Summary: | Heap::allocate should search a word at a time | ||||||||
---|---|---|---|---|---|---|---|---|---|
Product: | WebKit | Reporter: | Nathan Lawrence <nlawrence> | ||||||
Component: | JavaScriptCore | Assignee: | Nathan Lawrence <nlawrence> | ||||||
Status: | RESOLVED FIXED | ||||||||
Severity: | Normal | CC: | commit-queue, ggaren, nlawrence, slewis | ||||||
Priority: | P2 | ||||||||
Version: | 528+ (Nightly build) | ||||||||
Hardware: | All | ||||||||
OS: | All | ||||||||
Attachments: |
|
Description
Nathan Lawrence
2010-05-26 12:59:00 PDT
Created attachment 57126 [details]
suggested patch
Comment on attachment 57126 [details]
suggested patch
+ if (!~block->marked.bits[m_heap.nextCell >> 5])
+ m_heap.nextCell = (m_heap.nextCell & (~0x1F)) + 32;
I think this code might be wrong on 64bit.
Also, would be nice to encapsulate these bitops in the CollectorBitmap interface.
Let's talk about this in person.
Created attachment 57132 [details]
Moved the advance logic from Heap::allocate to CollectorBitmap::advanceToNextPossibleFreeCell.
Hey, I uploaded the patch to the correct bug this time!
Comment on attachment 57132 [details]
Moved the advance logic from Heap::allocate to CollectorBitmap::advanceToNextPossibleFreeCell.
r=me
Comment on attachment 57132 [details]
Moved the advance logic from Heap::allocate to CollectorBitmap::advanceToNextPossibleFreeCell.
Let's have auto-Eric take a crack at committing this.
Comment on attachment 57132 [details] Moved the advance logic from Heap::allocate to CollectorBitmap::advanceToNextPossibleFreeCell. Clearing flags on attachment: 57132 Committed r60323: <http://trac.webkit.org/changeset/60323> All reviewed patches have been landed. Closing bug. |