Bug 28676 – Safari 4 not releasing memory back to the operating system

• Status: ASSIGNED
• Component: New Bugs

[Jake Archibald]

Once Safari has claimed memory from the operating system, it doesn't seem to give it back, even though it's no longer using it.

Running the first test on the following page http://www.jakearchibald.co.uk/jsperformance/closurememory/1.html, this creates a 1mb string every second.

http://www.flickr.com/photos/jaffathecake/3851167461/sizes/o/

Chrome (and other browsers) release the memory back to the operating system when the test is ended, whereas Safari keeps it, even if the tab is closed. Safari will reuse the memory it has reserved, but it doesn't give it back to the operating system, even if it has claimed 700mb+.

[Jake Archibald]

This is an issue in 526 (not tested the nightly builds) but the option wasn't available. Not sure which component this should be part of either.

[Mark Rowe (bdash)]

Please test in a nightly build.  The memory allocator used by Safari on Windows has been improved so that memory is returned to the OS at a more aggressive rate.

[Jake Archibald]

http://farm3.static.flickr.com/2554/3853034847_806c897794_o.png

The above is the same test in a nightly build downloaded today. Although it starts releasing memory, it's not at a useful rate.

In this test (and previous tests) one pixel on the graph == 1 second.

[Mark Rowe (bdash)]

<rdar://problem/7165917>

[Geoffrey Garen]

*** Bug 22275 has been marked as a duplicate of this bug. ***

[Geoffrey Garen]

Created attachment 48950 [details]
patch

[Eric Seidel (no email)]

Attachment 48950 [details] did not build on mac:
Build output: http://webkit-commit-queue.appspot.com/results/278140

[WebKit Review Bot]

Attachment 48950 [details] did not pass style-queue:

Failed to run "WebKitTools/Scripts/check-webkit-style" exit_code: 1
JavaScriptCore/wtf/FastMalloc.cpp:1501:  TCMalloc_PageHeap::scavenge is incorrectly named. Don't use underscores in your identifier names.  [readability/naming] [4]
Total errors found: 1


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

[WebKit Review Bot]

Attachment 48950 [details] did not build on chromium:
Build output: http://webkit-commit-queue.appspot.com/results/279134

[WebKit Review Bot]

Attachment 48950 [details] did not build on qt:
Build output: http://webkit-commit-queue.appspot.com/results/279154

[WebKit Review Bot]

Attachment 48950 [details] did not build on gtk:
Build output: http://webkit-commit-queue.appspot.com/results/280012

[Oliver Hunt]

Comment on attachment 48950 [details]
patch

you have removed two #endifs but only one #if this would break stuff

[Geoffrey Garen]

Created attachment 48978 [details]
patch

Now with more buildy goodness.

[WebKit Review Bot]

Attachment 48978 [details] did not pass style-queue:

Failed to run "WebKitTools/Scripts/check-webkit-style" exit_code: 1
JavaScriptCore/wtf/FastMalloc.cpp:1503:  TCMalloc_PageHeap::scavenge is incorrectly named. Don't use underscores in your identifier names.  [readability/naming] [4]
Total errors found: 1 in 2 files


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

[Oliver Hunt]

Comment on attachment 48978 [details]
patch

r=me

[Geoffrey Garen]

Fixed that edge case: Committed revision 54988.

Still analyzing other causes.

Clearing review flag.

[Geoffrey Garen]

Created attachment 50579 [details]
patch

Fixed a bug where the old code would release only one item from each size class per scavenge, potentially leaving large numbers of large-sized objects unreleased for a long time.

[WebKit Review Bot]

Attachment 50579 [details] did not pass style-queue:

Failed to run "WebKitTools/Scripts/check-webkit-style" exit_code: 1
JavaScriptCore/wtf/FastMalloc.cpp:1385:  Weird number of spaces at line-start.  Are you using a 4-space indent?  [whitespace/indent] [3]
JavaScriptCore/wtf/FastMalloc.cpp:1386:  Weird number of spaces at line-start.  Are you using a 4-space indent?  [whitespace/indent] [3]
JavaScriptCore/wtf/FastMalloc.cpp:1387:  Weird number of spaces at line-start.  Are you using a 4-space indent?  [whitespace/indent] [3]
JavaScriptCore/wtf/FastMalloc.cpp:1387:  min_free_committed_pages_since_last_scavenge_ is incorrectly named. Don't use underscores in your identifier names.  [readability/naming] [4]
JavaScriptCore/wtf/FastMalloc.cpp:1432:  Weird number of spaces at line-start.  Are you using a 4-space indent?  [whitespace/indent] [3]
JavaScriptCore/wtf/FastMalloc.cpp:1438:  Weird number of spaces at line-start.  Are you using a 4-space indent?  [whitespace/indent] [3]
JavaScriptCore/wtf/FastMalloc.cpp:1464:  Weird number of spaces at line-start.  Are you using a 4-space indent?  [whitespace/indent] [3]
JavaScriptCore/wtf/FastMalloc.cpp:1507:  Weird number of spaces at line-start.  Are you using a 4-space indent?  [whitespace/indent] [3]
JavaScriptCore/wtf/FastMalloc.cpp:1517:  Weird number of spaces at line-start.  Are you using a 4-space indent?  [whitespace/indent] [3]
JavaScriptCore/wtf/FastMalloc.cpp:1518:  Weird number of spaces at line-start.  Are you using a 4-space indent?  [whitespace/indent] [3]
JavaScriptCore/wtf/FastMalloc.cpp:1525:  Weird number of spaces at line-start.  Are you using a 4-space indent?  [whitespace/indent] [3]
JavaScriptCore/wtf/FastMalloc.cpp:2401:  Weird number of spaces at line-start.  Are you using a 4-space indent?  [whitespace/indent] [3]
Total errors found: 12 in 2 files


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

[WebKit Review Bot]

Attachment 50579 [details] did not build on gtk:
Build output: http://webkit-commit-queue.appspot.com/results/631033

[WebKit Review Bot]

Attachment 50579 [details] did not build on qt:
Build output: http://webkit-commit-queue.appspot.com/results/664010

[Geoffrey Garen]

Committed revision 56028.

Still analyzing other causes.

Clearing review flag.