Architectures that do not support memory hotplugging will fail when
memory limits are set because that amount is hotplugged. Issue a warning
instead. The long-term solution is virtio-mem.
Fixes: #1412
Signed-off-by: Jakob Naucke <jakob.naucke@ibm.com>
Currently the virtio-mem device is hotplugged on the root bus.
This doesn't work for PCIe machines like q35.
Hotplug the virtio-mem device into the pci bridge instead.
Fixes#1953
Signed-off-by: Marcel Apfelbaum <marcel@redhat.com>
Keeping around two different x86 machines has no added value
and require more tests and maintenance. Prefer the q35 machine
since it has more features and drop the pc machine.
Fixes#1953
Depends-on: github.com/kata-containers/tests#3586
Signed-off-by: Marcel Apfelbaum <marcel@redhat.com>
`memory_offset` is used to increase the maximum amount of memory
supported in a VM, this offset is equal to the NVDIMM/PMEM device that
is hot added, in real use case workloads such devices are bigger than
4G, which is the current limit (uint32).
fixes#2006
Signed-off-by: Julio Montes <julio.montes@intel.com>
Add sandbox, container, and hypervisor IDs to trace spans. Note that
some spans in sandbox.go are created with a trace() call from api.go.
These spans have additional attributes set after span creation to
overwrite the api attributes.
Fixes#1878
Signed-off-by: Chelsea Mafrica <chelsea.e.mafrica@intel.com>
Export proc stats for virtiofsd.
This commit only adds for hypervisors that have support for it.
- qemu
- cloud-hypervisor
Fixes: #1926
Signed-off-by: Carlos Venegas <jos.c.venegas.munoz@intel.com>
CloudHypervisor is using virtiofsd.go to manage virtiofsd process,
but qemu has its code in qemu.go. This commit let qemu to re-use
code in virtiofsd.go to reduce code and improve maintenanceability.
Fixes: #1933
Signed-off-by: bin <bin@hyper.sh>
Span attributes (tags) are not consistent in runtime tracing, so
designate and use core attributes such source, package, subsystem, and
type as span metadata for more understandable output.
Use WithAttributes() during span creation to reduce calls to
SetAttributes().
Modify Trace() in katautils to accept slice of attributes so multiple
functions using different attributes can use it.
Fixes#1852
Signed-off-by: Chelsea Mafrica <chelsea.e.mafrica@intel.com>
Define the structure and functions needed to support confidential
guests, this commit doesn't add support for any specific technology,
support for TDX, SEV, PEF and others will be added in following
commits.
Signed-off-by: Julio Montes <julio.montes@intel.com>
Got:
FATA[0000] run pod sandbox: rpc error: code = Unknown desc = failed to
create containerd task: Add 189759MB virtio-mem-pci fail QMP command
failed: backend memory size must be multiple of 0x200000: unknown
This commit let sizeMB be multiple of 2Mib to fix the issue.
Fixes: #1796
Signed-off-by: Hui Zhu <teawater@antfin.com>
Allow and configure vhost-user-fs devices (virtio-fs) on s390x. As a
consequence, appendVhostUserDevice now takes a context, which affects
its signature for other architectures.
Fixes: #1753
Signed-off-by: Jakob Naucke <jakob.naucke@ibm.com>
This reverts commit 7f60911333.
Patch allowed other vhost user devices besides FS not supported on s390x
and failed to attach a CCW device number, which results in the
inavailability to use more devices after vhost-user-fs-ccw.
Signed-off-by: Jakob Naucke <jakob.naucke@ibm.com>
If the QEMU VMM fails to launch, we currently fail to kill virtiofsd,
resulting in leftover processes running on the host. Let's make sure we
kill these, and explicitly cleanup the virtiofs socket on the
filesystem.
Ideally we'll migrate QEMU to utilize the same virtiofsd interface that
CLH uses, but let's fix this bug as a first step.
Fixes: #1755
Signed-off-by: Eric Ernst <eric_ernst@apple.com>
Remove the prohibition of vhost-user devices on s390x, which are by now
supported (e.g. vhost-user-fs-ccw). As a consequence,
appendVhostUserDevice no longer needs an error in its signature.
This enables virtio-fs support on s390x.
Fixes: #1469
Signed-off-by: Jakob Naucke <jakob.naucke@ibm.com>
Fixed logic used to handle static agent tracing.
For a standard (untraced) hypervisor shutdown, the runtime kills the VM
process once the workload has finished. But if static agent tracing is
enabled, the agent running inside the VM is responsible for the
shutdown. The existing code handled this scenario but did not wait for
the hypervisor process to end. The outcome of this being that the
console watcher thread was killed too early.
Although not a problem for an untraced system, if static agent tracing
was enabled, the logs from the hypervisor would be truncated, missing the
crucial final stages of the agents shutdown sequence.
The fix necessitated adding a new parameter to the `stopSandbox()` API,
which if true requests the runtime hypervisor logic simply to wait for
the hypervisor process to exit rather than killing it.
Fixes: #1696.
