AR750S UART and NAND issues

I have been running a build compiled from the github/gl-inet/openwrt sources without much fuss. It is a minimal openwrt with no LuCI, no storage kmods, just dropbear and wireless access point. After a few reboots, the router does not fully come back online.

My serial adapter is able to receive data from the AR750S uart but it cannot transmit any data to the AR750S - so I’m unable to interact with U-Boot or serial console in OpenWrt. I have confirmed that my serial adapter is working on other devices, but not the AR750S.

I am able to flash factory images through the U-Boot web interface and get the router back to working order, but it takes multiple attempts because the ethernet device resets and the device freezes while flashing.

Noticed that the NAND is already reporting a bad block? Also noticed it gets abnormally hot.

My AR750S was one of the earliest pre-ordered and I am worried I might have gotten a defective unit?

U-Boot 1.1.4-ga35cd45d-dirty (Jun 22 2018 - 12:48:24)

ar750s - Dragonfly 1.0DRAM:  
ath_ddr_initial_config(278): (ddr2 init)
ath_sys_frequency: cpu 775 ddr 650 ahb 258
Tap values = (0xf, 0xf, 0xf, 0xf)
128 MB
Flash Manuf Id 0xef, DeviceId0 0x40, DeviceId1 0x18
flash size 16MB, sector count = 256
*** Warning - bad CRC, using default environment

Power up PLL with outdiv = 0 then switch to 3
No valid address in Flash. Using fixed address
ath_gmac_enet_initialize: reset mask:c02200 
athr_mgmt_init ::done
Dragonfly  ----> S17 PHY *
athrs17_reg_init: complete
SGMII in forced mode
athr_gmac_sgmii_setup SGMII done
: cfg1 0x80000000 cfg2 0x7114
eth0: 00:03:7f:09:0b:ad
eth0 up
Qualcomm Atheros SPI NAND Driver, Version 0.1 (c) 2014  Qualcomm Atheros Inc.
====== NAND Parameters ======
sc = 0x87ff6230 page = 0x800 block = 0x20000
Setting 0x181162c0 to 0x4b962100
Hit 'gl' to stop autoboot:  0 

Device 0 bad blocks:
Found ART,Checking calibration status...
Device have calibrated,Checking device test status...
Device have tested,Checking MAC address...
Device have MAC address,Checking device flash status...
Device have nor and nand flash,Booting standard firmware from nor flash...
Booting image at: 0x9F060000
Trying eth0
dup 1 speed 1000
Using eth0 device
ping failed; host is not alive
Unknown command 'ERROR!' - try 'help'
## Booting image at 9f060000 ...
   Image Name:   MIPS OpenWrt Linux-4.9.120

For UART, are you sure the wiring of tx pin and GND pin are correct?

Besides, where is the device freeze while flashing?

Wanted to add further findings:

It has never booted from NOR since I first got it. There should be a factory image present on NOR. Perhaps this is why when the NAND image stops working, it won’t boot from NOR?

When I try to flash through the U-Boot web, the upload gets interrupted and over UART shows ARP timeout and becomes unresponsive until I power cycle. I can retry until it eventually gets a full upload.

This is difficult for me to investigate further without being able to send anything to the AR750S over UART - Can only receive from AR750S.

Yup, it won’t boot from NOR. There isn’t NOR firmware as well. Both of NOR and NAND use the same kernel part, you only choose one.

Good to know. Thanks. Is there any reason why UART RX pin disabled or not working? I can use the same cable on my AR300M and other devices but not AR750S. Would try to figure out what is causing the router to fail to boot after a few days. This issue also happen with the official GL firmware.

Sorry, Not any clue for UART.

Too bad. I shall consider this AR750S defective NAND part and UART (maybe U-boot is incorrect version from factory and has wrong pin mapping). It can only hold the official GL firmware for a few days before it stops booting, then has to be flashed again.

Had high hopes for this device. It is not usable like this. Wasted the money.

You can contact the seller and get a refund. Never seen this happen on any other units.

Contacted seller. I suspect this AR750S has a bad part. Is the only thing that explains why it show erratic behaviors both with the GL firmware and any build I compile from the GL sources. Hopefully we can resolve it.

Yeah, seriously- I wouldn’t be surprised if GL themselves contacts you about a replacement if only so they can post-mortem the unit you’ve got to verify there’s not manufacturing defects.

Yes! :slight_smile:
They are going to let me return it for a replacement. Something is definitely weird with this one. It won’t hold a flash for more than a few days - then it starts failing to boot and I have to re-flash.