lib: Set up cpuset controller for restrictive numatune
The aim of 'restrictive' numatune mode is to rely solely on
CGroups to have QEMU running on configured NUMA nodes. However,
we were never setting the cpuset controller when a domain was
starting up. We are doing so only when
virDomainSetNumaParameters() is called (aka live pinning).
This is obviously wrong. Fortunately, fix is simple as
'restrictive' is similar to 'strict' - every location where
VIR_DOMAIN_NUMATUNE_MEM_STRICT occurs can be audited and
VIR_DOMAIN_NUMATUNE_MEM_RESTRICTIVE case can be added.
Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=2070380 Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
With strict numatune we can't guarantee that all memory is moved
to new location. Therefore, let's forbid moving memory in that
case. However, allow it for restrictive mode, which is documented
to be best effort.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
hypervisor: Drop dead code in virDomainCgroupSetupGlobalCpuCgroup()
Since its introduction in v1.3.2-43-gef1fa55e46 there is a dead
code in virDomainCgroupSetupGlobalCpuCgroup() (well,
qemuSetupGlobalCpuCgroup() back then). The code formats NUMA
nodeset but never sets it.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Note that we attempt to remove logs only if virtlogd is in use.
Otherwise we do not know the pattern for rotated files.
For example for VM named "foo" we can not use "foo.log*" pattern to
remove rotated logs as we can have VM named "foo.log" with log
"foo.log.log". We can add extra check that filename does not end with
".log" but for VM "foo.log" we can have rotated log "foo.log.log.1". Ok
let's check we don't have "log" in filename part corresponging to * but
what if someone will use logrotate with "%Y.log-%m-%d" 'dateformat'
option. In this case the check will exclude proper rotated files.
Yes, the last example if quite artificial but it shows it is difficult
to find out correctly rotated files when rotated files pattern is not
known. Thus the above decision only to support case with virtlogd when
we know the pattern.
Another reason for not removing log files when logrotate is present is
that due to races some files can escape deletion. For example foo.log.3
will be rotated to foo.log.4 after removing function will read directory
files and thus foo.log.4 will not be deleted.
Signed-off-by: Nikolay Shirokovskiy <nshirokovskiy@virtuozzo.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
qemu_command: Generate cmd line for virtio-net.rss hash properties
Also, validate that the requested feature is supported by QEMU.
Signed-off-by: Andrew Melnychenko <andrew@daynix.com> Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
qemu_capabilities: Add capability for virtio-net "rss" and "hash".
Introduce QEMU_CAPS_VIRTIO_RSS capability which tracks
virtio-net.rss attribute introduced in qemu-5.2.
Signed-off-by: Andrew Melnychenko <andrew@daynix.com> Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
domain_conf: Add configs for virtio net RSS and Hash report.
Added "rss" and "rss_hash_report" configuration that should be
used with qemu virtio RSS. Both options are triswitches. Used as
"driver" options and affects only NIC with model type "virtio".
In other patches - options should turn on virtio-net RSS and hash
properties.
Signed-off-by: Andrew Melnychenko <andrew@daynix.com> Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Andrea Bolognani [Thu, 24 Mar 2022 18:32:13 +0000 (19:32 +0100)]
network: Convert managed property of hostdev-pci ports correctly
virNetworkForwardDef.managed is a bool but
virNetworkPortDef.hostdevpci.managed is a virTristateBool, which
means that the current code performs the following incorrect
conversion:
false -> BOOL_ABSENT
true -> BOOL_YES
Using the virTristateBoolFromBool() helper solves the issue.
Fixes: 6cb0ec48bd95c95489a987e05a88e8bcf1f9109c Signed-off-by: Andrea Bolognani <abologna@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Andrea Bolognani [Thu, 24 Mar 2022 18:07:40 +0000 (19:07 +0100)]
conf: Format managed property of hostdev-pci ports correctly
The property is parsed using virTristateBoolTypeFromString() but
formatted as if it was a regular bool, which results in the
following incorrect conversion:
Use the virTristateBoolTypeToString() helper to ensure the
setting can survive a roundtrip conversion.
Fixes: 4b4a981d60d3372fb73bb3546ab1052844b2e3d5 Signed-off-by: Andrea Bolognani <abologna@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Andrea Bolognani [Thu, 24 Mar 2022 19:15:54 +0000 (20:15 +0100)]
conf: Restore error checking in VideoAccelDefParseXML()
The check that ensures that at least one property among accel3d,
accel2d and rendernode has been provided by the user had been
lost during the conversion to virXMLPropTristateBool().
Fixes: 0fe2d8dd335054fae38b46bbbac58a4662e1a1d0 Signed-off-by: Andrea Bolognani <abologna@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Andrea Bolognani [Thu, 24 Mar 2022 18:47:16 +0000 (19:47 +0100)]
qemu: Accept <allowReboot value='default'/>
Up until a few commits ago, libvirt produced this XML and so
we need to be able to read it back to prevent a bunch of
error : virXMLPropEnumInternal:516 : XML error: Invalid value
for attribute 'value' in element 'allowReboot': 'default'
messages from being logged on daemon upgrade when there are
running guests.
Fixes: 0fe2d8dd335054fae38b46bbbac58a4662e1a1d0 Signed-off-by: Andrea Bolognani <abologna@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Peter Krempa [Thu, 10 Feb 2022 11:57:26 +0000 (12:57 +0100)]
qemu: command: Always assume QEMU_CAPS_OVERCOMMIT
Starting with qemu-3.1 we always have the '-overcommit' argument and use
it instead of '-realtime'. Remove the capability check and fix all
fake-caps tests.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
The flag was based on a version check which no longer made sense. Remove
the flag by replacing it's only use by an arch-check which is equivalent
at this point.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
All qemu versions now support FD passing either directly or via FDset.
