Difference between revisions of "Router Recovery"

From Embedded Xinu
Jump to navigation Jump to search
Line 113: Line 113:
 
Issuing Processor / Peripheral Reset ... Done
 
Issuing Processor / Peripheral Reset ... Done
 
</pre></code>
 
</pre></code>
You probably don't have the combinations of options for your router.  Play with the different switches available before attempting to read/write from flash.  
+
You probably don't have the correct combination of options for your router.  Play with the different switches available before attempting to read/write from flash.  
  
  
Line 135: Line 135:
 
     4) Unrecognized CPU Chip ID.
 
     4) Unrecognized CPU Chip ID.
 
</pre></code>
 
</pre></code>
There's probably something wrong with your JTAG cable, you don't have correct permissions, the header is soldered poorly, or you forgot the <tt>/wiggler</tt> switch (active cable only).  
+
There's probably something wrong with your JTAG cable, you don't have correct permissions, the header is soldered poorly, or you forgot the <tt>/wiggler</tt> switch (active cable only).
  
 
== External References ==
 
== External References ==

Revision as of 20:25, 20 June 2011

So you've created yourself an expensive paperweight...

Not to worry! If you aren't breaking things, then chances are you aren't making hardcore progress. This page serves as a knowledge pool for methods to revive routers that are corrupted or otherwise considered non-functional. The information below is mostly specific to the WRT54GL as it is our most popular and well understood platform at this time. The process is somewhat similar for other models/platforms; however, some of the utilities are limited to specific platforms and commands vary slightly between bootloaders (ex. [U-Boot] vs CFE).

Sometimes the router won't boot because of a corrupted NVRAM variable and a simple factory reset will resolve the problem. Chances are if you've sought out this page you are in much deeper and probably need a more serious TFTP or JTAG recovery. We'll start with the simple solutions and work our way to the more intense recovery methods.

Before You Begin

If you haven't already, backup your router's configuration. Hopefully you did this earlier so you can restore to a "known good" working state. If you didn't, it is still a good idea to do that now; it can always get worse. You'll also want to grab yourself a copy of some reliable firmware. The default firmware that shipped with your router is a good place to start (generally available from the manufacturer's website). Otherwise, a stable release of your favorite embedded Linux distribution is a good alternative.

Factory Reset

Don't get your hopes up for this one, but sometimes [Occam's razor] applies to router recovery. To do a factory reset, hold down the reset switch for about 10 seconds while the unit is powered on, then unplug it. Let it rest for a little while then power it back up. If this isn't working for you, try the dd-wrt [30/30/30] reset method.

There's another option if you have access to a serial console, but your router isn't necessarily readily accessible (say locked in a rack somewhere with a pool of backends). Access the CFE as you normally would. Then, issue the command

CFE> nvram erase

and then reboot. Some models do not properly reinitialize their NVRAM variables automatically, so be careful with this method. The WRT54GL does recover them conveniently from a separate stored location in flash.

If you haven't caught on by now, these methods erase any custom settings that were stored in NVRAM. Don't forget to re-configure and commit your network settings, if applicable.

Serial Console

Now is a good time and check that your serial console is up and running. If you don't see any text coming across the serial link, then you should double check your transceiver is working properly. Swap with a working one or at least try the one in question on another working router, if you have one available. If you can access the web interface (assuming the OS in flash has one) at either the default IP address or the one you configured, then your flash image is probably fine--fix your serial console.

If you can see text and access the CFE but your router won't boot, try re-flashing the firmware with TFTP.

If you are confident the serial interface hardware is working properly but your router appears dead, then proceed to the next section.

JTAG

In a nutshell, JTAG is a interface that allows external control of an SoC and its memory. You can read more about this on our EJTAG page. JTAG allows us to recover routers that are completely unresponsive (aka debricking). Before continuing, you'll need a JTAG cable (active or passive will do) and a header soldered to the JTAG connector on your router. See the references below for some suggestions on this bit.

Software

For WRT54GL recovery, the popular programs are the original HairyDairyMaid utility and a port called TJTAG. If you purchase a commercial cable, it may come with a hardware specific recovery tool. This guide focuses on TJTAG. Download a copy of the source (linked below) and compile as usual.

If you built your own cable, chances are it uses a legacy printer style parallel port interface. You'll need adequate permissions to use this device in order to run the JTAG software. In *nix style systems usually means the user you execute TJTAG as must be a member of the lp group.

