>>99
Sorry, I can't find the original post (Goddamn, Internet search is so shitty and it's getting shittier by the day). But you can read
http://code.google.com/p/chromium/issues/detail?id=7823 if you feel like it, comment #46 and some others are examples of this obnoxious behavior (fix: block third party advertisement/tracking networks, or in general just do aggressive adblocking as you should already)
>>100
The entire browser does not deadlock (talking about Chrome here, right?), but even if it did, it's trivial to find which process is at fault: it'll be consuming 100% CPU (of one core).
I've only seen it deadlock with plugins, that's to be expected because the communication mechanism in plugins isn't designed to be block-resistant (actually it's only designed to suck).
Even if this happens, killing the plugin process will solve the deadlock. The plugin process can be a bit more difficult to find if it's not spinlocking, but it can be done since unlike the renderer processes it is not sandboxed and therefore has different privileges.