iBook and ATI video

So I installed ATI Displays for Mac OS X version 4.5.6….

Little did I know….

My iBook (G4 late 2004, 1.2Ghz 12″) started randomly locking up. Easiest way to trigger the lockup was by starting up matlab.

After some panicing on my part I’ve decided that maybe it’s my RAM that is failing – matlab tends to gobble up all the RAM you have and then some. Of course this happened just when I urgently needed to graph something that was due in a couple of hours.

I’ve re-seated the RAM in the iBook,
downloaded and installed memtest for Mac OS X , booted into single user mode (CMD-S on bootup), and run memtest all 3 -L.

Memtest took forever (1 gig stick + quarter gig on board), but didn’t find any problems.

So while in single user mode, I’ve started looking into /var/log/system.log.

Found this little gem:

Nov 23 11:53:32 gilva kernel[0]: ** ASIC Hang Log Start **
Nov 23 11:53:32 gilva kernel[0]: 0x01005c63 4f000217 00000007 00000003
Nov 23 11:53:32 gilva kernel[0]: 0x0200a859 c0001c04 00000002 00000008
Nov 23 11:53:32 gilva kernel[0]: 0x00004443 01e1f827 00000e0e 80010140
Nov 23 11:53:32 gilva kernel[0]: 0x4000ffff 001e0000 51b3a220 72001005
Nov 23 11:53:32 gilva kernel[0]: 0x080a0f00 00000000 040100f8 80000003
Nov 23 11:53:32 gilva kernel[0]: 0x0008bbbb 00000002
Nov 23 11:53:32 gilva kernel[0]: 0:0x000101ce
Nov 23 11:53:32 gilva kernel[0]: 1:0x10014020
Nov 23 11:53:32 gilva kernel[0]: 2:0x00000002
Nov 23 11:53:32 gilva kernel[0]: 3:0x000101ce

[about 1020 lines more of similar kernel messages]

Nov 23 11:53:42 gilva kernel[0]: 1021:0xffffffff
Nov 23 11:53:42 gilva kernel[0]: 1022:0xffffffff
Nov 23 11:53:42 gilva kernel[0]: 1023:0xffffffff
Nov 23 11:53:42 gilva kernel[0]: 0x56500bb3
Nov 23 11:53:42 gilva kernel[0]: ** ASIC Hang Log End **
Nov 23 11:53:42 gilva kernel[0]: ATIRadeon::submit_buffer: Overflowed block waiting for FIFO space.   Have 5, need 6. RBBM_STATUS 0x80010140. VAP_CNTL_STATUS 0x00000002
Nov 23 11:53:53 gilva kernel[0]: ** ASIC Hang Log Start **
Nov 23 11:53:53 gilva kernel[0]: 0x01005c63 4f000217 00000007 00000003
Nov 23 11:53:53 gilva kernel[0]: 0x0200a859 c0001c04 00000002 00000008
Nov 23 11:53:53 gilva kernel[0]: 0x00004443 01e1f827 00000e0e 80010140
Nov 23 11:53:53 gilva kernel[0]: 0x4000ffff 001e0000 51b3a220 72001005
Nov 23 11:53:53 gilva kernel[0]: 0x080a0f00 00000000 040100f8 80000003
Nov 23 11:53:53 gilva kernel[0]: 0x0008bbbb 00000002
Nov 23 11:53:53 gilva kernel[0]: 0:0x000101ce
Nov 23 11:53:53 gilva kernel[0]: 1:0x10014020
Nov 23 11:53:53 gilva kernel[0]: 2:0x00000002
Nov 23 11:53:53 gilva kernel[0]: 3:0x000101ce
Nov 23 11:53:53 gilva kernel[0]: 4:0x10016020

[ ditto ]

