lkml.org 
[lkml]   [2008]   [Mar]   [9]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
    /
    Date
    From
    SubjectRe: [PATCH] x86: Change x86 to use generic find_next_bit

    * Alexander van Heukelum <heukelum@mailshack.com> wrote:

    > x86: Change x86 to use the generic find_next_bit implementation
    >
    > The versions with inline assembly are in fact slower on the machines I
    > tested them on (in userspace) (Athlon XP 2800+, p4-like Xeon 2.8GHz,
    > AMD Opteron 270). The i386-version needed a fix similar to 06024f21 to
    > avoid crashing the benchmark.
    >
    > Benchmark using: gcc -fomit-frame-pointer -Os. For each bitmap size
    > 1...512, for each possible bitmap with one bit set, for each possible
    > offset: find the position of the first bit starting at offset. If you
    > follow ;). Times include setup of the bitmap and checking of the
    > results.
    >
    > Athlon Xeon Opteron 32/64bit
    > x86-specific: 0m3.692s 0m2.820s 0m3.196s / 0m2.480s
    > generic: 0m2.622s 0m1.662s 0m2.100s / 0m1.572s

    ok, that's rather convincing.

    the generic version in lib/find_next_bit.c is open-coded C which gcc can
    optimize pretty nicely.

    the hand-coded assembly versions in arch/x86/lib/bitops_32.c mostly use
    the special x86 'bit search forward' (BSF) instruction - which i know
    from the days when the scheduler relied on it has some non-trivial setup
    costs. So especially when there's _small_ bitmasks involved, it's more
    expensive.

    > If the bitmap size is not a multiple of BITS_PER_LONG, and no set
    > (cleared) bit is found, find_next_bit (find_next_zero_bit) returns a
    > value outside of the range [0,size]. The generic version always
    > returns exactly size. The generic version also uses unsigned long
    > everywhere, while the x86 versions use a mishmash of int, unsigned
    > (int), long and unsigned long.

    i'm not surprised that the hand-coded assembly versions had a bug ...

    [ this means we have to test it quite carefully though, as lots of code
    only ever gets tested on x86 so code could have built dependency on
    the buggy behavior. ]

    > Using the generic version does give a slightly bigger kernel, though.
    >
    > defconfig: text data bss dec hex filename
    > x86-specific: 4738555 481232 626688 5846475 5935cb vmlinux (32 bit)
    > generic: 4738621 481232 626688 5846541 59360d vmlinux (32 bit)
    > x86-specific: 5392395 846568 724424 6963387 6a40bb vmlinux (64 bit)
    > generic: 5392458 846568 724424 6963450 6a40fa vmlinux (64 bit)

    i'd not worry about that too much. Have you tried to build with:

    CONFIG_CC_OPTIMIZE_FOR_SIZE=y
    CONFIG_OPTIMIZE_INLINING=y

    (the latter only available in x86.git)

    > Patch is against -x86#testing. It compiles.

    i've picked it up into x86.git, lets see how it goes in practice.

    Ingo


    \
     
     \ /
      Last update: 2008-03-09 21:13    [W:0.028 / U:89.024 seconds]
    ©2003-2016 Jasper Spaans. hosted at Digital OceanAdvertise on this site