solaris mutexes?
Roy Gordon wrote: > Is the following true: Solaris mutexes only exist in the user address > space (including any shared memory space); they have no associated > kernel data structure. True. For non-shared-memory mutexes. Ditto CVs & unnamed semaphores. > > If true, this would be opposed to system V semaphores. Yup. > > > Also, if true, then a given mutex could be moved to a different > address (suitably aligned) and as long as all threads (or processes, as > the case may be) reference it at that address, then it would continue > functioning as if it hadn’t been moved. > > Is this correct too (if the initial assumption is correct, that is)? You mean like a compacting garbage collector? Yeah. ‘Matter of fact, I believe that’s what Java does on some platforms.
