BULLETIN NUMBER: SWB-041

DATE ISSUED: December 6, 2004

DATE CLOSED: December 20, 2004

AFFECTED SYSTEMS: All PowerMAX OS architectures with RAMiX 675 Boards.

RELEASE LEVEL: All PowerMAX OS running with the rmxf package.

EXPLANATION: rmxf does not reset the 675 boards properly when the system 
is shutdown via an init 0 or an init 6. During the next boot, the kernel gets:
 
 - many stray interrupts for the rmxf board, then
 - many garbage lines are output to the console, then
 - the kernel hangs in rmxf_675_drain_outbound_port+0xA0 waiting for the
 board to send back a -1 that signifies that it is finished 
 
This problem is occurring because even though there is a function to reset the
675 rmxf board, the function does not do anything but return to the caller. As a
result the rmxf board is left running even though the kernel is going down. 
When the kernel is rebooted, it receives a bunch of rmxf receive interrupts
before the driver has been initialized properly.
 
RESOLUTION: To workaround this problem, you must use init 0 instead of init 6
AND, you must either powercycle the system, OR, you must issue a PCI reset to
the console using a '~p' after the init 0.
 
The problem has been resolved and verified in the next releases of the rmxf product. 
The releases are 4.3.2, 5.1.2, 6.1.1, and 6.2.1 for PowerMAX OS 4.3, 5.1, 6.1, and 
6.2 respectively. The new releases can be ordered from Software Distribution 
through http://www.ccur.com/isd_support_ordersoftware.asp?r5=1 .