Hi,
I noticed there was a change in page I/O accounting from kernel-2.4.3, as
mentioned in the pre7 section of ChangeLog-2.4.3.
Before the change(like in kernel-2.2.x), the "kstat.pgpgin" and
"kstat.pgpgout" are incremented (by one) for every block-read/write.
(in "driver/block/ll_rw_blk.c")
However, starting from kernel-2.4.3 onwards, they are incremented for every
sector-read/write. And, then the numbers are both divided by 2 before printing
to "/proc/stat" (in "fs/proc/proc_misc.c"), so they become the number of
1K-units read/written finially.
In my system, block size is always 4K. I think the original way of counting
for page I/O is more meaningful as PAGE_SIZE is also 4K.
So is "kstat.pgpgin" still meant to be the number of pages read or sth else?
Can sb explain to me why the accouting method was changed, as I would like
to collect the page I/O statistics correctly ?
Thanks in advance!
--
regards,
ZM
--
There is no mechanical problem so difficult that it cannot be solved by
brute strength and ignorance. -- William's Law
--
Kernelnewbies: Help each other learn about the Linux kernel.
Archive: http://mail.nl.linux.org/kernelnewbies/
FAQ: http://kernelnewbies.org/faq/
