1043NDv2.1 Debricking

I've got a hold of a bricked 1043NDv2 to try to fix and to add to my network. It was put up on FB Marketplace as junk parts for almost free. I've tried the TFTP "Holding Reset" button on TP-Link Devices, but it does not initialize. It's almost as if the reset button does nothing since there isn't a change of any LEDs. It always lights up as Power, Activity (the one that looks like *), LAN1, and the WPS button. Wireshark does not capture any packets with useful information.

I've soldered a serial header on and have successfully gotten a log to show up on PuTTY. It shows:

U-Boot 1.1.4 (Jun 13 2014 - 15:14:01)

ap135 - Scorpion 1.0DRAM:
sri
Scorpion 1.0
ath_ddr_initial_config(211): (16bit) ddr1 init
tap = 0x00000002
Tap (low, high) = (0xaa55aa55, 0x0)
Tap values = (0x8, 0x8, 0x8, 0x8)
 4MB

Which I know is already different from the Wiki, given that it should be like this:

U-Boot 1.1.4 (Jul 17 2015 - 14:31:22)

ap135 - Scorpion 1.0DRAM:
sri
Scorpion 1.0
ath_ddr_initial_config(178): (16bit) ddr2 init
tap = 0x00000003
Tap (low, high) = (0x0, 0x1e)
Tap values = (0xf, 0xf, 0xf, 0xf)
64 MB

From then on in, it does nothing apart from normal - lighting up the following LEDs again: Power, Activity (the one that looks like *), LAN1, and the WPS button. I cannot do the standard wait and type tpl

I got a chat with the owner of the bricked router and he stated that he played around with generating his own OpenWRT image and it all went to crap when he installed an image he generated that had ASLR on. Could this have done something to Uboot-Env?

My question is this: Am I missing something? Can I do anything else besides JTAG or reprogramming the SPI/Flash Chip? I don't think that it's worth the effort doing either option since I don't have a fine enough solder tip or the steady hands required for finer soldering if that's the case.

Looks like Dram configuration is gone. Your is showing only 4MB. The least complicated thing is to me to reprogram the SPI Flash. I would get a hold of a new SPI chip, program it and exchange it (desolder and solder a new one).

If I were to theoretically do that, what would the necessary tools/process be?

I can only really think of:

  • SPI Programmer (I already have a Raspberry Pi that can be loaded up with FlashROM/Libreboot)
  • New SPI chip (Winbond W25Q64FVSIG)
  • Heat Gun (to remove/replace the SPI chip)

I would then also need to somehow read this data off the old SPI chip:

  • U-Boot (Yes/No?)
  • Atheros Radio Test (ART)

If I somehow did all of this, I would then need to create/gather an image and then load it up on the new flash chip.

This honestly sounds like an incredibly huge undertaking of both time and effort. How is it possible that a single image managed to do all this damage? From my experience it's REALLY hard to brick a router given that there are so many means of recovery - but to somehow have only the most resource intensive methods of SPI reprogramming or JTAG as the sole options left seems crazy.

For desoldering you will need a soldering iron and solder sucker.

A video with desoldering you will find here: https://www.youtube.com/watch?v=l6rw1zo4A2c&app=desktop

I have a TL-wr1043nd v2.1 and can supply you the u-boot and art if you want.

Maybe a 16MB flash together with u-boot from https://github.com/pepe2k/u-boot_mod ?

Forgive me since, I've only ever done TFTP and Serial Recovery - but doesn't the ART contain the specific MAC address of the router? If it doesn't, I would really like a copy of the ART partition.

Thanks for the video, I'm not German but it did help. However, I think I'd have to ask a friend who has better/more available tools for the actual transfer of the chip.

So that leaves me with just generating an image with the contents of /proc/mtd. From the video you sent, even though it is another router model, what or do I need to get anything from the old flash chip?

  1. u-boot <-- will need this
  2. kernel <-- comes from OpenWrt or stock image
  3. rootfs <-- comes from OpenWrt or stock image
  4. roofs_data <-- comes from OpenWrt or stock image
  5. art <-- from my old chip or (potentially) from you (thank you!!!)
  6. firmware <-- from my old chip (I don't know if this partition exists in the 1043NDv2.1 though)

U-boot is here: http://www.mediafire.com/file/p4809agj26tmtp0/mtd0.zip/file

Try to extract the art from the old chip, if you do not succeed, come back.

1 Like

The firmware partition is a union overlaying kernel+rootfs+rootfs_data. It's not a separate partition. In other words it is an alias for the middle part of the flash that is not bootloader or ART.

I thought the Atheros chips probe the RAM size from ROM, before starting the bootloader.

There is no further output on serial after it says "4MB" ?

This will take me some time to do given the season and the lack of tools. But I'll try and do it.

Nope, there is no output whatsoever. Serial connection is clean given that the initial output is not garbled. All I know is that the seller had tried to flash a custom generated OpenWrt image with ASLR, resulting in this current brick.