Help - is my Cisco 678 a paperweight now?
bgeer
would hang requiring a power cycle. So I downloaded 2.4.8 firmware
from /archive/dsl &, carefully followed the upgrade instructions last
night.
It won't reboot - nothing happens - no red "alarm" light flash, just a
steady green power light. I can see a red led shining thru the vent
holes in the top.
So, is my 678 a paperweight now?
Any suggestions mucho appreciated...Bob
---
Here's a transcript of what I did - beorn is my Linux box:
---
I ran telnet in one window, tftp in another.
TELNET WINDOW:
beorn:cisco% telnet 10.0.0.1
Trying 10.0.0.1...
Connected to 10.0.0.1.
Escape character is '^]'.
cbos>enable
Password:
cbos#show tftp
TFTP Configuration
Is not enabled
Currently accepts connections from any host
Currently uses port 69
cbos#set tftp enabled
TFTP is enabled
cbos#set telnet timeout off
Telnet timeout disabled
cbos#write
NVRAM written.
[here I ran tftp in another window, shown below]
cbos#
Downloading in progress...... done.
Saving image................. done.
Please reboot the CPE for the new download to take effect
cbos#
TFTP WINDOW:
beorn:cisco% cd wan/dsl/cisco
beorn:cisco% tftp 10.0.0.1
tftp> mode binary
tftp> put c678dmt.full.2.4.8.bin
Sent 1015088 bytes in 9.2 seconds
tftp> quit
--
<> Robert Geer & Donna Tomky | |||| We sure |||| <>
<> bg...@xmission.com | == == find it == == <>
<> dto...@xmission.com | == == enchanting == == <>
<> Albuquerque, NM USA | |||| here! |||| <>
Brad M
> It won't reboot - nothing happens - no red "alarm" light flash, just a
> steady green power light. I can see a red led shining thru the vent
> holes in the top.
I had the same thing happen to me. From the searches I performed on
Google, at this point it is a paper weight. :\ I made the mistake
of first uploading the wrong firmware, then uploading the correct
one, but didn't let it finish properly.
Brad
Eric Jorgensen
That's interesting.
As i understand it, these boxes have a two stage bootloader. It would
make sense to basically never make it possible to erase the sector that
has the first stage loader. That's the way upgradeable embedded devices
are usually - but not always - set up. The one exception I've run into
was a PPC evaluation board where the documentation we'd been given said
that we had a 4 meg flash area with 128k erase sectors and the
bootloader resided in the first 128k erase sector. We bricked two of
them before finding out that we had a 2 meg flash area with the boot
loader in the last 128k sector, and we were erasing the bootloader.
So if you power it on and hit ctrl-c on the management console,
nothing happens?
Technically it should be feasable to hook up some sort of outboard
debugger and flash through that - you can get free tools for coldfire
processors - but that's a little beyond my knowhow.
bgeer
>On Thu, 5 Feb 2004 19:50:09 +0000 (UTC)
>Brad M <bpha...@xmission.com> wrote:
>> bgeer <bg...@xmission.com> wrote:
>> > It won't reboot - nothing happens - no red "alarm" light flash, just
>> > a steady green power light. I can see a red led shining thru the
>> > vent holes in the top.
>>
>> I had the same thing happen to me. From the searches I performed on
>> Google, at this point it is a paper weight. :\ I made the mistake
>> of first uploading the wrong firmware, then uploading the correct
>> one, but didn't let it finish properly.
I used c678dmt.full.2.4.8.bin. I know I need dmt, don't know what the
difference is between the "nsrouter" version & this one. I assume
"ns" is for netspeed, tho my device sez nothing about netspeed on it
or in it.
> As i understand it, these boxes have a two stage bootloader. It would
>make sense to basically never make it possible to erase the sector that
>has the first stage loader. That's the way upgradeable embedded devices
>are usually - but not always - set up.
I've had tons of experience flashing various SBCs with VxWorks. The
vendor-supplied flash function was so cumbersome to use that I ended
up writing my own. It had a near perfect record, too...:-) For the
record, tho, I never blew the vendor function. I've flashed modems &
pc bioses, too. Anyways, I know about being careful about flashing.
> So if you power it on and hit ctrl-c on the management console,
>nothing happens?
Nope, nada! Tried different uart speeds, etc, too.
> Technically it should be feasable to hook up some sort of outboard
>debugger and flash through that - you can get free tools for coldfire
>processors - but that's a little beyond my knowhow.
Yeah, me too. After trying everything I could think of or that I've
heard about, I pop'd the case & there are no service connectors on the
pcb.
Michael Wilkinson
>
> As i understand it, these boxes have a two stage bootloader. It would
> make sense to basically never make it possible to erase the sector that
> has the first stage loader. That's the way upgradeable embedded devices
> are usually - but not always - set up. The one exception I've run into
> was a PPC evaluation board where the documentation we'd been given said
> that we had a 4 meg flash area with 128k erase sectors and the
> bootloader resided in the first 128k erase sector. We bricked two of
> them before finding out that we had a 2 meg flash area with the boot
> loader in the last 128k sector, and we were erasing the bootloader.
