lkml.org 
[lkml]   [2005]   [Jul]   [25]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
    Patch in this message
    /
    Date
    From
    SubjectRe: [PATCH NFS 3/3] Replace nfs_block_bits() with roundup_pow_of_two()
    On Sun, Jul 24, 2005 at 07:24:23PM -0400, Trond Myklebust wrote:
    > su den 24.07.2005 Klokka 19:09 (-0400) skreiv Trond Myklebust:
    > > What non-power-of-two target? Anything _not_ aligned to a power of two
    > > boundary is a BUG!

    So we could simply replace the loop in nfs_block_bits() with call to
    BUG()? :->

    > Furthermore, rounding UP in order to "correct" this non-alignment would
    > definitely be a bug.
    >
    > If users choose to override the default rsize/wsize, that is almost
    > always in order to limit the UDP fragmentation per read/write request on
    > lossy networks. By rounding up, you are doubling the number of fragments
    > that the user requested instead of respecting the limit.

    OK. Either way, this function can be cleaned up. Apparently the Plan 9
    filesystem guys thought it rounds up.

    What's your opinion of the following patch, which explicitly states the
    intent of nfs_block_bits()? It still needs patch 1 and 2.

    Thanks,
    Rene



    [PATCH 3/4] Simplify and rename nfs_block_bits() to rounddown_pow_of_two()

    nfs_block_bits() doesn't have a lot to do with bits anymore. It can
    also be implemented simpler and clearer with fls().

    Signed-off-by: Rene Scharfe <rene.scharfe@lsrfire.ath.cx>
    ---

    fs/nfs/inode.c | 21 +++++----------------
    1 files changed, 5 insertions(+), 16 deletions(-)

    1388b63224334877b1b154e38ad9ee17f1726bca
    diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c
    --- a/fs/nfs/inode.c
    +++ b/fs/nfs/inode.c
    @@ -185,20 +185,9 @@ nfs_umount_begin(struct super_block *sb)
    rpc_killall_tasks(rpc);
    }

    -
    -static inline unsigned long
    -nfs_block_bits(unsigned long bsize)
    +static inline unsigned long rounddown_pow_of_two(unsigned long x)
    {
    - /* make sure blocksize is a power of two */
    - if (bsize & (bsize - 1)) {
    - unsigned char nrbits;
    -
    - for (nrbits = 31; nrbits && !(bsize & (1 << nrbits)); nrbits--)
    - ;
    - bsize = 1 << nrbits;
    - }
    -
    - return bsize;
    + return x ? (1UL << (fls(x) - 1)) : 0;
    }

    /*
    @@ -222,7 +211,7 @@ nfs_block_size(unsigned long bsize)
    else if (bsize >= NFS_MAX_FILE_IO_BUFFER_SIZE)
    bsize = NFS_MAX_FILE_IO_BUFFER_SIZE;

    - return nfs_block_bits(bsize);
    + return rounddown_pow_of_two(bsize);
    }

    /*
    @@ -319,10 +308,10 @@ nfs_sb_init(struct super_block *sb, rpc_
    }

    if (sb->s_blocksize == 0) {
    - sb->s_blocksize = nfs_block_bits(server->wsize);
    + sb->s_blocksize = rounddown_pow_of_two(server->wsize);
    sb->s_blocksize_bits = fls(sb->s_blocksize - 1);
    }
    - server->wtmult = nfs_block_bits(fsinfo.wtmult);
    + server->wtmult = rounddown_pow_of_two(fsinfo.wtmult);

    server->dtsize = nfs_block_size(fsinfo.dtpref);
    if (server->dtsize > PAGE_CACHE_SIZE)
    -
    To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
    the body of a message to majordomo@vger.kernel.org
    More majordomo info at http://vger.kernel.org/majordomo-info.html
    Please read the FAQ at http://www.tux.org/lkml/
    \
     
     \ /
      Last update: 2005-07-25 18:01    [W:0.025 / U:128.584 seconds]
    ©2003-2016 Jasper Spaans. hosted at Digital OceanAdvertise on this site