xfs: increase readdir buffer size

While doing some testing of readdir perf a while back,
I noticed that the buffer size we're using internally is
smaller than what glibc gives us by default.  Upping this
size helped a bit, and seems safe.

glibc's __alloc_dir() does:

  const size_t default_allocation = (4 * BUFSIZ < sizeof (struct dirent64)
                                     ? sizeof (struct dirent64) : 4 * BUFSIZ);
  const size_t small_allocation = (BUFSIZ < sizeof (struct dirent64)
                                   ? sizeof (struct dirent64) : BUFSIZ);
  size_t allocation = default_allocation;
#ifdef _STATBUF_ST_BLKSIZE
  if (statp != NULL && default_allocation < statp->st_blksize)
    allocation = statp->st_blksize;
#endif

and

#define _G_BUFSIZ 8192
#define _IO_BUFSIZ _G_BUFSIZ
# define BUFSIZ _IO_BUFSIZ

so the default buffer is 4 * 8192 = 32768
(except in the unlikely case of blocks > 32k....)

Signed-off-by: Eric Sandeen <sandeen@sandeen.net>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Alex Elder <aelder@sgi.com>

diff --git a/fs/xfs/linux-2.6/xfs_file.c b/fs/xfs/linux-2.6/xfs_file.c
index e4caeb2..3805ada 100644 (file)
--- a/fs/xfs/linux-2.6/xfs_file.c
+++ b/fs/xfs/linux-2.6/xfs_file.c
@@ -203,9 +203,9 @@ xfs_file_readdir(
         *
         * Try to give it an estimate that's good enough, maybe at some
         * point we can change the ->readdir prototype to include the
-        * buffer size.
+        * buffer size.  For now we use the current glibc buffer size.
         */
-       bufsize = (size_t)min_t(loff_t, PAGE_SIZE, ip->i_d.di_size);
+       bufsize = (size_t)min_t(loff_t, 32768, ip->i_d.di_size);
 
        error = xfs_readdir(ip, dirent, bufsize,
                                (xfs_off_t *)&filp->f_pos, filldir);