There is something like a boot loader on these routers (I don't remember
the correct term). I too fraged my 678 back when they were still supported
by cisco, I was sent some intructions on loading a new OS through the
boot loader. I recall I had to connect to the management port and load
the new OS using xmodem (or kermit or something)...hmm, let me see
if I can find it...
Ah, here it is...hopefully with this information you can save your equipment:
http://www.cisco.com/warp/public/794/600swinstl.shtml
Good luck!
-Mike
--
Michael Wilkinson
Game Designer
Tower Ravens LLC
Eric Jorgensen
bgeer <bg...@xmission.com> wrote:
> Eric Jorgensen <al...@xmission.com> writes:
>
> >On Thu, 5 Feb 2004 19:50:09 +0000 (UTC)
> >Brad M <bpha...@xmission.com> wrote:
>
> >> bgeer <bg...@xmission.com> wrote:
> >> > It won't reboot - nothing happens - no red "alarm" light flash,
> >just> > a steady green power light. I can see a red led shining
> >thru the> > vent holes in the top.
> >>
> >> I had the same thing happen to me. From the searches I performed
> >on> Google, at this point it is a paper weight. :\ I made the
> >mistake> of first uploading the wrong firmware, then uploading the
> >correct> one, but didn't let it finish properly.
>
> I used c678dmt.full.2.4.8.bin. I know I need dmt, don't know what the
> difference is between the "nsrouter" version & this one. I assume
> "ns" is for netspeed, tho my device sez nothing about netspeed on it
> or in it.
iirc, the nsrouter version is for people upgrading from
pre-really-old-version firmware. These boxes were originally designed by
netspeed, acquired by cisco. At some point cisco changed the flash
format and 2nd stage boot loader. The last time i used an nsrouter file
to update my 675 was more than three years ago. Really thoughtful of
cisco to make sure you don't have to flash it twice. Cisco stopped
allowing downloads from their website and leaked the instructions for
flashing from the rom monitor after a few tens of thousands of end users
bricked their routers using the wrong version, though.
I did that once myself, but it wasn't hard to repair - just really
slow.
> > As i understand it, these boxes have a two stage bootloader. It
> > would
> >make sense to basically never make it possible to erase the sector
> >that has the first stage loader. That's the way upgradeable embedded
> >devices are usually - but not always - set up.
>
> I've had tons of experience flashing various SBCs with VxWorks. The
> vendor-supplied flash function was so cumbersome to use that I ended
> up writing my own. It had a near perfect record, too...:-) For the
> record, tho, I never blew the vendor function. I've flashed modems &
> pc bioses, too. Anyways, I know about being careful about flashing.
You think vxworks is bad, you should see the nightmare Motorola is
foisting off on unsuspecting customers.
"OK, do you have your $400 parallel port debugger dongle containing 40
cents worth of buffer chips? Good! Step one: install CodeWarrior. Two:
Apply for CodeWarrior license. Three: Wait a day or two for Metrowerks
to approve your application and send you a license key. Four: on the
CodeWarrior CD, locate and compile this project. we're not going to tell
you where it is, just what it is named. the CW user interface will not
be able to tell you if the file you selected is a project or not, let
alone whether it's the one we told you to build. Please ignore files
with similarly cryptic names. Five: Run the project. Did we mention it
will only run from inside CodeWarrior?. Six: watch the horizontal
barber's pole and pray. Oh, by the way, this won't give you a human
readable console. You will be required to run this project every
time you wish to communicate with the board.
The punchline - the real punchline here - is that motorola tells you
right to your face that every board ships with a pre-programmed flash
simm, and not a single flash simm will arrive pre-programmed. I think
this is so that you are forced to discover the great value and power of
code warrior in the most fruitful way possible.
> > So if you power it on and hit ctrl-c on the management console,
> >nothing happens?
>
> Nope, nada! Tried different uart speeds, etc, too.
Should be 38.4k, no flow control. Cisco says 9600bps should work too.
But you're probably not going to get anywhere with flow control enabled
in your terminal emulator.
> > Technically it should be feasable to hook up some sort of
> > outboard
> >debugger and flash through that - you can get free tools for
> >coldfire processors - but that's a little beyond my knowhow.
>
> Yeah, me too. After trying everything I could think of or that I've
> heard about, I pop'd the case & there are no service connectors on the
> pcb.
Yeah, if the rom monitor isn't healthy, maybe you're out of luck.
Apparently it resides in sector 7, at the end of flash. Some interesting
info here: http://zerospace.org/c675/
After reading that, I find myself wondering if anybody has tried
working up a build of the i960 port of uClinux to run on old 675's. The
DSL part will never do anything useful, but, hey, network attached
serial port. 2 megs of flash should be plenty for a uclinux system. It'd
be nice to be able to ssh into a serial console on some other device,
for example. Or i bet i could shoehorn the board from a 675 into my
vt220 .. . . .
There's got to be a way to hook up an external debugger, but it might
be more trouble than it's worth. Some processors - like strongarms -
require very little logic between the cpu and the jtag port. Some
require more. I've never fiddled with an i960, and i was sure multiple
people had told me they were coldfire based. Whatever.
