[split] U-Boot with NVMe support, keyboard & display working

[as400]

Thanks for the info guys. Well, it was kind of expected that problems will still be there. But hey - it's always worth trying.

I have some crazy theory regarding inconsistencies of booting. I mean for some people the device boots without problems and for others it doesn't. Or at least boots rarely without a hang.

My question is @Arwen , @belfastraven , @giamma1295 - from which batch your pbp is ?


And the image for sd emmc is here --->  SD/eMMC.

[belfastraven]

Thanks for the info guys. Well, it was kind of expected that problems will still be there. But hey - it's always worth trying.

I have some crazy theory regarding inconsistencies of booting. I mean for some people the device boots without problems and for others it doesn't. Or at least boots rarely without a hang.

My question is @Arwen , @belfastraven , @giamma1295 - from which batch your pbp is ?


And the image for sd emmc is here --->  SD/eMMC.

IMy pinebook was and early one from the second 100 sent out.  By the way,  It was coming down to not being able to boot at all.  I used an armbian build on sd through the serial cable to reflash pcm720's latest version.  I was just wondering whether you (or anyone else) knows the proper u-boot console command  to erase the spi flash.  I had no problem getting to the u-boot console with your build,  which was great,  and I know it would be sf erase "xxxxx" --just wasn't sure what the "xxxxx" was. (e.g. was it mtd0 0 0)  Although I disassembled this machine at least 30 times when I first got it (losing and replacing all the screws and standoffs--ha ha) and even took a hacksaw to the original nvme adapter,  I really would like to minimize the times that I need to take things apart,...

[wdt]

I'm wondering how well this handles sleep....

[as400]

Thanks @belfastraven - mine is from the second batch (first ansi ones).

@samueldr also has from the first batch and he also has problems booting. My suspicion is that the hardware batches differ slightly.
It doesn't have to be something big. Maybe quality of components or something like that.
I wonder from which batch is @giamma1295 - his apparently boots quite consistently. Well, let's wait.

As to the dissasemblies - welcome to the club I also lost screws, had to short pins and so on hundreds of times It's like neverending adventure.

@wdt - I wouldn't count on sleep.

[wdt]

>@wdt - I wouldn't count on sleep.
Well, you have that uboot installed, does it work or not?
If it works, should be less than 10%/d (when sleeping for that whole day)
Saves a lot of charging annoyance
AND makes it feasible to run off SD

[as400]

Can't tell you that because I have NVME installed. And sleep never worked with NVME drives.
AFAIK sleep problem with mainline kernels was not resolved as of now. Although there is one person working on it.

[samueldr]

If I recall properly, mine was from the first 100 too. I get something like 90-95% hangs without the workaround.

NVME is not installed.

Sleep with mainline kernel is partially-solved, with the proprietary trustzone blob from Rockchip.

[giamma1295]

Thanks @belfastraven - mine is from the second batch (first ansi ones).

@samueldr also has from the first batch and he also has problems booting. My suspicion is that the hardware batches differ slightly.
It doesn't have to be something big. Maybe quality of components or something like that.
I wonder from which batch is @giamma1295 - his apparently boots quite consistently. Well, let's wait.

As to the dissasemblies - welcome to the club I also lost screws, had to short pins and so on hundreds of times It's like neverending adventure.

@wdt - I wouldn't count on sleep.

Mine is from last batch.
Tests performed:
1) Manjaro NVME 5.7, sometimes before starting up I have to turn off and on again even 10 times, so it is not very consistent.
2) Debian on eMMC , installed using the script present in this forum, with the 5.5 kernel always booting correctly, with a 5.7 kernel compiled by me it boots very rarely.
3) Anything from USB does not boot

Unfortunately i don't have so much luck whit this bootloader

EDIT: I don't know what is happening, i erased the spi(flash_erase) so now my pbp should use the u-boot installed on eMMC, but it still boots from NVME (really fast and consistent too)...

[Arwen]

I've tried the latest U-Boot here with my ISO Pinebook Pro on the eMMC, because it still had the Mr Fixit Debian, and kernel 4.4. It appears to work great. Have to boot a bunch more times to see if it's reliable. This Pinebook Pro is from the first 100 batch.

One odd thing, during switch over from U-Boot to kernel, I get vertical black and white bars on the screen for a few seconds. Startled me the first time because I thought it was going to hang. Probably some interaction between the 2 video initializations.

Edit:
One note, the timeout specified in U-Boot configuration is in 1/10s of a second. So 50 in the example is 5 seconds. Not quite long enough for my purpose.

And, Manjaro 5.7 kernel does not boot, (out of 3 times). But, I made it a second partition and boot entry. The Debian 4.4 still boots fine, same Pinebook Pro. (Of course I could have made a mistake in the Manjaro copy...)

[as400]

@Arwen - the problems with booting kernels > 5.5 are known so no surprise here.
I wonder why I don't experience that inconsistency so painfully. Seems my theory of differences in hardware is not relevant.

@giamma1295 - wait, so you still have uboot from this thread even though you erased SPI ?

I wonder guys if someone of you could test my binary compiled 5.8.0rc7 kernel ? Maybe this happens because of patchset used ?
If somebody would be willing to test it I'll be more than happy to provide you with my kernel, initrd, modules and dtb file.

I will also try stock Manjaro kernel and see if this will give me problems like you have. Will report back.

EDIT:

So testing my kernel does not make sense since Manjaro stock kernel (5.7.0-3) boots fairly consistently for me.