Odroid-N2+ doesn't boot from eMMC (11.5+, other than Kingston eMMC) #3227
Comments
|
With HAOS 11.5 we updated the U-Boot bootloader to 2024.01, so this is probably bootloader related. FWIW, my ODROID-N2 test device here boots fine with HAOS 11.5 and 12.0, so this must be related to your particular unit. Do you have access to the serial console (e.g. using the USB-UART adapter from Hardkernel) to observe the output at boot time? |
|
Unfortunately, no. I feel like I was lucky to have bought the eMMC writer just in case. For what it's worth, this is a brand new unit so it sounds like Hardkernel has changed something in their file loads at the factory. Would there be anything I could help provide by flashing back to Ubuntu and stopping during Petitboot or files I could share? Since it took about a week and a half to get the delivery from Korea, would something like this work, provided the jumpers can be made to fit inside the header (and the appropriate TTL level is selected)? |
|
@agners However, moving back to 11.4 the following is spat out |
|
Looks like there is a similar issue with 12.1 maybe |
|
@scoobydrvr I see you linked my post, I actually got a similar output as you today. Ive been trying all sorts to get it working but having no luck. going to try 11.4 now and see if it works. Here's what I managed to get on output , looks very similar to what you are seeing. edit: 11.4 doesnt work for me either, same error about it not finding the mmc. If it flash ubuntu it works no problem.
|
|
What size is your MMC? I remembered I have a 16Gb card from an old Odroid device. I just threw 12.1 on it an, low and behold, the thing boots up! I had hoped that society had moved past storage device limits but it seems to somehow have an effect here... |
It's 32Gb, I just found another thread with a similar issue. What I noticed in your logs was that when its successful its booting from MMC 1, but when it fails its booting from MMC 2. Mine always seems to try boot from MMC 2 (hence its always failing). I don't know if there is a way to force this, but feels like its trying to boot the wrong device. |
|
Not sure it matters but I also noticed my mmc card revision is 0.5, a lot of the pictures ive seen show 0.4. So wonder if the new revision is having issues or something. I might have to install with SD card as right now I have no automation running, so my house is extra dumb. |
My "non-functional" 32Gb is also a rev 0.5 hmmm... |
Yeah, it doesnt make too much sense, if i can boot ubuntu I don't really see how the card could be the issue. However, maybe some kind of incompatibility has occurred and is causing issues. Its way above my knowledge level really, but as other people seem to be fine, it feels like I must have something different to them. |
|
Maybe someone with more intimate knowledge can sort out how v0.4 and v0.5 differentiate and why v0.5 doesn't seem to be working? https://wiki.odroid.com/accessory/emmc/reference_chart#orange_emmc_module |
|
I have at least one eMMC module with version v0.5 and it seems to work here, so I don't think it is the PCB revision. What PCB colors is yours? it seems that PCB color gives some indication what eMMC chip is on the PCB. I do have orange PCBs here. |
|
@agners Mine is an orange module |
|
I should add, ive tried both flashing with the USB 3.0 reader board, and flashing direct via the cable plugged into the N2+, both flash successfully but wont boot. |
Both of mine, the v0.4 16Gb and v0.5 32Gb, are orange cards |
|
I actually have an old model Home Assistant Blue which I based my recommendation to piercy on when he was selecting hardware. I went through the fresh install process just days before piercy and had no issues It's currently deployed 2 hours away, so I'll have to wait a few days before I can get any covers off to get photos / part numbers for comparison |
|
Happy to see this one, as there's nothing wrong my my board and/or eMMC after all :) I can confirm I was using an orange 32gb v0.5 eMMc |
Did this work? |
Yes, thankfully. See output in the comment above. Only caveat is the power and ground cables had to have the plastic connectors removed so they could fit inside the serial header on the board. Should also wrap them in electrical tape or similar to prevent shorts. Additionally, the dongle is powered by the board so you'll need to be quick with PuTTY or whatever serial terminal you use to capture the boot messages from the beginning. |
|
Any solution for this? I am having the same problem wit my N2+ I had a corrupted install of HA OS on an 32Gb eMMC card and tried reflashing the HA OS to the eMMC but do not see a blue light during boot. Is it possible that the partitions created by the flashing of HA create a problem? How can I "reset" the eMMC card so it is completely empty and has no partitions? |
|
My problem was the SD card, and after purchasing a New ODROID N2+, everything is fine. I am using a 256GB eMMC version .5 without any issue. |
|
Weird issue. |
I had mine boot once too, figured it had been resolved by an update (i'd tried a different versions). When i then shut it down to move it to another room, it didn't boot again. It certainly is a weird issue, seems like its starting to happen more often now though. Hopefully more examples can help diagnose and resolve it . |
While i could potentially do this, the fact I can boot into ubuntu fine suggest thats it isn't a problem with my eMMC or the N2+. I am currently using SD card rather that eMMC for HAOS but hoping a solution comes along soon. |
And if you wait for a night or so and then try starting it up? |
|
@lugaru90 There's another issue for Kingston modules, unfortunately there's no resolution for that yet, as it's probably a SoC issue that it's not able to work with them reliably. @smarthomefamilyverrips No, Blue should be fine. The issues with other eMMCs started to appear after we changed the clock rate to 40 MHz in order to fix compatibility with Kingston. That's now reverted as well. |
@sairon OK thank you for your quick response, then I know I can safely update to 13.0 👍 |
This comment was marked as off-topic.
This comment was marked as off-topic.
