Date: Thu, 5 Dec 1996 01:50:18 -0500 (EST) From: "John S. Dyson" <toor@dyson.iquest.net> To: ccsanady@friley216.res.iastate.edu (Chris Csanady) Cc: freebsd-smp@FreeBSD.org Subject: Re: make locking more generic? Message-ID: <199612050650.BAA04384@dyson.iquest.net> In-Reply-To: <199612050626.AAA03220@friley216.res.iastate.edu> from "Chris Csanady" at Dec 5, 96 00:26:56 am
next in thread | previous in thread | raw e-mail | index | archive | help
> Could we possibly make our global semaphore a bit more generic? I guess this > is what I had in mind.. > FYI, I have been thinking about locking in the VM and VFS subsystems. I am sure that you will need to be taking input from various sources, in the VM system in particular, it will be likely that we might need to re-organize some of the code to fix deadlocks. Eventually, if we want good fine grained locking, we'll need to work out a clean hierarchy. I am also very hopeful of being able to build the single cpu code with very very little additional overhead, and SMP code with whatever added overhead necessary for locking code. If any of you SMP guys have any words of wisdom, or documentation sources for SMP technology, I would really like to hear from you with pointers. Also, I would sure like to hear about the direction that you are planning to go, so I can learn what I need to help. I am going to be getting to the VM documentation, and that will be a good time to in parallel look at feasible locking hierarchies. When benchmarking and evaluating the performance of the system, assuming the images/files are already in memory the VM routines are pretty high on the list as to cpu usage. John dyson@freebsd.org
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199612050650.BAA04384>