The linux kernel feature RANDOMIZE_BASE improved the security and at
the same time increased the memory footprint of a kata container,
this feature was enabled in kata-containers/packaging#1006.
In order to mitigate this increase in memory consumption, we can
boot container using the uncompressed kernel.
Reduce boot time by ~5%
Reduce KSM memory footprint by ~14%
Reduce noKSM memory footprint by ~27%
fixes#669
Signed-off-by: Julio Montes <julio.montes@intel.com>
I noticed the spelling mistake while reviewing another change and
doing a "grep" for "privilege" that turned up nothing.
Fixes: #671
Signed-off-by: Christophe de Dinechin <dinechin@redhat.com>
Currently, isPCIeDevice() attempts to determine if a (host) device is
PCI-Express capable by looking up its link speed via the PCI slots
information in sysfs. This is a) complicated and b) wrong. PCI-e
devices don't have to have slots information, so this frequently fails.
Instead determine if devices are PCI-e by checking for the presence of
PCIe extended configuration space by looking at the size of the "config"
file in sysfs.
Forward ported from 6bf93b23 in the Kata 1.x runtime repository.
Fixes: #611
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Add guide on how to pass a VFIO-AP device, such as Crypto Express cards
on IBM Z mainframes, to a Kata container. Like the documentation for
VFIO-PCI, this was put in the virtcontainers README.
Fixes: #658
Signed-off-by: Jakob-Naucke <jakob.naucke@ibm.com>
Recognise when a device to be hot-plugged is an IBM Adjunct Processor
(AP) device and execute VFIO AP hot-plug accordingly. Includes unittest
for recognising and uses CCW for addDeviceToBridge in hotplugVFIODevice
if appropriate.
Fixes: #491
Signed-off-by: Jakob-Naucke <jakob.naucke@ibm.com>
Co-authored-by: Julio Montes <julio.montes@intel.com>
Reviewed-by: Alice Frosi <afrosi@redhat.com>
Since we always build musl kata-agent, there is no need to build
it inside a musl container. We can just build on the host and then
copy the binary to the target rootfs.
There are still a lot to clean up and it should be made so for ALL
target distros instead of just alpine. But this is at least working
for alpine first.
Signed-off-by: Peng Tao <bergwolf@hyper.sh>
for s390x virtio devices
Add iommu_platform annotations for qemu for ccw,
other supported devices can also make use of that.
Fixes#603
Signed-off-by: Qi Feng Huo <huoqif@cn.ibm.com>
We kept observing instabilities from CLH CI jobs periodically (kata
1.x). To separate the random failures caused by `seccomp` from other
failures, this patch disables the 'seccomp' option from clh in kata for
now. We will bring this option back after completing the 'seccomp'
filter lists based on Kata's CI workload. Details are tracked in the
following two issues:
https://github.com/kata-containers/runtime/issues/2899 and
https://github.com/kata-containers/runtime/issues/2901
We are facing the similar challenge to stabilize CI jobs related to
cloud-hypervisor in Kata 2.0. We are disabling the `seccomp` option here
for the same reason. Related issue:
https://github.com/kata-containers/tests/issues/2813Fixes: #614
Signed-off-by: Bo Chen <chen.bo@intel.com>
This is a re-vendor of intel/govmm, with support for hot-plugging IBM
Adjunct Processor (AP) devices over VFIO. This is necessary for
enabling AP device pass-through in Kata (see #491).
39c372a Add support for hot-plugging IBM VFIO-AP devices
f5bdd53 travis: disable amd64 jobs
1af1c0d github: enable github actions
4831c6e travis: Run coveralls after success
cf0f05d qemu: add iommu_platform knob for qemuParams
175ac49 typo fix
Fixes: #565
Signed-off-by: Jakob-Naucke <jakob.naucke@ibm.com>
When a one-shot pod dies in CRI-O, the shimv2 process isn't killed until
the pod is actually deleted, even though the VM is shut down. In this
case, the shim appears to busyloop when attempting to talk to the (now
dead) agent via VSOCK. To address this, we disconnect from the agent
after the VM is shut down.
