Re: bin/55815: tar opens device files

[Martin Husemann <martin%duskware.de@localhost>]

The following reply was made to PR bin/55815; it has been noted by GNATS.

From: Martin Husemann <martin%duskware.de@localhost>
To: gnats-bugs%netbsd.org@localhost
Cc: 
Subject: Re: bin/55815: tar opens device files
Date: Wed, 25 Nov 2020 10:19:16 +0100

 On Wed, Nov 25, 2020 at 10:10:43AM +0100, Joerg Sonnenberger wrote:
 > >  Those should be the default then, or maybe check (is there a mount flag
 > >  or something in statvfs?) whether any ACLs or ext attrs could be there at
 > >  all.
 > 
 > That would be the ostrich approach of ignoring that open with side
 > effects is fundamentally a bad idea. It also doesn't match what other
 > tar implementations do and violates POLA just as well.
 
 I agree that device open having an effect on the state of the device is
 a bad thing in general, but it has been the case in Unix since the beginning,
 or am I misremembering something? At least it was that way when dealing with
 all tapes I ever dealt with.
 
 But besides that: if there *could* be no extattr or ACL on the file system
 (like I *knew* for sure for the case at hand, it was an anciend FFSv1
 file system way too old for any of that) and making that information
 available (like with a mount flag or a fs superblock flag or whatever) -
 why is skipping the unneccessary open in that case an ostrich aproach?
 
 Or why would it violate POLA? I would call it a smart optimization instead,
 and would be very suprised to see thousands of totally unneeded open()
 calls.
 
 Martin