Date: Tue, 21 Dec 2004 22:41:45 -0500 (EST) From: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> To: joe@zircon.seattle.wa.us Cc: arch@FreeBSD.org Subject: Re: Fixing Posix semaphores Message-ID: <200412220341.iBM3fj5j043952@khavrinen.lcs.mit.edu> In-Reply-To: <1103681460.30309.799.camel@zircon.zircon.seattle.wa.us> References: <1102975803.30309.196.camel@zircon.zircon.seattle.wa.us> <Pine.NEB.3.96L.1041221235624.62809A-100000@fledge.watson.org> <20041222011506.GG801@straylight.m.ringlet.net>
next in thread | previous in thread | raw e-mail | index | archive | help
In article <1103681460.30309.799.camel@zircon.zircon.seattle.wa.us> you write:
>So, there are words there that can be interpreted many different ways.
>The most restricted way to view them is as 14-character names optionally
>beginning with a slash. That also seems to me to be the stupidest way
>to view them. Robert's idea of semfs seems brilliant, allowing multiple
>name spaces for jailed processes. I plan to start thinking and working
>on that idea shortly.
Here is an even easier way that avoids creating a special file type...
sem_open() can be implemented as:
fd = shm_open(name, oflag, omode);
if (oflag & O_CREAT) {
ftruncate(fd, sizeof(struct sem_private));
}
sem_private = mmap(fd, ...);
sem_private->fd = fd;
sem = malloc(sizeof *sem);
_sem_init(sem, sem_private, value);
return (sem);
The underlying semaphore can then be implemented in the standard way
with a mutex, a condition variable (or simulation thereof), and a
counter in the shared-memory region -- no (additional) kernel code
required. (This avoids the overhead of a system call if there is no
need to wait.)
Alternatively, given the current kernel implementation, you can very
simply convert any vnode into a 20-byte flat name for the file using
VFS_VPTOFH() -- this would require only minimal changes to the
existing code.
-GAWollman
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200412220341.iBM3fj5j043952>