Optimize division by / modulo of gPageSize

When running under the page-agnostic configuration, while the global
constant gPageSize is always set to a power-of-two value at static
initialization time, it may not be recognised as a guaranteed
power-of-two by the compiler for the purpose of optimizations. This
means that divisions by gPageSize may not be replaced by an efficient
right-shift, and applications of modulo gPageSize may not be optimized
via a bitwise-AND.

This patch introduces two functions: one to perform a gPageSize division
as a logical right-shift, and the other to compute modulo gPageSize via
bitwise-AND.

Inlining these optimized implementations is then done everywhere that
requires division or modulo of gPageSize. As they are inlined, the
compiler is able to optimize the reuse of registers when multiple
divisions or modulos are required (for example, the result of
WhichPowerOf2 on the page size is stored when subsequent divisions are
necessary, which then each become just a single right-shift).

The tests were run for legacy 4K, page size agnostic 4K and 16K.

Test: art/tools/run-gtests.sh
Test: art/test/testrunner/testrunner.py --target --64
Test: art/tools/run-libcore-tests.sh --mode=device --variant=X64
Test: art/tools/run-libjdwp-tests.sh --mode=device --variant=X64

Change-Id: I01bd5fa89b88806c84660d4a2d62ddcba061678c

0 files changed, 0 insertions, 0 deletions