util: virhostcpu: Fail when fetching CPU Stats for invalid cpu
virHostCPUGetStatsLinux walks through every cpu in /proc/stat until it
finds cpu%cpuNum that matches with the requested cpu.
If none is found it logs the error but it should return -1, instead of 0.
Otherwise virsh nodecpustats --cpu <invalid cpu number> and API bindings
don't fail properly, printing a blank line instead of an error message.
This patch also includes an additional test for virhostcputest to avoid
this regression to happen again in the future.
Fixes: 93af79fba3fd75a8df6b7ca608719dd97f9511a0 Reported-by: Satheesh Rajendran <satheera@in.ibm.com> Signed-off-by: Mauro S. M. Rodrigues <maurosr@linux.vnet.ibm.com> Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Christian Ehrhardt <christian.ehrhardt@canonical.com> Tested-by: Christian Ehrhardt <christian.ehrhardt@canonical.com>
Using the 'uuid' element for ppc64 NVDIMM memory added in the
previous patch, use it in qemuBuildMemoryDeviceStr() to pass
it over to QEMU.
Another ppc64 restriction is the necessity of a mem->labelsize,
given than ppc64 only support label-area backed NVDIMMs.
Finally, we don't want ppc64 NVDIMMs to align up due to the
high risk of going beyond the end of file with a 256MiB
increment that the user didn't predict. Align it down
instead. If target size is less than the minimum of
256MiB + labelsize, error out since QEMU will error out
if we attempt to round it up to the minimum.
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com> Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
conf: Introduce optional 'uuid' element for NVDIMM memory
ppc64 NVDIMM support was implemented in QEMU by commit [1].
The support is similar to what x86 already does, aside from
an extra 'uuid' element.
This patch introduces a new optional 'uuid' element for the
NVDIMM memory model. This element behaves like the 'uuid'
element of the domain definition - if absent, we'll create
a new one, otherwise use the one provided by the XML.
The 'uuid' element is exclusive to pseries guests and are
unavailable for other architectures.
Next patch will use this new element to add NVDIMM support
for ppc64.
qemu: capabilities: update qemu-5.0.0 capabilities for ppc64
Update ppc64 capabilities to pick up the new NVDIMM capability
support for ppc64.
Since the ppc64 capabilities weren't updated for some time, the
bulk of the changes here are related to the blockdev support
(see commit c6a9e54ce3 for info) that we are picking up just
now.
Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Michal Privoznik [Mon, 23 Mar 2020 14:49:59 +0000 (15:49 +0100)]
news: Document removal of support for .ini style of comments
In previous patches virKeyFile was replaced with its GLib
counterpart which created an incompatible change: comments can
now begin only with a number sign (#). While this won't probably
affect anyone, mention it in the release notes.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Rafael Fonseca [Mon, 23 Mar 2020 13:19:13 +0000 (14:19 +0100)]
util: remove virKeyFile
The functionality is now provided by glib's GKeyFile.
Signed-off-by: Rafael Fonseca <r4f4rfs@gmail.com> Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Rafael Fonseca [Mon, 23 Mar 2020 13:19:12 +0000 (14:19 +0100)]
util: authconfig: use g_key_file_*
Replace libvirt's virKeyFile by glib's GKeyFile.
Signed-off-by: Rafael Fonseca <r4f4rfs@gmail.com> Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Jonathon Jongsma [Fri, 20 Mar 2020 14:43:13 +0000 (09:43 -0500)]
api: disallow virDomainAgentSetResponseTimeout() on read-only connections
This function changes the amount of time that libvirt waits for a
response from the guest agent for all guest agent commands. Since this
is a configuration change, it should not be allowed on read-only
connections.
Signed-off-by: Jonathon Jongsma <jjongsma@redhat.com> Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Jonathon Jongsma [Fri, 20 Mar 2020 22:28:10 +0000 (17:28 -0500)]
qemu: fix response timeout for agent guest-sync
The agent 'guest-sync' command historically had a 5s response timeout
which was different from other agent commands, which waited forever.
When we added the ability to customize the response timeout for guest
agent commands, we intended to continue to use 5s for 'guest-sync' when
the user specified a response timeout greater than 5s, and use the
user-specified timeout if it was below 5s. Unfortunately, when
attempting to determine whether the user-specified timeout was less than
5s, we were comparing against an enum value of
VIR_DOMAIN_QEMU_AGENT_COMMAND_DEFAULT (which is -1) rather than against
the actual time value that it represented (5).
