From owner-freebsd-bugs  Thu Jul 27 12:59:01 1995
Return-Path: bugs-owner
Received: (from majordom@localhost)
          by freefall.cdrom.com (8.6.11/8.6.6) id MAA04283
          for bugs-outgoing; Thu, 27 Jul 1995 12:59:01 -0700
Received: from Root.COM (implode.Root.COM [198.145.90.1])
          by freefall.cdrom.com (8.6.11/8.6.6) with ESMTP id MAA04277
          for <bugs@freebsd.org>; Thu, 27 Jul 1995 12:58:55 -0700
Received: from corbin.Root.COM (corbin [198.145.90.18]) by Root.COM (8.6.11/8.6.5) with ESMTP id MAA09861; Thu, 27 Jul 1995 12:58:12 -0700
Received: from localhost (localhost [127.0.0.1]) by corbin.Root.COM (8.6.11/8.6.5) with SMTP id MAA00425; Thu, 27 Jul 1995 12:59:28 -0700
Message-Id: <199507271959.MAA00425@corbin.Root.COM>
To: Matt Dillon <dillon@blob.best.net>
cc: bugs@freebsd.org
Subject: Re: More on our shell2 crashes 
In-reply-to: Your message of "Thu, 27 Jul 95 10:30:09 PDT."
             <199507271730.KAA27241@blob.best.net> 
From: David Greenman <davidg@Root.COM>
Reply-To: davidg@Root.COM
Date: Thu, 27 Jul 1995 12:59:27 -0700
Sender: bugs-owner@freebsd.org
Precedence: bulk

>    This time around it crashed with a panic("nfsreq nogrps")

   That's a known bug. It is caused by someone doing the equivilent of
"setgroups(0, &foo)". Old versions of Smail are alleged to do this. We haven't
fixed this bug yet as there are several conflicting opinions about how it
should be fixed. Since setgroups() can only be executed by the Superuser, it
hasn't been a high priority thing to fix.

>    I am also seeing this:
>
>Jul 27 08:40:31 shell2 /kernel: vnode_pager_getpages: I/O read error
>Jul 27 08:40:32 shell2 /kernel: vm_fault: pager input (probably hardware) error,
> PID 690 failure
>Jul 27 08:40:32 shell2 /kernel: vnode_pager_getpages: I/O read error
>Jul 27 08:40:32 shell2 /kernel: vm_fault: pager input (probably hardware) error,
> PID 690 failure
>Jul 27 08:40:32 shell2 /kernel: vnode_pager_getpages: I/O read error
>Jul 27 08:40:32 shell2 /kernel: vm_fault: pager input (probably hardware) error,
> PID 690 failure
>Jul 27 08:40:32 shell2 /kernel: vnode_pager_getpages: I/O read error
>Jul 27 08:40:32 shell2 /kernel: vm_fault: pager input (probably hardware) error,
> PID 690 failure
>Jul 27 08:40:32 shell2 /kernel: pid 690: ftpd: uid 1905: exited on signal 6
>Jul 27 08:40:29 shell2 ftpd[690]: exiting on signal 11
>
>    This type of thing is occuring a whole lot... every couple of hours.
>    It always seems to be coming from FTPD.
>
>    The ftpd executable in this case is *ON* an NFS partition.  While the
>    messages are not really verbose enough, I believe the failure is occuring
>    when shell2 tries to page in some pages from the ftpd executable over NFS.
>
>    Another thing of interest: ftpd is modes 550 root.wheel, meaning that a 
>    screwup in the ucred *could* be causing the read failure.  Thus there
>    could be a relationship with the panic and the vnode_pager_getpages
>    fazilures if ucred is somehow getting corrupted or it's reference count
>    is getting messed up.

   I believe the problem here is that ftpd isn't allowed read permission on
the server, yet the process is "root" on the client. This means that the
client thinks it can exec the binary but the server refuses to allow reads to
the file during the page fault (file read). One work-around for this is to add
"-root=0" as an option in your server's exports file. Another is to fix the
permissions on the file.

-DG