STR (mostly copied from bug 1120115): 1. Do a local debug build of m-c 2. mach run 3. Load https://code.google.com/p/ozz-animation/ in a tab 4. Open a new window (cmd+n) and load about:memory 5. Hit "measure" on about:memory Expected: Get an about:memory report Actual: Crash. Stack attached

This crash happens on Linux as well.

It's an assertion failure, not a crash. Things are fine in an opt build. Here's the relevant parts of a better stack trace: #5 <signal handler called> #6 mozilla::WebGLBindable<StrongGLenum<BufferBindingDetails> >::Target ( this=0x7f24832d8a98) at ../../../dom/canvas/WebGLBindableName.h:37 #7 0x00007f24a315d6eb in mozilla::WebGLMemoryTracker::GetBufferCacheMemoryUsed () at ../../../dom/canvas/WebGLContextReporter.cpp:129 #8 0x00007f24a315caf6 in mozilla::WebGLMemoryTracker::CollectReports ( this=0x7f248c5341c0, handleReport=0x7f248198f760, data=0x7f2481ad56e0) at ../../../dom/canvas/WebGLContextReporter.cpp:46 Looks like an easy fix -- we should just check if HasEverBeenBound() is true before calling Target().

jgilbert: 1 week review ping. This is a trivial patch.

Comment on attachment 8547816 [details] [diff] [review] Fix an assertion failure in the WebGL memory reporter Review of attachment 8547816 [details] [diff] [review]: ----------------------------------------------------------------- This seems reasonable to me. If Jeff want's to check it he can do it post check-in.