Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 23 Dec 1999 22:05:14 -0500
From:      "Vladimir N. Silyaev" <vsilyaev@mindspring.com>
To:        Guido van Rooij <guido@gvr.org>
Cc:        Peter Wemm <peter@netplex.com.au>, Hidetoshi Shimokawa <simokawa@sat.t.u-tokyo.ac.jp>, vsilyaev@mindspring.com, dillon@freebsd.org, freebsd-emulation@freebsd.org, dbutter@wireless.net
Subject:   Re: VMware: Questions...
Message-ID:  <19991223220514.A1504@jupiter.delta.ny.us>
In-Reply-To: <19991222224803.A410@gvr.gvr.org>; from guido@gvr.org on Wed, Dec 22, 1999 at 10:48:03PM %2B0100
References:  <simokawa@sat.t.u-tokyo.ac.jp> <19991220153149.AD13C1CCE@overcee.netplex.com.au> <19991222224803.A410@gvr.gvr.org>

next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, Dec 22, 1999 at 10:48:03PM +0100, Guido van Rooij wrote:
> On Mon, Dec 20, 1999 at 11:31:49PM +0800, Peter Wemm wrote:
> > > 
> > > This is because mmaped file is written every 30 second by sync daemon.
> > > The file is usually named /var/tmp/ram0 but it's unlinked right after
> > > opened so you cannot see it by 'ls' although it exits.
> > 
> > It would be nice if the VFS/VM system detected this automatically and
> > switched on NOSYNC for files that got unlinked...  I wouldn't be suprised if
> > this is what Linux does.  Matt, is this possible?
> > 
> 
> I havent seen an answer to this question yet. If it is possible, that
> would be very nice. I doubt it though, but my knowledge on that part
> of the system is rather limited.
> A quick workaround could be to look at the ref count of the underlying
> inode of the fd passed to the linux mmap
> If the refcount is one then clearly the inode is no being referenced through
> a directory entry in the file system. This could even be done for the general
> mmap call (provided a regular file of course). But it might be a very
> specific situation because one usually would not used a file backed
> mmap in FreeBSD, yet use an anonymous mmap. 
My personal suggestion about this, that in the Linux emulator we can
safety use the original Hidetoshi patch, because I assume that Linux
never do syncing a mmaped files. I'm not discover the linux code about
that, and don't do any experiment.

Of course, this don't cancel the proposition about NO_ASYNC and unreferenced
files.

-- 
Vladimir Silyaev


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-emulation" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?19991223220514.A1504>