> Judging by the nonsequential way the block numbers were
> varying in bdflush in my tests I think it could be a big
> performance win then. I can see how to implement the
> clustering in bdflush but I'm not sure what to do about
> the swapping -- seems like there's a potential for
It's quite sure to give interference, but swap-outs
are also being clustered. That is, the disk only has
to seek to one position on the swap partition and does
all it's writing there (up to 128kB sequential).
> interference. It could be done at a lower level --
> buffer block writes and then do them in an efficiently
> ordered way but are there any other side effects from
> doing that?
FreeBSD has done this for a long time, and the SCSI
subsystem also processes and reorders it's own requests
(in hardware, that is), so I don't see any reason why
there can't be a general software solution for more
efficiency and operation on non-optimizing hardware.
Rik.
+-----------------------------+------------------------------+
| For Linux mm-patches, go to | "I'm busy managing memory.." |
| my homepage (via LinuxHQ). | H.H.vanRiel@fys.ruu.nl |
| ...submissions welcome... | http://www.fys.ruu.nl/~riel/ |
+-----------------------------+------------------------------+
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu