Re: [PATCH][5/?] count writeback pages in nr_scanned

[Andrea Arcangeli]

On Thu, Jan 06, 2005 at 03:55:34PM +1100, Nick Piggin wrote:
> However, if you had a plain io_schedule_timeout there, at least you
> would sleep for the full extend of the specified timeout.

I agree it sure would be safer but OTOH it may screwup performance by
waiting for unnecessary long times on fast stroage.

So it's ok for a test, but still it wouldn't be a final fix since the
timeout may be still too short in some case.

Waiting on one (or more) PG_writeback bitflags to go away should fix it
completely. This is how 2.4 throttles on the writeback I/O too of
course. How we choose which random page to pick may vary though.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/