Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 12 Sep 1995 13:42:51 -0700 (MST)
From:      Terry Lambert <terry@lambert.org>
To:        rgrimes@GndRsh.aac.dev.com (Rodney W. Grimes)
Cc:        terry@lambert.org, current@FreeBSD.org
Subject:   Re: Statistics request
Message-ID:  <199509122042.NAA02216@phaeton.artisoft.com>
In-Reply-To: <199509121919.MAA04151@GndRsh.aac.dev.com> from "Rodney W. Grimes" at Sep 12, 95 12:19:03 pm

next in thread | previous in thread | raw e-mail | index | archive | help
> Those are just spot selecting the non-0 ones...
> 
> Now fire up 3 copies of find:
>         namei     2     1K      6K  8873K   168480    0     0  16,32,64,1K
>         namei     2     1K      6K  8873K   172299    0     0  16,32,64,1K
>         namei     1     1K      6K  8873K   172415    0     0  16,32,64,1K

The question is whether it self-zero's.  If it does, then care has been
taken to doubly-break the buffer allocation code in nfs_namei(), which
is what I'm trying to determine.

I would expect that if this were the case, it would re-zero when idle.

If it is only singly broken, I'd expect the number of buffers to increase
and keep increasing, or to start freeing stuff that isn't allocated.

I haven't determined what statistic free unallocated buffers would bump;
unfortunately, I think it would be hidden in the noise on anything that
could reproduce the error cases.

Garrett has reported 0's in the idle case on his box.

I will probably need to write some shell test cases to specifically
exercise the bogus code.

Anyone have any idea what statistic will show preterbation when I free
a buffer twice?  I'd like to watch that one too.


					Terry Lambert
					terry@lambert.org
---
Any opinions in this posting are my own and not those of my present
or previous employers.



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199509122042.NAA02216>