Assume that we always have this capability so that we can simplify
chardev handling in many cases.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Peter Krempa [Thu, 3 Feb 2022 12:14:29 +0000 (13:14 +0100)]
qemu: Formally deprecate support for qemu < 3.1
As of April 23 2022, Ubuntu 20.04 will be out for two years, which means
we no longer have to support Ubuntu 18.04 along with qemu-2.11 shipped
with it.
This then brings the minimum qemu version we have to support to
qemu-3.1:
virtio-iommu needs to be an integrated device, and our address
assignment code will make sure that is the case. If the user has
provided an explicit address, however, we should make sure any
addresses pointing to a different bus are rejected.
Signed-off-by: Andrea Bolognani <abologna@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Andrea Bolognani [Thu, 23 Sep 2021 14:28:15 +0000 (16:28 +0200)]
conf: Add virDomainDeviceInfo to virDomainIOMMUDef
This is needed so that IOMMU devices can have addresses.
Existing IOMMU devices (intel-iommu and SMMUv3) are system
devices and as such don't have an address associated to them, but
virtio-iommu is a PCI device and needs one.
Signed-off-by: Andrea Bolognani <abologna@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
This capability detects the availability of the boot-bypass
property of the virtio-iommu-pci device.
This property was only introduced in QEMU 7.0 but, since the
device has been around for much longer, we end up querying its
properties for several more releases. As I don't have convenient
access to the 10+ binaries necessary to regenerate the replies,
I just put some fake data in there.
Signed-off-by: Andrea Bolognani <abologna@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Andrea Bolognani [Wed, 22 Sep 2021 17:15:01 +0000 (19:15 +0200)]
qemu: Introduce QEMU_CAPS_DEVICE_VIRTIO_IOMMU_PCI
This capability detects the availability of the virtio-iommu-pci
device.
Note that, while this device is present even in somewhat old
versions of QEMU, it's only some recent changes that made it
actually usable for our purposes.
Signed-off-by: Andrea Bolognani <abologna@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Andrea Bolognani [Thu, 23 Sep 2021 14:44:42 +0000 (16:44 +0200)]
qemu: Tweak some code
The altered code is functionally equivalent to the previous one,
but it's already laid down in a way that will make further
changes easier and less messy.
Signed-off-by: Andrea Bolognani <abologna@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Andrea Bolognani [Mon, 27 Sep 2021 13:49:23 +0000 (15:49 +0200)]
conf: Introduce VIR_PCI_CONNECT_INTEGRATED
This new flag can be used to convince the PCI address assignment
algorithm to place a device directly on the root bus. It will be
used to implement support for virtio-iommu, which needs to be an
integrated device in order to work correctly.
Signed-off-by: Andrea Bolognani <abologna@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Andrea Bolognani [Tue, 29 Mar 2022 15:36:31 +0000 (17:36 +0200)]
tests: Update capabilities for QEMU 7.0.0 on ppc64
The QEMU binary is built from the v7.0.0-rc2 tag.
Some of the additional capabilities that show up are a
consequence of more features being enabled in this build than
in the one used to generate the replies initially.
Signed-off-by: Andrea Bolognani <abologna@redhat.com>
Andrea Bolognani [Thu, 31 Mar 2022 09:01:24 +0000 (11:01 +0200)]
qemu: Dissolve virQEMUCapsFindBinaryForArch()
With the recent changes, virQEMUCapsGetDefaultEmulator() has
become a trivial wrapper around this function, as well as its
only caller. Clean up the situation by merging the two.
Signed-off-by: Andrea Bolognani <abologna@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Tested-by: Jim Fehlig <jfehlig@suse.com>
Andrea Bolognani [Thu, 31 Mar 2022 09:00:38 +0000 (11:00 +0200)]
qemu: Don't assume that /usr/libexec/qemu-kvm exists
On a machine where no QEMU binary is installed, we end up logging
libvirtd: Cannot check QEMU binary /usr/libexec/qemu-kvm:
No such file or directory
which is not very useful in general, and downright misleading in
the case of operating systems that are not derived from RHEL.
This is a consequence of treating that specific path in a different
way from all other possible QEMU binary paths, and specifically of
not checking whether the file actually exists but sort of assuming
that it must do if we haven't found another QEMU binary earlier.
Address the issue by trying this path out in
virQEMUCapsFindBinaryForArch(), along with all the other possible
ones, and making sure it exists before returning it.
Reported-by: Jim Fehlig <jfehlig@suse.com> Signed-off-by: Andrea Bolognani <abologna@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Tested-by: Jim Fehlig <jfehlig@suse.com>
Andrea Bolognani [Thu, 31 Mar 2022 08:56:55 +0000 (10:56 +0200)]
qemu: Clean up virQEMUCapsFindBinaryForArch()
If we get to the bottom of the function we know that none of the
attempts to locate a QEMU binary has been successful, so we can
simply return NULL directly.
This makes it unnecessary variable used to store the path, for
which we can use a more descriptive name.
Lastly, comparing with NULL explicitly is somewhat uncommon in
libvirt and more verbose than the equivalent implicit comparison,
so get rid of it.
Signed-off-by: Andrea Bolognani <abologna@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Tested-by: Jim Fehlig <jfehlig@suse.com>
projects / libvirt.git / log