This change makes it so that 'guest-sync' now uses the user-specified
tiemout if it is less than 5s.
Signed-off-by: Jonathon Jongsma <jjongsma@redhat.com> Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Michal Privoznik [Wed, 18 Mar 2020 09:18:46 +0000 (10:18 +0100)]
security: Try harder to run transactions
When a QEMU process dies in the middle of a hotplug, then we fail
to restore the seclabels on the device. The problem is that if
the thread doing hotplug locks the domain object first and thus
blocks the thread that wants to do qemuProcessStop(), the
seclabel cleanup code will see vm->pid still set and mount
namespace used and therefore try to enter the namespace
represented by the PID. But the PID is gone really and thus
entering will fail and no restore is done. What we can do is to
try enter the namespace (if requested to do so) but if entering
fails, fall back to no NS mode.
Michal Privoznik [Wed, 18 Mar 2020 11:59:08 +0000 (12:59 +0100)]
virprocess: Passthru error from virProcessRunInForkHelper
When running a function in a forked child, so far the only thing
we could report is exit status of the child and the error
message. However, it may be beneficial to the caller to know the
actual error that happened in the child.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Pavel Mores <pmores@redhat.com>
virfile: Handle directories in virFileBindMountDevice()
The @src is not always a file. It may also be a directory (for
instance qemuDomainCreateDeviceRecursive() assumes that) - even
though it doesn't happen usually. Anyway, mount() can mount only
a dir onto a dir and a file onto a file.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Pavel Mores <pmores@redhat.com>
The @devPath variable is not modifiable. It merely just points to
string containing path where private devtmpfs is being
constructed. Make it const so it doesn't look weird that it's not
freed.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Pavel Mores <pmores@redhat.com>
qemuDomainBuildNamespace: Try harder to remove temp directories
If building namespace fails somewhere in the middle (that is some
files exists under devMountsSavePath[i]), then plain rmdir() is
not enough to remove dir. Umount the temp location and use
virFileDeleteTree() to remove the directory.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Pavel Mores <pmores@redhat.com>
qemuDomainCreateDeviceRecursive: Report error if mkdir() fails
The virFileMakePathWithMode() which is our recursive version of
mkdir() fails, it simply just returns a negative value with errno
set. No error is reported (as compared to virFileTouch() for
instance).
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Pavel Mores <pmores@redhat.com>
Peter Krempa [Thu, 14 Dec 2017 17:03:04 +0000 (18:03 +0100)]
virjson: Deflatten arrays generated by the json->commandline generator
For the few instances where we'd generate an array in dotted syntax we
should be able to parse it back. Add another step in deflattening of the
dotted syntax which reconstructs the arrays so that the backing store
parser can parse it.
Peter Krempa [Thu, 19 Mar 2020 16:23:33 +0000 (17:23 +0100)]
qemu: block: Split up formatting of JSON props for 'raw' and 'luks' drivers
qemuBlockStorageSourceGetFormatRawProps aggregated both formats but
since we now have props specific for either of those formats it's
unwanted to aggregate the code such way. Split out the 'luks' props
formatter into qemuBlockStorageSourceGetFormatLUKSProps.
The wrong separation demonstrates istself on formatting of the 'size'
and 'offset' attributes for the 'luks' driver which does not conform
to the qapi schema.
Peter Krempa [Thu, 19 Mar 2020 15:54:52 +0000 (16:54 +0100)]
qemuBlockStorageSourceNeedsStorageSliceLayer: Deal with 'luks' files
The 'luks' driver in qemu is as any other non-raw format driver and thus
doesn't support the properties for 'slice'. Since libvirt considers
luks files to be raw+encryption we need to special case them when
dealing with the slice.
Michal Privoznik [Thu, 19 Mar 2020 11:51:55 +0000 (12:51 +0100)]
qemu: Don't crash when getting targets for a multipath
In one of my previous commits I've introduced code that creates
all devices for given (possible) multipath target. But I've made
a mistake there - the code accesses 'next->path' without checking
if the disk source is local. Note that the 'next->path' is
NULL/doesn't make sense for VIR_STORAGE_TYPE_NVME.
Pino Toscano [Wed, 18 Mar 2020 09:46:55 +0000 (10:46 +0100)]
vmx: make 'fileName' optional for CD-ROMs
It seems like CD-ROMs may have no 'fileName' property specified in case
there is nothing configured as attachment for the drive. Hence, make
sure that virVMXParseDisk() do not consider it mandatory anymore,
considering it an empty block cdrom device. Sadly virVMXParseDisk() is
used also to parse disk and floppies, so make sure that a NULL fileName
is handled in cdrom- and floppy-related paths.