Nov 23 11:54:03 gilva kernel[0]: 1021:0xffffffff
Nov 23 11:54:03 gilva kernel[0]: 1022:0xffffffff
Nov 23 11:54:03 gilva kernel[0]: 1023:0xffffffff
Nov 23 11:54:03 gilva kernel[0]: 0x56500bb3
Nov 23 11:54:03 gilva kernel[0]: ** ASIC Hang Log End **
Nov 23 11:54:03 gilva kernel[0]: ATIRadeon::submit_buffer: Overflowed block waiting for FIFO space.   Have 5, need 6. RBBM_STATUS 0x80010140. VAP_CNTL_STATUS 0x00000002
Nov 23 11:54:15 gilva kernel[0]: ** ASIC Hang Log Start **
Nov 23 11:54:15 gilva kernel[0]: 0x01005c63 4f000217 00000007 00000003
Nov 23 11:54:15 gilva kernel[0]: 0x0200a859 c0001c04 00000002 00000008
Nov 23 11:54:15 gilva kernel[0]: 0x00004443 01e1f823 00000e0e 80010140
Nov 23 11:54:15 gilva kernel[0]: 0x4000ffff 001e0000 51b3a220 72001005
Nov 23 11:54:15 gilva kernel[0]: 0x080a0f00 00000000 040100f8 80000003
Nov 23 11:54:15 gilva kernel[0]: 0x0008bbbb 00000002
Nov 23 11:54:15 gilva kernel[0]: 0:0x000101ce
Nov 23 11:54:15 gilva kernel[0]: 1:0x10014020
Nov 23 11:54:15 gilva kernel[0]: 2:0x00000002
Nov 23 11:54:15 gilva kernel[0]: 3:0x000101ce
Nov 23 11:54:15 gilva kernel[0]: 4:0x10016020
[...]

Based on what I understand, right now the problem is either in the hardware of the system (I’ll run the system through AHT paces once I make it home), or in the ATI Displays driver.

Joys. Somehow my bet is on ATI Displays kernel drivers being buggy. This agrees with me:

On Dec 29, 2004, at 9:37 AM, Avelino Santa Ana Jr. wrote:

Dec 27 21:52:27 localhost kernel: ATIRadeon::submit_buffer: Overflowed block waiting for FIFO space. Have 4, need 6. RBBM_STATUS 0x80116100. VAP_CNTL_STATUS 0xd4f40002


Howard Shere
http://blogs.greendragon.com/index.php/gdc
Altair 8800a to Mac OS X so far...

Hi,
Are you using vertex programs in your port? What hardware are you using? Is it an R200 based chipset (Radeon 8500-9200)? If so, I recall seeing similar issues with a couple of games I beta tested. The problem had to do with the R200 drivers and vertex programs. I'm not sure how the developers solved them.

I just glanced in the archives and Nicholas Francis noted a cause in his app (fog and vertex programs). This was in May 18, 2004 "Re: ARB vartex program crashbug"

[...]

I’ve run 4.5.1 for months with no problems, so I guess I will be reverting to that version, and will see if the problem went away.

Update 20051207: I broght iBook home that day, and run the extended tests using Apple Hardware Diagnostics disk. It passed all tests without any problems, although a gig of Kingston RAM took about 20 minutes to check.

Once I booted the system back up, it locked up about a minute after loging in.
I rebooted, and it locked up at the blue screen that shows for a few seconds while various services are loading (first graphic screen, essentially).

At that point I were still thinking that maybe something is wrong with my install of the OS.

I tried booting from a bootable CD of 10.3.4 (Some repair disk I downloaded a good long while ago, that essentially was starting up, creating RAM disk, and starting 10.3.4 to Finder, with a few diagnostic applications and Terminal.app), but system also locked up as soon as graphic mode started. By this point I knew that the probem is not with the OS. When I booted from the 10.3.5 install DVD that was originally shipped with the system, and it locked up, I knew for certain that the problem is with the video circuitry on the logic board.

I broght it to my local authorized service provider (who gets no link and remains nameless, because they are not great), their technican checked it out, confirmed that the logic board is malfunctioning, and ordered a replacement logic board. At this time system is less then a year old (bought late december 2004).

Cisco: Disabling CiscoPro only sofware loads

Introduction

Based on what I understand, CiscoPro was Cisco’s attempt at selling routers with crippled OS loads with discounts through resellers. It was the case in the late 90s, and certain models of routers (100x, 25xx, 45xx, 16xx) were available in the olive green of Cisco, and in white. White boxen were the CiscoPro variety, and would only run so called “CiscoPro” featuresets only. By IOS release 11.2 Cisco abandoned that practice, and offered a way to convert all existing CiscoPro routers into routers that would accept standard Cisco IOS load.

Hardware-wise both CiscoPro white boxes and reguar Cisco boxes in olive green are identical, and the only difference is in the non-volatile memory.

