There was no response to this note on the neomagic list, so I thought I might
forward it to xpert. The XFree86 developers may be interested to know that the
Cardbus bridge code in the CVS head causes an irretrievable loss of the
console on FreeBSD 4.6-RELEASE on my Sony SR7K laptop.
#ifdef'ing out the Cardbus bridge switch cases allows the CVS head to work for
me once more.
Is there interest in committing an #ifdef so that FreeBSD 4.X doesn't get the
Cardbus bridge code? 4.X doesn't support Cardbus, which is likely the root of
the hang.
-Brad, [EMAIL PROTECTED]
------- Forwarded Message
Return-Path: [EMAIL PROTECTED]
Delivery-Date: Thu Jun 27 17:03:21 2002
Return-Path: <[EMAIL PROTECTED]>
Received: from wyvern.icir.org (wyvern.icir.org [192.150.187.14])
by carp.icir.org (8.11.6/8.11.3) with ESMTP id g5S03Lf47543
for <[EMAIL PROTECTED]>; Thu, 27 Jun 2002 17:03:21 -0700 (PDT)
(envelope-from [EMAIL PROTECTED])
Received: from ICSI.Berkeley.EDU (fruitcake.ICSI.Berkeley.EDU [192.150.186.11])
by wyvern.icir.org (8.11.6/8.11.6) with ESMTP id g5S03Kq29523
for <[EMAIL PROTECTED]>; Thu, 27 Jun 2002 17:03:20 -0700 (PDT)
(envelope-from [EMAIL PROTECTED])
Received: from carp.icir.org (carp.icir.org [192.150.187.71])
by ICSI.Berkeley.EDU (8.9.0/8.9.0) with ESMTP id RAA20291
for <[EMAIL PROTECTED]>; Thu, 27 Jun 2002 17:03:20 -0700 (PDT)
Received: from carp.icir.org (localhost [127.0.0.1])
by carp.icir.org (8.11.6/8.11.3) with ESMTP id g5S03Ff47532;
Thu, 27 Jun 2002 17:03:15 -0700 (PDT)
(envelope-from [EMAIL PROTECTED])
Message-Id: <[EMAIL PROTECTED]>
X-Mailer: exmh version 2.5 07/13/2001 with nmh-1.0.3
From: Brad Karp <[EMAIL PROTECTED]>
To: [EMAIL PROTECTED]
Cc: [EMAIL PROTECTED]
Subject: found problem: CVS head on FreeBSD-4.6 hangs hard
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Date: Thu, 27 Jun 2002 17:03:14 -0700
Sender: [EMAIL PROTECTED]
I posted recently that on my Sony SR7K (NeoMagic NM2380) running FreeBSD
4.6-RELEASE, a build of XFree86 from the CVS head of 6/13/02 locks the machine
hard; the machine doesn't respond to VT switches anymore, and the console in
unrecoverable, though Ctrl-Alt-DEL succeeds in triggering a clean shutdown
(with no further console output).
Egbert Eich suggested I diff through sources to find the commit that causes
the breakage.
Done.
Revision 3.51 of xc/programs/Xserver/hw/xfree86/common/xf86pciBus.c (committed
on 4/4/02 by eich) introduces code to notice CardBus bridges and treat them
specially (several switch statements with new PCI_SUBCLASS_BRIDGE_CARDBUS
cases in them).
It's this CardBus-specific code that causes the hang; if I #ifdef out all such
CardBus-specific cases, the build runs as expected.
Note that it's *not* {save,restore}PciCardBusState() that cause the hang; with
the code in those commented out, the hang still occurs.
Under FreeBSD 4.X, there is no CardBus support; only NEWCARD in 5.X supports
CardBus. It's likely that diddling the CardBus controller under 4.X ist streng
verboten.
Suggestions:
- Try it on a 5.X machine (I can't; don't have time to reinstall the
OS) with NEWCARD CardBus support, to see if it works there.
- #ifdef out the CardBus code for FreeBSD 4.X, and for 5.X, too, if
it fails on 5.X. This code likely causes the CVS head of XFree86
to fail on all FreeBSD 4.X laptops.
- -Brad, [EMAIL PROTECTED]
------- End of Forwarded Message
_______________________________________________
Xpert mailing list
[EMAIL PROTECTED]
http://XFree86.Org/mailman/listinfo/xpert
