http://build.webkit.org/LeaksViewer/?url=http%3A%2F%2Fbuild.webkit.org%2F%2Fresults%2FSnowLeopard%20Intel%20Leaks%2Fr98879%20(19849)%2F malloc_zone_malloc > malloc > fastMalloc > RenderRegion::setRenderBoxRegionInfo We're leaking in this function. Looks like we're allocating a new RenderBoxRegionInfo, storing it in a HashMap, and never deleting it.
<rdar://problem/10374417>
*** Bug 71260 has been marked as a duplicate of this bug. ***
I have taken a look at both methods RenderRegion::setRenderBoxRegionInfo and RenderFlowThread::setRegionRangeForBox. Using our current tests, i was not able to find a place where we leak in these functions and their associated data. What i have found is that in JavaScriptCore/Parser.cpp, Parser class destructor, we are not freeing the Lexer object that was allocated in the Parser ctor. I will make a patch with that fix unless there is a reason for doing that after https://bugs.webkit.org/show_bug.cgi?id=71138.
Leaks are reported if one loads fast/regions/overflow-in-uniform-regions-dynamic.html.
In RenderFlowThread::logicalWidthChangedInRegions, oldInfo is deleted only inside the condition: if (!newInfo || newInfo->logicalWidth() != oldInfo->logicalWidth()) { .... } Otherwise, it is left undeleted and leaked.
The other leak is in function RenderRegion::removeRenderBoxRegionInfo, when the RenderBoxRegionInfo is removed from the map but not deleted.
Darin, I can make a patch for this one and another for 71260 if this is ok with you.
Sure, if you want to handle this, that’s fine.
Created attachment 113846 [details] Patch
Comment on attachment 113846 [details] Patch Clearing flags on attachment: 113846 Committed r99467: <http://trac.webkit.org/changeset/99467>
All reviewed patches have been landed. Closing bug.