From owner-freebsd-current  Sun Feb  4 11:38:18 1996
Return-Path: owner-current
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.3/8.7.3) id LAA26948
          for current-outgoing; Sun, 4 Feb 1996 11:38:18 -0800 (PST)
Received: from Root.COM (implode.Root.COM [198.145.90.17])
          by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id LAA26943
          for <current@freebsd.org>; Sun, 4 Feb 1996 11:38:15 -0800 (PST)
Received: from localhost (localhost [127.0.0.1]) by Root.COM (8.6.12/8.6.5) with SMTP id LAA00753; Sun, 4 Feb 1996 11:38:21 -0800
Message-Id: <199602041938.LAA00753@Root.COM>
X-Authentication-Warning: implode.Root.COM: Host localhost didn't use HELO protocol
To: KATO Takenori <kato@eclogite.eps.nagoya-u.ac.jp>
cc: current@freebsd.org
Subject: Re: kmem_malloc called at splimp 
In-reply-to: Your message of "Mon, 05 Feb 1996 01:05:41 +0900."
             <199602041605.BAA00550@marble.eps.nagoya-u.ac.jp> 
From: David Greenman <davidg@Root.COM>
Reply-To: davidg@Root.COM
Date: Sun, 04 Feb 1996 11:38:21 -0800
Sender: owner-current@freebsd.org
Precedence: bulk

>The function kmem_malloc() is called at splimp when mbuffer is
>initialized.  Is it safe?
>
>1. The function mbinit() calls m_clalloc() at splimp.
>2. The function m_clalloc() calls kmem_malloc() without changing
>   software interruption mask.
>3. Therefore, kmem_malloc() is called at splimp.
>
>4. The comment of kmem_malloc says that `this still only works in a
>   uni-processor environment and when called at splhigh().'
>                                 ^^^^^^^^^^^^^^^^^^^^^^^^
>Because splimp of  FreeBSD doesn't block disk I/O, above operation may 
>clobber vm map IF disk I/O interruption occurs.  (I have not checked
>whether it occurs or not, yet.)

   Our kernel malloc currently only supports interrupt-time allocations for
network buffers (splimp). We've considered changing this to be safe to use
in all cases, but we haven't done this yet. For this reason, only network
drivers are allowed to use malloc at interrupt time. If you know of a specific
case where a malloc is occuring during a non-network interrupt, please let us
know!

-DG

David Greenman
Core Team/Principal Architect, The FreeBSD Project