Peter Krempa [Tue, 20 Jul 2021 07:28:51 +0000 (09:28 +0200)]
qemu: process: Ignore 'RESET' event during startup
In cases when we are adding a <transient/> disk with sharing backend
(and thus hotplugging it) we need to re-initialize ACPI tables so that
the VM boots from the correct device.
This has a side-effect of emitting the RESET event and forwarding it to
the clients which is not correct.
Fix this by ignoring RESET events during startup of the VM.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Peter Krempa [Fri, 20 Aug 2021 13:40:18 +0000 (15:40 +0200)]
qemuDomainSetLifecycleAction: Properly update 'onReboot' action in qemu
When qemu supports 'set-action' command we can update what happens on
reboot. Additionally we can fully relax the checks as we now properly
update the lifecycle actions.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Peter Krempa [Thu, 19 Aug 2021 11:50:16 +0000 (13:50 +0200)]
qemu: command: Always use '-no-shutdown'
The '-no-shutdown' flag prevents qemu from terminating if a shutdown was
requested. Libvirt will handle the termination of the qemu process
anyways and using this consistently will allow greater flexibility for
the virDomainSetLifecycleAction API as well as will allow using
the 'system-reset' QMP command during startup to reinitiate devices
exported to the firmware.
Peter Krempa [Mon, 23 Aug 2021 14:40:46 +0000 (16:40 +0200)]
qemuProcessLaunch: Setup handling of 'on_reboot' via QMP when starting the process
Rather than using '-no-reboot' use the QMP command to update the
lifecycle action of 'on_reboot'.
This will be identical to how we set the behaviour during lifetime and
also avoids problems with use of the 'system-reset' QMP command during
bringup of the VM (used to update the firmware table of disks when disks
were hotplugged as part of startup).
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Peter Krempa [Fri, 20 Aug 2021 14:16:46 +0000 (16:16 +0200)]
qemuProcessHandleReset: Don't emulate lifecycle actions for RESET event
The RESET event is delivered by qemu only when the guest OS is actually
allowed to reboot ('-no-reboot' or equivalent is not used) and due to
the nature of async handling of the events VM is actually already
executing guest code after the reboot, until our code gets to killing
it.
In general it should have been impossible to reach a state where the
reboot action is 'destroy' but we didn't use '-no-reboot' but due to
various bugs it was.
Due to the fact that this was not a desired operation and additionally
guest code already is executing I think the best option is not to kill
the VM any more (possible data loss?) and rely for the proper fix where
we use the new 'set-action' QMP command to enable an equivalent
behaviour to '-no-reboot' during runtime.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Peter Krempa [Fri, 20 Aug 2021 14:46:25 +0000 (16:46 +0200)]
qemuDomainSetLifecycleAction: Forbid live update of 'on_reboot'
Without the ability to tell qemu to change the behaviour on reboot of
the guest it's fundamentally unsafe to change the action as the guest
would be able to execute instructions after the reboot before libvirt
terminates it due to the async nature of QMP events.
Stricten the code for now until we implement support for 'set-action'
QMP command.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Peter Krempa [Fri, 20 Aug 2021 14:11:05 +0000 (16:11 +0200)]
qemu: migration: Don't transfer 'allowReboot' flag
The original idea was to ensure that the destination has the same
original state of the '-no-reboot' flag to ensure identical behaviour of
the 'vidDomainModifyLifecycleAction' API.
With newer qemu's we'll be able to modify the behaviour using the
monitor so old daemons won't be able to keep up anyways.
Remove this feature as it's not very useful and will be replaced by a
proper solution.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Peter Krempa [Thu, 19 Aug 2021 12:54:18 +0000 (14:54 +0200)]
qemuDomainObjPrivate: Annotate 'allowReboot' field
Save further readers the headache of determining what it actually does
and note that it's not used with qemu version supporting the
'set-action' command.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Peter Krempa [Thu, 19 Aug 2021 12:44:51 +0000 (14:44 +0200)]
qemu: Honor 'restart' action for 'on_poweroff'
We simply terminate qemu instead of issuing a reset as the semantics of
the setting dictate.
Fix it by handling it identically to 'fake reboot'.
We need to forbid the combination of 'onReboot' -> 'destroy' and
'onPoweroff' -> reboot though as the handling would be hairy and it
honetly makes no sense.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Peter Krempa [Tue, 24 Aug 2021 11:44:00 +0000 (13:44 +0200)]
qemu: driver: Validate lifecycle actions in 'qemuDomainSetLifecycleAction'
Some actions are not supported by qemu. Use the recently added
'qemuValidateLifecycleAction' helper to ensure that the API does the
same validation as we do on startup in the validation callbacks.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
In addition when 'on_reboot' is 'destroy' in addition to 'on_poweroff'
being 'rename-restart' the guest is able to execute instructions after
issuing a reset before libvirt terminates it. This will be addressed
separately later.
