richlowe: I have spent an unhealthy amount of time in this corner of the C++ type system and ABI lol

there's some truly bizzare stuff in there

hm, with UTM the emulated x86 illumos boot hungs sometimes/often on boot while in start_other_cpus+0x260

probably because of some other things did fail before....

andyf how urgent do you need that pcie test data? I'm pretty busy and probably won't have time until weekend after next.

It isn't urgent, we need to collect a bit more testing data before going ahead with integration, but I'd rather we got more data. Some of the changes in the set still need review too. I am particularly interested in data from your systems :)

andyf: do you need data from systems that actually have devices plugged into pcie slots? I don't know if I have any slots populated on my main dev box

Both would be useful - we're basically going to be looking for differences between the way things are enumerated or programmed. Even with nothing plugged into slots there are PCIe components like root ports, bridges, USB controllers, onboard NICs, NVMe drives and so on. A lot of integrated graphics uses a PCIe-to-PCI bridge too.

if i can remember tonight or tomorrow, I can throw it on my home server

(supermicro 5028-TN4)

not sure how odd it is, but it did find that cpu deadlock that'd been sitting in the code for 15+ years :)

if we get -1 from ddi_get_instance(dip) -- that should be error, right?

NOTICE: iommulib_nexus_register: rootnex-1: Succesfully registered NEXUS i86pc nexops=fffffffffbe9a100

it's supposed to be infallible

It is also sometimes up to parent nexi to not screw up, I've found.

the root has a -1 instance, though.

see `create_devinfo_tree`

I screwed it up in the amdzen nexus driver. It depends a ubnch on what's going on.

@andyf --> got anyone with the X470 chipset volunteering yet? I can try your stuff out on HDC 3 if you need me.

( kebe.com/blog/?p=502 )

ping

Okay, phew I'm here again.