LGTM. Thanks, it fixes the kernel crashing with QEMU -M ppce500.

During tests I found that when a eligible NIC exists and network isn't prepared for bootp, it cannot load the local root file system.
More investigation is required:

edit: commented on wrong revision

Suspending this since it requires changes to continue.
The OS revision part is now being discussed in D34429.

Updated with gjb's suggestion

In D34429#780080, @imp wrote:

I love this idea, btw. I hope my reaction my reaction to one small detail doesn't obscure this :)

updated with imp's suggestion

Abandoned due to https://cgit.freebsd.org/src/commit/?id=298c3e8d6b6405cbea13a7a1a1fbc1ad5a45c378

add ports package versions and adjust comments

Makes sense. I've been using powerpc64le on Raptor Blackbird and QEMU and it looks solid.

• style change commited with hash 8d1ab5ad8449e6f51f081e261ed6202a0483a595
• removed COMPAT_FREEBSD8 from config files, since it doesn't exist.

updated after discussion on IRC

Patch updated with @pauamma_gundo.com comments. Thanks!

@pauamma_gundo.com fixes

Finished full tests (configure petitboot autoboot, install disc1.iso and boot).

assume it's a draft until proper tests are in progress

LGTM. Would be good to add arm group for review and smoke test this change

I boot-tested this patch and no regression found on:

• QEMU VM POWER9 with hardware acceleration (hw.ncpu: 144)
• bare metal POWER9 hw.ncpu: 16

correct comments wording

Justin, I tested this patch on Blackbird and it looks good, just need to be rebased to latest main. Also tested it along with D30626.

address reviewer comments

addresses commiter's suggestions

Great job @bdragon and @luporl!
Tested a powerpc64 disc1.iso with this patch on QEMU6.0 and installation went fine. Also tested make buildkernel/installkernel with no issues.

I tested this patch with D31232 few weeks ago on the Blackbird and it looks good to me once it's approved.

In D31804#717525, @dim wrote:

Let's do this for now to get the PowerPC builds working again. But also be aware that the bisected upstream commit is only exposing an underlying issue: most likely the root cause is in the PowerPC backend.

In D31698#715220, @fbsd-phab_maskray.me wrote:

This is not suitable for the upstream. The root cause behind dso_local needs to be found and fixed.

Abandoned in favor of D31698

In D31646#713731, @emaste wrote:

Move logic to BROKEN_OPTIONS as suggested by @jhibbits

This will work for src itself, but I think it won't work with ports that use /usr/share/mk?

In D31646#713717, @emaste wrote:

is there a PR to reference here?

Move logic to BROKEN_OPTIONS as suggested by @jhibbits

Great job luporl!

After discussion with @imp , we decided to try the "reroot" approach instead. It appears to be generic solution for other modules as well, and can be used on other platforms.
A new review entry will be created when it's ready for review.

For some reason this breaks the boot with VGA console on Blackbird.
I'm using D30626 locally, but reverted it as well. With D30626 reverted and adding kernel parameter 'hw.ofwfb.physaddr=0x000600c280000000' the kernel hangs as well with no output.