https://bugzilla.redhat.com/show_bug.cgi?id=1624223
There are two ways to request memory preallocation on cmd line:
-mem-prealloc and .prealloc attribute for a memory-backend-file.
However, as it turns out it's not safe to use both at the same
time. If -mem-prealloc is used then qemu will fully allocate the
memory (this is done by actually touching every page that has
been allocated). Then, if .prealloc=yes is specified,
mbind(flags = MPOL_MF_STRICT | MPOL_MF_MOVE) is called which:
a) has to (possibly) move the memory to a different NUMA node,
b) can have no effect when hugepages are in play (thus ignoring user
request to place memory on desired NUMA nodes).
Prefer -mem-prealloc as it is more backward compatible
compared to switching to "-numa node,memdev= + -object
memory-backend-file".
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: John Ferlan <jferlan@redhat.com>
Remove the disk from tests focusing on other aspects so that change to
-blockdev will touch less tests.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Depending on whether QEMU actually supports the option, we can put the
'rendernode' on the '-display egl-headless' cmdline.
https://bugzilla.redhat.com/show_bug.cgi?id=1628892
Signed-off-by: Erik Skultety <eskultet@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Unlike with SPICE and SDL which use the <gl> subelement to enable OpenGL
acceleration, specifying egl-headless graphics in the XML has
essentially the same meaning, thus in case of egl-headless we don't have
a need for the 'enable' element attribute and we'll only be interested
in the 'rendernode' one further down the road.
Signed-off-by: Erik Skultety <eskultet@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Up until now, we formatted 'rendernode=' onto QEMU cmdline only if the
user specified it in the XML, otherwise we let QEMU do it for us. This
causes permission issues because by default the /dev/dri/renderDX
permissions are as follows:
crw-rw----. 1 root video
There's literally no reason why it shouldn't be libvirt picking the DRM
render node instead of QEMU, that way (and because we're using
namespaces by default), we can safely relabel the device within the
namespace.
Signed-off-by: Erik Skultety <eskultet@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Add a new memoryBacking source type "memfd", supported by QEMU (when
the capability is available).
A memfd is a specialized anonymous memory kind. As such, an anonymous
source type could be automatically using a memfd. However, there are
some complications when migrating from different memory backends in
qemu (mainly due to the internal object naming at this point, but
there could be more). For now, it is simpler and safer to simply
introduce a new source type "memfd". Eventually, the "anonymous" type
could learn to use memfd transparently in a separate change.
The main benefits are that it doesn't need to create filesystem files,
and it also enforces sealing, providing a bit more safety.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Add new functions to generate zPCI command string and append it to
QEMU command line. And the related tests are added.
Signed-off-by: Yi Min Zhao <zyimin@linux.ibm.com>
Reviewed-by: Boris Fiuczynski <fiuczy@linux.ibm.com>
Reviewed-by: Stefan Zimmermann <stzi@linux.ibm.com>
Reviewed-by: Bjoern Walk <bwalk@linux.ibm.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Andrea Bolognani <abologna@redhat.com>
We should ensure that QEMU supports zPCI when a zPCI address is defined
in XML and otherwise report an error. This patch introduces a generic
validation function qemuDomainDeviceDefValidateAddress() which calls
qemuDomainDeviceDefValidateZPCIAddress() if address type is PCI address.
Signed-off-by: Yi Min Zhao <zyimin@linux.ibm.com>
Reviewed-by: Boris Fiuczynski <fiuczy@linux.ibm.com>
Reviewed-by: Andrea Bolognani <abologna@redhat.com>
This patch introduces new XML parser/formatter functions. Uid is
16-bit and non-zero. Fid is 32-bit. They are the two attributes of zpci
which is introduced as PCI address element. Zpci element is parsed and
formatted along with PCI address. And add the related test cases.
Signed-off-by: Yi Min Zhao <zyimin@linux.ibm.com>
Reviewed-by: Boris Fiuczynski <fiuczy@linux.ibm.com>
Reviewed-by: Stefan Zimmermann <stzi@linux.ibm.com>
Reviewed-by: Bjoern Walk <bwalk@linux.ibm.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Andrea Bolognani <abologna@redhat.com>
QEMU 3.1 supports Hyper-V-style PV IPIs making it cheaper for Windows
guests to send an IPI, especially when it targets many CPUs.
Reviewed-by: Andrea Bolognani <abologna@redhat.com>
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
https://bugzilla.redhat.com/show_bug.cgi?id=1632833
When doing a SCSI passthrough we don't put format= onto the
command line. This causes qemu to probe the format automatically
which ends up in a warning in the domain log and possible qemu
disabling writes to the first block (according to the warning
message).
Based-on-work-of: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Commit ed5aa85f37
qemu: don't use chardev FD passing for vhostuser backend
altered the legacy DO_TEST macro.
Run the test against capabilities of QEMU 2.5.0 (which did not
support QEMU_CAPS_CHARDEV_FD_PASS) as well as the latest version.
