Messages in this thread |  | | Date | Fri, 15 Sep 2000 22:00:18 +0200 | From | Andrea Arcangeli <> | Subject | Re: (reiserfs) Re: More on 2.2.18pre2aa2 |
| |
On Fri, Sep 15, 2000 at 09:08:25PM -0400, Giuliano Pochini wrote: > Which is tightly dependent on the way we insert new rqs.
Sure the implementation would differ in function of the way we order the requests during inserction, but the conceptual algorithm of the latency control could remain the same even if we'll sligtly change the ordering algorithm.
The reason the implementation would differ is because they are implemented together to browse the queue the less times possible.
> Rqs are added according to the IN_ORDER() condition which enforces ascending > ordering. What does CSCAN mean ?
CSCAN means Cyclic scan. It enforces ascending order and when it reaches the last cylinder it causes a large seek to the head.
> [..] When we insert 15 the list becomes 12 13 14 1 2 3 15.
You're right.
Infact the only reason we need the latency control stuff is that 14 have to become a barrier eventually (and 15 have to become a barrier eventually as well) if we want them to be served by the lowlevel driver otherwise they could be passed for way too long time (just the time to write some Tera of data to disk) by requests with lower blocknr. This was the elevator latency bugfix.
Andrea - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org Please read the FAQ at http://www.tux.org/lkml/
|  |