Pino Toscano [Wed, 18 Mar 2020 09:29:51 +0000 (10:29 +0100)]
vmx: shortcut earlier few 'ignore' cases in virVMXParseDisk()
Move earlier the checks for skipping a hard disk when parsing a CD-DROM,
and for skipping a CD-ROM when parsing a hard disk. This should have no
behaviour changes, and avoids to add repeated checks in following
commits.
Signed-off-by: Pino Toscano <ptoscano@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com> Tested-by: Richard W.M. Jones <rjones@redhat.com>
Peter Krempa [Wed, 18 Mar 2020 11:24:40 +0000 (12:24 +0100)]
qemu: Suppress error reporting from qemuMonitorDelObject in cleanup paths
Many calls of qemuMonitorDelObject don't actually check the return value
or report the error from the object deletion itself since they are on
cleanup paths. In some cases this can lead to reporting of spurious
errors e.g. when qemuMonitorDelObject is used to clean up a possibly
pre-existing objects.
Add a new argument for qemuMonitorDelObject which controls whether
the internals report errors from qemu and fix all callers accordingly.
Note that some of the cases on device unplug which check the error code
don't in fact propagate the error to the user, but in this case it is
important to add the log entry anyways for tracing that the device
deletion failed.
Peter Krempa [Wed, 18 Mar 2020 10:08:58 +0000 (11:08 +0100)]
qemuMonitorJSONCheckError: Allow suppressing of error reporting
In some cases we'll need to check whether there was an error but avoid
reporting an actual libvirt error. Rename qemuMonitorJSONCheckError to
qemuMonitorJSONCheckErrorFull with a new flag to suppress the error
reporting and add a wrapper with the original name so that callers don't
need to be fixed.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
When changing media we'd attempt to remove the managed pr daemon even if
neither of the images involved in the media change used it. This caused
libvirtd to log a spurious error:
2020-03-18 01:41:19.832+0000: 643207: error : qemuMonitorJSONCheckError:412 : internal error: unable to execute QEMU command 'object-del': object 'pr-helper0' not found
With this patch we completely avoid calling the deletion code.
The loop which checks whether the vcpus are in proper configuration for
the requested hot(un)plug skips the first modified vcpu. This means
that 'firstvcpu' which is used to print the error message in case the
configuration is not suitable would never point to the first modified
vcpu.
Michal Privoznik [Wed, 11 Mar 2020 16:56:12 +0000 (17:56 +0100)]
conf: Don't generate clashing machine names for embed driver
So far, when using the qemu:///embed driver, management
applications can't chose whether they want to register their
domains in machined or not. While having that option is certainly
desired, it will require more work. What we can do meanwhile is
to generate names that include part of hash of the root
directory. This is to ensure that if two applications using
different roots but the same domain name (and ID) start the
domain no clashing name for machined is generated.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Andrea Bolognani <abologna@redhat.com>
Michal Privoznik [Wed, 11 Mar 2020 17:33:34 +0000 (18:33 +0100)]
qemu_conf: Track embed root dir
When initializing virQEMUDriverConfig structure we are given the
root directory for possible embed connection. Save it for future
use. While we could get it later from @uri member, it's not as
easy as dereferencing a pointer (virURIParse() +
virURIGetParam() + error reporting).
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Andrea Bolognani <abologna@redhat.com>
Jiri Denemark [Mon, 9 Mar 2020 13:14:04 +0000 (14:14 +0100)]
cpu: Honor check='full' for host-passthrough CPUs
The check attribute was completely ignored for host-passthrough CPUs
even if they explicitly requested some features to be enabled. For
example, a domain with the following CPU definition
would happily start even when 'svm' cannot be enabled.
Let's call virCPUArchUpdateLive for host-passthrough CPUs with
VIR_CPU_CHECK_FULL to make sure the architecture specific code can
validate the provided virtual CPU against the desired definition.
Jiri Denemark [Mon, 9 Mar 2020 09:30:39 +0000 (10:30 +0100)]
cpu: Change control flow in virCPUUpdateLive
The updateLive CPU sub-driver function is supposed to be called only for
a subset of CPU definitions. Let's make it more obvious by turning a
negative test and return into a positive check.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Julio Faracco [Mon, 2 Mar 2020 00:54:13 +0000 (21:54 -0300)]
lxc: Add HPET device into allowed devices
This commit is related to RTC timer device too. HPET is being shared
from host device through `localtime` clock. This timer is available
creating a new timer using `hpet` name.
