webkit's fastmalloc is built on top of tc-malloc and all allocations call to fastmalloc must fallback on to tc-malloc's impl of malloc including tryFastMalloc
I think u must have missed this line #define malloc fastMalloc Try doing a -E (preprocess) instead of -c on the FastMalloc.cpp On Wed, Mar 28, 2012 at 5:20 AM, Fabrice Robinet <cmg...@motorola.com>wrote: > Hi group, > > In short: I wonder where tryFastMalloc should be used, and what's *really* > implemented in it (or should be). > > Longer story: > According to the .cpp, it is supposed to use a cache, which would justify > the "fast" in the function name. > > From FastMalloc.cpp > > // A malloc that uses a per-thread cache to satisfy small malloc requests. > > // (The time for malloc/free of a small object drops from 300 ns to 50 ns.) > > But looking at the current source tryFastMalloc looks to just wrap > malloc. (I just updated Chromium and thus WebCore too...), > > I have seen tryFastMalloc used in WebCore code that is not too old (like > WebGL), > so I believe it is not deprecated and *should* be used, so this raises a > couple of questions: > > Is current state of tryFastMalloc temporary in the sense that a cache will > be used eventually ? > Did I miss something about this cache in the implementation ? > Are there (updated) guidelines about the usage of tryFastMalloc ? > > Thanks, > Fabrice. > > _______________________________________________ > webkit-help mailing list > webkit-help@lists.webkit.org > http://lists.webkit.org/mailman/listinfo.cgi/webkit-help > > -- Sriram Neelakandan Author - Embedded Linux System Design And Development ( http://tinyurl.com/2doosu)
_______________________________________________ webkit-help mailing list webkit-help@lists.webkit.org http://lists.webkit.org/mailman/listinfo.cgi/webkit-help
