Bug 184176 – bmalloc should do partial scavenges more frequently

• Component: bmalloc
• Status: RESOLVED FIXED

[Saam Barati]

...

[Saam Barati]

Created attachment 337055 [details]
WIP

[Saam Barati]

Created attachment 337141 [details]
WIP

I'm playing around with the heuristic of when to drive the scavenger. I'm tuning against JetStream on a MBP. It's tricky to do anything that isn't a performance regression but a memory progression since ToT essentially never runs the scavenger in ToT.

[Saam Barati]

Created attachment 337220 [details]
WIP

Added some preliminary support for calling madvise while not holding the global heap lock.

[Saam Barati]

Created attachment 337252 [details]
WIP

Trying out more stuff. Giving Fil's idea of keeping track of high water mark and scavenging anything above that point a go.

[Saam Barati]

(In reply to Saam Barati from comment #4)
> Created attachment 337252 [details]
> WIP
> 
> Trying out more stuff. Giving Fil's idea of keeping track of high water mark
> and scavenging anything above that point a go.

High water mark is the way to go.

[Saam Barati]

Created attachment 337256 [details]
WIP

[EWS Watchlist]

Attachment 337256 [details] did not pass style-queue:


ERROR: Source/bmalloc/bmalloc/Scavenger.cpp:39:  Bad include order. Mixing system and custom headers.  [build/include_order] [4]
ERROR: Source/bmalloc/bmalloc/Heap.cpp:40:  Alphabetical sorting problem.  [build/include_order] [4]
ERROR: Source/bmalloc/bmalloc/Heap.cpp:624:  More than one command on the same line  [whitespace/newline] [4]
ERROR: Source/bmalloc/bmalloc/Scavenger.h:34:  Alphabetical sorting problem.  [build/include_order] [4]
Total errors found: 4 in 17 files


If any of these errors are false positives, please file a bug against check-webkit-style.

[EWS Watchlist]

Comment on attachment 337256 [details]
WIP

Attachment 337256 [details] did not pass jsc-ews (mac):
Output: http://webkit-queues.webkit.org/results/7212687

New failing tests:
jsc-layout-tests.yaml/js/script-tests/stringimpl-to-jsstring-on-large-strings-2.js.layout-no-llint
jsc-layout-tests.yaml/js/script-tests/stringimpl-to-jsstring-on-large-strings-2.js.layout-dfg-eager-no-cjit
jsc-layout-tests.yaml/js/script-tests/stringimpl-to-jsstring-on-large-strings-2.js.layout-ftl-eager-no-cjit
jsc-layout-tests.yaml/js/script-tests/stringimpl-to-jsstring-on-large-strings-2.js.layout-ftl-no-cjit
jsc-layout-tests.yaml/js/script-tests/stringimpl-to-jsstring-on-large-strings-2.js.layout

[Saam Barati]

Created attachment 337310 [details]
patch

[Saam Barati]

Created attachment 337312 [details]
patch

[EWS Watchlist]

Attachment 337312 [details] did not pass style-queue:


ERROR: Source/bmalloc/bmalloc/Heap.cpp:41:  Alphabetical sorting problem.  [build/include_order] [4]
ERROR: Source/bmalloc/bmalloc/Heap.cpp:570:  More than one command on the same line  [whitespace/newline] [4]
ERROR: Source/bmalloc/bmalloc/Scavenger.cpp:40:  Bad include order. Mixing system and custom headers.  [build/include_order] [4]
ERROR: Source/bmalloc/bmalloc/Scavenger.h:34:  Alphabetical sorting problem.  [build/include_order] [4]
Total errors found: 4 in 21 files


If any of these errors are false positives, please file a bug against check-webkit-style.

[Saam Barati]

Created attachment 337594 [details]
patch

This patch addresses some feedback Fil gave me in person. IsoHeapImpl now tracks its directory high water mark and scavenges that directory and any directory with a larger index than it.

[EWS Watchlist]

Attachment 337594 [details] did not pass style-queue:


ERROR: Source/bmalloc/bmalloc/Heap.cpp:570:  More than one command on the same line  [whitespace/newline] [4]
Total errors found: 1 in 21 files


If any of these errors are false positives, please file a bug against check-webkit-style.

[EWS Watchlist]

Comment on attachment 337594 [details]
patch

Attachment 337594 [details] did not pass mac-ews (mac):
Output: http://webkit-queues.webkit.org/results/7264889

New failing tests:
imported/w3c/web-platform-tests/workers/name-property.html

[EWS Watchlist]

Created attachment 337596 [details]
Archive of layout-test-results from ews100 for mac-sierra

The attached test failures were seen while running run-webkit-tests on the mac-ews.
Bot: ews100  Port: mac-sierra  Platform: Mac OS X 10.12.6

[Saam Barati]

(In reply to Build Bot from comment #14)
> Comment on attachment 337594 [details]
> patch
> 
> Attachment 337594 [details] did not pass mac-ews (mac):
> Output: http://webkit-queues.webkit.org/results/7264889
> 
> New failing tests:
> imported/w3c/web-platform-tests/workers/name-property.html

This seems like a flaky or broken test.

[Saam Barati]

Created attachment 337633 [details]
patch for landing

[EWS Watchlist]

Attachment 337633 [details] did not pass style-queue:


ERROR: Source/bmalloc/bmalloc/Heap.cpp:570:  More than one command on the same line  [whitespace/newline] [4]
Total errors found: 1 in 21 files


If any of these errors are false positives, please file a bug against check-webkit-style.

[Saam Barati]

Created attachment 337634 [details]
patch for landing

[EWS Watchlist]

Attachment 337634 [details] did not pass style-queue:


ERROR: Source/bmalloc/bmalloc/Heap.cpp:570:  More than one command on the same line  [whitespace/newline] [4]
Total errors found: 1 in 21 files


If any of these errors are false positives, please file a bug against check-webkit-style.

[WebKit Commit Bot]

Comment on attachment 337634 [details]
patch for landing

Clearing flags on attachment: 337634

Committed r230501: <https://trac.webkit.org/changeset/230501>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.

[Radar WebKit Bug Importer]

<rdar://problem/39331070>