From owner-freebsd-bugs  Mon Oct 21 12:41:41 1996
Return-Path: owner-bugs
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.5/8.7.3) id MAA06192
          for bugs-outgoing; Mon, 21 Oct 1996 12:41:41 -0700 (PDT)
Received: from godzilla.zeta.org.au (godzilla.zeta.org.au [203.2.228.19])
          by freefall.freebsd.org (8.7.5/8.7.3) with ESMTP id MAA06187
          for <bugs@freebsd.org>; Mon, 21 Oct 1996 12:41:38 -0700 (PDT)
Received: (from bde@localhost) by godzilla.zeta.org.au (8.7.6/8.6.9) id FAA26416; Tue, 22 Oct 1996 05:37:19 +1000
Date: Tue, 22 Oct 1996 05:37:19 +1000
From: Bruce Evans <bde@zeta.org.au>
Message-Id: <199610211937.FAA26416@godzilla.zeta.org.au>
To: bugs@freebsd.org, hosokawa@mt.cs.keio.ac.jp
Subject: Re: type of dk_flags (wd.c) should be longer
Sender: owner-bugs@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

>In sys/i386/isa/wd.c:
>
>        short   dk_flags;       /* drive characteristics found */
>#define DKFL_SINGLE     0x00004 /* sector at a time mode */
>#define DKFL_ERROR      0x00008 /* processing a disk error */
>#define DKFL_LABELLING  0x00080 /* readdisklabel() in progress */
>#define DKFL_32BIT      0x00100 /* use 32-bit i/o mode */
>#define DKFL_MULTI      0x00200 /* use multi-i/o mode */
>#define DKFL_BADSCAN    0x00400 /* report all errors */
>
>It seems bit operation to dk_flags results in invalid behavior.  I
>think that the type of dk_flags should be longer (ex. long).

Erm, 0x400 fits in only 11 bits.  Perhaps dk_flags should be `int'
since accesses to shorts are slow.  It doesn't matter much, because
most accesses are optimized to byte accesses (and those that aren't
are pessimizations - (du->dk_flags & BIT) is optimized to a testb
inctruction, and (du->dk_flags |= BIT) is pessimized to an orw
instruction).

Bruce