- 23 Jun, 2017 1 commit
-
-
Ulan Degenbaev authored
This prepares ground for switching mark-compactor to use Worklist data-structure instead of the existing marking deque. BUG=chromium:694255 Change-Id: I0ac4c563018a9619962fb4bf388b5f3cceffb86d Reviewed-on: https://chromium-review.googlesource.com/544933Reviewed-by:
Michael Lippautz <mlippautz@chromium.org> Commit-Queue: Ulan Degenbaev <ulan@chromium.org> Cr-Commit-Position: refs/heads/master@{#46178}
-
- 23 Dec, 2016 2 commits
-
-
mlippautz authored
1) Alternate between processing v8 and wrappers 2) Once v8 is empty, try 3 rounds of finding the fixpoint between v8 and wrappers 3) After that, finalize once v8 marking deque is empty again Reland fixed: Toggle needs to be IncrementalMarking global as we need to properly alternate tracing v8 and wrappers. BUG=chromium:468240, chromium:668164 Review-Url: https://codereview.chromium.org/2599283002 Cr-Commit-Position: refs/heads/master@{#41940}
-
mlippautz authored
Revert of Reland "[heap] Ensure progress when incrementally marking wrappers" (patchset #8 id:140001 of https://codereview.chromium.org/2591383004/ ) Reason for revert: Breaks webkit-unit-tests. Investigating.. Original issue's description: > Reland "[heap] Ensure progress when incrementally marking wrappers" > > 1) Alternate between processing v8 and wrappers > 2) Once v8 is empty, try 3 rounds of finding the fixpoint between v8 and wrappers > 3) After that, finalize once v8 marking deque is empty again > > BUG= > > Review-Url: https://codereview.chromium.org/2591383004 > Cr-Commit-Position: refs/heads/master@{#41932} > Committed: https://chromium.googlesource.com/v8/v8/+/61a55548c50e01d84ed4aefa396324cbb4039b51 TBR=hpayer@chromium.org,ulan@chromium.org # Skipping CQ checks because original CL landed less than 1 days ago. NOPRESUBMIT=true NOTREECHECKS=true NOTRY=true BUG= Review-Url: https://codereview.chromium.org/2592393003 Cr-Commit-Position: refs/heads/master@{#41936}
-
- 22 Dec, 2016 3 commits
-
-
mlippautz authored
1) Alternate between processing v8 and wrappers 2) Once v8 is empty, try 3 rounds of finding the fixpoint between v8 and wrappers 3) After that, finalize once v8 marking deque is empty again BUG= Review-Url: https://codereview.chromium.org/2591383004 Cr-Commit-Position: refs/heads/master@{#41932}
-
mlippautz authored
Revert of [heap] Ensure progress when incrementally marking wrappers (patchset #3 id:60001 of https://codereview.chromium.org/2592403002/ ) Reason for revert: This won't work because the finalization still checks whether both marking deques are empty, also calling into blink. So we never proceed there. Original issue's description: > [heap] Ensure progress when incrementally marking wrappers > > The problem here is estimating the marking step size for wrapper tracing. If the > steps are too small, we cannot keep up with the mutator creating new wrappers. > The result is an endless stream of incremental marking steps, alternating v8 and > wrappers tracing, without ever finalizing in a GC. > > The mitigation here is to abort finding the fix point after 10 incremental > iterations. > > A proper solution would track newly created wrappers on the blink side during > wrapper tracing. Will give this more thought after the holidays. > > BUG=chromium:668164, chromium:468240 > > Review-Url: https://codereview.chromium.org/2592403002 > Cr-Commit-Position: refs/heads/master@{#41923} > Committed: https://chromium.googlesource.com/v8/v8/+/a47417b89373c615f9256800cfc803d84ba58378 TBR=ulan@chromium.org # Skipping CQ checks because original CL landed less than 1 days ago. NOPRESUBMIT=true NOTREECHECKS=true NOTRY=true BUG=chromium:668164, chromium:468240 Review-Url: https://codereview.chromium.org/2602433002 Cr-Commit-Position: refs/heads/master@{#41924}
-
mlippautz authored
The problem here is estimating the marking step size for wrapper tracing. If the steps are too small, we cannot keep up with the mutator creating new wrappers. The result is an endless stream of incremental marking steps, alternating v8 and wrappers tracing, without ever finalizing in a GC. The mitigation here is to abort finding the fix point after 10 incremental iterations. A proper solution would track newly created wrappers on the blink side during wrapper tracing. Will give this more thought after the holidays. BUG=chromium:668164, chromium:468240 Review-Url: https://codereview.chromium.org/2592403002 Cr-Commit-Position: refs/heads/master@{#41923}
-
- 20 Dec, 2016 1 commit
-
-
mlippautz authored
BUG=chromium:468240 Review-Url: https://codereview.chromium.org/2576453002 Cr-Commit-Position: refs/heads/master@{#41837}
-