Signed-off-by: James O. D. Hunt <james.o.hunt@intel.com>
#1389 has added a context for many signatures to improve trace spans.
Functions specific to s390x lack this. Add context where required. This
affects some common code signatures, since some functions that do not
require context on other architectures do require it on s390x.
Also remove an unnecessary import in test_qemu_s390x.go.
Fixes: #1562
Signed-off-by: Jakob Naucke <jakob.naucke@ibm.com>
It turns out we have managed to break the static checker in many
difference places with the absence of static checker in github action.
Let's fix them while enabling static checker in github actions...
Signed-off-by: Peng Tao <bergwolf@hyper.sh>
A significant number of trace calls did not use a parent context that
would create proper span ordering in trace output. Add local context to
functions for use in trace calls to facilitate proper span ordering.
Additionally, change whether trace function returns context in some
functions in virtcontainers and use existing context rather than
background context in bindMount() so that span exists as a child of a
parent span.
Fixes#1355
Signed-off-by: Chelsea Mafrica <chelsea.e.mafrica@intel.com>
VhostUserDeviceAttrs::PCIAddr didn't actually store a PCI address
(DDDD:BB:DD.F), but rather a PCI path. Use the PciPath type and
rename things to make that clearer.
TestHandleBlockVolume previously used the bizarre value "0001:01"
which is neither a PCI address nor a PCI path for this value. Change
it to a valid PCI path - it appears the actual value didn't matter for
that test, as long as it was consistent.
Forward port of
3596058c67fixes#1040
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
BlockDrive::PCIAddr doesn't actually store a PCI address
(DDDD:BB:DD.F) but a PCI path. Use the PciPath type and rename things
to make that clearer.
TestHandleBlockVolume() previously used a bizarre value "0002:01" for
the "PCI address" which was neither an actual PCI address, nor a PCI
path. Update it to use a PCI path - the actual value appears not to
matter in this test, as long as its consistent throughout.
Forward port of
64751f377bfixes#1040
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
The "PCI address" returned by Endpoint::PciPath() isn't actually a PCI
address (DDDD:BB:DD.F), but rather a PCI path. Rename and use the
PciPath type to clean this up and the various parts of the network
code connected to it.
Forward port of
3e589713cffixes#1040
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
This commit includes two changes:
- migrate from opentracing to opentelemetry
- add jaeger configuration items
Fixes: #1351
Signed-off-by: bin <bin@hyper.sh>
acpi is enabled for kata 1.x, port and rebase code for 2.x
including:
runtime: enable pflash;
agent: add acpi support for pci bus path;
packaging: enable CONFIG_RTC_DRV_EFI;
Fixes: #1317
Signed-off-by: Jianyong Wu <jianyong.wu@arm.com>
On pod delete, we were looking to read files that we had just deleted. In particular,
stopSandbox for QEMU was called (we cleanup up vmpath), and then QEMU's
save function was called, which immediately checks for the PID file.
Let's only update the persist store for QEMU if QEMU is actually
running. This'll avoid Error messages being displayed when we are
stopping and deleting a sandbox:
```
level=error msg="Could not read qemu pid file"
```
I reviewed CLH, and it looks like it is already taking appropriate
action, so no changes needed.
Ideally we won't spend much time saving state to persist.json unless
there's an actual error during stop/delete/shutdown path, as the persist will
also be removed after the pod is removed. We may want to optimize this,
as currently we are doing a persist store when deleting each container
(after the sandbox is stopped, VM is killed), and when we stop the sandbox.
This'll require more rework... tracked in:
https://github.com/kata-containers/kata-containers/issues/1181Fixes: #1179
Signed-off-by: Eric Ernst <eric.g.ernst@gmail.com>
Remove global sandbox variable, and save *Sandbox to hypervisor struct.
For some needs, hypervisor may need to use methods from Sandbox.
Signed-off-by: bin liu <bin@hyper.sh>
Use `if err := q.qmpSetup(); err != nil` to reduce code and make it easy
to read. And remove checking err if last function call also return an error,
return the function call directly.
Fixes: #1081
Signed-off-by: bin liu <bin@hyper.sh>
When guest panic, dump guest kernel memory to host filesystem.
And also includes:
- hypervisor config
- hypervisor version
- and state of sandbox
Fixes: #1012
Signed-off-by: bin liu <bin@hyper.sh>
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>
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>
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>
Import new console watcher to monitor guest console outputs, and will be
only effective when we turn on enable_debug option.
Guest console outputs may include guest kernel debug info, agent debug info,
etc.
Fixes: #389
Signed-off-by: Penny Zheng penny.zheng@arm.com
With kata containers moving to 2.0, (hybrid-)vsock will be the only
way to directly communicate between host and agent.
And kata-proxy as additional component to handle the multiplexing on
serial port is also no longer needed.
Cleaning up related unit tests, and also add another mock socket type
`MockHybridVSock` to deal with ttrpc-based hybrid-vsock mock server.
Fixes: #389
Signed-off-by: Penny Zheng penny.zheng@arm.com
Print a warning message when the device to append to a QEMU VM is not
supported. This change is just to improve debuggability.
