aarch64 broken on qemu

[Maxime Villard <max%m00nbsd.net@localhost>]

Is it normal that netbsd-aarch64 doesn't work on qemu?

Launching the latest current with:

	qemu-system-aarch64 -M virt -cpu max -m 4g -drive if=none,file=arm64.img,id=hd0 -device virtio-blk-device,drive=hd0 -bios QEMU_EFI.fd

Gives:

	[   1.0000000] panic: kernel diagnostic assertion "dv != NULL" failed: file "/home/source/ab/HEAD/src/sys/kern/subr_prf.c", line 757
	[   1.0000000] cpu0: Begin traceback...
	[   1.0000000] trace fp ffffc00000f89680
	[   1.0000000] fp ffffc00000f896a0 vpanic() at ffffc000004b3fe4 netbsd:vpanic+0x15c
	[   1.0000000] fp ffffc00000f89710 kern_assert() at ffffc000007d33ac netbsd:kern_assert+0x5c
	[   1.0000000] fp ffffc00000f897a0 aprint_normal_dev() at ffffc000004b5310 netbsd:aprint_normal_dev+0x90
	[   1.0000000] fp ffffc00000f89830 gic_v2m_acpi_find_msi_frame() at ffffc0000000f8a0 netbsd:gic_v2m_acpi_find_msi_frame+0x1b0
	[   1.0000000] fp ffffc00000f89880 acpi_madt_walk() at ffffc0000006ac18 netbsd:acpi_madt_walk+0x60
	[   1.0000000] fp ffffc00000f898b0 gic_acpi_attach() at ffffc0000000f100 netbsd:gic_acpi_attach+0xf0
	[   1.0000000] fp ffffc00000f89920 config_attach_loc() at ffffc00000497604 netbsd:config_attach_loc+0x184
	[   1.0000000] fp ffffc00000f89970 acpi_md_madt_probe_gic() at ffffc0000000bf28 netbsd:acpi_md_madt_probe_gic+0x30
	[   1.0000000] fp ffffc00000f89980 acpi_madt_walk() at ffffc0000006ac18 netbsd:acpi_madt_walk+0x60
	[   1.0000000] fp ffffc00000f899b0 acpi_md_callback() at ffffc0000000c7ac netbsd:acpi_md_callback+0x64
	[   1.0000000] fp ffffc00000f899e0 acpi_attach() at ffffc0000006b270 netbsd:acpi_attach+0x250
	[   1.0000000] fp ffffc00000f89a40 config_attach_loc() at ffffc00000497604 netbsd:config_attach_loc+0x184
	[   1.0000000] fp ffffc00000f89ac0 acpi_fdt_attach() at ffffc00000069218 netbsd:acpi_fdt_attach+0xb8
	[   1.0000000] fp ffffc00000f89b40 config_attach_loc() at ffffc00000497604 netbsd:config_attach_loc+0x184
	[   1.0000000] fp ffffc00000f89b90 fdt_scan() at ffffc000005b1fc8 netbsd:fdt_scan+0xe0
	[   1.0000000] fp ffffc00000f89d00 fdt_rescan() at ffffc000005b20d4 netbsd:fdt_rescan+0x54
	[   1.0000000] fp ffffc00000f89d30 config_attach_loc() at ffffc00000497604 netbsd:config_attach_loc+0x184
	[   1.0000000] fp ffffc00000f89d80 arm_fdt_attach() at ffffc00000064484 netbsd:arm_fdt_attach+0x64
	[   1.0000000] fp ffffc00000f89de0 config_attach_loc() at ffffc00000497604 netbsd:config_attach_loc+0x184
	[   1.0000000] fp ffffc00000f89e30 cpu_configure() at ffffc0000006012c netbsd:cpu_configure+0x44
	[   1.0000000] fp ffffc00000f89e60 main() at ffffc000007d364c netbsd:main+0x29c
	[   1.0000000] fp 0000000000000000 aarch64_start() at ffffc00000001870 netbsd:aarch64_start+0x1070
	[   1.0000000] cpu0: End traceback...
	Stopped in pid 0.0 (system) at  netbsd:cpu_Debugger+0x4:        ret
	db{0}>

"-cpu max" is triggers this bug. It was working fine a few months ago.

So I switched to cortex-a53, with this time networking:

	qemu-system-aarch64 -M virt -cpu cortex-a53 -m 4g -drive if=none,file=arm64.img,id=hd0 -device virtio-blk-device,drive=hd0 -netdev user,id=mynet0,hostfwd=tcp:127.0.0.1:2222-:22 -device e1000,netdev=mynet0 -bios QEMU_EFI.fd

And it gives a deadlock at boot time:

	Adding interface aliases:.
	Waiting for duplicate address detection to finish...
	Starting dhcpcd.
	[UNRESPONSIVE, HITTING "ENTER" DOES NOT EVEN DISPLAY '\n']

The network interface is causing this deadlock. If I disable dhcpcd, it boots,
but then when I launch it manually, it deadlocks again. It was working fine a
few months ago.

The combination of the two means it is impossible to use netbsd-aarch64 on Qemu:
cannot test recent hardware, cannot SSH/SFTP to the VM

Maxime