7e58278858
It appears that the PCI BAR memory is slow to access with remap_pfn_range
and that it should instead be faulted in one page at a time.
The commit
|
||
---|---|---|
.. | ||
dkms.conf | ||
kvmfr.c | ||
kvmfr.h | ||
Makefile | ||
README.md | ||
test.c | ||
test.expected |
This kernel module implements a basic interface to the IVSHMEM device for LookingGlass when using LookingGlass in VM->VM mode.
Additionally, in VM->host mode, it can be used to generate a shared memory device on the host machine that supports dmabuf.
Compiling (Manual)
Make sure you have your kernel headers installed first, on Debian/Ubuntu use the following command.
apt-get install linux-headers-$(uname -r)
Then simply run make
and you're done.
Loading
For VM->VM mode, simply run:
insmod kvmfr.ko
For VM->host mode with dmabuf, instead of creating a shared memory file, load
this module with the parameter static_size_mb
. For example, a 128 MB shared
memory device can be created with:
insmod kvmfr.ko static_size_mb=128
Multiple devices can be created by separating the sizes with commas. For
example, static_size_mb=128,64
would create two kvmfr devices: kvmfr0
would be 128 MB and kvmfr1
would be 64 MB.
Compiling & Installing (DKMS)
You can install this module into DKMS so that it persists across kernel upgrades. Simply run:
dkms install .
Loading
For VM->VM, simply modprobe the module:
modprobe kvmfr
For VM->host with dmabuf, modprobe with the parameter static_size_mb
:
modprobe kvmfr static_size_mb=128
Just like above, multiple devices can be created by separating the sizes with commas.
Usage
This will create the /dev/kvmfr0
node that represents the KVMFR interface.
To use the interface you need permission to access it by either creating a
udev rule to ensure your user can read and write to it, or simply change its
ownership manually, ie:
sudo chown user:user /dev/kvmfr0
An example udev rule, which you can put in /etc/udev/rules.d/99-kvmfr.rules
,
is (replace user
with your username):
SUBSYSTEM=="kvmfr", OWNER="user", GROUP="kvm", MODE="0660"
Usage with looking glass is simple, you only need to specify the path to the device node, for example:
./looking-glass-client -f /dev/kvmfr0
You may also use a config file: ~/.looking-glass-client.ini
, or /etc/looking-glass-client.ini
.
[app]
shmFile=/dev/kvmfr0
VM->Host
In VM->host mode, use this device in place of the shared memory file.
For example, with qemu
, you would use the following arguments:
-device ivshmem-plain,id=shmem0,memdev=looking-glass
-object memory-backend-file,id=looking-glass,mem-path=/dev/kvmfr0,size=128M,share=yes
Note that the size
argument must be the same size as what you passed
to static_size_mb
argument for the kernel module.
libvirt
With libvirt
, you can use the following XML block:
<qemu:commandline>
<qemu:arg value='-device'/>
<qemu:arg value='ivshmem-plain,id=shmem0,memdev=looking-glass'/>
<qemu:arg value='-object'/>
<qemu:arg value='memory-backend-file,id=looking-glass,mem-path=/dev/kvmfr0,size=128M,share=yes'/>
</qemu:commandline>
Remember to add xmlns:qemu='http://libvirt.org/schemas/domain/qemu/1.0'
to
the <domain>
.
On certain distros, running libvirt this way poses issues with apparmor and cgroups.
For apparmor, in /etc/apparmor.d/abstractions/libvirt-qemu
, append:
# Looking Glass
/dev/kvmfr0 rw,
For cgroups, in /etc/libvirt/qemu.conf
, uncomment the cgroup_device_acl
block and add /dev/kvmfr0
to the list. Then restart libvirtd
:
sudo systemctl restart libvirtd.service