We will need to store two more host CPU models and nested structs look
better than separate items with long complicated names.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
This new internal API makes a copy of virCPUDef while removing all
features which would block migration. It uses cpu_map.xml as a database
of such features, which should only be used as a fallback when we cannot
get the data from a hypervisor. The main goal of this API is to decouple
this filtering from virCPUUpdate so that the hypervisor driver can
filter the features according to the hypervisor.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
If formatting NUMA topology fails, the function returns immediatelly,
but the buffer structure allocated on the stack references lot of
heap-allocated memory and that would get lost in such case.
Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
qemuProcessVerifyHypervFeatures is supposed to check whether all
requested hyperv features were actually honored by QEMU/KVM. This is
done by checking the corresponding CPUID bits reported by the virtual
CPU. In other words, it doesn't work for string properties, such as
VIR_DOMAIN_HYPERV_VENDOR_ID (there is no CPUID bit we could check). We
could theoretically check all 96 bits corresponding to the vendor
string, but luckily we don't have to check the feature at all. If QEMU
is too old to support hyperv features, the domain won't even start.
Otherwise, it is always supported.
Without this patch, libvirt refuses to start a domain which contains
<features>
<hyperv>
<vendor_id state='on' value='...'/>
</hyperv>
</features>
reporting internal error: "unknown CPU feature __kvm_hv_vendor_id.
This regression was introduced by commit v3.1.0-186-ge9dbe7011, which
(by fixing the virCPUDataCheckFeature condition in
qemuProcessVerifyHypervFeatures) revealed an old bug in the feature
verification code. It's been there ever since the verification was
implemented by commit v1.3.3-rc1-5-g95bbe4bf5, which effectively did not
check VIR_DOMAIN_HYPERV_VENDOR_ID at all.
https://bugzilla.redhat.com/show_bug.cgi?id=1439424
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Introduce STRICT_FRAME_LIMIT_CFLAGS that will be used for
production code and RELAXED_FRAME_LIMIT_CFLAGS for tests.
Raising the limit for tests allows building them with clang
with optimizations disabled.
This header file has been created so that we can expose
internal functions to the test suite without making them
public: those in qemu_capabilities.h bearing the comment
/* Only for use by test suite */
are obvious candidates for being moved over.
This function runs an iscsi command and parses its output.
However, due to the nature of things, virISCSIExtractSession()
callback can be called multiple times. In each run it would
allocate new memory and overwrite the variable where we keep
pointer to it and thus leaking old allocations.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Even though the virMacMap object is not necessarily created at
the same time as the network object, the former makes no sense
without the latter and thus should be unref'd in the network
object dispose function.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Imagine that this function is called twice over the same disk
source. While in the first run all allocated memory is freed, not
all pointers are set to NULL (e.g. def->srcpool). So when called
again, these poitners are freed again resulting in double free.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
After restart of libvirtd the 'checkPool' method is supposed to validate
that the pool is online. Since libvirt then refreshes the pool contents
anyways just return whether the pool was supposed to be online so that
the code can be reached. This is necessary since if a pool does not
implement the method it's automatically considered as inactive.
Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=1436065
qemu requires that the topology equals to the maximum vcpu count.
Document this along with the API to set maximum vcpu count and the XML
element.
Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=1426220
Use the relative lookup specifier rather than the global one. Otherwise
only the first name would be looked up. Add a test case to cover the
scenario.
Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=1436574
The native gluster pool source list data differs from the data used for
attaching gluster volumes as netfs pools. Currently the only difference
was the format. Since native pools don't use it and later there will be
more differences add a more deterministic way to switch between the
types instead.
Similar to commit b202c39 ignore the warning that breaks the build
with clang:
util/virnetlink.c:365:52: error: cast from 'char *' to 'struct nlmsghdr *'
increases required alignment from 1 to 4 [-Werror,-Wcast-align]
for (msg = resp; NLMSG_OK(msg, len); msg = NLMSG_NEXT(msg, len)) {
^~~~~~~~~~~~~~~~~~~~
/usr/include/linux/netlink.h:87:7: note: expanded from macro 'NLMSG_NEXT'
(struct nlmsghdr*)(((char*)(nlh)) + NLMSG_ALIGN((nlh)->nlmsg_len)))
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Buggy condition meant that vcpu0 would not be iterated in the checks.
Since it's not hotpluggable anyways we would not be able to break the
configuration of a live VM.
Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=1437013
Like all devices, add the 'id' option for mdevs as well. Patch also
adjusts the test accordingly.
Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=1438431
Signed-off-by: Erik Skultety <eskultet@redhat.com>
https://bugzilla.redhat.com/show_bug.cgi?id=1371892
The 'capacity' value (e.g. guest logical size) for a LUKS volume is
smaller than the 'physical' value of the file in the file system, so
we need to account for that.
When peeking at the encryption information about the volume add a fetch
of the payload_offset which is described as the offset to the start of
the volume data (in 512 byte sectors) in QEMU's QCryptoBlockLUKSHeader.
Then adjust the ->capacity appropriately when we determine that the
volume target encryption has a payload_offset value.
Add check for more than one RTA_OIF, even though this is rather
unlikely.
Get rid of the buggy switch / break as this code won't need to
handle more attributes.
Use VIR_WARNINGS_NO_CAST_ALIGN to fix impossible to fix
util/virnetdevip.c:560:17: error: cast increases required alignment of target type [-Werror=cast-align]
Depending on the architecture, requirements for ACPI and UEFI can
be different; more specifically, while on x86 UEFI requires ACPI,
on aarch64 it's the other way around.
Enforce these requirements when validating the domain, and make
the error message more accurate by mentioning that they're not
necessarily applicable to all architectures.
Several aarch64 test cases had to be tweaked because they would
have failed the validation step otherwise.
The capabilities used in test cases should match those used
during normal operation for the tests to make any sense.
This results in the generated command line for a few test
cases (most notably non-x86 test cases that were wrongly
assuming they could use -no-acpi) changing.
Instead of having a single function that probes the
architecture from the monitor and then sets a bunch of
basic capabilities based on it, have a separate function
for each part: virQEMUCapsInitQMPArch() only sets the
architecture, and virQEMUCapsInitQMPBasicArch() only sets
the capabilities.
This split will be useful later on, when we will want to
set basic capabilities from the test suite without having
to go through the pain of mocking the monitor.
If a transient storage pool is deemed inactive after libvirtd restart it
would not be deleted from the list. Reuse virStoragePoolUpdateInactive
along with a refactor necessary to properly update the state.
Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=1242801
After a pool is made inactive the definition objects need to be updated
(if a new definition is prepared) and transient pools need to be
completely removed. Split out the code doing these steps into a separate
function for later reuse.
When registering a storage poll backend, the code would use
virStorageTypeToString instead of virStoragePoolTypeToString. The
following message would be logged:
virDriverLoadModuleFunc:71 : Lookup function 'virStorageBackendSCSIRegister'
virStorageBackendRegister:174 : Registering storage backend '(null)'
Currently, if we want to zero out disk source (e,g, due to
startupPolicy when starting up a domain) we use
virDomainDiskSetSource(disk, NULL). This works well for file
based storage (storage type file, dir, or block). But it doesn't
work at all for other types like volume and network.
So imagine that you have a domain that has a CDROM configured
which source is a volume from an inactive pool. Because it is
startupPolicy='optional', the CDROM is empty when the domain
starts. However, the source element is not cleared out in the
status XML and thus when the daemon restarts and tries to
reconnect to the domain it refreshes the disks (which fails - the
storage pool is still not running) and thus the domain is killed.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
The virMacMap module is there for dumping [domain, <list of is
MACs>] pairs into a file so that libvirt_guest NSS module can use
it. Whenever a interface is allocated from network (e.g. on
domain<F2> startup or NIC hotplug), network is notified and so is
virMacMap module subsequently. The module update functions
networkMacMgrAdd() and networkMacMgrDel() gracefully handle the
case when there's no module. The problem is, the module is
created if and only if network is freshly started, or if the
daemon restarts and network previously had the module.
This is not very user friendly - if users want to use the NSS
module they need to destroy their network and bring it up again
(and subsequently all the domains using it).
One disadvantage of this approach implemented here is that one
may get just partial results: any already running network does
not record mac maps, thus only newly plugged domains will be
stored in the module. The network restart scenario is not touched
by this of course. But one can argue that older libvirts had
never recorded the mac maps anyway.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
So far our code is full of the following pattern:
dom = virGetDomain(conn, name, uuid)
if (dom)
dom->id = 42;
There is no reasong why it couldn't be just:
dom = virGetDomain(conn, name, uuid, id);
After all, client domain representation consists of tuple (name,
uuid, id).
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
There was an unhandled 'open' call which resulted in:
"error: Library function returned error but did not set virError"
Even if this happens during the daemon's start when we still don't have
any set of outputs defined yet, we can safely report an error, since we
automatically fallback to stderr which is fine even for both
running as a daemonized process, since this happens before the daemon
forks into the background, and running as a systemd service, since
systemd re-directs std outputs to journald by default.
Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=1436060
Signed-off-by: Erik Skultety <eskultet@redhat.com>
In 9e2465834 a check that denies internal snapshots when pflash
based loader is configured for the domain. However, if there's
none and an user tries to do an internal snapshot they will
witness daemon crash as in that case vm->def->os.loader is NULL
and we dereference it unconditionally.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
CPU features which change their value from disabled to enabled between
two calls to query-cpu-model-expansion (the first with no extra
properties set and the second with 'migratable' property set to false)
can be marked as enabled and non-migratable in qemuMonitorCPUModelInfo.
Since the code consuming qemuMonitorCPUModelInfo currently ignores the
migratable flag, this change is effectively changing the CPU model
advertised in domain capabilities to contain all features (even those
which block migration). And this matches what we do for QEMU older than
2.9.0, when we detect all CPUID bits ourselves without asking QEMU.
As a result of this change
<cpu mode='host-model'>
<feature name='invtsc' policy='require'/>
</cpu>
will work with all QEMU versions. Such CPU definition would be forbidden
with QEMU >= 2.9.0 without this patch.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
If calling query-cpu-model-expansion on the 'host'/'max' CPU model with
'migratable' property set to false succeeds, we know QEMU is able to
tell us which features would disable migration. Thus we can mark all
enabled features as migratable.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
QEMU is able to tell us whether a CPU feature would block migration or
not. This patch adds support for storing such features in
qemuMonitorCPUModelInfo.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
When idx is 0 virStorageFileChainLookup returns the base (bottom) of the
backing chain rather than the top. This is expected by the callers of
qemuDomainGetStorageSourceByDevstr.
Add a special case for idx == 0
One of the problems with our virGetDomain function is that it
copies just domain name and domain UUID. Therefore it's very
easy to forget aboud domain ID. This can cause some bugs, like
virConnectGetAllDomainStats not reporting proper domain IDs.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
https://bugzilla.redhat.com/show_bug.cgi?id=1434882
Imagine the following scenario:
1) virsh net-start default
2) virsh start myFavouriteDomain
3) virsh net-destroy default
4) virsh destroy myFavouriteDomain
(assuming myFavouriteDomain has an interface from default
network)
Regardless of how unlikely this scenario looks like, we should
not crash. The problem is, on net-destroy in
networkShutdownNetworkVirtual() the virMacMap module is unrefed,
but the stale pointer is kept around. Thus when the domain
destroy procedure comes in, networkReleaseActualDevice() and
subsequently networkMacMgrDel() is called. This function sees the
stale pointer and starts calling the virMacMap module APIs which
work over freed memory.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
off_t is signed and it's size is the same as long only on 64b archs.
Thus it cannot be formatted as %lu.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
The value we use internally to represent the lack of a memory
locking limit, VIR_DOMAIN_MEMORY_PARAM_UNLIMITED, doesn't
match the value setrlimit() and prlimit() use for the same
purpose, RLIM_INFINITY, so we have to handle the translation
ourselves.
Partially-resolves: https://bugzilla.redhat.com/1431793
For guests that use <memoryBacking><locked>, our only option
is to remove the memory locking limit altogether.
Partially-resolves: https://bugzilla.redhat.com/1431793
Instead of having a separate function, we can simply return
zero from the existing qemuDomainGetMemLockLimitBytes() to
signal the caller that the memory locking limit doesn't need
to be set for the guest.
Having a single function instead of two makes it less likely
that we will use the wrong value, which is exactly what
happened when we started applying the limit that was meant
for VFIO-using guests to <memoryBacking><locked>-using
guests.
This reverts commit c2e60ad0e5.
Turns out this check is excessively strict: there are ways
other than <memtune><hard_limit> to raise the memory locking
limit for QEMU processes, one prominent example being
tweaking /etc/security/limits.conf.
Partially-resolves: https://bugzilla.redhat.com/1431793
If if_indextoname is not defined, the whole function using it should
not be defined either. Add stub to fix build on mingw.
Caused by 5dd607059d
Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
Creating a copy of the definition we want to add in a migration cookie
makes the code cleaner and less prone to memory leaks or double free
errors.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
https://bugzilla.redhat.com/show_bug.cgi?id=1398087
Clean up the virsh man page description for --pool-create-as in order
to better describe how the various arguments are used when creating
(or defining) a logical pool.
Also modify the storage pool XML parsing algorithm to check for the
mismatched "name" and "source-name".
