If this is b2g only then it's probably because we don't register secondary threads on b2g to avoid allocating a pseduostack to lower memory.

(In reply to Benoit Girard (:BenWa) from comment #1) > If this is b2g only then it's probably because we don't register secondary > threads on b2g to avoid allocating a pseduostack to lower memory. Thanks; that makes sense. This patch makes us assert having a pseudostack on non-b2g platforms only.

Comment on attachment 8343240 [details] [diff] [review] Don't assert mPseudoStack on B2G (v1) Review of attachment 8343240 [details] [diff] [review]: ----------------------------------------------------------------- ::: xpcom/threads/ThreadStackHelper.cpp @@ +155,5 @@ > + /* Normally, provided the profiler is enabled, it would be an error if we > + don't have a pseudostack here (the thread probably forgot to call > + profiler_register_thread). However, on B2G, profiling secondary threads > + may be disabled despite profiler being enabled. This is a limitation of > + the profiler and not an error. */ Not a limitation of profiler. It's an explicit decision to save memory.

(In reply to Benoit Girard (:BenWa) from comment #3) > Comment on attachment 8343240 [details] [diff] [review] > Don't assert mPseudoStack on B2G (v1) > > Review of attachment 8343240 [details] [diff] [review]: > ----------------------------------------------------------------- > > ::: xpcom/threads/ThreadStackHelper.cpp > @@ +155,5 @@ > > + /* Normally, provided the profiler is enabled, it would be an error if we > > + don't have a pseudostack here (the thread probably forgot to call > > + profiler_register_thread). However, on B2G, profiling secondary threads > > + may be disabled despite profiler being enabled. This is a limitation of > > + the profiler and not an error. */ > > Not a limitation of profiler. It's an explicit decision to save memory. Okay. Changed "limitation" to "by-design"