[JSC] WeakMapImpl do not need to take cellLock in visitOutputConstraint and main thread
Created attachment 375026 [details] Patch
Created attachment 453059 [details] Patch
Created attachment 453060 [details] Patch
Comment on attachment 453060 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=453060&action=review r=me > Source/JavaScriptCore/ChangeLog:8 > + WeakMapImpl::visitOutputConstraints is called in the constraint solver, and it does not modify the buffer of WeakMapImpl. Can you remove this part: "and it does not modify the buffer of WeakMapImpl"? It is not relevant, correct? The only relevant detail is what you stated below i.e. the constraint solver runs and rehash can never be run concurrently. Also, if concurrency were an issue, m_buffer isn't the only thing that needs to be protected: m_capacity does too. So, it seems strange to draw attention to the buffer here.
Comment on attachment 453060 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=453060&action=review >> Source/JavaScriptCore/ChangeLog:8 >> + WeakMapImpl::visitOutputConstraints is called in the constraint solver, and it does not modify the buffer of WeakMapImpl. > > Can you remove this part: "and it does not modify the buffer of WeakMapImpl"? It is not relevant, correct? The only relevant detail is what you stated below i.e. the constraint solver runs and rehash can never be run concurrently. > > Also, if concurrency were an issue, m_buffer isn't the only thing that needs to be protected: m_capacity does too. So, it seems strange to draw attention to the buffer here. OK, dropped.
Committed r290416 (247724@trunk): <https://commits.webkit.org/247724@trunk>
<rdar://problem/89405434>