tree b4474ffba397cbd05bd448e22595c44024a305ca
parent 6e7e0ddf1dda35570cd9ed23751d9306f7353d7f
author Vladimir Marko <vmarko@google.com> 1506612744 +0100
committer Vladimir Marko <vmarko@google.com> 1506680691 +0100

ART: Generalize "x >> (s & m)" simplification.

Extend the simplification from
  x SHIFT (s & mask), mask contains relevant bits
where SHIFT is Shl/Shr/UShr/Ror to
  x SHIFT (s | mask), mask does not contain relevant bits,
  x SHIFT (s ^ mask), mask does not contain relevant bits,
  x SHIFT (s + mask), mask does not contain relevant bits,
  x SHIFT (s - mask), mask does not contain relevant bits,
  x SHIFT ((type)s), s is non-64-bit integral type.

The simplification for the TypeConversion case is motivated
by the work to introduce Uint8 and convert '& 0xff' to
TypeConversion. This CL makes sure the old simplifications
shall still work. The simplifications for Or/Xor/Add/Sub
are added just because we can.

Test: Add tests to 548-checker-instruct-simplification.
Test: m test-art-host-gtest
Test: testrunner.py --host --optimizing
Bug: 23964345
Change-Id: I71210cb63496fae607a457a627f115260669c2c9