As most CiscoPro units still in the field have been converted, I’ll try to document what needs to be done to a CiscoPro router to make it happier, and potentially useful. As it happened, I got my hands on two Cisco routers with white paintjob – CPA2503 (sh ver below) and CPA2520, which correspond to Cisco 2503 and 2520.

Basic Idea

Expect CiscoPro router to have much less flash and RAM then Cisco lists as standard for the model in question. If this is a router you will use for CCNA tests, do shell out some money for flash and RAM upgrades, as they will be necessary to run versions of IOS that you can test on.

Grab the aproprite upgrade file from ftp://ftp-sj.cisco.com/pub/access/ and a copy of the upgrade image. For CiscoPro CPA25xx routers you need cpa25-up.bin.

Configure the route so it can talk to your tfpt server (DO password recovery if needed, it’s on Cisco web site), back up existing IOS, copy the cpa-25-up.bin into the flash of the router, and get it to reboot with that image. cpa25-up.bin will modify the non-volatile memory to remove the CiscoPro only requirement, and which point you can boot from the ROM, and flash in a modern IOS (or at least IOS that you backed up initially).

Step by Step procedure for Windows users

If you boot up a Cisco, and at “sh ver” it shows the following:

SDN-Link>sh ver
Cisco Internetwork Operating System Software
IOS (tm) 3000 Software (CPA25-Y-L), Version 11.1(24), RELEASE SOFTWARE (fc1)
Copyright (c) 1986-1999 by cisco Systems, Inc.
Compiled Mon 04-Jan-99 20:00 by richv
Image text-base: 0x030206C0, data-base: 0x00001000

ROM: System Bootstrap, Version 5.2(8a), RELEASE SOFTWARE
ROM: 3000 Bootstrap Software (IGS-RXBOOT), Version 10.2(8a), RELEASE SOFTWARE (fc1)

ISDN-Link uptime is 2 minutes
System restarted by power-on
System image file is "flash:cpa25-y-l.111-24.bin", booted via flash

cisco CPA2500 (68030) processor (revision F) with 2048K/2048K bytes of memory.
Processor board ID 04595024, with hardware revision 00000000
Bridging software.
X.25 software, Version 2.0, NET2, BFE and GOSIP compliant.
Basic Rate ISDN software, Version 1.0.
Authorized for CiscoPro software set only.
1 Ethernet/IEEE 802.3 interface.
2 Serial network interfaces.
1 ISDN Basic Rate interface.
32K bytes of non-volatile configuration memory.
4096K bytes of processor board System flash (Read ONLY)

Configuration register is 0x2102

ISDN-Link>

You have on your hands an affected router that needs some tender loving care.

For starters, expect that CiscoPro router will have minimum DRAM and flash. Cisco states that they shipped 2500s with 8 megs of DRAM and Flash. As you can see, this 2503 has only 4 megs of each. DRAM for 2500 is simple – it’s standard 72pin PC RAM, so if you have a 16 meg, single sided, it will work. Flash is much harder to obtain, and folks like MemoryX want ~80 USD for a 16 meg kit.

2500 platform is limited to 16 megs of flash and 16 megs of DRAM, but that would allow you to run IOS 12.3.16, which is the latest at the time of this writing. In other words, with over a million 2500s in the field, they will be supported for a while longer.

Next, you need to get yourself a PC with Windows. I did my upgrade under Windows 2000, although you might get away with a Unix/MacOS X box with tip, minicom, ZTerm.app, etc.

Under Windows you need to get yourself a copy of Cisco RSL (Router Software Loader) from ftp://ftp-sj.cisco.com/pub/access/ and a copy of the upgrade image. From the Cisco FTP site I grabbed cpa25-up.bin and RSL 7.11.

Then I configured the console cables so I could talk to the router over hyperterminal. In process I needed to do password recovery (drop to bootrom, set the confreg to 0x2142, reboot, say no at the original config) on one of the routers. I’ve used Cisco TFTP server for windows that ships with RSL (look for installer in the RSL711 directory) to back up the cpa25-xxxx.bin IOS loads, as they are hard to find and fit in 4 megs of flash requirement that I currently have. I also had to look around, and find an even older firmware image for 2500 that would fit into flash. At http://www.tfr.org/cisco-ios/ I grabbed router-2500-serial.bin which is IOS ™ 3000 Software (IGS-I-L), Version 10.3(5), RELEASE SOFTWARE (fc1). Of course 2500 and 3000 are basically the same router, so it was a fair deal.

