Date: Thu, 16 May 1996 11:18:35 -0700 (MST) From: Terry Lambert <terry@lambert.org> To: dyson@FreeBSD.ORG Cc: rminnich@Sarnoff.COM, toor@dyson.iquest.net, jgreco@brasil.moneng.mei.com, hackers@FreeBSD.ORG Subject: Re: A question for the VM gurus..! Message-ID: <199605161818.LAA17487@phaeton.artisoft.com> In-Reply-To: <199605161424.JAA13589@dyson.iquest.net> from "John S. Dyson" at May 16, 96 09:24:08 am
next in thread | previous in thread | raw e-mail | index | archive | help
> > > > MADV_WILLNEED appears to fault pages (again, asynchrnously, from > > > > what I can tell). > > > > > > Hmmm... time to implement kernel threads. > > > > actually, sunos had async. read-ahead from 1988 on, long before kernel > > threads. They just queued an i/o and did not wait for the result. So you > > don't absolutely have to have kernel threads. > > > We used to have async readahead in the VM system (vnode_pager.) It was a bit > tricky, and kernel threads might make it simpler. Maybe not? A kernel thread scheduled against a "work to do" queue would seem to be the correct way to handle async. But async wants to be more general than a simple implementation; it wants to be enough to implement async calls at user level. Rather than special casing the code in the read/write path to enable the aioread/aiowrite/aiowait/aiocancel, I'd like to see an alternate trap gate for async calls... much better for LWP support. So I guess the same changes needed for kernel multithreading would be needed regardless.... 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?199605161818.LAA17487>