[kernel] bug#148: Kernel hangs in do_gettimeofday

X-PA-RISC Linux-PR-Message: report 148
X-PA-RISC Linux-PR-Package: kernel
X-Loop: daniel_frazier@hp.com
Received: via spool by bugs@bugs.parisc-linux.org id=B.100354065018587
          (code B ref -1); Sat, 20 Oct 2001 01:18:01 GMT
Date: Fri, 19 Oct 2001 18:17:23 -0700
From: Randolph Chung <randolph@tausq.org>
To: submit@bugs.parisc-linux.org
Message-ID: <20011019181723.T537@tausq.org>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
User-Agent: Mutt/1.3.20i
X-PGP: for PGP key, see http://www.tausq.org/pgp.txt
X-GPG: for GPG key, see http://www.tausq.org/gpg.txt

Package: kernel
Version: 2.4.9-pa58
Severity: serious

There appears to be a bug in the timer handling functions such that in
do_gettimeofday, gettimeroffset sometimes returns a negative value that
is interpreted as a huge number by the caller and causes the kernel to
spin.

I can reproduce this consistently with the following steps:
On a c3k running a 64-bit kernel:
1. compile with CONFIG_MODULES turned on
2. run "nmap -O localhost" as root

For some odd reason this doesn't happen when CONFIG_MODULES is not
turned on.

jsm has volunteered to look into this more. :)

more details at
http://lists.parisc-linux.org/pipermail/parisc-linux/2001-October/014215.html

randolph
-- 
   @..@                                         http://www.TauSq.org/
  (----)
 ( >__< )
 ^^ ~~ ^^