I got a bit bored by how much `cp /dev/zero /tmp' does stall the machine
(all reads precisely). So I fixed the problem. But with my approch the
speed of writes to disk while there are also read in progress or while
there's some buffers that needs to be read, drops. Also scheduling rate
increase a lot under such conditions...

But the interactive feeling is perfect now. I _can't_ notice anymore a `cp
/dev/zero /tmp` in background :-).

Right now I had no good idea on how to be iteractive without dropping
write performances.

If you try it let me know if you are happy or not with it (you can CC also
to the list). The thing is merged in my jumbo patch, too late to extract
it from my CVS tree now (I'll do that + implementation details tomorrow if
somebody is interested).

	ftp://e-mind.com/pub/andrea/kernel/pre-2.3.4-2_andrea1.bz2

(against pre-2.3.4-2 of course :)

Andrea Arcangeli


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/