Diags Report A Failure On 82596 CHIP
EtherExpress Flash 32 adapter
Diagnostics (v1.0) reports an incorrect failure of the 82596 in certain
machines. This message is seen occasionally with 60- or 66-MHZ
Problem with the 82596 self test in the diagnostic program. The drivers
are not affected by this message.
ERROR MESSAGE: "DISK BOOT FAILURE", OR "BOOT DEVICE NOT FOUND" WITH ADAPTEC SCSI DISK CONTROLLER.
Obtain the updated FL32DIAG v1.1 diagnostic program. This file
FL32DIAG.EXE is available through Intel's online services.
EtherExpress Flash32 adapter
Error Message: "Disk Boot Failure", or "Boot Device not found" with
Adaptec SCSI disk controller.
EISA CONFIGURATION OR HARD DISK APPEARS TO BE DAMAGED BY FLASHSET
Installing the Flash32 with SCSI disk driver (ie Adaptec 1740). With
Adaptec's controllers, when there has been any EISA configuration
change, the ROM (default location CC000h) will not initialize the SCSI
Continue so that FlashSet comes up. It will inform you that it detects
an Adaptec and that it has to reboot the computer. FlashSet will
configure all Flash32s in the system NVRAM and reboot, setting the
FlashSet bit to run again. This will then give you a normal boot
sequence, bringing up FlashSet so you can modify the configuration.
EISA configuration or hard disk appears to be damaged by FlashSet
An Adaptec SCSI controller can be the cause of this problem. You install
a Flash32 and move another card in the process (to make room.). You get
the "boot failure" message. Instead of pressing F1 to continue, you
power down and remove the Flash32. You still get the boot failure
message (because the card you moved is in a different slot and is still
creating a configuration error.)
Another cause could be that the NVRAM was incorrectly read into memory
FLASHSET SCREEN LOOPING WITH MESSAGE THAT FLASHSET WAS SUCCESSFUL.
Run the ECU from a floppy. If a floppy isn't available, EISA machines
generally have a switch or jumper to enable the default configuration.
If your most recent configuration is on the hard drive, usually the
default configuration (via floppy or motherboard switch) has the hard
drive configured. After configuring the system so it recognizes the hard
drive, run the ecu from the hard drive.
FlashSet screen looping with message that FlashSet was successful.
- The bios of the PC conflicts with int 19 which the Flash uses on boot
- FlashSet runs when it detects there is a configuration error. So, if
there is a card in the machine that creates a configuration error and it
is NOT a Flash32, FlashSet will run over and over and over. This is
because Flashset configures all the Flash32 cards, but cannot configure
the other card(s), so there is STILL a configuration error.
This scenario could occur if the user installs more than one card at the
same time (our Flash32 and a modem card for example). Just moving an
existing card to another slot could also cause the problem.
Netware* Server Hanging, Workstations Lose Connection To EtherExpress Flash32 Adapter
- For bios related problem, contact the computer manufacturer or your
Network Services Supplier for updated bios.
- Install the Flash32(s) first, boot and run Flashset, then install
other cards and run the ECU. The user could also run the ECU before
installing any cards, and disable the flash on the Flash32 cards to
prevent Flashset from running.
NetWare* server hanging, workstations lose connection to Flash32.
This problem could be caused by long wait states on the EISA bus. The
situation occurs when the PC gives long wait states when the memory
controller is doing some cache line manipulations while a bus master
owns the EISA bus. These long wait states cause the Flash32 to hold the
bus too long and generate an EISA busmaster time out.
There are a two options to try.
For NetWare server driver, the syntax would be as follows.
- Check the CMOS of the PC and look for settings for WRITE BACK CACHE
or WRITE THROUGH CACHE. Enable the WRITE THROUGH. Changing this
parameter should eliminate the problem of long wait states, although it
may increase CPU utilization..
- Another method to avoid the bus time out problem is by using ON/OFF
parameters for the Flash32 to get it on and off the EISA bus earlier.
The ON/OFF timers provide a throttling mechanism to control how the
82596 requests bus ownership. The units for the number is local
clock periods (25MHz clock). The ON parameter limits the number of clock
cycles that the card will hold the bus for on data transfers. The
default value is 200. The OFF parameter sets the minimum number of clock
cycles that the card will remain off the bus between data transfers.
