The new API is based on containerd's cgroups package.
With that conversion we can simpligy the virtcontainers sandbox code and
also uniformize our cgroups external API dependency. We now only depend
on containerd/cgroups for everything cgroups related.
Depends-on: github.com/kata-containers/tests#3805
Signed-off-by: Samuel Ortiz <samuel.e.ortiz@protonmail.com>
Signed-off-by: Eric Ernst <eric_ernst@apple.com>
Eventually, we will convert the virtcontainers and the whole Kata
runtime code base to only rely on that package.
This will make Kata only depends on the simpler containerd cgroups API.
Signed-off-by: Samuel Ortiz <samuel.e.ortiz@protonmail.com>
The only process we are adding there is the container host one, and
there is no such thing anymore.
Signed-off-by: Samuel Ortiz <samuel.e.ortiz@protonmail.com>
This is a simplification of the host cgroup handling by partitioning the
host cgroups into 2: A sandbox cgroup and an overhead cgroup.
The sandbox cgroup is always created and initialized. The overhead
cgroup is only available when sandbox_cgroup_only is unset, and is
unconstrained on all controllers. The goal of having an overhead cgroup
is to be more flexible on how we manage a pod overhead. Having such
cgroup will allow for setting a fixed overhead per pod, for a subset of
controllers, while at the same time not having the pod being accounted
for those resources.
When sandbox_cgroup_only is not set, we move all non vCPU threads
to the overhead cgroup and let them run unconstrained. When it is set,
all pod related processes and threads will run in the sandbox cgroup.
Signed-off-by: Samuel Ortiz <samuel.e.ortiz@protonmail.com>
Regardless of the sandbox_cgroup_only setting, we create the sandbox
cgroup manager and set the sandbox cgroup path at the same time.
Without doing this, the hypervisor constraint routine is mostly a NOP as
the sandbox state cgroup path is not initialized.
Fixes#2184
Signed-off-by: Samuel Ortiz <samuel.e.ortiz@protonmail.com>
Sync the virtcontainers api.md document, add `ConfidentialGuest` `EntropySourceList` `GuestSwap` three
fields to the HypervisorConfig API.
Fixes#2625
Signed-off-by: wangyongchao.bj <wangyongchao.bj@inspur.com>
sync the virtcontainers api.md document, add SandboxBindMounts field to the SandboxConfig API.
And update the order of the SandboxConfig API fields.
Fixes#2621
Signed-off-by: wangyongchao.bj <wangyongchao.bj@inspur.com>
If the device has no permission, such as /dev/null, /dev/urandom,
it needs to be added into cgroup.
Fixes: #2615
Signed-off-by: Binbin Zhang <binbin36520@gmail.com>
This brings it back into line with the normal qemu version. We refer to
v6.1.0 by full SHA in versions.yaml, rather than the tag, so that
apply_patches.sh sees it as different and applies the virtiofs DAX patches
which is what the experimental version is actually about having.
The virtiofs DAX patches themselves are updated to the version from
https://gitlab.com/virtio-fs/qemu, virtio-fs-dev branch as of commit
3620cb0a.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
We need qemu-6.1 for ACPI PCI hotplug support for the q35 machine. At the
moment qemu will use SHPC hotplug under the PCIe to PCI bridge on q35.
SHPC is too slow to use for our purposes (it requires a 5s delay).
Update the qemu version to v6.1.0. This leaves the experimental version
*older* than the normal version, but we'll fix that up later.
We also need to tweak the snapcraft.yaml, since the location for configs
has changed in the new qemu version.
fixes#1691
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
In getThreadIDs(), the cpuID variable is derived from a string that
already contains a whitespace. As a result, strings.SplitAfter returns
the cpuID with a leading space. This makes any go variant of string to int
fail (strconv.ParseInt() in our case). This patch makes sure that the
leading space character is removed so the string passed to
strconv.ParseInt() is "CPUID" and not " CPUID".
This has been caused by a change in the naming scheme of vcpu threads
for Firecracker after v0.19.1.
Fixes: #2592
Signed-off-by: Anastassios Nanos <ananos@nubificus.co.uk>
If the script doesn't find a patches directory it expects, it gives an
error saying to create a dummy 'no_patches' file if you really don't want
any patches applied for that version.
But actual practice in the tree is to call the dummy file 'no_patches.txt'
rather than simply 'no_patches'. Correct the message to match existing
practice.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Given the fix to the bugs of the openapi spec file is included in the
Cloud Hypervisor v18.0 [1], this patch reverts the workaround we carried
in the CLH driver.
This reverts commit 932ee41b3f.
[1] https://github.com/cloud-hypervisor/cloud-hypervisor/pull/3029
Signed-off-by: Bo Chen <chen.bo@intel.com>
Highlights from the Cloud Hypervisor release v18.0: 1) Experimental User
Device (vfio-user) support; 2) Migration support for vhost-user devices;
3) VHDX disk image support; 4) Device pass through on MSHV hypervisor;
5) AArch64 for support virtio-mem; 6) Live migration on MSHV hypervisor;
7) AArch64 CPU topology support; 8) Power button support on AArch64; 9)
Various bug fixes on PTY, TTY, signal handling, and live-migration on
AArch64.
Details can be found: https://github.com/cloud-hypervisor/cloud-hypervisor/releases/tag/v18.0Fixes: #2543
Signed-off-by: Bo Chen <chen.bo@intel.com>
make kata-tarball is the main way to
build a kata in a single host. Lets
test it to make sure it works on every PR.
Fixes: #2416
Signed-off-by: Carlos Venegas <jose.carlos.venegas.munoz@intel.com>
- Run kata-deploy tarball generation action on every PR.
- Use kata-deploy makefile targets.
Signed-off-by: Carlos Venegas <jose.carlos.venegas.munoz@intel.com>
If a binary tarball for a single component is done,
the logs will be shown in stdout.
e.g.
make kernel-tarball
To build all a the same time still store logs in files.
make kata-tarball
Signed-off-by: Carlos Venegas <jose.carlos.venegas.munoz@intel.com>
Now that local-build kata-deploy makefile is inlucded in toplevel
makefile, lets use the suffix `-tarball` to avoid name collitions
and identify the tarball releted targets.
Signed-off-by: Carlos Venegas <jose.carlos.venegas.munoz@intel.com>
Use kata-deploy targets from toplevel.
This will help if want to build and
reinstall just one single kata component.
Signed-off-by: Carlos Venegas <jose.carlos.venegas.munoz@intel.com>
Add how-to-setup-swap-devices-in-guest-kernel.md to how-to to introduce
how to setup swap device in guest kernel.
Fixes: #2326
Signed-off-by: Hui Zhu <teawater@antfin.com>
adds the default devices for unix such as /dev/null, /dev/urandom to
the container's resource cgroup spec
Fixes: #2539
Signed-off-by: Binbin Zhang <binbin36520@gmail.com>
This adds fields to BridgeDevice struct to allow qemu's io-reserve,
mem-reserve and pref64-reserve properties to be set for PCI bridges.
This is needed for Kata's upcoming change to ACPI hotplug.
fixes#200
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
zsh doesn't support "==" as equal comparison operator, so
replace "==" with "=" to make the script more portable
Fixes: #2584
Signed-off-by: Binbin Zhang <binbin36520@gmail.com>
The current default config in qemu for arm64 doesn't suit for qemu
version 5.1+, so remove them here.
Fixes: #2595
Signed-off-by: Jianyong Wu <jianyong.wu@arm.com>
