Re: Running Mezzano in bhyve

Turns out, panic always happens in the same place when Mezzano is trying to
initialize an AHCI controller. Qemu uses an old PATA controller by default,
which works. When forcing SATA, it fails just like bhyve. Virtio block
devices are also claimed to work, but using virtio-blk option in bhyve also
results in panic. When I tried Mezzano on my PC (bare metal), I used a USB
stick. Can a USB stick be emulated in bhyve? I guess, only using USB
controller passthrough?

ср, 9 окт. 2024 г. в 12:18, Vasily Postnicov <shamaz.mazum@gmail.com>:

> Thanks! It worked, but Mezanno crashes after a short period of time:
>
> Begin PCI scan.
> PCI:0:1F:0 8086:7000 Intel Corporation - 82371SB PIIX3 ISA [Natoma/Triton
> II] 6:1:0 rid: 0 hdr: 0 intr: FF
> PCI:0:1D:0 FB5D:40FB NIL - NIL 3:0:0 rid: 0 hdr: 0 intr: FF
>     40: MSI 84
> PCI:0:5:0 1AF4:1000 Red Hat, Inc - Virtio network device 2:0:0 rid: 0 hdr:
> 0 intr: A
>     40: Unknown capability 11
>     4C: MSI 80
> PCI:0:3:0 8086:2821 Intel Corporation - 82801HR/HO/HH (ICH8R/DO/DH) 6 port
> SATA Controller [AHCI mode] 1:6:1 rid: 0 hdr: 0 intr: B
>     40: MSI 88
> Detected AHCI ABAR at C1002000
> AHCI IRQ is B
> Host Capabilities FF30FF3F
> Global Host Control 80000000
> Interrupt Status 0
> Ports Implemented 1
> Version 10300
> Command Completion Coalescing Control 0
> Command Completion Coalescing Ports 0
> Enclosure Management Location 0
> Enclosure Management Control 0
> Host Capabilities Extended 4
> BIOS/OS Handoff Control and Status 0
> AHCI HBA version 1.300
> Handler: 0
> Config register: 17
> ----- PANIC -----
> Early page fault on address 80000035B0
> Local CPU is #<7A4AF9>
> Run queues:
> Thread #<1468079 Thread Initial thread> holds the world
> Run queue #<1498139>/SUPERVISOR:
> Run queue #<1498159>/HIGH:
> Run queue #<1498179>/NORMAL:
> Run queue #<1498199>/LOW:
> IRQ state:
> IRQ #<1497DF9 Irq :Number 0> - 0 (86 delivered)
>   #<1499119> PIT [exclusive]
> IRQ #<1497E29 Irq :Number 1> - 1 (1 delivered)
>   #<1499699> INTEL-8042-KEYBOARD [exclusive]
> IRQ #<1498009 Irq :Number B> - B (0 delivered)
>   #<149C959> #<149C909>
> IRQ #<1498039 Irq :Number C> - C (0 delivered)
>   #<14993D9> INTEL-8042-AUXILIARY [exclusive]
> Active timers: (current time is 147260)
> Thread #<1468079 Thread Initial thread> ACTIVE #<1498669 Wait-Queue
> Heartbeat wait queue>
> TIFH: 1 TPFH: NIL
> 20000121FB90 7FFF892B07 DEBUG-DUMP
> 20000121FD20 7FFF800FAE (LAMBDA IN PANIC-1)
> 20000121FE30 7FFF80073F PANIC-1
> 20000121FE50 7FFF800633 PANIC
> 20000121FEC0 7FFF8820B1 FATAL-PAGE-FAULT
> 20000121FEF0 7FFF88114B %PAGE-FAULT-HANDLER
> 20000121FF40 7FFF87F3BB %%INTERRUPT-SERVICE-ROUTINES
> 20000121FFD0 80000035B0 #<unknown>
> 200000A1FFF0 7FFF800841 %CALL-ON-WIRED-STACK-WITHOUT-INTERRUPTS
> 20000221FE60 7FFF958E90 (SETF PCI-CONFIG/16)
> 20000221FE90 7FFF9B5670 AHCI-PCI-REGISTER
> 20000221FEF0 7FFF95B9C9 (FLET DX-LAMBDA IN PCI-DETECT)
> 20000221FF50 7FFF95B22E MAP-PCI-DEVICES
> 20000221FF90 7FFF955671 PCI-DETECT
> 20000221FFA0 7FFF93C8F4 INITIALIZE-PLATFORM
> 20000221FFC0 7FFF8D7F03 BOOTLOADER-ENTRY-POINT
>
> Looks like it executes an interrupt handler with calling something "on a
> wired stack without interrupts". Postmortem rflags:
> root@vonbraun:~ # bhyvectl --vm=ubuntu --get-rflags
> rflags[0] 0x0000000000000046
>
> I do not know what to think, that %cal-on-wired-stack-without-interrupts
> seems to be legit to me:
> https://github.com/froggey/Mezzano/blob/master/supervisor/x86-64/interrupts.lisp
> Anyway, thanks again for the help!
>
> ср, 9 окт. 2024 г. в 11:20, Peter Grehan <grehan@freebsd.org>:
>
>> > Hi! Mezzano is an operating system written in Common lisp. I would like
>> > to test it in bhyve. When I launch it, it spins infinitely in a loop
>> > measuring how many CPU cycles it makes in a unit of time (I understood
>> > it by adding various debug messages).
>> >
>> > The timer it uses is something called a PIT:
>> > https://wiki.osdev.org/Programmable_Interval_Timer
>> > <https://wiki.osdev.org/Programmable_Interval_Timer>
>> >
>> > The problem is, once the timer is configured it fires an interrupt only
>> > once, while it obviously must fire interrupts constantly at some
>> > frequency (100 Hz, as I understand from the code).
>> >
>> > Is the PIT supported by bhyve? Does FreeBSD use it anywhere in its
>> kernel?
>>
>>   Yes, though mode 3 of the timer isn't supported by the device
>> emulation. It appears that Mezzano uses that mode
>>
>>
>> https://github.com/froggey/Mezzano/blob/master/supervisor/x86-64/time.lisp#L20
>>
>>   Try this patch to the PIT code to see if it helps
>>      https://people.freebsd.org/~grehan/bhyve_vatpit.diff
>>
>> later,
>>
>> Peter.
>>
>>