Re: [Lse-tech] lockstat results with fast-walk patch running dbench 16 on 8-way

[Rick L. <ric...@us...>]

    When metering a spin_lock(), lockmeter first does a spin_trylock().  If
    that's successful, then "hold" time begins.  If the spin_trylock() is
    not successful, then the "wait" start-time is captured and an unmetered
    spin_lock() is done.  When that acquisition completes, then the "wait"
    end-time is captured.  So the answer is 'no', the "wait" time doesn't
    count that initial spin_trylock().  And the "hold" time doesn't count
    that initial spin_trylock(), either.  If people think this is important,
    I can easily change it.

No need to do it on my account :)  It strikes me as the wrong tool for
measuring cache thrash.  If it could already tell us that, there's no
reason to discard the data, but it sounds like there may be other tools
better suited for that.

Rick