blob: 540d39f2ec21b8195557916b27ba4d2941fdda6e (
plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
|
This is an *experimental* port of OVMF for the QEMU microvm
machine type.
microvm background info
-----------------------
microvm is designed for modern, virtio-based workloads. Most legacy
lpc/isa devices like pit and pic can be turned off. virtio-mmio
(i.e. '-device virtio-{blk,net,scsi,...}-device') is used for
storage/network/etc.
Optional pcie support is available and any pcie device supported by
QEMU can be plugged in (including virtio-pci if you prefer that over
virtio-mmio).
https://qemu.readthedocs.io/en/latest/system/i386/microvm.html
https://www.kraxel.org/blog/2020/10/qemu-microvm-acpi/
design issues
-------------
Not fully clear yet how to do hardware detection best. Right now
using device tree to find virtio-mmio devices and pcie host bridge,
can reuse existing ArmVirtPkg code that way. Needs patched QEMU.
features
--------
[working] serial console
[working] direct kernel boot
[working] virtio-mmio support
[in progress] pcie support
known limitations
-----------------
* rtc=on is required for now.
* can't use separate code/vars (actually an microvm limitation,
there is no pflash support).
* transitional virtio-pci devices do not work. microvm doesn't
support ioports on pcie, and ovmf doesn't initialize pcie devices
with ioports if there is no address space for them (even though
pcie devices are required to be functional without ioports).
usage
-----
qemu-system-x86_64 \
-nographic \
-machine microvm,acpi=on,pit=off,pic=off,rtc=on \
-bios /path/to/MICROVM.fd \
[ ... more args here ... ]
|