I created a directory “images” in RSL711 directory, and dropped cpa25-y-l.111-24.bin that I recovered from the router, router-2500-serial.bin that I grabbed from above site, and cpa25-up.bin

Then I disconnected HyperTerminal, and started RSL. RSL found my 2503, queried it, and asked me what IOS image I’d like to run on it. I pointed at the IOS 10.3, at which point RSL told me that I need to rename cpa25-up to a different name, as it can’t find it. I renamed cpa25-up.bin to what it asked me, and told RSL to continue. It went on for a while occasionaly looking , and about 5 minutes later it told me that it is done.

NOTE: Tell RSL to NOT back up any files. That’s why you backed up IOS as the first step. If you do tell it to back up, it will not finish running and just sit there, because it is expecting somewhat different prompts then what IOS on my routers was offering it.

I shut RSL down, and connected to the router over hyperterminal. It was happily running IOS 10.3(5). At that point I’ve started Cisco TFTP server again, and flashed in original IOS (You can use RSL for that too, it will work. Again, no backups, as you already did them).

So after a RAM upgrade, and this convoluted procedure I were left with:

Router#sh ver
Cisco Internetwork Operating System Software
IOS (tm) 3000 Software (CPA25-Y-L), Version 11.1(24), RELEASE SOFTWARE (fc1)
Copyright (c) 1986-1999 by cisco Systems, Inc.
Compiled Mon 04-Jan-99 20:00 by richv
Image text-base: 0x030206C0, data-base: 0x00001000

ROM: System Bootstrap, Version 5.2(8a), RELEASE SOFTWARE
ROM: 3000 Bootstrap Software (IGS-RXBOOT), Version 10.2(8a), RELEASE SOFTWARE (fc1)

Router uptime is 0 minutes
System restarted by power-on
System image file is "flash:cpa25-y-l.111-24.bin", booted via flash

cisco 2500 (68030) processor (revision F) with 16384K/2048K bytes of memory.
Processor board ID 04595024, with hardware revision 00000000
Bridging software.
X.25 software, Version 2.0, NET2, BFE and GOSIP compliant.
Basic Rate ISDN software, Version 1.0.
1 Ethernet/IEEE 802.3 interface.
2 Serial network interfaces.
1 ISDN Basic Rate interface.
32K bytes of non-volatile configuration memory.
4096K bytes of processor board System flash (Read ONLY)

Configuration register is 0x2102

Router#

And everything was well in the world…

Cisco: Cisco 26xx/36xx network module compatibility.

Short story

NM-2E anything nor any NM-XFE will work in Cisco 2620. This includes NM-2E2W, NM-1FE, NM-2FE, etc. Same with HPPI and HSSI cards. Here is a list of supported NMs.

To me this looks like a deliberate crippling of 26xx platform just to convince people to buy bigger boxes, because NM-4E is supported under 2620, so in NM-2E2W it’s not a backplane speed issue.

Long story

I ended up with a Cisco 2620 handed to me. It came with only a single WIC slot populated with WIC-1DSU-T1, so it had an empty NM slot, and an empty second WIC slot, plus built in 10/100 FastEthernet port.

I figured that I have three NM-2E2W interfaces in my Cisco 3640, yet I am only using 4 uplinks, so I should just move one NM-2E2W to the 2620, turning it into a mean routing machine, with 3 ethernet interfaces, plus a serial interface. Then I could cross-connect 2620 and 3640 over serial, and use HSRP or something fancy, or, if worst comes to worst, just sell 3640 – I can get away with just 3 interfaces.

So I transfered NM-2E2W to the 2620.

It booted up, and I saw a curious syslog message:

00:00:08: SERVICE_MODULE(Serial0/0): self test finished: Passed
00:00:09: %PA-2-UNDEFPA: Undefined Port Adaptor type 30 in bay 1
00:00:13: %LINK-3-UPDOWN: Interface Serial0/0, changed state to down

That’s weird, I thought. I got distracted, however, as I had to drop to rommon, change config register to 0x2401, write a new config file in to the router, etc. But sh ver was showing me only

cisco 2620 (MPC860) processor (revision 0x102) with 61440K/4096K bytes of memory.
Processor board ID XXXXXXXXXXX (XXXXXXXXX)
M860 processor: part number 0, mask 49
Bridging software.
X.25 software, Version 3.0.0.
1 FastEthernet/IEEE 802.3 interface(s)
1 Serial network interface(s)
32K bytes of non-volatile configuration memory.
16384K bytes of processor board System flash (Read/Write)

which is not exactly useful as I generally use ethernet, and short of doing something fancy, like dot1q, router in it’s current state is not that functional.