While parsing if the storage source is not present, then a defaultFormat
was not set. This could lead to oddities such as seeing "unknown" format
in output for the "logical" pool even though the only format the pool could
support would be "lvm2".
This does "put a label" on other pool defaults as follows:
File System: FS_AUTO
Network File System: NETFS_AUTO
Disk: UNKNOWN
Each of which is the "0" value for their respective pools and thus
would be no "real" change.
QEMU allows for TSC frequency to be explicitly set to enable migration
with invtsc (migration fails if the destination QEMU cannot set the
exact same frequency used when starting the domain on the source host).
Libvirt already supports setting the TSC frequency in the XML using
<clock>
<timer name='tsc' frequency='1234567890'/>
</clock>
which will be transformed into
-cpu Model,tsc-frequency=1234567890
QEMU command line.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
The frequency is documented and formatted as an attribute of the <timer>
element rather than a nested <frequency> element expected by the parser.
Luckily enough, timer frequency has not been used by any driver so far.
And users were not able to set it in the XML either.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
- Make virMediatedDeviceNew() stub args match its prototype
- Fix typo: virRerportError -> virReportError
- Move MDEV_SYSFS_DEVICES definition out of the #ifdef __linux__ block
so we don't have to stub virMediatedDeviceGetSysfsPath()
https://bugzilla.redhat.com/show_bug.cgi?id=1430679
As it turns out some file headers (e.g. ext4) may be larger/longer than
the 512 bytes of zeros being written prior to a pvcreate, so let's write
out 2048 bytes similar to how the pvcreate sources would peek at the first
4 sectors of the device.
Make sure there is at enough bytes on the device to clear before doing
doing the clear - just to be sure.
This adds a few validations to the devices listed for a hostdev network:
* devices must be listed by PCI address, not by netdev name
* listing a device by PCI address is valid only for hostdev networks, not
for other types of network (e.g. macvtap passthrough).
* each device in a hostdev pool must be an SR-IOV VF
Resolves: https://bugzilla.redhat.com/1004676
Previously, this function must've been called only on Linux in order
to fail gracefully. That lead to #ifdef mess in callers, so the
function was redesigned so it failed gracefully on non-existing
files. However that commit forgot to define the function outside the
__linux__ ifdef, it broke non-Linux builds.
Caused by c67e04e25f.
Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
The public API flags are handled by the cpuBaselineXML wrapper. The
internal cpuBaseline API only needs to know whether it is supposed to
drop non-migratable features.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
cpuBaseline is responsible for computing a baseline CPU while feature
expansion is done by virCPUExpandFeatures. The cpuBaselineXML wrapper
(used by hypervisor drivers to implement virConnectBaselineCPU API)
calls cpuBaseline followed by virCPUExpandFeatures if requested by
VIR_CONNECT_BASELINE_CPU_EXPAND_FEATURES flag.
The features in the three changed test files had to be sorted using
"sort -k 3" because virCPUExpandFeatures returns a sorted list of
features.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
Having to use cpuBaseline with VIR_CONNECT_BASELINE_CPU_EXPAND_FEATURES
flag to expand CPU features is strange. Not to mention that cpuBaseline
can only expand host CPU definitions (i.e., it completely ignores
feature policies). The new virCPUExpandFeatures API is designed to work
with both host and guest CPU definitions.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
This is the maximum for many reasons, for starters because index ==
bus number, and a controller's bus number is 8 bits.
This incidentally resolves: https://bugzilla.redhat.com/1329090
Having this information available will make it easier to determine the
culprit when MAC or vlan tag appear to not be set, eg.:
https://bugzilla.redhat.com/1364073
(This patch doesn't fix that bug, just makes it easier to diagnose)
If an SRIOV VF has previously been used for VFIO device assignment,
the "admin MAC" that is stored in the PF driver's table of VF info
will have been set to the MAC address that the virtual machine wanted
the device to have. Setting the admin MAC for a VF also sets a flag in
the PF that is loosely called the "administratively set" flag. Once
that flag is set, it is no longer possible for the net driver of the
VF (either on the host or in a virtual machine) to directly set the
VF's MAC again; this flag isn't reset until the *PF* driver is
restarted, and that requires taking *all* VFs offline, so it's not
really feasible to do.
If the same SRIOV VF is later used for macvtap passthrough mode, the
VF's MAC address must be set, but normally we don't unbind the VF from
its host net driver (since we actually need the host net driver in
this case). Since setting the VF MAC directly will fail, in the past
"we" ("I") had tried to fix the problem by simply setting the admin MAC
(via the PF) instead. This *appeared* to work (and might have at one
time, due to promiscuous mode being turned on somewhere or something),
but it currently creates a non-working interface because only the
value for admin MAC is set to the desired value, *not* the actual MAC
that the VF is using.