Forbid the flag in the qemu def validator and update the documentation
to be factual.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Peter Krempa [Fri, 20 Aug 2021 13:26:47 +0000 (15:26 +0200)]
qemu: driver: Use 'qemuDomainSaveStatus' for saving status XML
We've got multiple random open-coded versions. Switch to the helper
function which doesn't report errors as they'd be mostly wrong as the
operation was indeed successful.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Peter Krempa [Wed, 18 Aug 2021 12:28:23 +0000 (14:28 +0200)]
qemuxml2argvtest: Add 'cpu-host-model' cases for all x86-64 real qemu caps versions
The host model expansion depends on the capability data, so in this case
it makes sense to have specific invocations of the test for all qemu
versions we have.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Peter Krempa [Wed, 18 Aug 2021 12:02:36 +0000 (14:02 +0200)]
qemuxml2argvtest: Unify 'cpu-tsc-frequency' and 'cpu-tsc-high-frequency' case
Use the larger number in the original test to avoid having two files.
Additionally this avoids use of 'host-model' with DO_TEST_CAPS_LATEST in
cases when it isn't necessary for the purpose of the test as the CPU
model tends to change.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Peter Krempa [Wed, 18 Aug 2021 11:45:30 +0000 (13:45 +0200)]
qemuxml2argvdata: Don't use 'host-model' cpu in 'hugepages-memaccess3'
The test case doesn't really test anything about the specific CPU. Using
a host-model cpu with DO_TEST_CAPS_LATEST results in commandline changes
every time qemu updates the cpu definiton.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
storage_conf: add validation against schema in pool define
We need to validate the XML against schema if option '--validate'
was passed to the virsh command. This patch also includes
propagation of flags into the virStoragePoolDefParse() function.
Signed-off-by: Kristina Hanicova <khanicov@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Ján Tomko [Sun, 4 Oct 2020 22:20:27 +0000 (00:20 +0200)]
API: discourage usage of non-ListAll APIs
They require the caller to provide the maximum number
of array elements upfront, leading to either incomplete
results or violations of the zero-one-infinity rule.
Signed-off-by: Ján Tomko <jtomko@redhat.com> Reviewed-by: Martin Kletzander <mkletzan@redhat.com>
Simon Rowe [Mon, 23 Aug 2021 15:40:47 +0000 (15:40 +0000)]
iohelper: skip lseek() and ftruncate() on block devices
Signed-off-by: Simon Rowe <simon.rowe@nutanix.com> Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
src: add driver support for networkDefineXMLFlags()
I have added new driver functions which define network with given
flags. I have also replaced definitions of the functions without
flags with function calls to the new ones.
Signed-off-by: Kristina Hanicova <khanicov@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Peter Krempa [Mon, 23 Aug 2021 12:14:55 +0000 (14:14 +0200)]
conf: snapshot/checkpoint: Rewrite 'AlignDisk' logic to appease clang
New clang has a false-positive about value of 'olddisks' being unused
after being set. This is clearly wrong because we want to use
'g_autofree' to clear it later.
While I'm against modifying good code for the sake of bad static
analysis in this case it's not obvious that we depend on the lifetime of
'olddisks' being needed until the end of the function as we store
pointers into it into the hash table and later copy them out.
Rewrite the code by assigning to 'olddisks' earlier and then using
'olddisks' in the loop, so it's clear where the lifetime of the objects
ends, and this should also silence the warning.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Michal Privoznik [Thu, 19 Aug 2021 13:54:31 +0000 (15:54 +0200)]
virhostmem: Handle numactl-less build in hugepages allocation/reporting
When using 'virsh freepages' or 'virsh allocpages' then
virHostMemGetFreePages() or virHostMemAllocPages() is called,
respectively. But the following may happen: libvirt was built
without numactl support and thus a fake NUMA node was constructed
for capabilities, which means that startCell is going to be 0.
But we can't blindly pass startCell = 0 to virNumaGetPageInfo()
nor virNumaSetPagePoolSize() because they would operate over node
specific path (/sys/devices/system/node/nodeX) rather than NUMA
agnostic path (/sys/kernel/mm/hugepages/) and we are not
guaranteed that the former exists (kernel might have been built
without NUMA support).
