commit ec8f7322826fb233062cb15cbb6df81141a3d12c
author Omer Katz <omerkatz@chromium.org> Fri Jul 12 09:03:42 2019
committer Commit Bot <commit-bot@chromium.org> Fri Jul 12 09:03:42 2019
tree 1b8bb28fb20884c661636488e935a4bfe10cf4ed
parent 8386aa4561bbb10d3fa4ec2236b809ab923bb90a

heap: Optimize PreFinalizer traversal

This CL optimizes pre-finalizer invocation.

Previously, we would maintain pre-finalizers in a LinkedHashSet and invoke as follows:
1. Create a reversed copy of pre-finalizers in a Vector
2. Iterate the vector
2.a Invoke pre-finalizers on dead objects
2.b Remove dead objects from HashSet

Instead, we now maintain pre-finalizers in a Deque and invoke as follows:
1. Iterate the deque backwards
2. Store non-pre-finalized (live) objects in a new Deque
3. Move the new Deque over.

This improves performance on selected microbenchmarks by at least 25% (and up to 65%).

Bug: 981043
Change-Id: I0f3e3bbbc9c75ca017b11c6d60f991b1a68b44be
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/1698465
Reviewed-by: Michael Lippautz <mlippautz@chromium.org>
Reviewed-by: Kentaro Hara <haraken@chromium.org>
Commit-Queue: Omer Katz <omerkatz@chromium.org>
Cr-Commit-Position: refs/heads/master@{#676801}