Re: Problem booting Orange Pi One on NetBSD 10.0_BETA

[Dave Tyson <dtyson%anduin.org.uk@localhost>]

On Wed, 2023-07-26 at 10:42 +0200, Martin Husemann wrote:
> 
> Have you tried the -10 image from there?
> That would tell you immediately if the problem is in -10 or the u-
> boot
> part (which is identical for all images for each board).
> 
> Martin

OK, I tried the NetBSD-10 images from armbsd. The Orange Pi One image
fails booting in the same way as my initial report - basically no root
filesystem.

Also, although the NetBSD-9 image from armbsd booted OK, trying to do
any useful work (using vi to edit config files) resulted in a crash and
drop into ddb - some kind of translation error.

The Orange Pi One worked fine 5 years ago, but I wanted to eliminate a 
possible hardware fault and so wrote the appropriate NetBSD-10 image
from armbsd onto a Orange Pi zero. It booted OK, extended the
filesystem and on rebooting successfully. Plugging in a RAL wifi stick 
produced:

armv7# [ 592.5310430] rum0 at uhub2 port 1
[ 592.5310430] rum0: Ralink (0x07d1) 802.11 bg WLAN (0x3c03), rev
2.00/0.01, addr 2
[ 592.6610763] rum0: MAC/BBP RT2573 (rev 0x2573a), RF RT2528, address
00:22:b0:69:b5:78

[ 592.6710782] uvm_fault(0x9193f898, 0, 1) -> e
[ 592.6810827] Fatal kernel mode data abort: 'Translation Fault (S)'
[ 592.6810827] trapframe: 0xa847dbc8
[ 592.6910752] FSR=00000005, FAR=00000000, spsr=600c0013
[ 592.6910752] r0 =00000000, r1 =a847de40, r2 =00000216, r3 =00000001
[ 592.7010737] r4 =9167e300, r5 =00000000, r6 =c0286938, r7 =a847de40
[ 592.7010737] r8 =00000000, r9 =00000000, r10=00000000, r11=a847dc7c
[ 592.7110742] r12=a847dc80, ssp=a847dc18, slr=805e6af0, pc =804cd3a4

Stopped in pid 601.601 (dhcpcd)
at      netbsd:mutex_enter+0x34:        ldr     

This sort of translation fault is the same error I was getting with
NetBSD-9 as noted above.

So the bottom line seems to be that support for the Orange Pi systems
is broken at present...

I suppose I could try a few other arm devices (Banana Pi etc) but am
surprised how something which was rock solid in 2017 has now become
broken or unstable.

I will try and do a bit more testing in a week to try and diagnose what
the problems are. 

If a lot of changes are being done in the arm low level code (e.g
scanning the dtb tree to pick up devices) then maybe developers should
post out on current-users to ask people to test so that bugs faults can
be picked up quickly rather than languishing and biting people later.

Dave