OSDN Git Service

ART: Rework ARM64 entry sequence
authorAndreas Gampe <agampe@google.com>
Tue, 29 Jul 2014 07:35:00 +0000 (00:35 -0700)
committerAndreas Gampe <agampe@google.com>
Mon, 4 Aug 2014 16:30:07 +0000 (09:30 -0700)
commitf29ecd69af2743a425314baa4abd6c44d8d88649
tree4b54ebb2c2bc7509b8081d524d858bf315f07e68
parent00d1ef41d0d0d438d604d102183a6bfac58643e7
ART: Rework ARM64 entry sequence

Try to fold one sub of SP in the ARM64 entry sequence. When the
framesize is small, generate a sub over the full frame-size, and
adjust the spill offsets accordingly. If the framesize is too
large, use a pre-indexed store and fill upwards from there.

Change-Id: I1c15ac6276fb62b8164372de02fd92437f605938
compiler/dex/quick/arm64/arm64_lir.h
compiler/dex/quick/arm64/assemble_arm64.cc
compiler/dex/quick/arm64/call_arm64.cc
compiler/dex/quick/arm64/codegen_arm64.h
compiler/dex/quick/arm64/int_arm64.cc