The default value is 2. Thus a low ON number causes the 596 to transfer
data in a series of short bus-ownership cycles (arbitration cycles)
rather than fewer, longer transfers. Although this may seem less
efficient, the individual data transfer cycles actually run at the same
speed either way, so there is very little affect on throughput. One may
have to try different values for these numbers. Suggested values are
ON=10, OFF=10 or ON=20, OFF=2
LOAD FL32 SLOT=3 ON=10 OFF=10
Note for Multiple Frame Types:
The FL32.LAN driver will first parse the command line parameters to
obtain the ON/OFF values but the given parameters are used to initialize
the 82596 only for the first instance of the driver (the first frame
This is true for both v5.23 and v6.02. In v6.02, however, the
DriverReset routine will reinitialize the network chip with the stored
parameter values each time it is called, in which case, the last stored
parameter values (the values given for the last driver instance) will be
used to initialize the chip.
ON/OFF values should ONLY be given for the first frame type loaded.
Prompts For Driv And Glob Paramters When Loading The Netware* Server Driver
Prompts for DRIV and GLOB parameters when loading the NetWare* server
The incorrect slot number is specified on the load line.
Error Message When Loading Flash32 Netware* Server Driver: Fl32-Nw-091: The Hardware Configuration Conflicts Module Initialization Failed
Verify the slot number that the Flash32 is installed in and specify that
on the load line for the server driver.
Error Message when loading Flash32 NetWare* server driver: FL32-NW-091:
The hardware configuration conflicts Module Initialization failed
Loading the driver with a name whose first four letters are "inte"
cause the error message.
App Notes: Etherexpress Flash32 Adapter And Adaptec SCSI
Remove the "e" or change it to another letter and the driver will load.
Here is a summary of how the Flash32 adapter, Adaptec SCSI disk
controller and EISA configuration interact. The behavior described
applies to Adaptec 1740 and 2740 disk controllers, hereafter referred to
as "Adaptec controllers".
Adaptec controllers have a BIOS ROM located at CC000h (default). This
ROM is executed every time you boot. If there has been any sort of EISA
configuration change (card added, moved, or removed), the ROM does not
initialize the SCSI controller, and you will get a "boot failed, press
F1 to continue, F2 for setup" type of message (message text will vary
from system to system). This behavior occurs regardless of type of
adapters installed in the machine. Because the Adaptecs work this way,
any EISA configuration change forces you to run the ECU from a floppy.
If you insert one or more Flash32 adapter(s) into a system that includes
an Adaptec controller, here's what happens:
Error: Manchester Encoder Chip Failure
- The Adaptec ROM runs first. On 2740s, you get a "BIOS not
installed!" error and a "press F1 to continue" type message because
the addition of the Flash32(s). On the 2740, if you must press F1
to continue. The 1740 tested continued with no key press.
- Now Flashset runs. Flashset detects the presence of any Adaptec
controller (EISA ids beginning with the letters ADP). It puts up
a message explaining the it has detected an Adaptec controller and
that it is going to reboot. Flashset then configures all the
Flash32 adapters in the system and updates the EISA configuration
NVRAM. It reboots the system, and it has set the option to have
FlashSet run again. Why does it reboot? To get the Adaptec
controller initializes so that a normal hard disk boot can proceed
after Flashset runs a second time (step 4 and 5 below).
- The system reboots, and the Adaptec ROM runs again. This time
there is no configuration error, and the Adaptec initializes
- Now Flashset runs again, normally this time. It runs normally
because there is no configuration error. Flash32 Flashset uses the
presence of an Adaptec controller and a configuration error to
trigger the special behavior described in step 2. If there is an
Adaptec and a configuration error, you get the "reboot" behavior in
step 2. If there is no configuration error, or no Adaptec, you get
- Now (still in Flashset), you save the configuration, and if you
tell Flashset not to run again, you will not see it at the next
Product: EtherExpress 32 adapter
Other: DOS 6.2
Using a workstation running DOS 6.2, the EtherExpress 32 card fails the
internal loopback diags. Running diags again without rebooting, the
card will pass. Run once more, the card will fail again. During
testing, the workstation was able to connect to the network even though
it failed diags. The message "The board's Manchester encoder chip is
not working correctly" is displayed.
Error: Busmaster Interface Chip Failed
Boot the workstation with a different version of DOS to run diags.
This is an error message from the EtherExpress 32 network adapter
diagnostics. In some systems this error is confined to the diagnostics
and the board will work fine when you load the driver. In other
systems, the driver will load, but the system will not be able to
communicate with the network. If possible, we recommend you resolve
this problem before loading the driver.
There are several possible causes and Solutions.
DOS 5.0 / DOS 6.x Problem
The memory management for DOS 5.0 or 6.0 appears to cause a conflict
with the Diagnostic software for the EtherExpress 32 adapter. You can
correct this conflict by loading HIMEM.SYS in the CONFIG.SYS file.
Not in a Bus Master Slot
Not all EISA slots are bus mastering slots. Try another slot.
The 82586 is failing. Replace the board.