This comment was marked as off-topic.
This comment was marked as off-topic.
This comment was marked as off-topic.
This comment was marked as off-topic.
This comment was marked as off-topic.
This comment was marked as off-topic.
This comment was marked as off-topic.
This comment was marked as off-topic.
|
I would encourage you guys to use the below issue for kingston issues, as maybe once enough people have the problem we will start to get somewhere. Currently we haven't reached critical mass, but I suspect these emmc's aren't going away, so its only a matter of time before it becomes more widespread Admins it would be good if the duplicate tag could be removed from the below issue, as it isn't a duplicate. It was considered a dupe until it was realized that this issue 3227 and 3250 were different. |
This comment was marked as off-topic.
This comment was marked as off-topic.
This comment was marked as off-topic.
This comment was marked as off-topic.
This comment was marked as off-topic.
This comment was marked as off-topic.
|
@piercy @smarthomefamilyverrips Guys, please use another place for discussions, you're creating unnecessary noise for all people in this issue report. I will answer piercy's concerns in the other issue, for general discussions please use GH Discussions or the Community forums 🙏 |
@sairon OK, sorry not was my intention to clutter this issue report 👍 |
@sairon Just to confirm, I updated my HA Blue to version 13.0 without any problems, all still working as expexted. 💯 👍 |
|
Updated to 13.0 and it caused a boot issue, as others have experienced. I should have checked before but it turns out I have the Kingston eMMC module as well... Glad I made that backup this morning!
|
|
So updating to OS 13.0 is also not a good idea. Hopefully the issue get's solved quickly. |
|
Kingston 32GB device does not work on multiple N2+ systems. I have tried every single HAOS version from 11.3 up to the latest 13.1. None worked. Device tries to boot from network with "serverIP not set" console message. Just hangs. Other OS (Ubuntu) flashed on this 32GB eMMC does work without problems. |
You should be using #3250 for kingston issues. I realize the OP has now realized his is a kingston, but developers have asked to put kingston information on 3250. It seems to be a dead thread but maybe if we fill it up it start to get some attention. |
|
Ok. The fact that #3250 is a dead thread is concerning. I am sure there are many with same problem. Appreciate your feeback. Will move it to that thead |
Testing shows that the 24MHz cap is problematic for Kingston. Bumping this cap to 40 MHz in #3319 fixed Kingston but reintroduced issues with some other eMMCs that the frequency cap was supposed to fix. This commit adds another patch that selects a different clock for eMMC. While it's still unclear why Kingston fails with the frequencies of slightly below 24 MHz (see patch message why it's below), using 24 MHz from xtal seems to be fine and hopefully shouldn't introduce regressions for old eMMCs. Alternative approach would be using the same clock source but setting the frequency cap to 25 MHz. This results in exact 25 MHz from PLL which is even the defined frequency for some eMMC modes, Kingston eMMC works fine with it and the difference from the previously used value wouldn't be too high, but with 24 MHz we're even closer. refs #3250, refs #3227
…ROID-N2 Replace the original patch that limits the frequency to 24 MHz with a more sophisticated version that first tries to initialize the eMMC at this limited frequency, and if it fails, then tries disabling the limit to use one defined in the upstream driver. While still hacky, this seems to finally fix the Kingston eMMC issues, and with the current implementation, the initialization should appear the same as on older HAOS versions, as 24 MHz is used first. Better solution that addresses the root cause on some eMMCs (128 GB Microns?) would be still nice though... refs #3250, refs #3227
Replace the original patch that limits the frequency to 24 MHz with a more sophisticated version that first tries to initialize the eMMC at this limited frequency, and if it fails, then tries disabling the limit to use one defined in the upstream driver. While still hacky, this seems to finally fix the Kingston eMMC issues, and with the current implementation, the initialization should appear the same as on older HAOS versions, as 24 MHz is used first. Better solution that addresses the root cause on some eMMCs (128 GB Microns?) would be still nice though... refs #3250, refs #3227
Replace the original patch that limits the frequency to 24 MHz with a more sophisticated version that first tries to initialize the eMMC at this limited frequency, and if it fails, then tries disabling the limit to use one defined in the upstream driver. While still hacky, this seems to finally fix the Kingston eMMC issues, and with the current implementation, the initialization should appear the same as on older HAOS versions, as 24 MHz is used first. Better solution that addresses the root cause on some eMMCs (128 GB Microns?) would be still nice though... refs #3250, refs #3227
|
There hasn't been any activity on this issue recently. To keep our backlog manageable we have to clean old issues, as many of them have already been resolved with the latest updates. |
|
Closing this - some issues with Micron 128 GB eMMCs were resolved by reverting the change of the default clock rate, other turned out to be Kingston sticks. |
Describe the issue you are experiencing
Clean install by writing image to eMMC card. Device does not boot to HA OS with version 11.5 or 12.0. If 11.4 is flashed, device will boot and behave as expected. Updating to 11.5 causes device to fail to boot. Further information - #3226
What operating system image do you use?
odroid-n2 (Hardkernel ODROID-N2/N2+)
What version of Home Assistant Operating System is installed?
11.4
Did you upgrade the Operating System.
Yes
Steps to reproduce the issue
OR
Anything in the Supervisor logs that might be useful for us?
Anything in the Host logs that might be useful for us?
System information
No response
Additional information
Noticed that there is a kernel change between 11.4 and 11.5. Could this be the root cause?
The text was updated successfully, but these errors were encountered: