ExecuteSystemPowerdown issues `system_powerdown` and waits
for `SHUTDOWN`. The event emitted is `POWERDOWN` per spec.
Without this we get an error even though the VM has shutdown
gracefully.
Per QEMU spec:
```
POWERDOWN (Event)
Emitted when the virtual machine is powered down through the power
control system, such as via ACPI.
Since
0.12
Example
<- { "event": "POWERDOWN",
"timestamp": { "seconds": 1267040730, "microseconds": 682951 } }
SHUTDOWN (Event)
Emitted when the virtual machine has shut down, indicating that qemu is
about to exit.
Arguments
guest: boolean
If true, the shutdown was triggered by a guest request (such as a
guest-initiated ACPI shutdown request or other hardware-specific action)
rather than a host request (such as sending qemu a SIGINT). (since 2.10)
reason: ShutdownCause
The ShutdownCause which resulted in the SHUTDOWN. (since 4.0)
Note
If the command-line option “-no-shutdown” has been specified, qemu will
not exit, and a STOP event will eventually follow the SHUTDOWN event
Since
0.12
Example
<- { "event": "SHUTDOWN", "data": { "guest": true },
"timestamp": { "seconds": 1267040730, "microseconds": 682951 } }
```
Signed-off-by: Manohar Castelino <mcastelino@apple.com>
ExecuteBlockdevAddWithDriverCache has three one parameter driver
than ExecuteBlockdevAddWithCache.
Parameter driver can set the driver of block device.
Fixes: #198
Signed-off-by: Hui Zhu <teawater@antfin.com>
Use of the 'props' argument to 'object-add' has been deprecated since QEMU
5.0 (commit 5f07c4d60d09) in favor of flattening the properties directly
into the 'object-add' arguments. Support for 'props' is removed entirely
in qemu 6.0 (commit 50243407457a).
fixes#193
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Kata requires version 5.2 (or 5.1 on ARM) anyway. Simplify code by
dropping support for older versions. In any case explicit checks against
version number aren't necessarily reliable for patched qemu versions.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
ExecuteBlockdevAdd() and ExecuteBlockdevAddWithCache() both appear to be
intended to create block devices in the guest which backend onto a block
device in the host. That seems to be the way that Kata always uses it.
However blockdevAddBaseArgs(), used by both those functions always uses the
"file" driver, which is only intended for use with regular file backends.
Use of the "file" driver for host block devices was deprecated in qemu-3.0,
and has been removed entirely in qemu-6.0 (commit 8d17adf34f5). We should
be using the "host_device" driver instead.
fixes#191
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Currently virtio-mem-pci devices can be hotplugged only on the root bus.
This doesn't work for PCIe machines like q35.
Extend the API to optionally support hotplugging on PCI bridges.
Fixes: #176
Signed-off-by: Marcel Apfelbaum <marcel@redhat.com>
In Kata, we are getting a *lot* of logs at runtime from QMP, in particular `read from QMP: xxxx`
Ideally we'd set this to only be visible for trace, but I did not see this working when adding a
V(7) check around these prints. To avoid filling journal with info that isn't useful, let's drop.
Fixes: #165
Signed-off-by: Eric Ernst <eric.g.ernst@gmail.com>
Add ExecuteAPVFIOMediatedDeviceAdd to qmp.go, which executes a hotplug
for an IBM Adjunct processor (AP) VFIO device (see also
https://www.kernel.org/doc/html/latest/s390/vfio-ap.html )
Also includes the respective unittest and adds the VfioAP DeviceDriver
constant to qemu.go.
Pushing again due to incidental CI failure
Fixes: #133
Signed-off-by: Jakob-Naucke <jakob.naucke@ibm.com>
Reviewed-by: alicefr <afrosi@redhat.com>
vhost is a Netdev Tap Option used to configure a host TAP network interface
backend, according to the QMP API documentation the type for such option must
be a boolean. Use boolean type for vhost option to fix the following
error on recent versions of QEMU:
```
Invalid parameter type for 'vhost', expected: boolean
```
Signed-off-by: Julio Montes <julio.montes@intel.com>
According to QEMU's nvdimm documentation: When 'pmem' is 'on' and QEMU is
built with libpmem support, QEMU will take necessary operations to guarantee
the persistence of its own writes to the vNVDIMM backend.