Signed-off-by: Julio Montes <julio.montes@intel.com>
[ port from runtime commit 379f19f7ccd71ebe938d9d6fe3cfe5f05f4f02bf ]
Add default value for Clock, otherwise rtc parameter will be dropped
by Valid function. "host" is the default value in qemu for rtc clock.
Signed-off-by: Shuicheng Lin <shuicheng.lin@intel.com>
Signed-off-by: Peng Tao <bergwolf@hyper.sh>
[ port from runtime commit 0100af18a2afdd6dfcc95129ec6237ba4915b3e5 ]
To control whether guest can enable/disable some CPU features. E.g. pmu=off,
vmx=off. As discussed in the thread [1], the best approach is to let users
specify them. How about adding a new option in the configuration file.
Currently this patch only supports this option in qemu,no other vmm.
[1] https://github.com/kata-containers/runtime/pull/2559#issuecomment-603998256
Signed-off-by: Jia He <justin.he@arm.com>
Signed-off-by: Peng Tao <bergwolf@hyper.sh>
[ port from runtime commit 7b269ff7aa2d62fe12593ff7040798e6c9bd5d65 ]
If we take one of the error paths from setupVirtiofsd() after
opening the fd variable, the fd.Close() function is not called.
Signed-off-by: Christophe de Dinechin <dinechin@redhat.com>
Signed-off-by: Peng Tao <bergwolf@hyper.sh>
The qemuPaths field in qemuArchBase maps from machine type to the default
qemu path. But, by the time we construct it, we already know the machine
type, so that entry ends up being the only one we care about.
So, collapse the map into a single path. As a bonus, the qemuPath()
method can no longer fail.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
The supportedQemuMachines array in qemuArchBase has a list of all the
qemu machine types supported for the architecture, with the options
for each. But, the machineType field already tells us which of the
machine types we're actually using, and that's the only entry we
actually care about.
So, drop the table, and just have a single value with the machine type
we're actually using. As a bonus that means the machine() method can
no longer fail, so no longer needs an error return.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Currently, newQemuArch() doesn't return an error. So, if passed an invalid
machine type, it will return a technically valid, but unusable qemuArch
object, which will probably fail with other errors shortly down the track.
Change this, to more cleanly fail the newQemuArch itself, letting us
detect a bad machine type earlier.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
As for some hypervisors, like firecracker, they support built-in rate limiter
to control network I/O bandwidth on VMM level. And for some hypervisors, like qemu,
they don't.
Fixes: #250
Signed-off-by: Penny Zheng <penny.zheng@arm.com>
The virtiofs daemon may run into errors other than the file
not existing, e.g. the file may not be executable.
Fixes: #2682
Message is now:
virtiofs daemon /usr/local/bin/hello returned with error:
fork/exec /usr/local/bin/virtiofsd: permission denied
instead of
panic: runtime error: invalid memory address or nil
Fixes: #2582
Message is now:
virtiofs daemon /usr/local/bin/hello-not-found returned with error:
fork/exec /usr/local/bin/hello-not-found: no such file or directory
instead of:
virtiofsd path (/usr/local/bin/hello-no-found) does not exist
Signed-off-by: Christophe de Dinechin <dinechin@redhat.com>
Signed-off-by: Fabiano Fidêncio <fidencio@redhat.com>
We need to make sure containers cannot modify host path unless it is explicitly shared to it. Right now we expose an additional top level shared directory to the guest and allow it to be modified. This is less ideal and can be enhanced by following method:
1. create two directories for each sandbox:
-. /run/kata-containers/shared/sandboxes/$sbx_id/mounts/, a directory to hold all host/guest shared mounts
-. /run/kata-containers/shared/sandboxes/$sbx_id/shared/, a host/guest shared directory (9pfs/virtiofs source dir)
2. /run/kata-containers/shared/sandboxes/$sbx_id/mounts/ is bind mounted readonly to /run/kata-containers/shared/sandboxes/$sbx_id/shared/, so guest cannot modify it
3. host-guest shared files/directories are mounted one-level under /run/kata-containers/shared/sandboxes/$sbx_id/mounts/ and thus present to guest at one level under /run/kata-containers/shared/sandboxes/$sbx_id/shared/
Signed-off-by: Peng Tao <bergwolf@hyper.sh>
Add a configuration option and a Pod Annotation
If activated:
- Add kernel parameters to load iommu
- Add irqchip=split in the kvm options
- Add a vIOMMU to the VM
Fixes#2694
Signed-off-by: Adrian Moreno <amorenoz@redhat.com>
Signed-off-by: Fabiano Fidêncio <fidencio@redhat.com>
To use the kata-containers repo path.
Most of the change is generated by script:
find . -type f -name "*.go" |xargs sed -i -e \
's|github.com/kata-containers/runtime|github.com/kata-containers/kata-containers/src/runtime|g'
Fixes: #201
Signed-off-by: Peng Tao <bergwolf@hyper.sh>
