Until now, runtime-rs couldn't be compiled on s390x.
We need to lift those restrictions in Makefile first.
Fixes: #8446
Signed-off-by: Hyounggyu Choi <Hyounggyu.Choi@ibm.com>
It fails to compile virt_container because Dragonball is only
used in the implementation of the trait method Persist::restore().
As the hypervisor is not compiled on s390x and QEMU implements
the trait method, this commit is to let the method use QEMUi's.
Signed-off-by: Hyounggyu Choi <Hyounggyu.Choi@ibm.com>
Dragonball and cloud-hypervisor are not supported on s390x.
We need to exclude the plugins for these hypervisors from compilation.
Signed-off-by: Hyounggyu Choi <Hyounggyu.Choi@ibm.com>
Kata CI's pod-sandbox-vcpus-allocation.yaml ends with "---", so the
empty YAML document following that line should be ignored.
To test this fix:
genpolicy -u -y pod-sandbox-vcpus-allocation.yaml
Fixes: #8895
Signed-off-by: Dan Mihai <dmihai@microsoft.com>
Allow users to specify in genpolicy-settings.json a default cluster
namespace other than "default". For example, Kata CI uses as default
namespace: "kata-containers-k8s-tests".
Fixes: #8976
Signed-off-by: Dan Mihai <dmihai@microsoft.com>
There is a race condition in agent HVSOCK_STREAMS hashmap, where a
stream may be taken before it is inserted into the hashmap. This patch
add simple retry logic to the stream consumer to alleviate this issue.
Fixes: #6714
Signed-off-by: Zixuan Tan <tanzixuan.me@gmail.com>
Linux forbids opening an existing socket through /proc/<pid>/fd/<fd>,
making some images relying on the special file /dev/stdout(stderr),
/proc/self/fd/1(2) fail to boot in passfd io mode, where the
stdout/stderr of a container process is a vsock socket.
For back compatibility, a pipe is introduced between the process
and the socket, and its read end is set as stdout/stderr of the
container process instead of the socket. The agent will do the
forwarding between the pipe and the socket.
Fixes: #6714
Signed-off-by: Zixuan Tan <tanzixuan.me@gmail.com>
In passfd io mode, when not using a terminal, the stdout/stderr vsock
streams are directly used as the stdout/stderr of the child process.
These streams are non-blocking by default.
The stdout/stderr of the process should be blocking, otherwise
the process may encounter EAGAIN error when writing to stdout/stderr.
Fixes: #6714
Signed-off-by: Zixuan Tan <tanzixuan.me@gmail.com>
We want the io connection keep connected when the containerd closed
the io pipe, thus it can be attached on the io stream.
Signed-off-by: Fupan Li <fupan.lfp@antgroup.com>
Support the hybrid fd passthrough mode with passing pipe fd,
which can specify this connection kept even when the pipe
peer closed, and this connection can be reget wich re-opening
the pipe.
Signed-off-by: Fupan Li <fupan.lfp@antgroup.com>
In linux, when a FIFO is opened and there are no writers, the reader
will continuously receive the HUP event. This can be problematic
when creating containers in detached mode, as the stdin FIFO writer
is closed after the container is created, resulting in this situation.
In passfd io mode, open stdin fifo with O_RDWR|O_NONBLOCK to avoid the
HUP event.
Fixes: #6714
Signed-off-by: Zixuan Tan <tanzixuan.me@gmail.com>
When container exits, the agent should clean up the term master fd,
otherwise the fd will be leaked.
Fixes: kata-containers#6714
Signed-off-by: Zixuan Tan <tanzixuan.me@gmail.com>
When one end of the connection close, the epoll event will be triggered
forever. We should close the connection and kill the connection.
Fixes: #6714
Signed-off-by: Zixuan Tan <tanzixuan.me@gmail.com>
Currently in the kata container, every io read/write operation requires
an RPC request from the runtime to the agent. This process involves
data copying into/from an RPC request/response, which are high overhead.
To solve this issue, this commit utilize the vsock fd passthrough, a
newly introduced feature in the Dragonball hypervisor. This feature
allows other host programs to pass a file descriptor to the Dragonball
process, directly as the backend of an ordinary hybrid vsock connection.
The runtime-rs now utilizes this feature for container process io. It
open the stdin/stdout/stderr fifo from containerd, and pass them to
Dragonball, then don't bother with process io any more, eliminating
the need for an RPC for each io read/write operation.
In passfd io mode, the agent uses the vsock connections as the child
process's stdin/stdout/stderr, eliminating the need for a pipe
to bump data (in non-tty mode).
Fixes: #6714
Signed-off-by: Zixuan Tan <tanzixuan.me@gmail.com>
Two toml options, `use_passfd_io` and `passfd_listener_port` are introduced
to enable and configure dragonball's vsock fd passthrough io feature.
