVector<mlir::Value>> conjunctions; | ||
| helper(conjunctions, {constraint}, precondition); |
There was a problem hiding this comment.
why do you need the helper out of line here?
| } | ||
|
|
||
| //TODO | ||
| const int64_t min_int = -800; |
There was a problem hiding this comment.
i guess you were looking for std::numeric_limits<int64_t>::min()
| return KNOWN_VALUE; | ||
| } | ||
|
|
||
| mlir::Value DynamicArgumentAnalysis::compute(mlir::Value expression) { |
There was a problem hiding this comment.
can you provide a more informative name? compute is not very specific
|
the part of the code that emits the code that computes min and max is complex, can't it be somehow simplified by placing as much code as possible in the fuzzer standard library and calling it? |
…dy bound struct fields.
cemcebeci
force-pushed
the
dynamic-analysis
branch
from
521821f to
ffb82ec
Compare
rl-language
force-pushed
the
master
branch
2 times, most recently
from
9eb7c9a to
73de6fa
Compare
No description provided.