Peter Krempa [Mon, 22 Nov 2021 13:41:23 +0000 (14:41 +0100)]
qemu: Store TLS config options for chardevs in qemuDomainChrSourcePrivate
When setting up TLS options from config in qemuDomainPrepareChardevSourceOne
we can also extract the x509 certificate path and default tlsVerify
setting so that 'qemuBuildChardevCommand' doesn't need to access the
config object any more.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Peter Krempa [Fri, 19 Nov 2021 16:35:21 +0000 (17:35 +0100)]
qemuBuildChardevCommand: Split creation of the command and setup of other objects
Completely seprate the creation of the commandline string from the setup
of other objects instantiated on the commandline.
'qemuBuildChardevCommand' will aggregate the setup of individual
parameters such as -add-fd and setup of TLS and the -chardev parameter
itself while the code formatting the commandline will be moved into
qemuBuildChardevStr.
'fdset' names are then stored in qemuDomainChrSourcePrivate.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Peter Krempa [Thu, 4 Nov 2021 12:59:47 +0000 (13:59 +0100)]
qemuBuildChrChardevStr: Directly generate command line
'qemuBuildChrChardevStr' used a hybrid approach where some arguments
were directly added to '@cmd' while the commandline itself was returned
as a string.
This patch renames qemuBuildChrChardevStr to qemuBuildChardevCommand
and adds the argument directly to @cmd inside the function.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Peter Krempa [Mon, 15 Nov 2021 16:00:56 +0000 (17:00 +0100)]
qemuxml2argvtest: Add _LATEST version for 'name-escape' case
It was impossible to use _LATEST when commit d7c814f7f75 was modernizing
the cases as improper separation in the code caused that files were
created in the host during the testsuite run.
Now that the host manipulation when instantiating chardevs is separated
we can add the missing version.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Peter Krempa [Wed, 3 Nov 2021 13:12:16 +0000 (14:12 +0100)]
qemu: Store chardev 'wait' flag in chardev source private data
We have just one case when we wish to wait for incomming connections for
a listening socket and that is for vhost-user network devices.
Passing this via a flag to qemuBuildChrChardevStr is unwieldy. Add a
field to qemuDomainChrSourcePrivate and populate it for our special
case inside of qemuDomainPrepareChardevSourceOne.
Since we wait for incomming connections only on startup of a new VM we
also need to pass in a flag whether qemuDomainPrepareChardevSourceOne
is called on a new start or on hotplug.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Peter Krempa [Mon, 25 Oct 2021 10:42:16 +0000 (12:42 +0200)]
qemu: Move creation and opening of chardev backend FDs to host prepare step
The opening of files for FD passing for a chardev backend was
historically done in the function which is formatting the commandline.
This has multiple problems. Firstly the function takes a lot of
parameters which need to be passed through the commandline formatters.
This made the 'qemuBuildChrChardevStr' extremely unappealing to the
extent that we have multiple other custom formatters in places which
didn't really want to use the function.
Additionally the function is also creating files in the host in certain
configurations which is wrong for a commandline formatter to do. This
meant that e.g. not all chardev test cases can be converted to use
DO_TEST_CAPS_LATEST as we attempt to use such code path and attempt to
create files outside of the test directory.
This patch moves the opening of the filedescriptors from
'qemuBuildChrChardevFileStr' into a new helper
'qemuProcessPrepareHostBackendChardevOne' which is called using
'qemuDomainDeviceBackendChardevForeach'.
To preserve test behaviour we also have another instance
'testPrepareHostBackendChardevOne' which is populating mock
filedescriptors.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Peter Krempa [Tue, 2 Nov 2021 13:15:58 +0000 (14:15 +0100)]
qemu: domain: Introduce helpers for initializing chardev backend of devices
Introduce qemuDomainDeviceBackendChardevForeach(One) which calls the
callback if either given device has a chardev backend or for all chardev
backends of all devices.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Peter Krempa [Tue, 7 Dec 2021 16:17:13 +0000 (17:17 +0100)]
virDomainNumatuneParseXML: Refactor cleanup
Use automatic memory clearing for the temporary strings and bitmap and
remove the cleanup section. There are multiple temporary strings added
so that we don't reuse one.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Michal Privoznik [Fri, 10 Dec 2021 12:46:54 +0000 (13:46 +0100)]
Revert "lxc: controller: Fix container launch on cgroup v1"
Unfortunately, this fix breakes machinectl in a very nasty way,
for instance 'machinectl shell' drops into the host shell. It's
worse than being unable to start a container with CGroupsV1.
References: https://gitlab.com/libvirt/libvirt/-/issues/182 Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
remote_daemon: Set shutdown callbacks only after init is done
The initialization of drivers happens in a separate thread.
However, the main thread continues initialization and sets
shutdown callbacks (virStateShutdownPrepare() and
virStateShutdownWait()) even though the driver init thread is
still running. This is dangerous because if the daemon decides to
quit early (e.g. because SIGINT was delivered) the
shutdownPrepare and shutdownWait callback are called over
partially init drivers.
Set callbacks only after all drivers were initialized.
Peter Krempa [Wed, 1 Dec 2021 15:21:29 +0000 (16:21 +0100)]
virDomainMigrate: Introduce VIR_MIGRATE_NON_SHARED_SYNCHRONOUS_WRITES flag
Non-shared storage migration of guests which are disk I/O intensive and
have fast local storage may actually never converge if the guest happens
to dirty the disk faster than it can be copied.