Signed-off-by: Ján Tomko <jtomko@redhat.com>
Test CCID smartcard passthrough from a unix listen socket.
Use the capabilities of QEMU 2.5.0 which did not support
chardev FD passing and the latest one, which (at the time
of this commit) it does.
Signed-off-by: Ján Tomko <jtomko@redhat.com>
-net name= will be deprecated in QEMU 3.1:
commit 101625a4d4ac7e96227a156bc5f6d21a9cc383cd
net: Deprecate the "name" parameter of -net
git describe: v3.0.0-791-g101625a4d4
Use the id option instead, supported since QEMU 1.2:
commit 6687b79d636cd60ed9adb1177d0d946b58fa7717
convert net_client_init() to OptsVisitor
git describe: v1.0-3564-g6687b79d63 contains: v1.2.0-rc0~142^2~8
Thankfully, libvirt only uses -net for non-PCI, non-virtio NICs
on ARM.
Signed-off-by: Ján Tomko <jtomko@redhat.com>
Reviewed-by: Laine Stump <laine@laine.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
virDomainDefCollectBootOrder() is called for every item on the list
for each type of device. One of the checks it makes is to gather the
order attributes from the <boot> element of all devices, and assure
that no two devices have been given the same order.
Since (internally to libvirt, *not* in the domain XML) an <interface
type='hostdev'> is on both the list of hostdev devices and the list of
network devices, it will be counted twice, and the code that checks
for multiple devices with the same boot order will give a false
positive.
To remedy this, we make sure to return early for hostdev devices that
have a parent.type != NONE.
This was introduced in commit 5b75a4, which was first in libvirt-4.4.0.
Resolves: https://bugzilla.redhat.com/1601318
Signed-off-by: Laine Stump <laine@laine.org>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Attempting to use a chardev definition like
<serial type='unix'>
<target type='isa-serial'/>
</serial>
correctly results in an error being reported, since the source
path - a required piece of information - is missing; however,
the very similar
<serial type='unix'>
<target type='pci-serial'/>
</serial>
was happily accepted by libvirt, only to result in libvirtd
crashing as soon as the guest was started.
The issue was caused by checking the chardev's targetType
against whitelisted values from virDomainChrChannelTargetType
without first checking the chardev's deviceType to make sure
it is actually a channel, for which the check makes sense,
rather than a different type of chardev.
The only reason this wasn't spotted earlier is that the
whitelisted values just so happen to correspond to USB and
PCI serial devices and Xen and UML consoles respectively,
all of which are fairly uncommon.
https://bugzilla.redhat.com/show_bug.cgi?id=1609720
Signed-off-by: Andrea Bolognani <abologna@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
https://bugzilla.redhat.com/show_bug.cgi?id=1622455
If a domain is configured to use <source type='file'/> under
<memoryBacking/> we have to honour that setting and produce
-mem-path on the command line. We are not doing so if domain has
no guest NUMA nodes nor hugepages.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: John Ferlan <jferlan@redhat.com>
The new tests use DO_TEST_CAPS_ARCH_LATEST() with an input
XML describing a very simple headless guest and cover most
architectures and machine types we care about.
Signed-off-by: Andrea Bolognani <abologna@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
virtio-serial is an alias for virtio-serial-pci, which
should not have been used for a PCIe-less aarch64/virt
guest but it ended up being used anyway because the
virtio-mmio capability was missing and the algorithm
is buggy.
Fix the test case so that we can fix the algorithm next.
Signed-off-by: Andrea Bolognani <abologna@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Commit 6534b3c4 tried to raise an error when there is no numa
nodes by setting access='shared' in the domain config, but added
a helper called from qemuDomainDeviceDefValidate instead of a
helper called from qemuDomainDefValidate for XML:
<memoryBacking>
<hugepages/>
<access mode='shared'/>
</memoryBacking>
Since there are no memory devices in the test XML, there would
be no validation failure, but the test added was still failing.
Investigating that it turns out that unnecessary XML elements
were causing the failure (no need for <video>, <graphics>,
<pm>, usb controller model "piix3-uhci", disk attribute for
"discard='unmap'", <serial>, <console>, <channel> and a
memballoon model). Removing all those before moving the method
caused the test to succeed.
So this patch moves the validation to the right place and
removes all the unnecessary XML pieces that were causing
a false validation failure.
https://bugzilla.redhat.com/show_bug.cgi?id=1448149#c14
Signed-off-by: Luyao Huang <lhuang@redhat.com>
Reviewed-by: John Ferlan <jferlan@redhat.com>
The architecture is new enough that we don't need to
concern ourselves with backwards compatibility in any
capacity.
