Sorry to join late the party, but...a read only mmap isn't equally dangerous? If the page cache doesn't contain its mapping, reading from it sequentially (and madvice is just...an advice) won't save us from major/minor page faults anyway. Just 1 major page fault with transparent huge page + the JVM safepoint machinery could lead long TTSP...or am I missing something? FileChannel::read seems to me the only way to go to avoid long TTSP unless we are sure of what the OS page contain..
-- You received this message because you are subscribed to the Google Groups "mechanical-sympathy" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