Signed-off-by: Julio Montes <julio.montes@intel.com>
Currently, virtio transports for each device are determined with
architecture dependent build time conditionals. This isn't the ideal
solution, as virtio transports aren't exactly tied to the host's
architecture.
For example, aarch64 VMs do support both PCI and MMIO devices, and
after the recent introduction of the microvm machine type, that's also
the case for x86_64.
This patch extends each device that supports multiple transports with
a VirtioTransport field, so users of the library can manually specify
a transport for each device. To avoid breaking the compatibility, if
VirtioTransport is empty a behavior equivalent to the legacy one is
achieved by checking runtime.GOARCH and Config.Machine.Type.
Keeping support for isVirtioPCI/isVirtioCCW in qmp.go is a bit
tricky. Eventually, the hot-plug API should be extended so callers
must manually specify the transport for the device.
Signed-off-by: Sergio Lopez <slp@redhat.com>
Caller can hotplug vhost-user device via qmp.
The Qemu vhost-user device, like vhost-user-blk-pci and
vhost-user-scsi-pci can be hotplugged by qmp API:
ExecuteCharDevUnixSocketAdd() together with
ExecutePCIVhostUserDevAdd()
Signed-off-by: Liu Xiaodong <xiaodong.liu@intel.com>
The upper hyervisor manager application maybe need to wait some
QMP event to control boot sequence, but the event we wanted maybe
not exist in some older version, so we need query all QMP ABI and
check the event is supported or not.
related: kata-containers/runtime#1918
Signed-off-by: Ning Bo <ning.bo9@zte.com.cn>
Support for function isSocketIDSupported, isThreadIDSupported and isDieIDSupported.
The functions check if the cpu driver and the qemu version support the
id parameter.
Fixes: #102
Signed-off-by: Alice Frosi <afrosi@de.ibm.com>
In QEMU 4.1 the CPU topology for x86 will change to:
`socket > die > core > thread`.
Add `die-id` field to `CPUProperties` and include it in CPU hotplugging
Signed-off-by: Julio Montes <julio.montes@intel.com>
Hotplugged memory could be backed by a file on the host with sharing
turned on. This change allows qmp to pass that option to a govmm.
Signed-off-by: Ganesh Maharaj Mahalingam <ganesh.mahalingam@intel.com>
In newer versions of QEMU, like 4.0-rc2, QMP events can be thrown even before
the QMP-version response, one example of this behaviour is when a virtio serial
is closed and a VSERPORT_CHANGE event is thrown.
Re-implement mainLoop to check the data received from the VM channel, since
it's not a guarantee that the first data read from the VM channel is the
QMP version.
fixes https://github.com/kata-containers/runtime/issues/1474
Signed-off-by: Julio Montes <julio.montes@intel.com>
Since []byte channel type transfer slice info(include slice underlying array pointer, len, cap)
between channel sender and receiver. scanner.Bytes() function returned slice's underlying array
may point to data that will be overwritten by a subsequent call to Scan(reference from:
https://golang.org/pkg/bufio/#Scanner.Bytes), which may make consecutive scan() call write the
read data into the same underlying array which causes receiver read mixed data,so we need to
copy line to new allocated space and then send to channel receiver to solve this problem.
Fixes: #88
Signed-off-by: jiangpengfei <jiangpengfei9@huawei.com>
For vCPU hotplug to work on ppc64le, we need not
pass threadID and socketID. So conditionally pass
arguments when executing CPU device add.
Fixes: #83
Signed-off-by: Nitesh Konkar niteshkonkar@in.ibm.com
Static check was complaining about code that looked like
_ = <-ch
when it wants to see simply
<-ch
There was only one instance of this in govmm and this commit fixes
that instance.
