FreeBSD for Clusterboard (V1.1) [*WIP]

[paradise]

https://www.pine64.org/?product=clusterb...dule-slots
On the left, it's connected to node slot 1 it's using SD-MMC controller 3 check the Allwinner A64 PDF

Yes.. Loading it in SPI (I use SPL because you merge it with loader.efi) and place it in SPI flash indeed.

I use SPI less because it's confusing if you also have Serial Peripheral Interface's. :-)

Wow, hard to see the eMMC zif connetor.
But still I don't understand your patch, it should work fine with latest HEAD, it does on the Pine64-LTS
So for SPI did you make a u-boot driver for it ?

I found a u-boot that can access the SPI I flash my own u-boot to SPI flash with that. 
Yesterday I flashed the last 35 nodes with the patched u-boot. (added the patch to fix the realtek bug so I can boot from PXE)

I'll try head.. I started this project couple of weeks ago and I didn't update the tree sinds.


I think I have it 100% stable now .. Ethernet is no longer dropping link. (I messed up the syscon while testing other stuff)

Code:

Jan  1 00:08:52 NODE001 kernel: Jan  1 00:08:52 NODE001 su[671]: freebsd to root on /dev/pts/0
Jan  1 00:54:50 NODE001 freebsd[712]: test
Jan  1 00:55:07 NODE001 kernel:
Jan  1 00:55:09 NODE001 syslogd: last message repeated 3 times
Jan  1 01:49:50 NODE001 freebsd[752]: here we are

kernel:  (nothing) message is my boot is tricky because my rootfs is only 16mb the real root is chrooted and I should not dare to touch the console like I do. ;-)
As topic mentions it's still [WIP*]

You did make a few good points and i'll try and be more direct in what the patches are.. I just started hacking the code until stuff started working and I'm like hyper focused on it so I test patches at a rate you will not believe.
I test on 7 nodes (can tell my cluster master to reboot all nodes to test a new kernel)
In a few minutes I can reproduce a weird bug.

Also I use stuff to compile the kernel faster.. the only slow-down is currently getting the new kernel from the NAS at work to a machine that can talk to my machine at home and then getting it here and on the right NFS location.. :-)

Code:

# ./cluster -Q iamsure
You asked for it and now you have done it! -- All nodes will do an instant reboot now!
{"result":"ok, 6 nodes responded"}

:-)