RockPro64 has bad Memory (Software sogfaults and kernel panics)

[kop316]

Hello,

I just purchased a rockpro64, and I am having some very odd issues with it:

- I attempted to install the Debian official image twice, and I caused it to kernel panic with the official image.

- I also attempted to use Armbian, and when I tried to update it, I got a segmentation fault on dpkg.

I have looked around and I have not seen anyone complain about these issues.

EDIT:

This is caused bybad memory. I ran as root on Armbian minimal install:

Code:

#memtester 3400M 1

It is on Rockpro64 V2.1 2018-07-02, 4GB variant, date code 4719

Pine64 has get to RMA my device despite me using support and contacting sales over a week ago.

UPDATE 16 Oct:

I hate to say, I have not heard anything useful since I opened up this issue on 4 Oct. I filed a Paypal dispute in the hopes that will get them to respond. This is frankly beyond disappointing. I have been very happy with the Pinephone, and that was the reason I have the RockPro a chance over the RaspPi. This whole ordeal sours my interest in buying anything from Pine again.

[foresto]

I didn't run in to that problem when I installed Debian on a RockPro64.

Could it be related to the firmware accessing memory behind the kernel's back?

https://forum.pine64.org/showthread.php?...9#pid55839

https://wiki.pine64.org/index.php/ROCKPr...sed_memory

https://github.com/ayufan-rock64/linux-b...-578410380

Could it be a problem in the Debian kernel version you're using?

[kop316]

I didn't run in to that problem when I installed Debian on a RockPro64.

Could it be related to the firmware accessing memory behind the kernel's back?

https://forum.pine64.org/showthread.php?...9#pid55839

https://wiki.pine64.org/index.php/ROCKPr...sed_memory

https://github.com/ayufan-rock64/linux-b...-578410380

Could it be a problem in the Debian kernel version you're using?

I was using the offical debian image and I got my Rockpro64 a week ago....it had 5.8 being installed on it.

[kop316]

Well I figured out why. I ran a memory tester on my chip and there are hard errors on my DDR. I will need to exchange it.

[richese]

Well I figured out why. I ran a memory tester on my chip and there are hard errors on my DDR. I will need to exchange it.

Hi kop316,

Could you please detail how you tested the memory of your RockPro?

I am also seeing seg faults that always occur at some point regardless of the used image or SD card.

[richese]

I tried to run the memtester on my board with fresh image and sure enough there seem to be memory issues in the part of memory it tested.
Resuts: https://pastebin.pl/view/00e8b2d7

Hardware I am using:
Rockpro64 V2.1 2018-07-02, 4GB variant, date code 52 19
running of SD card, tall heatsink installed, powered by Pine 5A power supply

For OS image I used ayufan's buster-minimal-rockpro64-0.10.12-1184-arm64 image.
I believe this one uses recent enough u-boot that does not use the binary blob in RAM.

Command used:

Code:

sudo memtester 3G 1

I tried the same on other SBC's I have with no errors so there is definitely something going on with this RockPRO board.

[kop316]

I tried to run the memtester on my board with fresh image and sure enough there seem to be memory issues in the part of memory it tested.
Resuts: https://pastebin.pl/view/00e8b2d7

Hardware I am using:
Rockpro64 V2.1 2018-07-02, 4GB variant, date code 52 19
running of SD card, tall heatsink installed, powered by Pine 5A power supply

For OS image I used ayufan's buster-minimal-rockpro64-0.10.12-1184-arm64 image.
I believe this one uses recent enough u-boot that does not use the binary blob in RAM.

Command used:

Code:

sudo memtester 3G 1

I tried the same on other SBC's I have with no errors so there is definitely something going on with this RockPRO board.

Sorry I just saw this. I did the same thing you did. I got a minimal Armbian image and run as root:

Code:

memtester 3400M 1

And I got DDR Errors. My RockPro64
Rockpro64 V2.1 2018-07-02, 4GB variant, date code 4719

[wildering]

I'm also experiencing the same issue. ROCKPro64_v2.1, 2018-07-02, 4GB variant, date code 5219, powered by a known good 12V 5A power supply.

Running Armbian Focal or Buster, any time I go to use apt update I get a segmentation fault. I've been unable to run memtester as of yet as I need to find an image with it preinstalled as I can't download anything using apt.

I've got 2 other RockPro64s of the same revision, one with date code 4719 and one with date code 5119. I'll be testing those right now and reporting back.

[wildering]

I cracked open the two other boards and ran memtest on them with no errors reported. I then transplanted that eMMC module onto the date code 5219 board and ran the test again. I was presented with a slew of errors. It's evident that, that ROCKPro64 (v2.1, 2018-07-02 5219) is also defective and will warrant an RMA.

[LMM]

I cracked open the two other boards and ran memtest on them with no errors reported. I then transplanted that eMMC module onto the date code 5219 board and ran the test again. I was presented with a slew of errors. It's evident that, that ROCKPro64 (v2.1, 2018-07-02 5219) is also defective and will warrant an RMA.

I ran memtest with Debian an it seems ok. (v2.1, 2018-07-02). What is noticeable is the high temperature reached (70°C) in spite of a heatsink. Then I put it over a fan and it dropped below 50°C after 9 min

---

I cracked open the two other boards and ran memtest on them with no errors reported. I then transplanted that eMMC module onto the date code 5219 board and ran the test again. I was presented with a slew of errors. It's evident that, that ROCKPro64 (v2.1, 2018-07-02 5219) is also defective and will warrant an RMA.

I ran memtest with Debian an it seems ok. (v2.1, 2018-07-02). What is noticeable is the high temperature reached (70°C) in spite of a heatsink. Then I put it over a fan and it dropped below 50°C after 9 min

I don't know if it is a good practice (and a good idea) but I cut the conductive pad in order to be able to double the layer on the DDR chip to make the contact with the heatsink. Otherwise it probably does not