Signed-off-by: Julio Faracco <jcfaracco@gmail.com> Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Julio Faracco [Mon, 2 Mar 2020 00:54:12 +0000 (21:54 -0300)]
lxc: Add Real Time Clock device into allowed devices
This commit share host Real Time Clock device (rtc) into LXC containers
to support hardware clock. This should be available setting up a `rtc`
timer under clock section. Since this option is not emulated, it should
be available only for `localtime` clock. This option should be readonly
due to security reasons.
Before:
root# hwclock --verbose
hwclock from util-linux 2.32.1
System Time: 1581877557.598365
Trying to open: /dev/rtc0
Trying to open: /dev/rtc
Trying to open: /dev/misc/rtc
No usable clock interface found.
hwclock: Cannot access the Hardware Clock via any known method.
Now:
root# hwclock
2020-02-16 18:23:55.374134+00:00
root# hwclock -w
hwclock: ioctl(RTC_SET_TIME) to /dev/rtc to set the time failed:
Permission denied
Signed-off-by: Julio Faracco <jcfaracco@gmail.com> Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Peter Krempa [Tue, 17 Mar 2020 14:51:57 +0000 (15:51 +0100)]
qemublocktest: Add tests for re-enabling of bitmaps after commit
Some branches were not covered and thus we didn't catch that the bitmaps
are not re-enabled if nothing is merged into them. Two bitmaps are
necessary to reliably test the case due to hash table ordering.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Tested-by: Daniel P. Berrangé <berrange@redhat.com>
Peter Krempa [Tue, 17 Mar 2020 14:31:12 +0000 (15:31 +0100)]
qemuBlockBitmapsHandleCommitFinish: Use proper variable to iterate
The function repeatedly checked the first element rather than iterating
through the array.
Reported-by: Daniel P. Berrangé <berrange@redhat.com> Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Tested-by: Daniel P. Berrangé <berrange@redhat.com>
Peter Krempa [Tue, 17 Mar 2020 14:10:23 +0000 (15:10 +0100)]
qemuBlockBitmapsHandleCommitStart: Fix allocation of string list
Allocate space also for the terminating NULL.
Reported-by: Daniel P. Berrangé <berrange@redhat.com> Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com> Tested-by: Daniel P. Berrangé <berrange@redhat.com>
qemu: lookup node device against nodedev driver before getting XML
Some of the node device APIs are a little odd because they accept a
virNodeDevicePtr object but are still implemented by the virt drivers.
The first thing the virt drivers need to do is get the XML config
associated with the node device, and that means talking to the node
device driver.
This worked previously because with monolithic libvirtd, both the
virt driver and node device driver were in the same daemon and thus
a single virConnectPtr can talk to both drivers.
With the split daemon world though, the virNodeDevicePtr passed into
the APIs is associated with the QEMU driver virConnectPtr, which has
no ability to invoke APIs against the node device driver. We must thus
get a duplicate virNodeDevicePtr object which is associated with a
virConnectPtr for the node device driver.
Reviewed-by: Michal Privoznik <mprivozn@redhat.com> Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
rpc: avoid name lookup when dispatching node device APIs
The node device APIs are a little unusual because we don't use a
"remote_nonnull_node_device" object on the wire, instead we just
have a "remote_string" for the device name. This meant dispatcher
code generation needed special cases. In doing so we mistakenly
used the virNodeDeviceLookupByName() API which gets dispatched
into the driver, instead of get_nonnull_node_device() which
directly populates a virNodeDevicePtr object.
This wasn't a problem with monolithic libvirtd, as the
virNodeDeviceLookupByName() API call was trivially satisfied
by the registered driver, albeit with an extra (undesirable)
authentication check. With the split daemons, the call to
virNodeDeviceLookupByName() fails in virtqemud, because the
node device driver obviously doesn't exist in that daemon.
Reviewed-by: Michal Privoznik <mprivozn@redhat.com> Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
are all handled by the virt drivers, not the node device driver.
A bug in the RPC generator meant that these APIs were sent to
the nodedev driver for handling. This caused breakage with the
split daemons, since nothing was available to process them.