Error: Address Prom Has Incorrect Node Address
This error occurs when loading the driver. The most common cause is
loading the wrong driver or the driver is in conflict with another
Loading wrong driver
Novell* writes the NE3200 drivers that Intel includes on the
EtherExpress 32 adapter disk. Novell's older versions of the NE3200
driver (below v3.15) do not work with the EtherExpress 32 card. These
older versions of the driver check the EEPROM for a Novell ethernet node
address. If they do not find Novell's node address, they return this
error. To correct the problem, load the NE3200 driver (v3.18) included
on the EtherExpress 32 network adapter diskette. You should copy the
driver to your C: partition and change the autoexec.ncf to load the
driver from C:.
Conflict with another NLM
Customers have reported that if they load Pserver before the
EtherExpress 32 network adapter driver this error occurs. Loading the
driver before loading Pserver corrects this problem.
Using EtherExpress 32 Network Adapters With Odi Workstations
There are two problems that have been seen with the EtherExpress 32
cards and the ODI driver.
One of the problems is with the EtherExpress 32 network adapter DOS ODI
driver and applications that use extended memory or memory managers
(i.e. Microsoft Windows 3.x). Previous releases of this driver would
not protect memory regions from the above mentioned applications. This
could cause files that the EtherExpress 32 card stored in memory to be
overwritten (corrupted) if an application wrote to that address space.
The problem has been seen most often with shared Windows installations.
The Symptom: is corrupted files. There have been instances of printing
problems from Windows*.
Another problem is that when using the EtherExpress 32 network adapter
DOS ODI driver, a print job is sent to the print queue and it indicates
that 65,000 copies of the print job are to be printed.
There is an updated ODI driver, file E32NDISK.EXE. It contains NE3200.COM
(dated 5/4/92), NET.CFG and README.TXT. The keyword DOUBLE BUFFER must
be included in the NET.CFG with this version of the driver. This
version of the driver defaults to an 802.2 frame type. If a different
frame type is needed, a FRAME statement should be added in the NET.CFG
file includes a sample NET.CFG for both of these cases. These are
available through Intel's online services.
Example NET.CFG file:
LINK DRIVER NE3200
PROTOCOL IPX 0 ETHERNET_802.3
BIND NE3200 Bind 1, if using IPXODI v2.x or later
We recommend that everyone using the EtherExpress 32 card in an ODI
workstation use this (or later) ODI driver.
In a NET.CFG file, use a single space or tab for indents. The slot
number should match the slot in the EISA configuration file. Make sure
the NET.CFG file is located in the same directory as the driver and that
there is not a second NET.CFG in the root directory.
For more information about creating and formatting a NET.CFG file, see
your Novell NetWare* documentation.
Problems Logging In With An EtherExpress 32 Card In The Server
EtherExpress 32 LAN Adapter
Other HARDWARE/SOFTWARE: NetWare* 3.11, NE3200.LAN v3.18
Intermittent problems where users are not able to log in to the network
if an EtherExpress 32 card is in a NetWare 3.11 server.
- Transmit errors may cause the driver to hang. The packet receive
buffer will continue to increase because packets are coming in,
but the board will stop transmitting. These transmit packets will
pile up in the queue. You may also see the communication buffers
- Another busmaster card with more than one EtherExpress 32 card in
the computer can cause problems. A "channel" conflict between the
two busmaster cards may cause the EtherExpress 32 card to hang.
One of the more common cards that can cause this problem is the
Adaptec 1740 SCSI controller.
The NE3200.LAN driver v3.24 fixes these problems. Order document 6325
for instruction on installing v3.24 of this driver.
This is a self extracting archive containing the current production
diskette for the Intel EtherExpress 32. (disk p/n 304510-003)
The following software has changed from the previous production disk:
- Added OS/2 ODI driver for NetWare workstations.
- NetWare 3.11 server driver: Fixed a timeout problem with fast
and 80486 based computers.
- EISA configuration file: Now supports sharing interrupts.
- NDIS OS/2 and DOS drivers: Added LAN Manager 2.1 support.
LANDesk* Express software v2.60.
Windows-based network management tool that lets you monitor any DOS or
Windows* workstation attached to a NetWare network, without leaving your
Contains the diagnostics program for the EtherExpress Flash32 adapter
This is a self extracting archive containing the current production
diskette for the Intel EtherExpress Flash 32 LAN adapter. (Software p/n
651916-001, SW Release 3.0) The software drivers on this disk are:
Novell* DOS ODI ver 4.07; Novell OS/2 ODI ver 4.01; Novell 4.x server ver
6.11; Novell 3.1X server ver 3.25; NDIS 2.0.1 DOS & OS/2 ver 7.11; Win
NT 3.5 and WFW 3.11 ver 1.09.
This applies to: