From owner-cvs-sys  Mon Feb 20 13:25:36 1995
Return-Path: cvs-sys-owner
Received: (from majordom@localhost) by freefall.cdrom.com (8.6.9/8.6.6) id NAA09060 for cvs-sys-outgoing; Mon, 20 Feb 1995 13:25:36 -0800
Received: from godzilla.zeta.org.au (godzilla.zeta.org.au [203.2.228.19]) by freefall.cdrom.com (8.6.9/8.6.6) with ESMTP id NAA09023; Mon, 20 Feb 1995 13:25:15 -0800
Received: (from bde@localhost) by godzilla.zeta.org.au (8.6.9/8.6.9) id IAA31791; Tue, 21 Feb 1995 08:19:29 +1100
Date: Tue, 21 Feb 1995 08:19:29 +1100
From: Bruce Evans <bde@zeta.org.au>
Message-Id: <199502202119.IAA31791@godzilla.zeta.org.au>
To: bde@freefall.cdrom.com, davidg@Root.COM
Subject: Re: cvs commit: src/sys/i386/isa icu.s
Cc: CVS-commiters@freefall.cdrom.com, cvs-sys@freefall.cdrom.com
Sender: cvs-sys-owner@freebsd.org
Precedence: bulk

>>  Fix restoring cpl in the swi_ast_phantom case.  The system sometimes
>>  returned to user mode without enabling ASTs.  The problem fixed itself

>   Hmmm...this sounds like the same bug that is (was?) causing Larry McVoy's
>bw_tcp benchmark to 1) hang, and 2) after interrupting it the process would
>hang in the kernel bouncing between the low level AST processing and the upper
>layer TCP processing code. This didn't use to happen a few months ago - was
>something changed to cause this?

Not that I know of.  The bug is very old - it is in 1.1.5 according to
the sources.  lmbench seemed OK (except it doesn't automatically scale
to < 32MB memory) when I ran it here on Nov 29.

Bruce