Subject: Re: [need review] pax(1) change to fix PR #8868
To: None <tech-userlevel@netbsd.org, tech-kern@netbsd.org>
From: Bill Studenmund <wrstuden@nas.nasa.gov>
List: tech-userlevel
Date: 02/10/2000 14:34:59
On Thu, 10 Feb 2000, ITOH Yasufumi wrote:

> abs@netbsd.org wrote:
> 
> > 	Just curious - can the SIGINFO hack cause problems if pax is
> > 	being used to read/write to an actual tape device?
> 
> I don't think read/write is always atomic, but, ... well, ...
> I'm not sure --- this is the reason I'm sending to tech-kern. :-)

A successful read or write should be atomic. NFS kinda bends the rules a
bit, but other than that, your write should fully complete before (or
after) someone else's.

There's no guarantee of ordering, just atomiticity.

Take care,

Bill