Date: Wed, 22 Mar 1995 20:58:00 +0000 () From: "John S. Dyson" <toor@jsdinc.root.com> To: cs.weber.edu!terry@implode.root.com (Terry Lambert) Cc: toor@Root.COM, star-gate.com!hasty@implode.root.com, hackers@FreeBSD.org Subject: Re: Why IDE is bad Message-ID: <199503222058.UAA00587@jsdinc.root.com> In-Reply-To: <9503230134.AA15550@cs.weber.edu> from "Terry Lambert" at Mar 22, 95 06:34:54 pm
next in thread | previous in thread | raw e-mail | index | archive | help
> > What do you mean "async mounts"? > > Since kernel entry via system call is atomic unless you have SMP > or multithreading to support multiple kernel thread contexts, it > really can't be anything but synchronous. > > A UFS mount in the kernel goes to the device, reads the volume > header and superblock into core, gets a vnode for inode 2, and > arranges to have references to the vnode in the file system it > is mounted into return vnode 2 of the mounted fs. Then it > returns from the mount call. > > This is synchronous, but it is also guaranteed to be deterministically > quick, by definition. > > Am I missing something about what you mean when you say "async mounts"? > You are missing my point (listen to what I mean, not what I say :-)). What I meant was that the async option is essentially ignored. In essence, the UFS does not act differently whether or not the async option is specified. Hmm, true async mounts -- better hold up V2.1, sounds like a neat feature :-). (PLEASE DONT TAKE THIS SERIOUSLY). John dyson@root.com
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199503222058.UAA00587>