Dave Gilbert brough up that passing --thread-pool-size=1 to virtiofsd
may result in a performance improvement especially when using
`cache=none`. While our current default is `cache=auto`, Dave mentioned
that he seems no harm in having it set and he also mentiond that it may
use a lot less stack space on aarch/arm.
Fixes: #943
Signed-off-by: Fabiano Fidêncio <fidencio@redhat.com>
Vivek Goyal found out that using "shared" thread pool, instead of
"exclusive" results in better performance.
Knowning that and with the plan to have virtio-fs as the default fs for
the 2.0, let's bring this patch in for both 5.0 and 5.1.
Fixes: #944
Signed-off-by: Fabiano Fidêncio <fidencio@redhat.com>
Update the build kata containers kernel document for 2.0 release. Fixed
the 1.x release project paths and urls, using the kata-containers
project file paths and urls.
Fixes: #929
Signed-off-by: Ychau Wang <wangyongchao.bj@inspur.com>
To update device resource entries from host to guest, we search for
the right entry by host major:minor numbers, then later update it.
However block and character devices exist in separate major:minor
namespaces so we could have one block and one character device with
matching major:minor and thus incorrectly update both with the details
for whichever device is processed second.
Add a check on device type to prevent this.
Port from the Kata 1 Go agent
https://github.com/kata-containers/agent/commit/27ebdc9d2761Fixes: #703
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
The agent needs to update device entries in the OCI spec so that it
has the correct major:minor numbers for the guest, which may differ
from the host.
Entries in the main device list are looked up by device path, but
entries in the device resources list are looked up by (host)
major:minor. This is done one device at a time, updating as we go in
update_spec_device_list().
But since the host and guest have different namespaces, one device
might have the same major:minor as a different device on the host. In
that case we could update one resource entry to the correct guest
values, then mistakenly update it again because it now matches a
different host device.
To avoid this, rather than looking up and updating one by one, we make
all the lookups in advance, creating a map from (host) device path to
the indices in the spec where the device and resource entries can be
found.
Port from the Go agent in Kata 1,
https://github.com/kata-containers/agent/commit/d88d46849130Fixes: #703
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
The Kata 1 Go agent included a unit test for updateSpecDeviceList, but no
such unit test exists for the Rust agent's equivalent
update_spec_device_list(). Port the Kata1 test to Rust.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
If update_spec_device_list() is given a device that can't be found in the
OCI spec, it currently does nothing, and returns Ok(()). That doesn't
seem like what we'd expect and is not what the Go agent in Kata 1 does.
Change it to return an error in that case, like Kata 1.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Allow for constraining the cpuset as well as the devices-whitelist . Revert
sandbox constraints for cpu/memory, as they break the K8S use case. Can
re-add behind a non-default flag in the future.
The sandbox CPUSet should be updated every time a container is created,
updated, or removed.
To facilitate this without rewriting the 'non constrained cgroup'
handling, let's add to the Sandbox's cgroupsUpdate function.
Signed-off-by: Eric Ernst <eric.g.ernst@gmail.com>
We were assuming base 10 string before, when the block size from sysfs
is actually a hex string. Let's fix that.
Fixes: #908
Signed-off-by: Eric Ernst <eric.g.ernst@gmail.com>
Create a containerd installation guide and a new `kata-manager` script
for 2.0 that automated the steps outlined in the guide.
Also cleaned up and improved the installation documentation in various
ways, the most significant being:
- Added legacy install link for 1.x installs.
- Official packages section:
- Removed "Contact" column (since it was empty!)
- Reworded "Versions" column to clarify the versions are a minimum
(to reduce maintenance burden).
- Add a column to show which installation methods receive automatic updates.
- Modified order of installation options in table and document to
de-emphasise automatic installation and promote official packages
and snap more.
- Removed sections no longer relevant for 2.0.
Fixes: #738.
Signed-off-by: James O. D. Hunt <james.o.hunt@intel.com>
Sometimes `Option.or_or` and `Result.map_err` may be simpler
than match statement. Especially in rpc.rs, there are
many `ctr.get_process` and `sandbox.get_container` which
are using `match`.
Signed-off-by: bin liu <bin@hyper.sh>
Remove the build in shim and proxy desgin description from the
kata-api-design.md file.
Fixes: #912
Signed-off-by: Ychau Wang <wangyongchao.bj@inspur.com>
There are some uses/codes/struct fields are commented out, and
may not turn into un-comment these codes, so delete these comments.
Signed-off-by: bin liu <bin@hyper.sh>
Use rust `Result`'s `or_else`/`and_then` can write clean codes.
And can avoid early return by check wether the `Result`
is `Ok` or `Err`.
Signed-off-by: bin liu <bin@hyper.sh>
Qemu v5.1 was released with an affending commit 9b3a35ec82
(virtio: verify that legacy support is not accidentally on).
As a result, it breaks commandline compatiblilities for old qemu
users. Upstream qemu has fixed it but no release has been put out yet.
Let's apply these fixes by hand for now.
Refs: https://www.mail-archive.com/qemu-devel@nongnu.org/msg729556.html
Depends-on: github.com/kata-containers/tests#2945
Signed-off-by: Peng Tao <bergwolf@hyper.sh>
This commit includes:
- update comments that not matched the function name
- file path with doubled slash
Fixes: #922
Signed-off-by: bin liu <bin@hyper.sh>
In function parse_cmdline there are some similar codes, if we want
to add more commandline arguments, the code will grow too long.
Use macro can reduce some codes with the same logic/processing.
Fixes: #914
Signed-off-by: bin liu <bin@hyper.sh>
If no namespace field in config files, CRI-O will failed:
setting pod sandbox name and id: cannot generate pod name without namespace
Signed-off-by: bin liu <bin@hyper.sh>
Run the snap CI on every PR is not needed. Don't run the snap CI
on PRs that don't change the source code (*.go/*.rs), a configuration
file or Makefile.
fixes#896
Signed-off-by: Julio Montes <julio.montes@intel.com>
Attackers might use it to explore other containers in the same pod.
While it is still safe to allow it, we can just close the race window
like runc does.
Fixes: #885
Signed-off-by: Peng Tao <bergwolf@hyper.sh>
On old kernels (like v4.9), kernel applies CLOECEC in wrong order w.r.t.
dumpable task flags. As a result, we might leak guest file descriptor to
containers. This is a former runc CVE-2016-9962 and still applies to
kata agent. Although Kata container is still valid at protecting the
host, we should not leak extra resources to user containers.
This sets the init processes that join and setup the container's
namespaces as non-dumpable before they setns to the container's pid (or
any other ) namespace.
This settings is automatically reset to the default after the Exec in
the container so that it does not change functionality for the
applications that are running inside, just our init processes.
This prevents parent processes, the pid 1 of the container, to ptrace
the init process before it drops caps and other sets LSMs.
The order during the exec syscall is that the process is set back to
dumpable before O_CLOEXEC are processed.
Refs:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=613cc2b6f272c1a8ad33aefa21cad77af23139f7https://github.com/torvalds/linux/blob/v4.9/fs/exec.c#L1290-L1318opencontainers/runc@50a19c6https://nvd.nist.gov/vuln/detail/CVE-2016-9962Fixes: #890
Signed-off-by: Peng Tao <bergwolf@hyper.sh>
Simply running `make` would generate some cargo lock updates for
agent-ctl. Let's include them so that we have fixed dependencies.
Fixes: #883
Signed-off-by: Peng Tao <bergwolf@hyper.sh>