Signed-off-by: Andrea Bolognani <abologna@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
'metadata' and 'leases' are features internal to libvirt and thus don't
influence the generated QEMU command line. As they are not tested we
don't need the output files.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Erik Skultety <eskultet@redhat.com>
Now we assume the flag always so there's no use for this test. Probably
a leftover from the cleanup of the capability.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Erik Skultety <eskultet@redhat.com>
The test files were unused, but we don't have any other test for this
feature. Make use of the existing files by removing disks and using
DO_TEST_CAPS_LATEST to execute them. The legacy output files will be
dropped.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Erik Skultety <eskultet@redhat.com>
aarch64-acpi-nouefi and hostdev-scsi-boot are unused. Noticed when
checking whether '-nodefconfig' is still used by libvirt.
Unused since their introduction in commit deb38c4 and bab6ee6
respectively.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
All supported qemus support FD passing so modify the tests to test the
proper code path.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: John Ferlan <jferlan@redhat.com>
Similarly to backing store indexes which will become stable eventually
we need also to be able to format and store in the status XML for later
use the index for the top level of the backing chain.
Add XML formatter, parser, schema and docs.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
Add test data for nested backing chains with/without indexes (used in
status XMLs) which will excercise blockdev and the related work.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
The blockdev support will change existing approach to add disks to VMs
so all tests using the DO_TEST_CAPS_LATEST approach which have any disks
need to be forked so that the changes can be applied.
Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
In order to test SEV we need real QEMU capabilities. Ideally, this would
be tested with -latest capabilities, however, our capabilities are
currently tied to Intel HW, even the 2.12.0 containing SEV were edited by
hand, so we can only use that one for now, as splitting the capabilities
according to the vendor is a refactor for another day. The need for real
capabilities comes from the extended SEV platform data (PDH, cbitpos,
etc.) we'll need to cache/parse.
Signed-off-by: Erik Skultety <eskultet@redhat.com>
Acked-by: Peter Krempa <pkrempa@redhat.com>
Qemu-3.0 supports Hyper-V-style PV TLB flush, Windows guests can benefit
from this feature as KVM knows which vCPUs are not currently scheduled (and
thus don't require any immediate action).
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Reviewed-by: John Ferlan <jferlan@redhat.com>
Qemu-3.0 supports so-called 'Reenlightenment' notifications and this (in
conjunction with 'hv-frequencies') can be used make Hyper-V on KVM pass
stable TSC page clocksource to L2 guests.
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Reviewed-by: John Ferlan <jferlan@redhat.com>
Qemu-2.12 gained 'hv-frequencies' cpu flag to enable Hyper-V frequency
MSRs. These MSRs are required (but not sufficient) to make Hyper-V on
KVM pass stable TSC page clocksource to L2 guests.
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Reviewed-by: John Ferlan <jferlan@redhat.com>
If a domain has hugepages configured and we're currently building
memory-backend-file for a nvdimm device that domain has we will
put hugepages path onto the command line. It should have been
nvdimm path configured in the XML.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Previously we were ignoring "nodeset" attribute for hugepage pages
if there was no guest NUMA topology configured in the domain XML.
Commit <fa6bdf6afa878b8d7c5ed71664ee72be8967cdc5> partially fixed
that issue but it introduced a somehow valid regression.
In case that there is no guest NUMA topology configured and the
"nodeset" attribute is set to "0" it was accepted and was working
properly even though it was not completely valid XML.
This patch introduces a workaround that it will ignore the nodeset="0"
only in case that there is no guest NUMA topology in order not to
hit the validation error.
After this commit the following XML configuration is valid:
<memoryBacking>
<hugepages>
<page size='2048' unit='KiB' nodeset='0'/>
</hugepages>
</memoryBacking>
but this configuration remains invalid:
<memoryBacking>
<hugepages>
<page size='2048' unit='KiB' nodeset='0'/>
<page size='1048576' unit='KiB'/>
</hugepages>
</memoryBacking>
The issue with the second configuration is that it was originally
working, however changing the order of the <page> elements resolved
into using different page size for the guest. The code is written
in a way that it expect only one page configured and always uses only
the first page in case that there is no guest NUMA topology configured.
See qemuBuildMemPathStr() function for details.
Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=1591235
Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
We can safely validate the hugepage nodeset attribute at a define time.
This validation is not done for already existing domains when the daemon
is restarted.
All the changes to the tests are necessary because we move the error
from domain start into XML parse.
Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
This use-case was broken by commit
<fa6bdf6afa878b8d7c5ed71664ee72be8967cdc5>.
We allowed this configuration and it was working as expected therefore
we can consider it as regression. We should have never allowed such
configuration so now the best solution is in case of non-numa guest
silently ignore the 'nodeset' attribute if it's set to '0'.
That will be fixed by following patches.
Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
This test case is currently working but it uncovers existing issue
in our code that the generated QEMU commandline uses the default 1G
hugepage instead of the 2M hugepage specified for exact node.
The issue in our code is that for non-numa guests we take into account
only the first hugepage. This will be fixed as invalid configuration
since it doesn't make any sense to set default and specific hugepage
for non-numa guest.
Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
Michal Privoznik