I remembered times from debugging Cisco 7204s at Cyberus, where different IOS loads would recognize different PAs. 2620 had a c2600-ik9o3s3-mz.123 load that was a few years old, so I figured that maybe I need a new IOS. No problems, a recent security advisory provided me with an excuse to call up TAC and ask for new IOS.

When calling TAC in case like this, one has to be aware that TAC will only offer a like for like update. In other words, if you have 12.0.x, you will get a 12.0.Y image, where Y > X, but still, same feature set, same IOS revision. Makes sense, as I am not a paying customer.

Basic concern for me was to get an IOS load that has 3DES support. 3DES support means SSH capable load. Sadly, Cisco is in the States, which has a somewhat silly law about encryption being more dangerous then nukes. Luckily the image on the router was already with the right features. So I called up TAC, spent about 2 hours on hold with them, talking to their wonderful people. I like TAC – they talk to me, even though I don’t have a support contract.

End result was that they gave me links to download c2600-ik9o3s3-mz.123-15b.bin and c3640-ik9s-mz.123-16.bin .
I’ve attempted to load the former onto the 2620, however there was a disappointment – Cisco’s latest and greatest is a few hundred K larger then the 16meg flash I have in 2620:

c2620#sh flash

System flash directory:
File  Length   Name/status
  1   11072164  c2600-ik9s-mz.122-17a.bin  
  2   309208   crashinfo_19930301-000024  
[11381500 bytes used, 4871428 available, 16252928 total]
16384K bytes of processor board System flash (Read/Write)

c2620#
stany@gilva:/private/tftpboot/works[11:29 PM]$ ls -la c2600-ik9o3s3-mz.123-15b.bin 
-rw-r--r--   2 stany  501  16303356 Sep 15 20:49 c2600-ik9o3s3-mz.123-15b.bin
stany@gilva:/private/tftpboot/works[11:30 PM]$ 

Way to go Cisco!

No, problems, if this load would recognize my NM-2E2W, I will tftpboot it each time.
I dropped to rommon, set a bunch of variables, and did tftpdlnt.

On my MacOS X box I did:

root@gilva:/private/tftpboot[11:36 PM]# ln ~stany/c2600-ik9o3s3-mz.123-15b.bin test.bin
root@gilva:/private/tftpboot[11:36 PM]# /usr/libexec/tftpd -d -s /private/tftpboot/
root@gilva:/private/tftpboot[11:37 PM]# netstat -an | grep *.69
udp4       0      0  *.69                   *.*                    
udp6       0      0  *.69                   *.*                    
root@gilva:/private/tftpboot[11:37 PM]# 

On the Cisco I dropped into rommon by pressing break on the boot-up, and checked that my settings made sense:

rommon 1 > set
PS1=rommon ! >
IP_ADDRESS=192.168.82.249
IP_SUBNET_MASK=255.255.255.0
DEFAULT_GATEWAY=192.168.82.254
TFTP_SERVER=192.168.82.161
TFTP_FILE=test.bin
TFTP_VERBOSE=2
?=38
RET_2_RTS=19:08:22 EST Sun Feb 28 1993
RET_2_RCALTS=
CRASHINFO=flash:crashinfo_19930301-000024
BSI=0
RET_2_RUTC=0
rommon 2 >

All tftpdlnd cares about are valid IP_ADDRESS IP_SUBNET_MASK DEFAULT_GATEWAY TFTP_SERVER and TFTP_FILE. TFTP_VERBOSE can be set to 0 (no verbosity), 1 (somewhat verbose, default) and 2 (tell me everything).

test.bin is a reasonable name for the file – it’s short and easy to type.

