MayPurgeStep takes the lock a second time after purging an arena.
Categories
(Core :: Memory Allocator, enhancement)
Tracking
()
| Tracking | Status | |
|---|---|---|
| firefox138 | --- | fixed |
People
(Reporter: pbone, Assigned: pbone)
References
Details
Attachments
(8 files, 2 obsolete files)
|
48 bytes,
text/x-phabricator-request
|
Details | Review | |
|
48 bytes,
text/x-phabricator-request
|
Details | Review | |
|
48 bytes,
text/x-phabricator-request
|
Details | Review | |
|
48 bytes,
text/x-phabricator-request
|
Details | Review | |
|
48 bytes,
text/x-phabricator-request
|
Details | Review | |
|
48 bytes,
text/x-phabricator-request
|
Details | Review | |
|
48 bytes,
text/x-phabricator-request
|
Details | Review | |
|
48 bytes,
text/x-phabricator-request
|
Details | Review |
moz_may_purge_one_now takes the lock again to "peek" to determine if there are more purge requests queued. It's better to return true if there "may be" more requests and let the caller call Purge() again if it wants to.
| Assignee | ||
Updated•1 year ago
|
| Assignee | ||
Comment 1•1 year ago
|
||
Updated•1 year ago
|
| Assignee | ||
Comment 2•1 year ago
|
||
Processing purge requests won't fail if there are none, so there's no point
peeking before attempting to process any.
| Assignee | ||
Comment 3•1 year ago
|
||
| Assignee | ||
Updated•1 year ago
|
| Assignee | ||
Comment 4•1 year ago
|
||
Updated•1 year ago
|
| Assignee | ||
Comment 5•1 year ago
|
||
| Assignee | ||
Comment 6•1 year ago
|
||
Updated•1 year ago
|
Updated•1 year ago
|
Updated•1 year ago
|
| Assignee | ||
Comment 7•1 year ago
|
||
This avoids releasing the lock only to acquire it a few lines later.
Updated•1 year ago
|
Updated•1 year ago
|
Updated•1 year ago
|
Updated•1 year ago
|
Updated•1 year ago
|
| Assignee | ||
Comment 9•1 year ago
|
||
Updated•1 year ago
|
Comment 10•1 year ago
|
||
Comment 11•1 year ago
|
||
Backed out for causing SM bustages.
Backout link: https://hg.mozilla.org/integration/autoland/rev/2375604263792c9a7a48244f49828e1741eaed2a
Failure log: https://treeherder.mozilla.org/logviewer?job_id=497793976&repo=autoland&lineNumber=5748
| Assignee | ||
Comment 12•1 year ago
|
||
The dirty page functions only make sense with MOZ_MEMORY, without it they're
no-ops. A later patch will move them to a more appropriate section of
malloc_decls.h which will make the undefined without MOZ_MEMORY. Therefore
this patch ifdef's out code that uses them if MOZ_MEMORY isn't defined.
The patch also makes these file-local functions static.
| Assignee | ||
Comment 13•1 year ago
|
||
Another patch will move moz_set_max_dirty_page_modifier() into a different
section of malloc_decls.h where it will be unavailable without defining
MOZ_MEMORY. This patch places #ifdefs around some of its uses.
Updated•1 year ago
|
Updated•1 year ago
|
Comment 14•1 year ago
|
||
| Assignee | ||
Updated•1 year ago
|
Comment 15•1 year ago
|
||
| bugherder | ||
https://hg.mozilla.org/mozilla-central/rev/aaf41eb3d828
https://hg.mozilla.org/mozilla-central/rev/f320b975ef05
https://hg.mozilla.org/mozilla-central/rev/791361c24764
https://hg.mozilla.org/mozilla-central/rev/b2dec082e21a
https://hg.mozilla.org/mozilla-central/rev/31e229ff65b0
https://hg.mozilla.org/mozilla-central/rev/dd640a024c06
https://hg.mozilla.org/mozilla-central/rev/de6152e95629
https://hg.mozilla.org/mozilla-central/rev/e35dca5c1c1e
Description
•