Subject: kern/4783: bha driver messages: mbi not in round-robin order
To: None <>
From: None <cgd@NetBSD.ORG>
List: netbsd-bugs
Date: 01/05/1998 22:57:39
>Number:         4783
>Category:       kern
>Synopsis:       bha driver messages: mbi not in round-robin order
>Confidential:   no
>Severity:       serious
>Priority:       low
>Responsible:    kern-bug-people (Kernel Bug People)
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Mon Jan  5 15:05:01 1998
>Originator:     Chris G. Demetriou
We don't need no stinkin' organization!
>Release:        1.3
System: NetBSD 1.3 NetBSD 1.3 (KICKME) #0: Fri Jan  2 21:38:44 PST 1998 i386

	[ filed as kern because it's an MI driver.  filed as low priority
	because, to the best of my knowledge, it's caused no real problem
	either in terms of system uptime or data integrity.  filed
	as sw-bug because, of all the candidates, i think that's most
	likely. 8-]

	A custom kernel built from 1.3 sources for one of my systems
	occasionally outputs:

		bha0: mbi not in round-robin order

	seemingly at times of heavy use.  (I've not ever been around when
	it's happened, but times from syslog messages indicate that they
	probably happened during daily runs.)  Unfortunately, I can't
	duplicate the problem at will.

	The system was previously running 1.2E (from around May 25, 1997,
	or so), and didn't ever print these messages.

	The boot messages printed for the board are:

bha0 at isa0 port 0x330-0x333 irq 11 drq 5
bha0: model BT-542BH, firmware 3.35
bha0: async, parity
scsibus0 at bha0: 8 targets
sd0 at scsibus0 targ 0 lun 0: <GENERIC, SCSIDRIVE, 6066> SCSI2 0/direct fixed
sd0: 2048MB, 3703 cyl, 9 head, 125 sec, 512 bytes/sect x 4194685 sectors
cd0 at scsibus0 targ 6 lun 0: <TOSHIBA, CD-ROM XM-3501TA, 3054> SCSI2 5/cdrom removable

	(That firmware is new as of ~february or march; I needed to upgrade
	it so that the board would work with my Jaz drive, which isn't
	currently connected to the system.)

	I note that this condition is deemed anomalous enough that it's
	always printed about, when it happens (i.e. it seems that no
	debugging needs to be enabled to get the printouts), so it smells
	a bit like a bug.

	Unknown.  The messages just crop up occasionally on my system,
	during periods of time when i think there's probably relatively
	heavy load (but i've never been around to see them happen).

	Unknown.  (It doesn't look like it would be trivial to drop back
	to a ~may version of the driver.)