Subject: kern/12800: Possible bug in aha driver with new scsipi-code
To: None <>
From: None <>
List: netbsd-bugs
Date: 05/01/2001 20:06:56
>Number:         12800
>Category:       kern
>Synopsis:       page fault in supervisor mode with aha in -current/i386
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Tue May 01 11:07:00 PDT 2001
>Originator:     Tom Ivar Helbekkmo
>Release:        NetBSD-current 2001-04-30
NetBSD 1.5U NetBSD 1.5U (ATHENE) #2: Tue May 1 14:06:55 CEST 2001 i386
Architecture: i386
Machine: i386
It seems to me the most common panic on the i386 port is a UVM panic
that says there was a page fault in supervisor mode.  :-) With a very
current current, I get this when the kernel probes for devices on an
AHA-1542CF controller.  Removing the code (commenting out the aha line
from the config file) removes the problem.  The panic occurs after the
kernel has written the line stating that it's pausing to allow devices
on that bus to settle, but before any other output occurs.  Without
support for the AHA controller compiled in, the kernel is stable, as
Very repeatable by booting with an AHA controller + devices, and a
kernel that supports them.
Not known.  I'm guessing it's related to the introduction of the new
scsipi code, since the problem definitely was not there three weeks
earlier; the last time I updated this machine.