[5712] in testers
Re: patch release breaks Sun dev environment for non-updated
daemon@ATHENA.MIT.EDU (Greg Hudson)
Mon Mar 8 17:02:27 2004
From: Greg Hudson <ghudson@MIT.EDU>
To: Robert A Basch <rbasch@mit.edu>
Cc: testers@mit.edu, rel-eng@mit.edu
In-Reply-To: <1078778426.2729.14.camel@anhedonia.mit.edu>
Content-Type: text/plain
Content-Transfer-Encoding: 7bit
Message-Id: <1078783321.25713.78.camel@error-messages.mit.edu>
Mime-Version: 1.0
Date: Mon, 08 Mar 2004 17:02:02 -0500
On Mon, 2004-03-08 at 15:40, Robert A Basch wrote:
> ...or maybe just change the directive to include the explicit
> path "/os...". This seems slightly easier, and safer (it still
> works in the unlikely event that rwstlock.h changes).
That seems fine.
> If we decide to go this route, we'll need to find all the places
> where all of the new headers are included.
I think the only real problem case is <sys/vnode.h> including
<sys/rwstlock.h>. The only other potential problem is
<sys/lgrp_user.h>, which affects the following headers (indentation used
to reflect how things are affected indirectly, e.g. <sys/cyclic.h>
includes <sys/cpupart.h>):
sys/cpupart.h
sys/cyclic.h
sys/cyclic_impl.h
sys/lgrp.h -- only if _KERNEL or _KMEMUSER defined
vm/seg_spt.h
vm/seg_vn.h
sys/fs/tmpnode.h
sys/exec.h
None of those seem likely to be used by any significant amount of
user-space code, so I'd like to restrict the hacking to <sys/vnode.h>.