Signed-off-by: Mark Ryan <mark.d.ryan@intel.com>
staticcheck was complaining as some of the error messages returned by
govmm began with a capital letter. This commit fixes the issue.
Signed-off-by: Mark Ryan <mark.d.ryan@intel.com>
ExecuteNVDIMMDeviceAdd can add a nvdimm disk to qemu.
Not implement NVDIMM device delete function because qemu doesn't support it.
Signed-off-by: Hui Zhu <teawater@hyper.sh>
For devices that actually support the option disable-modern, this
current commit provides a proper flag to the caller. This will allow
for better support when used in nested environment as virtio-pci
devices should rely on virtio 0.9 instead of 1.0 due to a bug in
KVM.
Fixes#80
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
The PR adds the s390x support. It sets the CCW devices and sets to false
all the devices in the mapping isVirtioPCI. It reimplements the functions
QemuNetdevParam and QemuDeviceParam to print an error message if the vhost-user
devices are used. It introduces a new function ExecuteNetCCWDeviceAdd for qmp
for the CCW devices.
Fixes: #37
Co-authored-by: Yash D Jain <ydjainopensource@gmail.com>
Signed-off-by: Alice Frosi <afrosi@de.ibm.com>
This commit updates the headers in the Go source files to adhere
to the new guidelines in the CONTRIBUTING.md file.
Signed-off-by: Mark Ryan <mark.d.ryan@intel.com>
Only get 'QMP command failed' error message now when execute QMP
command by 'executeCommandWithResponse' failed. This patch will
output more error detail.
Signed-off-by: NingBo <ning.bo9@zte.com.cn>
This PR prepares for the s390x support. It introduces:
- a generalization of ccw and pci devices. The variables for the pci devices
have been renamed by removing the Pci suffix. They have been moved to the
qemu_arch_base.go
- the mapping isVirtioPCI has been move to qemu_arch_base.go because in
this way a different mapping can be added for other architecture (e.g
s390x)
- the functions QemuNetdevParam and QemuDeviceParam have been moved to
qemu_arch_base.go. In this way, they could be reimplemented for other
architecture for the case VHOSTUSER
- a function disableModern has been introduced to check if the device is
a pci device and then returns the right parameters. In the case of ccw
devices, they don't have the disable-modern flag
- a function mqParameter has been introduced to return the right
parameters for the mq case. The virtio-net-ccw device doesn't have the
vectors flag
- in qemu_arch_base_test.go contains the test and strings that can be
overwritten for other architectures (e.g s390). The devices names and
the flags for the devices can be overwritten.
- the string for the romfile has been replaced by a variable romfile
that could be left empty if the devices doesn't support a romfile as
for the ccw devices for s390.
- clean-up: the disable-modern=on/off options have been changed to
disable-modern=true/false. In the code there was a mixture of on/true
off/false
Fixes: #61
Co-authored-by: Yash D Jain <ydjainopensource@gmail.com>
Signed-off-by: Alice Frosi <afrosi@de.ibm.com>
ExecuteBlockdevAddWithCache has two more parameters direct and noFlush
than ExecuteBlockdevAdd.
They are cache-related options for block devices that are described in
https://github.com/qemu/qemu/blob/master/qapi/block-core.json.
direct denotes whether use of O_DIRECT (bypass the host page cache)
is enabled. noFlush denotes whether flush requests for the device are
ignored.
Signed-off-by: Hui Zhu <teawater@hyper.sh>
This patch fixes the wrong behavior of specifying a netdev, MAC
address or PCI address entry when those were empty. Instead, it
does not provide those entries if the content is empty.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
Any device inheriting from virtio-pci can specify a ROM file. This
option is provisioned by default with "efi-virtio.rom", but most
of the time, firmwares such as OVMF or seabios will already support
what is provided by this ROM file.
In order to reduce the "forced" dependency on such ROM file, govmm
should provide an empty path if the consumer of the library does not
provide one.
This patch reorganizes the list of devices, so that it gets easier to
list which devices inherit from virtio-pci, and then adds the romfile
option to every single device that support this option.
Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