Michal Privoznik [Thu, 19 Aug 2021 13:55:19 +0000 (15:55 +0200)]
virhostmem: Let caller pass max NUMA node to virHostMemAllocPages
In all three cases (LXC, QEMU and VBox drivers) the caller has
access to host capabilities and thus know the maximum NUMA node.
This means, that virHostMemAllocPages() doesn't have to query
it. Querying may fail if libvirt was compiled without numactl
support.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Martin Kletzander <mkletzan@redhat.com>
Michal Privoznik [Thu, 19 Aug 2021 13:53:39 +0000 (15:53 +0200)]
virhostmem: Let caller pass max NUMA node to virHostMemGetFreePages
In all three cases (LXC, QEMU and VBox drivers) the caller has
access to host capabilities and thus know the maximum NUMA node.
This means, that virHostMemGetFreePages() doesn't have to query
it. Querying may fail if libvirt was compiled without numactl
support.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Martin Kletzander <mkletzan@redhat.com>
Peter Krempa [Fri, 20 Aug 2021 11:26:13 +0000 (13:26 +0200)]
qemu: block: Use correct format name when formatting overlay of qcow2+luks
A logic bug in the code creating overlays on existing images resulted
into wrongly using "luks" instead of "qcow2" for the backing format if
the backing image is an luks-encrypted qcow2. The special format munging
is needed only for raw luks images.
In practice the impact is not as critical as to use encrypted images in
the backing chain the user must fully describe the backing chain
including backing images to provide encryption keys, which overrides the
metadata recorded in the qcow2 header.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
secret_conf: add validation against schema in define
We need to validate the XML against schema if option '--validate'
was passed to the virsh command. This patch also includes
propagation of flags into the virSecretDefParse() function.
Signed-off-by: Kristina Hanicova <khanicov@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com> Signed-off-by: Ján Tomko <jtomko@redhat.com>
We need to validate the XML against schema if option 'validate'
was passed to the 'iface-define' virsh command. For that we need
to allow validation flag and propagate flags to parse function.
Signed-off-by: Kristina Hanicova <khanicov@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com> Signed-off-by: Ján Tomko <jtomko@redhat.com>
I have added a new driver function which allows to define
nwfilter with given flags. I have also replaced definition of
nwfilterDefineXML() with function call to the new function.
Signed-off-by: Kristina Hanicova <khanicov@redhat.com> Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Michal Privoznik [Fri, 20 Aug 2021 08:41:41 +0000 (10:41 +0200)]
virsocket: Don't preserve errno in virSocketRecvFD()
When setting O_CLOEXEC flag on received FD fails the FD is closed
using VIR_FORCE_CLOSE(). But the call is wrapped in errno save
which is not necessary because VIR_FORCE_CLOSE() preserves errno
value.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Also remove the leftover assignment of test-lib.sh,
since any new code attempting to use the compare function
with reversed arguments should be rejected by review
for using shell instead of C or Python.
Signed-off-by: Ján Tomko <jtomko@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Ján Tomko [Thu, 19 Aug 2021 12:01:14 +0000 (14:01 +0200)]
syntax-check: drop useless useless-if-before-free
With most of new code using g_auto for cleanup, contributors
are used to most of the free fucntions handling NULL gracefully.
Also, despite finding some occurrences in current codebase:
avoid_if_before_free
~/libvirt/src/ch/ch_monitor.c: if (mon->vm)
virObjectUnref(mon->vm);
~/libvirt/src/util/virresctrl.c: if (a_type->masks[cache])
virBitmapFree(a_type->masks[cache]);
the check passes succesfully, because the script's logic:
Exit status:
0 one or more matches
1 no match
2 an error
does not play nicely with xargs:
xargs exits with the following status:
0 if it succeeds
123 if any invocation of the command exited with status 1-125
The list of functions is also out of date - e.g. qemuCapsFree has
been renamed since.
This also helps eliminate one more Perl script per our programming
languages strategy: https://libvirt.org/programming-languages.html
Signed-off-by: Ján Tomko <jtomko@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Ján Tomko [Thu, 19 Aug 2021 10:48:28 +0000 (12:48 +0200)]
syntax-check: replace vc-list-files
Directly invoke git ls-tree instead of the wrapper file which also:
* checks for other versioning systems
* prepends the source directory to all output lines
Since there is no srcdir prefix in the output anymore, also drop
the extra 'sed' invocation that removes it.
Signed-off-by: Ján Tomko <jtomko@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
projects / thirdparty / libvirt.git / log
