On Tue, Apr 26, 2011 at 12:45 AM, Randy Dunlap <randy.dunlap@xxxxxxxxxx> wrote: > On Mon, 25 Apr 2011 19:44:31 +0900 Geunsik Lim wrote: > >> From: Geunsik Lim <geunsik.lim@xxxxxxxxxxx> >> >> Support kbuild menu to select memory unmap operation size >> at build time. >> >> Signed-off-by: Geunsik Lim <geunsik.lim@xxxxxxxxxxx> >> Acked-by: Hyunjin Choi <hj89.choi@xxxxxxxxxxx> >> --- >> Âinit/Kconfig |  70 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ >> Âmm/memory.c Â|  21 +++++++++++----- >> Â2 files changed, 84 insertions(+), 7 deletions(-) >> >> diff --git a/init/Kconfig b/init/Kconfig >> index 56240e7..0983961 100644 >> --- a/init/Kconfig >> +++ b/init/Kconfig >> @@ -557,6 +557,76 @@ config LOG_BUF_SHIFT >>          Â13 => Â8 KB >>          Â12 => Â4 KB >> >> +config PREEMPT_OK_MUNMAP_RANGE >> +   int "Memory unmap unit on preemption mode (8 => 32KB)" >> +   depends on !PREEMPT_NONE >> +   range 8 2048 >> +   default 8 >> +   help >> +    unmap_vmas(=unmap a range of memory covered by a list of vma) is treading > >     Âunmap_vmas (= unmap a range ... > >> +    a delicate and uncomfortable line between hi-performance and low-latency. > >                          Âhigh performane and low latency. > >> +    We've chosen to improve performance at the expense of latency. > >     ÂThis option improves performance at the expense of latency. > >> + >> +    So although there may be no need to resched right now, > >                       Âreschedule > >> +    if we keep on gathering more and more without flushing, > >            Âgathering more and more <what> ? > >> +    we'll be very unresponsive when a resched is needed later on. > >                      Âreschedule > >> + >> +    Consider the best suitable result between high performance and low latency >> +    on preemption mode. >> +    Select optimal munmap size to return memory space that is allocated by mmap system call. >> + >> +    For example, For recording mass files, if we try to unmap memory that we allocated > >            for > >> +    with 100MB for recording in embedded devices, we have to wait for more than 3seconds to > >                                           Â3 seconds > > (but try not to put text over 80 columns, please) > >> +    change mode from play mode to recording mode. This results from the unit of memory >> +    unmapped size when we are recording mass files like camcorder particularly. >> + >> +     ÂThis value can be changed after boot using the >> +     Â/proc/sys/vm/munmap_unit_size tunable. > > Indent above with tab + 2 spaces. > >> + >> +    Examples: >> +         Â2048 => 8,388,608bytes : for straight-line efficiency >> +         Â1024 => 4,194,304bytes >> +          512 => 2,097,152bytes >> +          256 => 1,048,576bytes >> +          128 =>  524,288bytes >> +          Â64 =>  262,144bytes >> +          Â32 =>  131,072bytes >> +          Â16 =>  Â65,536bytes >> +           8 =>  Â32,768bytes : for low-latency (*default) > > All of above would be better with added space before "bytes", as, e.g.: >            Â8 =>  Â32,768 bytes > >> + >> +config PREEMPT_NO_MUNMAP_RANGE >> +   int "Memory unmap unit on non-preemption mode (1024 => 4MB)" >> +   depends on PREEMPT_NONE >> +   range 8 2048 >> +   default 1024 >> +   help >> + >> +    unmap_vmas(=unmap a range of memory covered by a list of vma) is treading > >     Âunmap_vmas (= unmap > >> +    a delicate and uncomfortable line between hi-performance and low-latency. > >                          Âhigh performance and low latency. > >> +    We've chosen to improve performance at the expense of latency. > >     ÂThis option improves performance at the expense of latency. > >> + >> +    So although there may be no need to resched right now, > >                       Âreschedule > >> +    if we keep on gathering more and more without flushing, > >                 Âmore and more what? > >> +    we'll be very unresponsive when a resched is needed later on. > >                      Âreschedule > >> + >> +    Consider the best suitable result between high performance and low latency >> +    on preemption mode. > > but this option is for non-preempt mode... so should that text above be modified? > >> +    Select optimal munmap size to return memory space that is allocated by mmap system call. >> + >> +     ÂThis value can be changed after boot using the >> +     Â/proc/sys/vm/munmap_unit_size tunable. > > Indent above with tab + 2 spaces. > >> + >> +    Examples: >> +         Â2048 => 8,388,608bytes : for straight-line efficiency >> +         Â1024 => 4,194,304bytes (*default) >> +          512 => 2,097,152bytes >> +          256 => 1,048,576bytes >> +          128 =>  524,288bytes >> +          Â64 =>  262,144bytes >> +          Â32 =>  131,072bytes >> +          Â16 =>  Â65,536bytes >> +           8 =>  Â32,768bytes : for low-latency > >        ÂUse space before "bytes" in table above, please. > >> + >> Â# >> Â# Architectures with an unreliable sched_clock() should select this: >> Â# > > > --- > ~Randy > *** Remember to use Documentation/SubmitChecklist when testing your code *** Randy Dunlap. Thanks a lot. I will modify contents that you commented. > -- Regards, Geunsik Lim ( Samsung Electronics ) Blog : http://blog.naver.com/invain/ e-Mail: geunsik.lim@xxxxxxxxxxx     Â leemgs@xxxxxxxxx , leemgs1@xxxxxxxxx -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/ -- To unsubscribe from this list: send the line "unsubscribe linux-rt-users" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
