[PATCH] D43835: Simplify more cases of logical ops of masked icmps.

[David Li via Phabricator via llvm-commits]

davidxl added inline comments.


================
Comment at: lib/Transforms/InstCombine/InstCombineAndOrXor.cpp:499
+  // For example,
+  // (icmp ne (A & 12), 0) & (icmp eq (A & 7), 1) -> (icmp eq (A & 15), 9)
+  // (icmp ne (A & 15), 0) & (icmp eq (A & 7), 0) -> (icmp eq (A & 15), 8)
----------------
This check can probably be extended:

1) B's bitset subtracts the intersection of B and D contains only one bit
2) B and D's intersection have zero values

In other words, D does not need to be power of 2


================
Comment at: lib/Transforms/InstCombine/InstCombineAndOrXor.cpp:530
+  // (icmp ne (A & 15), 0) & (icmp eq (A & 3), 0) -> no folding.
+  if (ECst->isZero()) {
+    if ((BCst->getValue() & DCst->getValue()) == BCst->getValue())
----------------
Can you introduce a small lambda here to be reused here and other places?

auto IsSubset = [](Value *V1, Value *V2) {
   ....
};

if (IsSubset(BCst, DCst))  ....


================
Comment at: lib/Transforms/InstCombine/InstCombineAndOrXor.cpp:536
+
+  // At this point, B, D, E aren't zero and (B & D) == B, (B & D) == D or B ==
+  // D. If B is a superset of (or equal to) D, since E is not zero, LHS is
----------------
Can you split the comment here for case in line 548 and line 552?


Repository:
  rL LLVM

https://reviews.llvm.org/D43835