Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 24 Jan 2004 21:05:33 -0500 (EST)
From:      Robert Watson <rwatson@FreeBSD.org>
To:        src-committers@FreeBSD.org, cvs-src@FreeBSD.org, cvs-all@FreeBSD.org
Subject:   Re: cvs commit: src/sys/sys _mutex.h src/sys/kern kern_mutex.c
Message-ID:  <Pine.NEB.3.96L.1040124210221.31483D-100000@fledge.watson.org>
In-Reply-To: <200401250159.i0P1xR4i039361@repoman.freebsd.org>

next in thread | previous in thread | raw e-mail | index | archive | help

On Sat, 24 Jan 2004, Robert Watson wrote:

>   Add some basic support for measuring sleep mutex contention to the
>   mutex profiling code.  As with existing mutex profiling, measurement
>   is done with respect to mtx_lock() instances in the code, as opposed
>   to specific mutexes.  In particular, measure two things:

FYI --

Mutex profiling is pretty neat, and very under-documented.  At least, as
far as I know, only DES's original commit message and comments in NOTES
document it.  If someone on the doc side feels moved to explore and
document the implementation, as well as provide/gather best practicies for
using it in the performance optimization process, that would be great. 
I.e., mutex_profiling(9) or the like.  Now that we're beginning to have
significant kernel subsystems running entirely free of Giant, we should be
starting to shift gears into more performance analysis and optimization,
and documentation is going to be vital for that.

Thanks,

Robert N M Watson             FreeBSD Core Team, TrustedBSD Projects
robert@fledge.watson.org      Senior Research Scientist, McAfee Research




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.NEB.3.96L.1040124210221.31483D-100000>