WebKit Bugzilla
New
Browse
Log In
×
Sign in with GitHub
or
Remember my login
Create Account
·
Forgot Password
Forgotten password account recovery
RESOLVED FIXED
189156
[JSC] Clean up StructureStubClearingWatchpoint
https://bugs.webkit.org/show_bug.cgi?id=189156
Summary
[JSC] Clean up StructureStubClearingWatchpoint
Yusuke Suzuki
Reported
2018-08-30 09:02:32 PDT
[JSC] Clean up StructureStubClearingWatchpoint
Attachments
Patch
(5.13 KB, patch)
2018-08-30 09:03 PDT
,
Yusuke Suzuki
no flags
Details
Formatted Diff
Diff
Patch
(5.06 KB, patch)
2018-08-31 23:20 PDT
,
Yusuke Suzuki
saam
: review+
Details
Formatted Diff
Diff
Show Obsolete
(1)
View All
Add attachment
proposed patch, testcase, etc.
Yusuke Suzuki
Comment 1
2018-08-30 09:03:52 PDT
Created
attachment 348502
[details]
Patch
Saam Barati
Comment 2
2018-08-30 13:23:57 PDT
Comment on
attachment 348502
[details]
Patch View in context:
https://bugs.webkit.org/attachment.cgi?id=348502&action=review
> Source/JavaScriptCore/bytecode/StructureStubClearingWatchpoint.h:-96 > - std::unique_ptr<StructureStubClearingWatchpoint> m_head;
Was there a reason for the old code?
Yusuke Suzuki
Comment 3
2018-08-31 02:21:29 PDT
Comment on
attachment 348502
[details]
Patch View in context:
https://bugs.webkit.org/attachment.cgi?id=348502&action=review
>> Source/JavaScriptCore/bytecode/StructureStubClearingWatchpoint.h:-96 >> - std::unique_ptr<StructureStubClearingWatchpoint> m_head; > > Was there a reason for the old code?
I can't see any reason. Vector<StructureStubClearingWatchpoint> is bad since each StructureStubClearingWatchpoint address can be changed. But Vector<std::unique_ptr<StructureStubClearingWatchpoint>> is enough.
Saam Barati
Comment 4
2018-08-31 11:44:48 PDT
Comment on
attachment 348502
[details]
Patch Maybe the linked list uses less memory on average for common programs?
Yusuke Suzuki
Comment 5
2018-08-31 23:16:23 PDT
(In reply to Saam Barati from
comment #4
)
> Comment on
attachment 348502
[details]
> Patch > > Maybe the linked list uses less memory on average for common programs?
In the Vector<> case, we can drop a link in StructureStubClearingWatchpoint instead of allocating storage for Vector. The memory consumption is so-so, larger in Vector case due to unused memory allocation for storages, or larger in linked-list case due to the representation of the enlarged objects in bmalloc (it depends on whether the given object size would fit to the bmalloc's segregated heap). We can use Bag<> here, it does not change memory usage. Changed.
Yusuke Suzuki
Comment 6
2018-08-31 23:19:50 PDT
(In reply to Yusuke Suzuki from
comment #5
)
> (In reply to Saam Barati from
comment #4
) > > Comment on
attachment 348502
[details]
> > Patch > > > > Maybe the linked list uses less memory on average for common programs? > > In the Vector<> case, we can drop a link in StructureStubClearingWatchpoint > instead of allocating storage for Vector. > The memory consumption is so-so, larger in Vector case due to unused memory > allocation for storages, or larger in linked-list case due to the > representation of the enlarged objects in bmalloc (it depends on whether the > given object size would fit to the bmalloc's segregated heap). > We can use Bag<> here, it does not change memory usage. Changed.
If allocations of this watchpoint in this vector/bag is so many, SegmentedVector<> can improve memory usage more. In the meantime, I'll use bag.
Yusuke Suzuki
Comment 7
2018-08-31 23:20:12 PDT
Created
attachment 348708
[details]
Patch
Yusuke Suzuki
Comment 8
2018-09-06 12:00:47 PDT
Ping?
Saam Barati
Comment 9
2018-09-06 12:50:21 PDT
Comment on
attachment 348708
[details]
Patch r=me
Yusuke Suzuki
Comment 10
2018-09-06 21:23:55 PDT
(In reply to Saam Barati from
comment #9
)
> Comment on
attachment 348708
[details]
> Patch > > r=me
Thanks!
Yusuke Suzuki
Comment 11
2018-09-06 21:24:24 PDT
Committed
r235776
: <
https://trac.webkit.org/changeset/235776
>
Radar WebKit Bug Importer
Comment 12
2018-09-06 21:25:32 PDT
<
rdar://problem/44211170
>
Note
You need to
log in
before you can comment on or make changes to this bug.
Top of Page
Format For Printing
XML
Clone This Bug