The result of the groups command should look something like this before you continue:

user@host:~$ groups
user lp dialout

Membership of the dialout group isn't strictly necessary; however, you will need this as well if you want to use a local serial console.

Establish a Connection

Now we must verify that TJTAG can properly connect to the JTAG interface on the router. Connect the JTAG cable to both your PC and the router but leave the power disconnected. On most routers you will be fighting the watchdog timer so it is a good idea to type out whatever command you want to execute (without hitting enter), then provide power, and finally quickly hit enter as soon as the router LEDs light up.

Active vs Passive

If you built the active buffered cable you need to add the /wiggler option to all of your TJTAG commands. The passive unbuffered cable does not require this option and you should leave it off when using this type of cable. If you anticipate needing to revive routers often, an active cable is surely worth the additional investment in time and parts so you aren't restricted to working within 6 inches of your parallel port. Otherwise, the unbuffered cable works great provided you can manage the logistics of the restricted cable length.

TJTAG Options

The next trick is to find the magical combination of optional TJTAG parameters which makes your router happy. Even within a single make/model this seems to vary greatly--most likely because of various flash chip manufacturers. For starters, we'll use the -probeonly option to guess and check which options will work before modifying the contents of flash. Usually something like

user@host:tjtag$ ./tjtag -probeonly /wiggler /noemw /noreset

will do the trick. If you are not getting the desired output (see below), try experimenting with the DMA, break, and reset switches. Once you've mastered the combinatorics game, you can move onto read/write operations.

Successful Output

When you've got the right combinations of parameters, you should see an output like this:

==============================================
 EJTAG Debrick Utility v3.0.1 Tornado-MOD 
==============================================

Probing bus ... Done

Instruction Length set to 8

CPU Chip ID: 00000101001101010010000101111111 (0535217F)
*** Found a Broadcom BCM5352 Rev 1 CPU chip ***

    - EJTAG IMPCODE ....... : 00000000100000000000100100000100 (00800904)
    - EJTAG Version ....... : 1 or 2.0
    - EJTAG DMA Support ... : Yes
    - EJTAG Implementation flags: R4k MIPS32

Issuing Processor / Peripheral Reset ... Skipped
Enabling Memory Writes ... Skipped
Halting Processor ... <Processor Entered Debug Mode!> ... Done
Clearing Watchdog ... Done

Probing Flash at (Flash Window: 0x1fc00000) ... 
Done

Flash Vendor ID: 00000000000000000000000011101100 (000000EC)
Flash Device ID: 00000000000000000010001010100010 (000022A2)
*** Found a K8D3216UBC  2Mx16 BotB     (4MB) Flash Chip ***

    - Flash Chip Window Start .... : 1fc00000
    - Flash Chip Window Length ... : 00400000
    - Selected Area Start ........ : 00000000
    - Selected Area Length ....... : 00000000



 *** REQUESTED OPERATION IS COMPLETE ***

The last line is important. Don't move on until you get this response.

Troubleshooting

If you see something like this:

==============================================
 EJTAG Debrick Utility v3.0.1 Tornado-MOD 
==============================================

Probing bus ... Done

Instruction Length set to 8

CPU Chip ID: 00000101001101010010000101111111 (0535217F)
*** Found a Broadcom BCM5352 Rev 1 CPU chip ***

    - EJTAG IMPCODE ....... : 00000000100000000000100100000100 (00800904)
    - EJTAG Version ....... : 1 or 2.0
    - EJTAG DMA Support ... : Yes
    - EJTAG Implementation flags: R4k MIPS32

Issuing Processor / Peripheral Reset ... Done

You probably don't have the correct combination of options for your router. Play with the different switches available before attempting to read/write from flash.


If you see something like this:

==============================================
 EJTAG Debrick Utility v3.0.1 Tornado-MOD 
==============================================

Probing bus ... Done

Instruction Length set to 5

CPU Chip ID: 11111111111111111111111111111111 (FFFFFFFF)
*** Unknown or NO CPU Chip ID Detected ***

*** Possible Causes:
    1) Device is not Connected.
    2) Device is not Powered On.
    3) Improper JTAG Cable.
    4) Unrecognized CPU Chip ID.

There's probably something wrong with your JTAG cable, you don't have correct permissions, the header is soldered poorly, or you forgot the /wiggler switch (active cable only).

External References