Re: [PATCH] properly account readahead file major faults

[Marcelo Tosatti]

Hi Hugh!

On Tue, Nov 22, 2005 at 12:55:02PM +0000, Hugh Dickins wrote:
> On Tue, 22 Nov 2005, Marcelo Tosatti wrote:
> > 
> > Pages which hit the first time in cache due to readahead _have_ caused
> > IO, and as such they should be counted as major faults.
> 
> Have caused IO, or have benefitted from IO which was done earlier?

Which caused IO, either synchronously or via (previously read)
readahead.

> It sounds debatable, each will have their own idea of what's major.

I see your point... and I much prefer the "majflt means IO performed"
definition :)

As a user I want to know how many pages have been read in from disk to
service my application requests.

>From the "time" manpage:

F      Number of major, or I/O-requiring, page faults  that  oc-
       curred  while  the process was running.  These are faults
       where the page has actually migrated out of primary memo-
       ry.

> Maybe PageUptodate at the time the entry is found in the page cache
> should come into it?  !PageUptodate implying that we'll be waiting
> for read to complete.

Hum, I still strongly feel that users care about IO performed and not
readahead effectiveness (which could be separate information).

I don't think the semantics are precisely defined anywhere are they?
-
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/