A true mainline Linux Kernel for the Pinebook Pro

[theotherjimmy]

Re: Suspend-to-ram,

It looks like TF-A master contains a fix that claims to fix hangs when resuming from a soft-reset, which may be related to the post-suspend hang.
link to the commit: https://git.trustedfirmware.org/TF-A/tru...516b8785ec
(Not super important link, merged to integration 7 days later: https://git.trustedfirmware.org/TF-A/tru...2672630011 )

If that does not fix it for you, you could capture the debug UART, looking for a TF-A or Linux Panic. If you find a TF-A panic, you could raise an issue on trustedfirmware.org here: https://developer.trustedfirmware.org/ma...uery/open/

PS: Hello Pine forums

[bsammon]

Can someone post/give-pointers-to a brief set of instructions on how to compile/install this kernel?

I used to compile kernels all the time 15+ years ago, but mostly on x86 (and a few times on my Powerbook)

I just dug up the following website
https://kernelnewbies.org/FAQ/KernelCompilation
which seems to match the way I remember compiling kernels back in the day.

Would those instructions fit this kernel here? I have my doubts about whether "make install" would do what I want/need it to do.

[xmixahlx]

the debian way is to build kernel packages

you need build deps, a kernel config, and 1-2 hours to build

what distro are you building for?

[bsammon]

So I did a "git clone https://gitlab.manjaro.org/tsys/linux-pinebook-pro.git" and it was 2.5 GB.
I did a "download as .tar.bz2" from https://gitlab.manjaro.org/tsys/linux-pinebook-pro and it was only 130 MB.

Just ... wow.

(maybe this is a "I'm pretty new to git" moment)

P.S. The time it took for both operations was (to an order of magnitude) proportional to those relative numbers.

[xmixahlx]

yes. for an idea, git has every revision of the repo ever in it...

you can use --depth=1 to get only the latest version and still benefit from the git system.

that is what i do most of the time if just needing source to build from.

[xmixahlx]

hi @tsys , thank you for your mainline work! (also @danielt for your attention)

regarding pinebook_pro_defconfig, please disable the new ARM erratum addition. if enabled the big cores won't come up. an interesting note: without the big cores i get a *970* octane2 score versus *10200* with!!!

# CONFIG_ARM64_ERRATUM_1319367 is not set

also confirming that 5.5-rc7-panfrost-fixes and v5.5 branches work very well on debian sid arm64 with mesa-git.


cheers

[tsys]

Hi @xmixahlx,

regarding pinebook_pro_defconfig, please disable the new ARM erratum addition. if enabled the big cores won't come up. an interesting note: without the big cores i get a *970* octane2 score versus *10200* with!!!

the BIG cores are online on my device with CONFIG_ARM64_ERRATUM_1319367=y. Tasks seem to be scheduled to them as well. Also corrupted TLBs don't sound particularly fun to me. Are you sure I should disable that option?

[as400]

I have the same problem with big cores not coming up on boot if I set maxcpus=4 on boot. After boot there is no way to bring them up.

[User 15997]

that's an effect of the BSP uboot the device comes with not bringing the big cores up properly. a newer version of uboot does that, making the maxcpus=4 hack and disabling mitigations unnecessary: https://forum.pine64.org/showthread.php?tid=8752

[as400]

that's an effect of the BSP uboot the device comes with not bringing the big cores up properly. a newer version of uboot does that, making the maxcpus=4 hack and disabling mitigations unnecessary: https://forum.pine64.org/showthread.php?tid=8752

Problem is that newer version of uboot doesn't support nvme which I'm booting from