Diagnosing faulty eMMCs?
#1
Hi,

I received a couple of months ago two ROCK64 4GBs as well as two 64GB eMMC modules. Yesterday, I finally began trying to get them up and running. I'm planning to run my home automation and NAS on them, so I'd be getting Armbian Stretch on them. I failed to order the eMMC USB adapters (mistake!), so getting the eMMCs flashed were a bit problematic. I finally did the following:

I installed these prerequisites:

https://github.com/ayufan-rock64/flash-tools

https://github.com/ayufan-rock64/rkdeveloptool

Next, I got myself a USB A-A cable (actually had an USB B-B adapter which worked well with a couple of USB A-B cables - the other alternative would've been to splice two USB cables).

Then, went through the following process:

  1. Placed a jumper on the maskrom headers (the ones closest to the Recovery button)
  2. Connected the USB cable to the upper (OTG) USB port
  3. Connected the power cable
  4. After a few seconds, removed the jumper
Then, I used the `rklinux_flash` script to flash an image:



Code:
sudo ./rklinux_flash Armbian_5.42_Rock64_Debian_stretch_default_4.4.124_desktop.img


This approach produced a bootable device.
Now, the question: this only seems to work on one of the eMMC modules. With both of them, flashing completes just fine. However, only one of the devices boot and become accessible (both via local console and SSH). With the other one, there's no HDMI output and the green Ethernet light keeps flashing suspiciously.
Both of the SBCs seem to be fine. Both of them boot and work with the good eMMC and give the same behaviour with the bad one, so I think the issue is with the eMMC module.
Now, how should I debug the issue further? Are the symptoms enough for me to request a replacement eMMC from PINE64? I already ordered the eMMC adapter but it'll probably take a couple of weeks to arrive. Could I get any additional information on the serial console? I have a bunch of ESP8266 related hardware which I could probably use to jury rig a 3.3V serial port adapter, but that's some trouble I'd prefer to avoid.
Thanks for any help!
#2
(08-19-2018, 02:50 AM)mairas Wrote: Hi,

I received a couple of months ago two ROCK64 4GBs as well as two 64GB eMMC modules. Yesterday, I finally began trying to get them up and running. I'm planning to run my home automation and NAS on them, so I'd be getting Armbian Stretch on them. I failed to order the eMMC USB adapters (mistake!), so getting the eMMCs flashed were a bit problematic. I finally did the following:

I installed these prerequisites:

https://github.com/ayufan-rock64/flash-tools

https://github.com/ayufan-rock64/rkdeveloptool

Next, I got myself a USB A-A cable (actually had an USB B-B adapter which worked well with a couple of USB A-B cables - the other alternative would've been to splice two USB cables).

Then, went through the following process:

  1. Placed a jumper on the maskrom headers (the ones closest to the Recovery button)
  2. Connected the USB cable to the upper (OTG) USB port
  3. Connected the power cable
  4. After a few seconds, removed the jumper
Then, I used the `rklinux_flash` script to flash an image:



Code:
sudo ./rklinux_flash Armbian_5.42_Rock64_Debian_stretch_default_4.4.124_desktop.img


This approach produced a bootable device.
Now, the question: this only seems to work on one of the eMMC modules. With both of them, flashing completes just fine. However, only one of the devices boot and become accessible (both via local console and SSH). With the other one, there's no HDMI output and the green Ethernet light keeps flashing suspiciously.
Both of the SBCs seem to be fine. Both of them boot and work with the good eMMC and give the same behaviour with the bad one, so I think the issue is with the eMMC module.
Now, how should I debug the issue further? Are the symptoms enough for me to request a replacement eMMC from PINE64? I already ordered the eMMC adapter but it'll probably take a couple of weeks to arrive. Could I get any additional information on the serial console? I have a bunch of ESP8266 related hardware which I could probably use to jury rig a 3.3V serial port adapter, but that's some trouble I'd prefer to avoid.
Thanks for any help!

For a start id suggest burning an image on the ""faulty"" one with a different flasher, as example i usually use etcher with exelent results...but im not sure if you have either an usb-emmc or microsd-emmc adapter...
nvm that, now i see that you dont have the adapter...tho somehow if its a new emmc module it still ""feels"" like the mbr or however is called nowadays to the very few sectors on a storage device is not correct...
#3
he said he didn't buy usb-mmc adapter.

try to reflash the offending emmc once again. then try it on both boards. if it keeps failing, congrats, you've got a fualty piece of hardware. if not, wahoo - build your things. Smile
ANT - my hobby OS for x86 and ARM.
#4
I was able to consistently flash the other eMMC using either board and failed consistently with the other. Tried several times, so, yeah. I don't think the flashing method I used had any verification whatsoever, though, and I'd love to get a bit more information on how/why it failed other than that the board didn't come online.

I was digging my coffers for USB serial adapters but the ones I found had CH340G and didn't seem to work with 1.5M bps. I did find one ESP32 board that had CP2102 on it. I tried the old Arduino trick of using it as the USB serial adapter by connecting RST (actually, EN) to GND but it didn't seem to work. I suspect ESP32 keeps pulling TX low even if the reset pin is pulled low.

I ordered the eMMC USB adapter and the serial adapter from PINE64 store. Once they arrive, I'll be able to decisively tell that the eMMC is defective. Meh. Hate waiting. :-)
#5
OK, got the eMMC adaptor. Etcher writes the image successfully but complains about corruption issues after verifying the write. So yes, got a faulty eMMC.


Thanks for everyone who chimed in!
#6
Possibly the method initially used damaged the eMMc card. Who knows at this stage though.
#7
(09-04-2018, 02:46 PM)Rocklobster Wrote: Possibly the method initially used damaged the eMMc card. Who knows at this stage though.

Why would it? The jumper method is the standard way of booting the device in maskrom mode and has no impact on the eMMC module.
#8
Just to give the issue a conclusion: I RMAed the module; it was verified as faulty and a replacement was shipped back to me. The new one works as intended. The customer support experience was decent: not fabulous by any means, but perfectly acceptable, and the end result was what I wanted.


Forum Jump:


Users browsing this thread: 1 Guest(s)