Follow up after bug 874437 sticks, get rid of boolean flags that have piled up in MessageChannel.

+ MOZ_ASSERT(aFlags & REQUIRE_URGENT_SCRIPT_BLOCKING ? NS_IsMainThread() : true); + mFlags |= aFlags; Can any flag changes happen while the channel is in operation? Because if any changes happen after initial channel setup, then this is a giant read/write footgun. mFlags isn't an atomic, and |= operations can't be atomic anyway.

Comment on attachment 8474228 [details] [diff] [review] patch v.1 Review of attachment 8474228 [details] [diff] [review]: ----------------------------------------------------------------- ::: ipc/glue/MessageChannel.cpp @@ +1709,5 @@ > PostErrorNotifyTask(); > } > > void > +MessageChannel::SetChannelFlags(uint32_t aFlags) Maybe we should assert in this function and the next that we haven't opened the channel yet. I think !mLink should work. Then we'll have to move the existing SetChannelFlags above the Open calls, but that looks easy. Hopefully that will allay bsmedberg's concern. @@ +1714,2 @@ > { > + MOZ_ASSERT(aFlags & REQUIRE_URGENT_SCRIPT_BLOCKING ? NS_IsMainThread() : true); Can you put this in ClearFlags as well? ::: ipc/glue/MessageChannel.h @@ +92,5 @@ > // WindowsMessageLoop code should enable deferred native message > // handling to prevent deadlocks. Should only be used for protocols > // that manage child processes which might create native UI, like > // plugins. > + REQUIRE_DEFERRED_MESSAGE_PROTECTION = 1 << 0, The REQUIRE_ prefix on these doesn't really seem necessary. Can we just remove it and change REQUIRE_DEFAULT to DEFAULT_FLAGS? Also, can you please put an empty line after each flag?