Re: [PATCH] Fixing soft NFS umount -f, round 1

[Thor Lancelot Simon <tls%panix.com@localhost>]

On Fri, Jul 03, 2015 at 10:07:28AM -0700, Chuck Silvers wrote:
> On Wed, Jul 01, 2015 at 09:10:40PM +0200, Emmanuel Dreyfus wrote:
> > David Holland <dholland-tech%netbsd.org@localhost> wrote:
> > 
> > > ...however, with a dead nfs server even sync() should time out and
> > > fail eventually.
> > 
> > The problem is that you can ask VFS_SYNC(9) to wait or not, but there is
> > no such option for sync(2). 
> 
> for a soft mount sync() should time out like everything else, yes.
> for a hard mount, our current semantics for sync() will cause it to
> wait indefinitely.  sync() is not required to wait for anything at all,

It's not required to, but I am glad it does.  How else can you actually
know all the buffers hit the disk before you hit the power switch?

Thor