Re: kern/38990: Unmounting a disk fails the first time

[David Holland <dholland-bugs%netbsd.org@localhost>]

The following reply was made to PR kern/38990; it has been noted by GNATS.

From: David Holland <dholland-bugs%netbsd.org@localhost>
To: Christos Zoulas <christos%zoulas.com@localhost>
Cc: David Holland <dholland-bugs%netbsd.org@localhost>, 
gnats-bugs%NetBSD.org@localhost,
        kern-bug-people%netbsd.org@localhost, gnats-admin%netbsd.org@localhost,
        netbsd-bugs%netbsd.org@localhost, apb%cequrux.com@localhost
Subject: Re: kern/38990: Unmounting a disk fails the first time
Date: Sat, 21 Jun 2008 23:23:14 +0000

 On Sat, Jun 21, 2008 at 05:38:48PM -0400, Christos Zoulas wrote:
  > |  >  It's due to the kernel's 'vrele' thread. Sometimes we have to defer
  > |  >  VOP_INACTIVE because of the horrible locking requirements. It always
  > |  >  gets done for layered file systems. The delay is about one second.
  > | 
  > | Is it worth installing a workaround into umount(8)?
  > 
  > It is worth fixing the syscall :-)
 
 Simplifying the locking isn't going to happen before branching.
 Afterwards, though... maybe.
 
 I was thinking that putting in a temporary hack of the form "sync();
 sleep(1); retry" is a lot easier at userlevel, that's all.
 
 -- 
 David A. Holland
 dholland%netbsd.org@localhost