Depending on the flash part used, it may actually be possible to
program that directly from a parallel port after soldering several
wires to the board. Again, more trouble than it's worth, unless you
really enjoy that sort of thing.
Neil Wagstaff
http://www.xmission.com/help/dsl/678cap1ip.html
Under item 3 you will find instructions on how to get to the 678 bios
upgrades.
There you will find a README file that says:
Cisco/Netspeed Cbos Upgrade
If your Netspeed is on version NSOS 1.40, stop right now. This model is
not
flash upgradable. Call Qwest and have them send you a new model
replacement.
Upgrading with XMODEM:
1. Connect serial management cable to DSL router.
2. Start XMODEM capable terminal on computer.
3. Speed should be set to 38400.
4. Enter your router password. If no password is set, simply press
the Enter key..
5. Type "set download code" at the prompt.
6. Send file with XMODEM.
7. If transfer fails, DO NOT REBOOT, try again until you succeed.
8. When finished, type "reboot" at the prompt.
9. Once it reboots, type "set web disabled" hit Enter, then type "set
web port 137"
hit Enter, then type "write" and choose Enter (this helps protect
against
crashes from exploits)
In the case your upgrade fails and your box will not boot, follow these
instructions:
Break into the router using ctrl-c during boot up, when the alarm light
blinks red.
(comments follow dashes below)
es 0 --erase sector 0
es 1 --erase sector 1
es 2 --erase sector 2
es 3 --erase sector 3
es 4 --erase sector 4
es 5 --erase sector 5
es 6 --erase sector 6 (running config)
--note this will require the router be reconfigured
df 10008000 --download the file to this location in memory
--start and xmodem download with the 675 image
--after transfer is complete you will see the # of bytes it
--has written - e.g. 00fe8001 - you will need this later
pb 10008000 fee00000 <size> --program the area of memory to flash area
--<size> is the # of bytes (no brackets)
the
--monitor reported being downloaded
m0 --turn off monitor
rb --reboot
------------------------------------------
Hope this helps
Neil
--
bgeer
> You think vxworks is bad, you should see the nightmare Motorola is
>foisting off on unsuspecting customers.
I have a motorola cell phone & my experience with that is about the same.
> "OK, do you have your $400 parallel port debugger dongle containing 40
>cents worth of buffer chips? Good! Step one: install CodeWarrior. Two:
>Apply for CodeWarrior license. Three: Wait a day or two for Metrowerks
>to approve your application and send you a license key.
Is that for VxWorks? I haven't done VxWorks for 4 years now, but back
then they were drifting further & further towards nazi user
relations.
<rant>
Argh...CodeWarrior...sounds like pandering to dweebs with delusions of
grandeur. A true(tm) code warrior only needs gcc, gdb, & emacs or vi.
</rant>
Anyways, thanks to all for the pointers to Cisco info. No amount of
screwing around with serial seems to help. The alarm light never
blinks which appears to be the prerequisite for ctrl-c.
I impulsively ordered a Zoom dsl modem. It's $69, ethernet, & I found
Zoom success stories on google groups. I'll post my experience with
it.
Not sure what I'll do with the 678...
Eric Jorgensen
bg...@xmission.xmission.com (bgeer) wrote:
> Is that for VxWorks? I haven't done VxWorks for 4 years now, but back
> then they were drifting further & further towards nazi user
> relations.
Nah, that's just to load the rom monitor and debugger onto the board's
flash simm so that it will actually boot. The first time i attempted it,
it took two days, and would have taken at least one since i had to wait
for a CW license. The quickest i saw it done while at Lineo was about 4
hours.
> <rant>
> Argh...CodeWarrior...sounds like pandering to dweebs with delusions of
> grandeur. A true(tm) code warrior only needs gcc, gdb, & emacs or vi.
> </rant>
CW started out as an advanced development environment for MacOS, and
was expanded to be a multi-platform thing. In the beginning, it
really was a good thing if you needed to code on a Mac. Motorola owns
the company that writes it, Metrowerks. The quality of the back-end, the
actual compiler, is pretty good, but there doesn't seem to be a way to
avoid their dev environment, which is an abomination.
Aside from being user-hostile and needlessly complex, they can't even
stick to any one user interface convention. Some things behave like
MacOS, some things behave like windows, some things behave like xwindows
with athena widgets, sometimes with motif widgets. Some file dialogs
accept fully-qualified paths only with forward slashes, some only with
backward slashes, some don't let you manually enter a file name at all.
And they use a horizontal bumblebee colored barber's pole animation
wherever possible. It's like being pandered to and insulted at the same
time.
I remember opening the big black development kit which was promised to
include free gifts. The gifts included a tshirt boringly advertising
Code Warrior, and a miniscule mult-tool including pliers that somehow
have no leverage and do not ply, a dull 2" blade, a #1 philips head
screwdriver on a 13mm shank perpendicular to and at the mid point of the
tool's 3" length, and a factory-rounded slot screwdriver. I felt that it
was well paired with CW. Especially as the side panels were vented with
slots at a 45 degree angle to match the CW barber's pole.