dtrace discuss - Dec 2008 - seems the output of fsrw.d doesn''t match the 128k recordsize of zfs, shouldn''t the size of fop_read/fop_write be multiple times of 128K?

The recordsize of all zfs is 128K, and there is only zfs on the system.
# zfs list -o recordsize
RECSIZE
   128K
   128K
   128K
      -
   128K
   128K
      -
Then I make a file with size 1K and then copy that file to another file with
the following commands:
# mkfile 1k test4
# cp test4 test5


bash-3.00# ./fsrw.d
Event            Device RW     Size Offset Path
    disk_io        zfs2  R     4096      0 <none>
    disk_io       cmdk0  R     4096      0 <none>
    disk_io        zfs2  R     4096      0 <none>
    disk_io       cmdk0  R     4096      0 <none>
sc-pread              .  R        4      0 /var/adm/wtmpx
  fop_read            .  R        4      0 /var/adm/wtmpx
    disk_io       cmdk0  R    27136      0 <none>
sc-read               .  R    32768      0 /test/test4
  fop_read            .  R    32768      0 /test/test4        * ==> Since
the recordsize is 128K, I think 128K should be shown instead of 32K*
sc-write              .  W     1024      0
/test/test5
  fop_write           .  W     1024      0 /test/test5          *==> Since
the recordsize is 128K, I think 128K should be shown instead of 1K, if not
,what the role does recordsize 128k play here?*
sc-read               .  R    32768      1 /test/test4
  fop_read            .  R    32768      1 /test/test4
^C
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://mail.opensolaris.org/pipermail/dtrace-discuss/attachments/20081226/f5dcf73b/attachment.html>