Reviewed-by: Michal Privoznik <mprivozn@redhat.com> Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
virhostcpu.c: fix 'die_id' parsing for Power hosts
Commit 7b79ee2f78 makes assumptions about die_id parsing in
the sysfs that aren't true for Power hosts. In both Power8
and Power9, running 5.6 and 4.18 kernel respectively,
'die_id' is set to -1:
This breaks virHostCPUGetDie() parsing because it is trying to
retrieve an unsigned integer, causing problems during VM start:
virFileReadValueUint:4128 : internal error: Invalid unsigned integer
value '-1' in file '/sys/devices/system/cpu/cpu0/topology/die_id'
This isn't necessarily a PowerPC only behavior. Linux kernel commit 0e344d8c70 added in the former Documentation/cputopology.txt, now
Documentation/admin-guide/cputopology.rst, that:
To be consistent on all architectures, include/linux/topology.h
provides default definitions for any of the above macros that are
not defined by include/asm-XXX/topology.h:
This means that it might be expected that an architecture that
does not implement the die_id element will mark it as -1 in
sysfs.
It is not required to change die_id implementation from uInt to
Int because of that. Instead, let's change the parsing of the
die_id in virHostCPUGetDie() to read an integer value and, in
case it's -1, default it to zero like in case of file not found.
This is enough to solve the issue Power hosts are experiencing.
Fixes: 7b79ee2f78bbf2af76df2f6466919e19ae05aeeb Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
nodedev: fix race in API usage vs initial device enumeration
During startup the udev node device driver impl uses a background thread
to populate the list of devices to avoid blocking the daemon startup
entirely. There is no synchronization to the public APIs, so it is
possible for an application to start calling APIs before the device
initialization is complete.
This was not a problem in the old approach where libvirtd was started
on boot, as initialization would easily complete before any APIs were
called.
With the use of socket activation, however, APIs are invoked from the
very moment the daemon starts. This is easily seen by doing a
'virsh -c nodedev:///system list'
the first time it runs it will only show one or two devices. The second
time it runs it will show all devices. The solution is to introduce a
flag and condition variable for APIs to synchronize against before
returning any data.
Reviewed-by: Michal Privoznik <mprivozn@redhat.com> Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Peter Krempa [Tue, 3 Mar 2020 13:14:28 +0000 (14:14 +0100)]
qemuDomainBlockPivot: Handle merging of bitmaps when pivoting an active block-commit
Active layer block commit makes the 'base' image the new top image of
the disk after it finishes. This means that all bitmap operations need
to be handled prior to this happening as we'd lose writes otherwise.
The ideal place is to handle it when pivoting to the new image as only
guest-writes would be happening after this point.
Use qemuBlockBitmapsHandleCommitFinish to calculate the merging
transaction.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com>
Peter Krempa [Tue, 3 Mar 2020 12:59:48 +0000 (13:59 +0100)]
qemuDomainBlockCommit: Handle bitmaps on start of commit
On start of the commit job, we need to disable any active bitmap in the
base. Use qemuBlockBitmapsHandleCommitStart to calculate which and call
the appropriate QMP APIs. We use blockdev-reopen to make the 'base'
writable to disable the bitmaps.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com>
Peter Krempa [Mon, 2 Mar 2020 16:16:17 +0000 (17:16 +0100)]
qemublocktest: Add tests for handling of bitmaps during block-commit
Add code for testing the two necessary steps of handling bitmaps during
block commit and exercise the code on the test data which we have for
bitmap handling.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com>
Peter Krempa [Mon, 2 Mar 2020 13:40:18 +0000 (14:40 +0100)]
qemu: block: Implement helpers for dealing with bitmaps during block commit
qemuBlockBitmapsHandleCommitStart prepares for disabling the bitmaps in
the 'base' of the commit job so that the bitmaps are not dirtied by the
commit job. This needs to be done prior to start of the commit job.
qemuBlockBitmapsHandleCommitFinish then calculates the necessary merges
that agregate all the bitmaps between the commited images and write them
into the base bitmap.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com>
Peter Krempa [Tue, 25 Feb 2020 06:28:44 +0000 (07:28 +0100)]
qemu: blockjob: Store list of bitmaps disabled prior to commit
Starting a commit job will require disabling bitmaps in the base image
so that they are not dirtied by the commit job. We need to store a list
of the bitmaps so that we can later re-enable them.