This is especially catastrophic for one-shot pods that may persist for
hours or days, but it also applies to any shimv2 pod where Kata is
configured to use VSOCK for communication.
See github.com/kata-containers/runtime#2719 for details.
Fixes#2719
Signed-off-by: Evan Foster <efoster@adobe.com>
Expand unit tests for virtcontainers/utils/utils.go to include testing
CleanupFds, CPU calculations, ID string creation, and memory alignment
functions.
Fixes#490
Signed-off-by: Chelsea Mafrica <chelsea.e.mafrica@intel.com>
[ Port from runtime commit 44b58e4151d1fc7debed41274b65c37233a437e3 ]
This patch enables kata+clh to unplug block devices, which is required
to pass cri-o integration tests.
Fixes: #461
Signed-off-by: Bo Chen <chen.bo@intel.com>
[ Port from runtime commit 03fb9c50c180d3359178c30e06f1122df312ae76 ]
To support unplug block device, we need to set the 'Id' explicitly while
hotplugging devices with cloud-hypervisor HTTP API.
Fixes: #461
Signed-off-by: Bo Chen <chen.bo@intel.com>
[ Port from runtime commit 39897867bc89667daaafdd141367ec4a5fdc9247 ]
API now requires cpu topology.
Fixes: #461
Signed-off-by: Jose Carlos Venegas Munoz <jose.carlos.venegas.munoz@intel.com>
Signed-off-by: Bo Chen <chen.bo@intel.com>
[ Port from runtime commit 40f49312a4881c904a1cbdace04c4c697bd2d429 ]
Update api geneated by openapi.
Fixes: #461
Signed-off-by: Jose Carlos Venegas Munoz <jose.carlos.venegas.munoz@intel.com>
Signed-off-by: Bo Chen <chen.bo@intel.com>
[ Port from runtime commit 30b40f5505fd46d23b89eb5fb38301d2f7454f35 ]
Along with the release of cloud-hypervisor v0.8.0, this option has been
deprecated. clh now enforces to use the alternative controls,
e.g. "shared" and "hugepages", which can infer the backing file
paths. Also, we don't use "hugepages" in kata, so we are fine now as the
"shared" control is already enabled.
Fixes: #461
Signed-off-by: Bo Chen <chen.bo@intel.com>
Some type declarations were changed. The example code here is outdated
according to the example_pod_run_test.go under virtcontainers directory.
And add the imports to make where the types from clear.
Fixes: #507
Signed-off-by: Li Ning <lining_yewu@cmss.chinamobile.com>
It should wait until the stdin io copy
termianted to close the process's io stream,
otherwise, it would miss forwarding some contents
to process stdin.
Fixes: #439
Signed-off-by: fupan.lfp <fupan.lfp@antgroup.com>
The Kata architecture does not support rebooting VMs (the lifecycle
being start/exec/kill) and if a VM is killed (e.g. using sysrq-trigger),
the VM does not exit fully and other layers do not notice the state change.
Set the NoReboot config Knob so that govmmQemu.LaunchQemu() runs QEMU
with the --no-reboot command-line option.
Fixes: #2866
Signed-off-by: Liam Merwick <liam.merwick@oracle.com>
Add unit tests for types/container.go. Tests were adapted from
sandbox_test.go since ContainerState is a sandbox state structure and
the transition tests are the same.
Fixes#451
Signed-off-by: Chelsea Mafrica <chelsea.e.mafrica@intel.com>
As the current qemu of arm64 is so old, the new multidev parameter
in 9pfsdev is not supported on arm64, so disabled it temporarily.
Fixes:#466
Signed-off-by: Jianyong Wu <jianyong.wu@arm.com>