This commit is a preparation for vsock fd passthrough io feature.
Fixes: #6714
Signed-off-by: Zixuan Tan <tanzixuan.me@gmail.com>
Using custom input paths with -i is counter-intuitive. Simplify path handling with explicit flags for rules.rego and genpolicy-settings.json.
Fixes: #8568
Signed-Off-By: Malte Poll <1780588+malt3@users.noreply.github.com>
Removed the setting of default values for runtime fields. Added explicit checks for missing or empty fields, reporting errors with clear messages.
Fixes: #8838
Signed-off-by: yaoyinnan <35447132+yaoyinnan@users.noreply.github.com>
The `noop-method-call` is a rustc lint that has existed since v1.52.0.
This lint has been moved to the warn by default lint level since v1.73.0.
Therefore build is failing with this version and above.
This commit removes the unnecessary call to `<&T as Deref>::deref` on `T: !Deref`.
Fixes: #8586
Signed-off-by: Kvlil <kalil.pelissier@gmail.com>
The auto-generated Policy already allows these volumes to be mounted,
regardless if they are:
- Present, or
- Missing and optional
Fixes: #8893
Signed-off-by: Dan Mihai <dmihai@microsoft.com>
Qemu stderr monitoring runs in its own asynchronous green thread.
For that, `stderr` is taken out of the Child representing the qemu child
process to avoid partial move and make it possible for the main thread
still to call functions on QemuInner::qemu_process (e.g. kill(), id()).
Fixes#8937
Signed-off-by: Pavel Mores <pmores@redhat.com>
We'll want to capture qemu's stderr in parallel with normal runtime-rs
execution. Tokio's primitives make this much easier than std's. This
also makes child process management more consistent across runtime-rs
(i.e. virtiofsd child process is already launched and managed using tokio).
Some changes were necessary due to tokio functions being slightly different
from their std counterparts. Child::kill() is now async and Child::id()
now returns an Option.
Signed-off-by: Pavel Mores <pmores@redhat.com>
Allow Kata CI's pod-nested-configmap-secret.yaml to work with
genpolicy and current cbl-mariner images:
1. Ignore the optional type field of Secret input YAML files.
It's possible that CoCo will need a more sophisticated Policy
for Secrets, but this change at least unblocks CI testing for
already-existing genpolicy features.
2. Adapt the value of the settings field below to fit current CI
images for testing on cbl-mariner Hosts:
"kata_config": {
"confidential_guest": false
},
Switching this value from true to false instructs genpolicy to
expect ConfigMap volume mounts similar to:
"configMap": {
"mount_type": "bind",
"mount_source": "$(sfprefix)",
"mount_point": "^$(cpath)/watchable/$(bundle-id)-[a-z0-9]{16}-",
"driver": "watchable-bind",
"fstype": "bind",
"options": [
"rbind",
"rprivate",
"ro"
]
},
instead of:
"confidential_configMap": {
"mount_type": "bind",
"mount_source": "$(sfprefix)",
"mount_point": "$(sfprefix)",
"driver": "local",
"fstype": "bind",
"options": [
"rbind",
"rprivate",
"ro"
]
}
},
This settings change unblocks CI testing for ConfigMaps.
Simple sanity testing for these changes:
genpolicy -u -y pod-nested-configmap-secret.yaml
kubectl apply -f pod-nested-configmap-secret.yaml
kubectl get pods | grep config
nested-configmap-secret-pod 1/1 Running 0 26s
Fixes: #8892
Signed-off-by: Dan Mihai <dmihai@microsoft.com>
This fixes a flaw pointed out in review of PR #8185. Creation of the
directory semantically fits better into VM preparation than VM launch.
Signed-off-by: Pavel Mores <pmores@redhat.com>
Validating the node name is currently outside the scope of the CoCo
policy.
This change unblocks testing using Kata CI's test-pod-file-volume.yaml
and pv-pod.yaml.
Fixes: #8888
Signed-off-by: Dan Mihai <dmihai@microsoft.com>
Add metadata containing the Policy annotation if the user didn't
provide any metadata in the input yaml file.
For a simple sanity test using a Kata CI YAML file:
genpolicy -u -y job.yaml
kubectl apply -f job.yaml
kubectl get pods | grep job
job-pi-test-64dxs 0/1 Completed 0 14s
Fixes: #8891
Signed-off-by: Dan Mihai <dmihai@microsoft.com>
SharedVersion fiel add a versiontable property that isn't supported by upstream QEMU.
This is dead code since virtcontainers isn't setting SharedVersions to true.
Fixes: #7720
Signed-off-by: Kvlil <kalil.pelissier@gmail.com>