Earlier patches in this series reverted that behavior, so that we once
again set the MAC of the VF itself for macvtap passthrough operation,
not the admin MAC. But that brings back the original bug - if the
interface has been used for VFIO device assignment, you can no longer
use it for macvtap passthrough.
This patch solves that problem by noticing when virNetDevSetMAC()
fails for a VF, and in that case it sets the desired MAC to the admin
MAC via the PF, then "bounces" the VF driver (by unbinding and the
immediately rebinding it to the VF). This causes the VF's MAC to be
reinitialized from the admin MAC, and everybody is happy (until the
*next* time someone wants to set the VF's MAC address, since the
"administratively set" bit is still turned on).
Some PF drivers allow setting the admin MAC (that is the MAC address
that the VF will be initialized to the next time the VF's driver is
loaded) to 00:00:00:00:00:00, and some don't. Multiple drivers
initialize the admin MACs to all 0, but don't allow setting it to that
very same value. It has been an uphill battle convincing the driver
people that it's reasonable to expect The argument that's used is
that an all 0 device MAC address on a device is invalid; however, from
an outsider's point of view, when the admin MAC is set to 0 at the
time the VF driver is loaded, the VF's MAC is *not* set to 0, but to a
random non-0 value. But that's beside the point - even if I could
convince one or two SRIOV driver maintainers to permit setting the
admin MAC to 0, there are still several other drivers.
So rather than fighting that losing battle, this patch checks for a
failure to set the admin MAC due to an all 0 value, and retries it
with 02:00:00:00:00:00. That won't result in a random value being set
in the VF MAC at next VF driver init, but that's okay, because we
always want to set a specific value anyway. Rather, the "almost 0"
setting makes it easy to visually detect from the output of "ip link
show" which VFs are currently in use and which are free.
The global functions virNetDevReplaceMacAddress(),
virNetDevReplaceNetConfig(), virNetDevRestoreMacAddress(), and
virNetDevRestoreNetConfig() are no longer used, as their functionality
has been replaced by virNetDev(Save|Read|Set)NetConfig().
The static functions virNetDevReplaceVfConfig() and
virNetDevRestoreVfConfig() were only used by the above-named global
functions that were removed.
It takes longer to explain this than to fix it...
In the past we weren't able to save the VF's own MAC address *at all*
when using it for hostdev assignment, because we had already unbound
the VF from the host net driver prior to saving its config. With the
previous patch, that problem has been solved, so we now have the VF's
MAC address saved and can move on to the *next* problem, which is twofold:
1) during teardown we restore the config before we've re-bound, so the
VF doesn't have a net driver, and thus we can't set its MAC address
directly.
2) even if we delay restoring the config until the VF is bound to a
net driver, the request to set its MAC address would fail, since
(during device setup) we had set the "admin MAC" for the VF via an
RTM_SETLINK to the PF - once you've set the admin MAC for a VF, the
VF driver (either on host or on guest) is not allowed to change the
VF's MAC address "forever" (well, until you reload the PF driver,
but that requires destroying and recreating every single VF, which
isn't something you can require).
The solution is to keep the restoration of config at the same place,
but to set the *admin MAC* to the address you want the VF to have -
when the VF net driver is later initialized (as a part of re-binding
to the VF net driver) its MAC will be initialized to the current value
of the admin MAC.
In order to properly restore the original state of an SRIOV VF when
we're finished with it, we need to save the MAC address of the VF
itself (not just the admin MAC address for the VF that is stored in
the PF). But that can only be done when the VF is still bound to the
host's netdev driver, and we have always done the saving of device
config after the VF is already bound to vfio-pci. This patch prepares
us for adding a save of the VF's MAC by calling the function that
saves netconfig earlier in the device preparation, before we've
unbound it from the host netdev driver.
These two operations will need to be separated so that saving of the
original config is done before detaching the host net driver, and
setting the new config is done after attaching vfio-pci. This patch
splits the single function into two, but for now calls them together
(to make bisecting easier if there is a regression).
virHostdevNetConfigReplace() and virHostdevNetConfigRestore() are
modified to use the new virNetDev*NetConfig() functions.
Note that due to the VF's original MAC addresses being saved after it
has already been un-bound from the host net driver, the actual current
VF MAC address won't be saved (because it no longer exists) - only the
"admin MAC" will be saved. This reflects existing behavior that will
be fixed in an upcoming patch.
