The Static Analysis tool Coverity added that member variable canWait_ is not initialized in the default constructor of the class: >>js::FutexRuntime::FutexRuntime() >> : cond_(nullptr), >> state_(Idle) >>{ CID 1347691 (#1 of 1): Uninitialized scalar field (UNINIT_CTOR) 2. uninit_member: Non-static class member canWait_ is not initialized in this constructor nor in any functions that it calls. >>} The above code it's a copy paste from Coverity Error log. Looking though code it's could be the case the canWait_ is read through: >>bool canWait() in 1. >>js::FutexRuntime::wait(JSContext* cx, double timeout_ms, AtomicsObject::FutexWaitResult* result) >>{ >> MOZ_ASSERT(&cx->runtime()->fx == this); >> MOZ_ASSERT(cx->runtime()->fx.canWait()); >> MOZ_ASSERT(lockHolder_ == PR_GetCurrentThread()); 2. >> if (!rt->fx.canWait()) >> return ReportCannotWait(cx); >> >> // This lock also protects the "waiters" field on SharedArrayRawBuffer, >> // and it provides the necessary memory fence. Now the setter of the variable is only called through: >>JS_SetFutexCanWait(JSRuntime* rt) >>{ >> rt->fx.setCanWait(true); >>} and it get's caleld from: >> JSRuntime* rt = Runtime(); >> MOZ_ASSERT(rt); >> >> JS_SetRuntimePrivate(rt, new WorkerThreadRuntimePrivate(aWorkerPrivate)); >> >> js::SetPreserveWrapperCallback(rt, PreserveWrapper); >> JS_InitDestroyPrincipalsCallback(rt, DestroyWorkerPrincipals); >> JS_SetWrapObjectCallbacks(rt, &WrapObjectCallbacks); >> if (aWorkerPrivate->IsDedicatedWorker()) { >> JS_SetFutexCanWait(rt); Now i can't figure for sure that setter gets called before the geter for canWait_, but surely can hurt to make it false in the constructor specially when the comment for this variable is: >> // If canWait() returns false (the default) then futexWait is disabled >> // on the runtime to which the FutexRuntime belongs.

Review commit: https://reviewboard.mozilla.org/r/30519/diff/#index_header See other reviews: https://reviewboard.mozilla.org/r/30519/

Comment on attachment 8706860 [details] MozReview Request: Bug 1238911 - initialize canWait with false from constructor, avoid using the variable without initialization. r=lhansen https://reviewboard.mozilla.org/r/30519/#review27243

Comment on attachment 8706860 [details] MozReview Request: Bug 1238911 - initialize canWait with false from constructor, avoid using the variable without initialization. r=lhansen Stealing review, it's my fault :)

Comment on attachment 8706860 [details] MozReview Request: Bug 1238911 - initialize canWait with false from constructor, avoid using the variable without initialization. r=lhansen Review request updated; see interdiff: https://reviewboard.mozilla.org/r/30519/diff/1-2/

https://reviewboard.mozilla.org/r/30519/#review27243 I've reupdated the patch since i've changed it's header in order to contain you as a reviewer. Hope this is ok.

Comment on attachment 8706860 [details] MozReview Request: Bug 1238911 - initialize canWait with false from constructor, avoid using the variable without initialization. r=lhansen https://reviewboard.mozilla.org/r/30519/#review27247 Bug 1238911 - initialize canWait with false from constructor, avoid using the variable without initialization. r=lhansen

Comment on attachment 8706860 [details] MozReview Request: Bug 1238911 - initialize canWait with false from constructor, avoid using the variable without initialization. r=lhansen https://reviewboard.mozilla.org/r/30519/#review35483 I don't know how to get the review request removed from bugzilla except to review this again. Oh, ReviewBoard. :-|