Once I verified that everything was in order, I proceeded to download the image and executing it in memory (-r option). This is, BTW, a very useful way of testing IOS images, when you have something in flash that you know works, and you don’t have space for multiple images on flash. This way you can test new image, knowing that a power-cycle will get rid of it.

rommon 2 > tftpdnld -r

         IP_ADDRESS: 192.168.82.249
      IP_SUBNET_MASK: 255.255.255.0
     DEFAULT_GATEWAY: 192.168.82.254
         TFTP_SERVER: 192.168.82.161
           TFTP_FILE: test.bin

Performing tftpdnld over Fast Enet.
Interface is operating at: 100Mbps/HALF DUPLEX
Initializing interface.
Interface link state up.
ARPing for 192.168.82.161
ARP reply for 192.168.82.161 received.  MAC address 00:0d:93:42:2a:96
Receiving test.bin from 192.168.82.161 !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!.!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
File reception completed.
program load complete, entry point: 0x80008000, size: 0xf8c3e0
Self decompressing the image : ########################################################################
#######################################################################################################
######################################################################################## [OK]

Smart Init is enabled
smart init is sizing iomem
  ID            MEMORY_REQ                 TYPE
000094          0X000B3B80 C2600 Single Fast Ethernet
00001E           UNKNOWN port adapter
                0X00098670 public buffer pools
                0X00211000 public particle pools
TOTAL:          0X0035D1F0

If any of the above Memory Requirements are
"UNKNOWN", you may be using an unsupported
configuration or there is a software problem and
system operation may be compromised.
Rounded IOMEM up to: 4Mb.
Using 6 percent iomem. [4Mb/64Mb]

              Restricted Rights Legend

Use, duplication, or disclosure by the Government is
subject to restrictions as set forth in subparagraph
(c) of the Commercial Computer Software - Restricted
Rights clause at FAR sec. 52.227-19 and subparagraph
(c) (1) (ii) of the Rights in Technical Data and Computer
Software clause at DFARS sec. 252.227-7013.

           cisco Systems, Inc.
           170 West Tasman Drive
           San Jose, California 95134-1706



Cisco Internetwork Operating System Software
IOS (tm) C2600 Software (C2600-IK9O3S3-M), Version 12.3(15b), RELEASE SOFTWARE (fc1)
Copyright (c) 1986-2005 by cisco Systems, Inc.
Compiled Thu 25-Aug-05 13:39 by ssearch
Image text-base: 0x80008098, data-base: 0x81A13C28


This product contains cryptographic features and is subject to United
States and local country laws governing import, export, transfer and
use. Delivery of Cisco cryptographic products does not imply
third-party authority to import, export, distribute or use encryption.
Importers, exporters, distributors and users are responsible for
compliance with U.S. and local country laws. By using this product you
agree to comply with applicable laws and regulations. If you are unable
to comply with U.S. and local laws, return this product immediately.

A summary of U.S. laws governing Cisco cryptographic products may be found at:
http://www.cisco.com/wwl/export/crypto/tool/stqrg.html

If you require further assistance please contact us by sending email to
export@cisco.com.

cisco 2620 (MPC860) processor (revision 0x102) with 61440K/4096K bytes of memory.
Processor board ID XXXXXXXXXXX (XXXXXXXXXX)
M860 processor: part number 0, mask 49
Bridging software.
X.25 software, Version 3.0.0.
1 FastEthernet/IEEE 802.3 interface(s)
2 Serial network interface(s)
32K bytes of non-volatile configuration memory.
16384K bytes of processor board System flash (Read/Write)

Uncompressed configuration from 1579 bytes to 2531 bytes


Press RETURN to get started!

Still no NM-2E2W!

I telnetted into Cisco, and did sh diag:

c2620#sh diag | beg 88 00
          0x30: 88 00 00 00 00 01 24 01 FF FF FF FF FF FF FF FF

Slot 1:
        Unknown (type 30) Port adapter
        Port adapter is disabled 
        Port adapter insertion time unknown
        EEPROM contents at hardware discovery:
        Hardware revision 1.2           Board revision E0
        Serial number     18783769      Part number    800-01171-05
        FRU Part Number:  NM-2E2W=

        Test history      0x0           RMA number     00-00-00
        EEPROM format version 1
        EEPROM contents (hex):
          0x00: 01 1E 01 02 01 1E 9E 19 50 04 93 05 00 00 00 00
          0x10: 70 00 00 00 00 02 29 17 FF FF FF FF FF FF FF FF


c2620#

In other words, it sees the NM-2E2W, just refuses to deal with it.

Argh!

Eventually, after more googling, I discovered this useful document, that tells me that cisco 2620 is a piece of crap, unless I spend a fortune on NM-4E. Argh.