Subject: port-mac68k/29179: uvm assertion triggered very early at boot
To: None <port-mac68k-maintainer@netbsd.org, gnats-admin@netbsd.org,>
From: Julio M. Merino Vidal <jmmv@menta.net>
List: netbsd-bugs
Date: 01/31/2005 23:55:00
>Number: 29179
>Category: port-mac68k
>Synopsis: uvm assertion triggered very early at boot
>Confidential: no
>Severity: critical
>Priority: high
>Responsible: port-mac68k-maintainer
>State: open
>Class: sw-bug
>Submitter-Id: net
>Arrival-Date: Mon Jan 31 23:55:00 +0000 2005
>Originator: jmmv@menta.net (Julio M. Merino Vidal)
>Release: NetBSD 2.99.15
>Organization:
>Environment:
System: NetBSD dawn.local 2.99.15 NetBSD 2.99.15 (DAWN) #4: Sat Jan 29 13:28:58 CET 2005 jmmv@dawn.local:/var/build/obj/home/jmmv/NetBSD/src/sys/arch/i386/compile/DAWN i386
Architecture: i386
Machine: i386
>Description:
Booting a very current kernel on a mac68k dies after printing the
available memory. It triggers the assertion in uvm/uvm_map.c,
line 472 as of version 1.183:
KASSERT(!VM_MAP_USE_KMAPENT(map) ||
(old_entry->flags & UVM_MAP_QUANTUM) || !UMR_EMPTY(umr));
This happens as a result of the call to uvm_map_protect in
sys/arch/mac68k/mac68k/machdep.c. I've seen other architectures
(mostly all *68k) do this too, so they are probably also affected
by the same problem.
I suspect this is caused by the changes introduced by yamt@ in
revision 1.174 of uvm_map.c (well, at least the assertion is new
in that revision, aside a lot of other code). This may be either
because there really is a problem or because they are effectively
catching a bug :)
I'm marking this as a critical issue because the installation
kernels come with DIAGNOSTIC enabled, so they cannot boot.
>How-To-Repeat:
Boot a recent mac68k kernel. I'm trying on a Performa 630 with
a 68LC040 chip, but that shouldn't matter. (In case it does,
I will provide any other required information.)
>Fix:
Unknown.
>Unformatted:
To: gnats-bugs@gnats.NetBSD.org
Subject:
From: jmmv@menta.net
Reply-To: jmmv@menta.net
X-send-pr-version: 3.95