Add a field and status XML handling code as well as a test.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com>
Peter Krempa [Tue, 3 Mar 2020 15:42:23 +0000 (16:42 +0100)]
qemuDomainBlockCommit: Move checks depending on capabilities after liveness check
Since capabilities are not present for inactive VMs we'd report that we
don't support '--delete' or committing while checkpoints exist rather
than the proper error.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com>
Peter Krempa [Mon, 24 Feb 2020 16:30:26 +0000 (17:30 +0100)]
qemuCheckpointDiscardBitmaps: Use correct field for checkpoint bitmap name
The code deleting checkpoints needs the name of the parent checkpoint's
disk's bitmap but was using the disk alias instead. This would create
wrong bitmaps after deleting some checkpoints.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com>
Peter Krempa [Thu, 13 Feb 2020 11:49:14 +0000 (12:49 +0100)]
qemuCheckpointDiscardBitmaps: Reopen images for bitmap modifications
Qemu's bitmap APIs don't reopen the appropriate images read-write for
modification. It's libvirt's duty to reopen them via blockdev-reopen
if we wish to modify the bitmaps.
Use the new helpers to reopen the images for bitmap manipulation.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com>
Peter Krempa [Thu, 13 Feb 2020 08:24:36 +0000 (09:24 +0100)]
qemu: block: implement helpers for blockdev-reopen
Introduce a set of helpers to call blockdev-reopen in certain scenarios
Libvirt will use the QMP command to turn certain members of the backing
chain read-write for bitmap manipulation and we'll also want to use it
to replace/install the backing chain of a qcow2 format node.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com>
Peter Krempa [Thu, 27 Feb 2020 11:23:29 +0000 (12:23 +0100)]
qemu: capabilities: Add QEMU_CAPS_BLOCKDEV_REOPEN
This capability will be asserted once qemu stabilizes 'blockdev-reopen'.
For now we just add the capability so that we can introduce some code
that will use the reopening call. This will show our willingness to
adopt use of reopen and help qemu developers stabilize it.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Eric Blake <eblake@redhat.com>
qemu: fix detection of vCPU pids when multiple dies are present
The logic for querying hotpluggable CPUs needs to sort the list
of CPUs returned by QEMU. Unfortunately our sorting method failed
to use the die_id field, so CPUs were not correctly sorted.
This is seen when configuring a guest with partially populated
CPUs
Peter Krempa [Mon, 9 Mar 2020 14:05:58 +0000 (15:05 +0100)]
qemu: Pass through arguments of 'ssh' block driver used by libguestfs
We currently don't model the 'ssh' protocol properties properly and
since it seems impossible for now (agent path passed via environment
variable). To allow libguestfs to work as it used in pre-blockdev era we
must carry the properties over to the command line. For this instance we
just store it internally and format it back.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Peter Krempa [Mon, 9 Mar 2020 16:06:10 +0000 (17:06 +0100)]
qemublocktest: Add JSON->JSON test cases for block device backends
Add testing of the interpretation of the JSON pseudo-protocol backing
store into JSON structs for blockdev. This will be used to test JSON
pseudo-URIs used by libguestfs while actually also validating the output
against the QMP schema. Since libguestfs uses obsolete/undocumented
values the outputs will differ and a benefit is that modern output is
used now.
The example test case covers the fields and values used by libguestfs
when using the https driver.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Peter Krempa [Mon, 9 Mar 2020 14:39:04 +0000 (15:39 +0100)]
qemublocktest: XMLjsonXML: Test formatting/parsing of modern JSON
The test was invoking the JSON formatter with the 'legacy' flag thus
formatting bunch of obsolete JSON blockdev definitions. We also should
test the modern ones. Add a boolean and re-run all the tests in both
cases.
Additionally for any modern invocation we should also validate that the
output conforms to the QAPI schema.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Peter Krempa [Fri, 6 Mar 2020 07:29:59 +0000 (08:29 +0100)]
virStorageSourceParseBackingJSONUri: Handle undocumented value 'off' for sslverify
libguestfs abuses a quirk of qemu's parser to accept also other variants
of the 'sslverify' field which would be valid on the command line but
are not documented in the QMP schema.
If we encounter the 'off' string instead of an boolean handle it rather
than erroring out to continue support of pre-blockdev configurations.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Peter Krempa [Mon, 9 Mar 2020 07:19:02 +0000 (08:19 +0100)]
qemu: domain: Store data for 'secret' object representing http cookies
The http cookies can have potentially sensitive values and thus should
not be leaked into the command line. This means that we'll need to
instantiate a 'secret' object in qemu to pass the value encrypted.
This patch adds infrastructure for storing of the alias in the status
XML.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
projects / libvirt.git / log