[lkml]   [2005]   [Dec]   [24]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
    SubjectHZ_200/1000 (configurable) Timer Question
    I was looking at the 2.6.14 kernel (then checked back to 2.6.9 and
    realized this is fairly new) and found a configurable Timer setting.
    The option is under "Processor Type and Features" then "Timer

    A quick grep of the source hasn't revealed anything obvious in terms of
    where this is implemented, so I suspect i'm missing something (it would
    probably be arch-specific).

    On another UNIX-based operating system, back in my employable days (late
    90s), I wrote up a specification on how to implement the tunable clock
    frequencies for a particular power-pc based platform (further - i
    implemented it). The timer chip that was used at the time was an i82378
    (intel) chip and the internal clock ran at a frequency of 1.19318 MHz,
    you just had to (at the OS/hardware-level) set the number of internal
    ticks had passed before it would fire an interrupt. This may be
    completely useless, incorrect, or maybe if correct obvious information,
    but this is what I have from /memory/:

    (1) to determine how many ticks to tell the chip to pass before firing,
    you performed elementary-school division:

    1,193,180 (i.e. #hz)
    -------------------------------------- = Timeout value
    Configured-by-user-frequency (hz)

    (2) How not to break everything in the kernel: In the particular kernel
    we used, there was just a constant value (#define HZ <value>) and that
    just had to be changed. If set correctly, everything else would work
    fine. We further gave the user the option of making (for testing
    purposes) the HZ value configurable at run-time, but that would slow
    down (potentially) the system since every call to get the value of HZ
    would be a system call.

    This /could/ be considered proprietary information, but on the other
    hand, they are now using and contributing to linux, and further I was
    told there was no problem talking about anything I did or learned in the
    past since it's all outdated (we're talking nearly 10 years ago now).

    I hope someone finds this useful. If not, ignore me, I'm probably wrong
    anyway. i was just reviewing some of my notes and I don't know if what
    I did was ever tested, used, or put into practice.


    To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
    the body of a message to
    More majordomo info at
    Please read the FAQ at

     \ /
      Last update: 2005-12-24 21:22    [W:0.021 / U:2.260 seconds]
    ©2003-2017 Jasper Spaans. hosted at Digital OceanAdvertise on this site