This patch introduces a new flag
'VIR_MIGRATE_NON_SHARED_SYNCHRONOUS_WRITES' which will instruct
hypervisors to synchronize local I/O writes with the writes to remote
storage used for migration so that the guest can't overwhelm the
migration. This comes at a cost of decreased local I/O performance for
guests which behave well on average.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Wire up the flag to enable the 'write-blocking' 'copy-mode' of
'blockdev-mirror'.
It's not supported by all qemu versions but it is with those which we
use -blockdev with so we can use that instead of adding another custom
capability as we use blockdev for some time now.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Peter Krempa [Wed, 1 Dec 2021 14:09:45 +0000 (15:09 +0100)]
virDomainBlockCopy: Introduce VIR_DOMAIN_BLOCK_COPY_SYNCHRONOUS_WRITES flag
In cases when the destination storage is slower than the normal VM
storage and the VM does intensive I/O to the disk a block copy job may
never converge.
Switching it to synchronous mode will ensure that all writes done by the
guest are propagated to the destination at the cost of slowing down I/O
of the guest to the synchronous speed.
This patch adds the new API flag and implements virsh support.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Peter Krempa [Tue, 30 Nov 2021 15:20:50 +0000 (16:20 +0100)]
qemuMonitorJSONGraphicsRelocate: Clean up command argument construction
Move the construction of the command from the variable declaration so
that it doesn't exceed the line length and we can also move the logic of
determining the protocol outside of the command construction.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Peter Krempa [Tue, 30 Nov 2021 15:12:00 +0000 (16:12 +0100)]
qemuMonitorJSONMigrate: Extract flags prior to constructing command
The migration API takes specific flags which are then converted to
boolean parameters for the command. Extract the flag into helper
variables rather than using ternary operators while constructing the
command itself.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Peter Krempa [Wed, 1 Dec 2021 08:21:33 +0000 (09:21 +0100)]
virLockDaemonClientPreExecRestart: Modernize JSON object construction
Use virJSONValueObjectAdd instead of step-by-step construction of the
object. This also removes a handful impossible to reach errors with
translatable messages.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Laine Stump [Thu, 2 Dec 2021 18:52:48 +0000 (13:52 -0500)]
util: call virNetDevGetPhysPortID() in less places
Whenever virPCIGetNetName() is called, it is either called with
physPortID = NULL, or with it set by the caller calling
virNetDevGetPhysPortID() soon before virPCIGetNetName(). The
physPortID is then used *only* in virPCIGetNetName().
Rather than replicating that same call to virNetDevGetPhysPortID() in
all the callers of virPCIGetNetName(), lets just have all those
callers send the NetDevName whose physPortID they want down to
virPCIGetNetName(), and let virPCIGetNetName() call
virNetDevGetPhysPortID().
Signed-off-by: Laine Stump <laine@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Laine Stump [Thu, 2 Dec 2021 19:13:18 +0000 (14:13 -0500)]
util: fix erroneous requirement for phys_port_id to get ifname of a VF
Commit 795e9e05c3 (libvirt-7.7.0) refactored the code in virpci.c and
virnetdev.c that gathered lists of the Virtual Functions (VF) of an
SRIOV Physical Function (PF) to simplify the code.
Unfortunately the simplification made the assumption, in the new
function virPCIGetVirtualFunctionsFull(), that a VF's netdev
interface name should only be retrieved if the PF had a valid
phys_port_id. That is an incorrect assumption - only a small handful
of (now previous-generation) Mellanox SRIOV cards actually use
phys_port_id (this is for an odd design where there are multiple
physical network ports on a single PCI address); all other SRIOV cards
(including new Mellanox cards) have a file in sysfs called
phys_port_id, but it can't be read, and so the pfPhysPortID string is
NULL.
The result of this logic error is that virtual networks that are a
pool of VFs to be used for macvtap connections will be unable to
start, giving an errror like this:
VF 0 of SRIOV PF enp130s0f0 couldn't be added to the interface pool because it isn't bound to a network driver - possibly in use elsewhere
This error message is misinformed - the caller of
virNetDevGetVirtualFunctionsFull() only *thinks* that the VF isn't
bound to a network driver because it doesn't see a netdev name for the
VF in the list. But that's only because
virNetDevGetVirtualFunctionsFull() didn't even try to get the names!
We do need a way for virPCIGetVirtualFunctionsFull() to sometimes
retrieve the netdev names and sometimes not. One way of doing that
would be to send down the netdev name of the PF whenever we also want
to know the netdev names of the VFs, but send a NULL when we
don't. This can conveniently be done by just *replacing* pfPhysPortID
in the arglist with pfNetDevName - pfPhysPortID is determined by
simply calling virNetDevGetPhysPortID(pfNetDevName) so we can just
make that call down in virPCIGetVirtualFunctionsFull() (when needed).
This solves the regression introduced by commit 795e9e05c3, and also
nicely sets us up to (in a subsequent commit) move the call to
virNetDevGetPhysPortID() down one layer further to virPCIGetNetName(),
where it really belongs!
Resolves: https://bugzilla.redhat.com/2025432 Fixes: 795e9e05c3b6b9ef3abe6f6078a6373a136ec23b Signed-off-by: Laine Stump <laine@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
The Homebrew package explicitly enables this driver despite us
disabling it by default on macOS, so it must be functional to
at least some extent and certainly can't be causing any build
failures.
Additionally, if the user has explicitly asked for the network
driver to be enabled but libvirtd is disabled for whatever
reason, we should error out instead of silently disabling the
network driver.
Signed-off-by: Andrea Bolognani <abologna@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
projects / libvirt.git / log