Re: [RFC] - Some notions that I would like comments on

Chuck Lever (cel@monkey.org)
Thu, 15 Jul 1999 15:26:42 -0400 (EDT)

On Thu, 15 Jul 1999, Stephen C. Tweedie wrote:
> A page should only ever be non-uptodate if either it is currently still
> being read in (in which case it will be locked, too), or if there was an
> IO error reading it in.

in 2.3.7+, in block_write_partial_page, the PageUptodate flag is not set
if all the buffers for a page are not uptodate. by my instrumentation, my
experimental update code (which is only triggered when the page is not
locked and not up to date) is hit for between 0.5% and 5% of all read
pages.

> In the latter case, you want to do an IO retry
> when the user next explicitly asks for that page, but you really want to
> avoid doing the retry every single time the readahead zone hits the
> page.

that makes sense. thanks.

- Chuck Lever

--
corporate:	<chuckl@netscape.com>
personal:	<chucklever@netscape.net> or <cel@monkey.org>


The Linux Scalability project:
	http://www.citi.umich.edu/projects/linux-scalability/




-
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/