Also restore the label to its original value after qemu is finished
with the file.
Prior to this patch, qemu domain restore did not function properly if
selinux was set to enforce.
If an active migration operation fails, or is cancelled by the
admin, the QEMU on the destination is shutdown and the one on
the source continues running. It is important in shutting down
the QEMU on the destination, the security drivers don't reset
the file labelling/permissions.
* src/qemu/qemu_driver.c: Don't reset labelling/permissions
on migration abort
The patches for shared storage migration were not correctly written
for json mode. Thus the 'blk' and 'inc' parameters were never being
set. In addition they didn't set the QEMU_MONITOR_MIGRATE_BACKGROUND
so migration was synchronous. Due to multiple bugs in QEMU's JSON
impl this wasn't noticed because it treated the sync migration requst
as asynchronous anyway. Finally 'background' parameter was converted
to take arbitrary flags but not renamed, and not all uses were changed
to unsigned int.
* src/qemu/qemu_driver.c: Set QEMU_MONITOR_MIGRATE_BACKGROUND in
doNativeMigrate
* src/qemu/qemu_monitor_json.c: Process QEMU_MONITOR_MIGRATE_NON_SHARED_DISK
and QEMU_MONITOR_MIGRATE_NON_SHARED_INC flags
* src/qemu/qemu_monitor.c, src/qemu/qemu_monitor.h,
src/qemu/qemu_monitor_json.h, src/qemu/qemu_monitor_text.c,
src/qemu/qemu_monitor_text.h: change 'int background' to
'unsigned int flags' in migration APIs. Add logging of flags
parameter
During incoming migration the QEMU monitor is not able to be
used. The incoming migration code did not keep hold of the
job lock because migration is split across multiple API calls.
This meant that further monitor commands on the guest would
hang until migration finished with no timeout.
In this change the qemuDomainMigratePrepare method sets the
job flag just before it returns. The qemuDomainMigrateFinish
method checks for this job flag & clears it once done. This
prevents any use of the monitor between prepare+finish steps.
The qemuDomainGetJobInfo method is also updated to refresh
the job elapsed time. This means that virsh domjobinfo can
return time data during incoming migration
* src/qemu/qemu_driver.c: Keep a job active during incoming
migration. Refresh job elapsed time when returning job info
When configuring serial, parallel, console or channel devices
with a file, dev or pipe backend type, it is necessary to label
the file path in the security drivers. For char devices of type
file, it is neccessary to pre-create (touch) the file if it does
not already exist since QEMU won't be allowed todo so itself.
dev/pipe configs already require the admin to pre-create before
starting the guest.
* src/qemu/qemu_security_dac.c: set file ownership for character
devices
* src/security/security_selinux.c: Set file labeling for character
devices
* src/qemu/qemu_driver.c: Add character devices to cgroup ACL
The current code pattern requires that callers of qemuMonitorClose
check for the return value == 0, and if so, set priv->mon = NULL
and release the reference held on the associated virDomainObjPtr
The change d84bb6d6a3 violated that
requirement, meaning that priv->mon never gets set to NULL, and
a reference count is leaked on virDomainObjPtr.
This design was a bad one, so remove the need to check the return
valueof qemuMonitorClose(). Instead allow registration of a
callback that's invoked just when the last reference on qemuMonitorPtr
is released.
Finally there was a potential reference leak in qemuConnectMonitor
in the failure path.
* src/qemu/qemu_monitor.c, src/qemu/qemu_monitor.h: Add a destroy
callback invoked from qemuMonitorFree
* src/qemu/qemu_driver.c: Use the destroy callback to release the
reference on virDomainObjPtr when the monitor is freed. Fix other
potential reference count leak in connecting to monitor
Before issuing monitor commands it is neccessary to check whether
the guest is still running. Most places use virDomainIsActive()
correctly, but a few relied on 'priv->mon != NULL'. In theory
these should be equivalent, but the release of the last reference
count on priv->mon can be delayed a small amount of time until
the event handler is finally deregistered. A further ref counting
bug also means that priv->mon might be never released. In such a
case, code could mistakenly issue a monitor command and wait for
a response that will never arrive, effectively leaving the QEMU
driver waiting on virCondWait() forever..
To protect against these possibilities, make sure all code uses
virDomainIsActive(), not 'priv->mon != NULL'
* src/qemu/qemu_driver.c: Replace 'priv->mon != NULL' with
calls to 'priv->mon != NULL'()
Following Daniel Berrange's multiple helpful suggestions for improving
this patch and introducing another driver interface, I now wrote the
below patch where the nwfilter driver registers the functions to
instantiate and teardown the nwfilters with a function in
conf/domain_nwfilter.c called virDomainConfNWFilterRegister. Previous
helper functions that were called from qemu_driver.c and qemu_conf.c
were move into conf/domain_nwfilter.h with slight renaming done for
consistency. Those functions now call the function expored by
domain_nwfilter.c, which in turn call the functions of the new driver
interface, if available.
If VM startup fails early enough (can't find a referenced USB device),
libvirtd will crash trying to clear the VNC port bit, since port = 0,
which overflows us out of the bitmap bounds.
Fix this by being more defensive in the bitmap operations, and only
clearing a previously set VNC port.
Signed-off-by: Cole Robinson <crobinso@redhat.com>
Followup to https://bugzilla.redhat.com/show_bug.cgi?id=599091,
commit 20206a4b, to reduce disk waste in padding.
* src/qemu/qemu_monitor.h (QEMU_MONITOR_MIGRATE_TO_FILE_BS): Drop
back to 4k.
(QEMU_MONITOR_MIGRATE_TO_FILE_TRANSFER_SIZE): New macro.
* src/qemu/qemu_driver.c (qemudDomainSaveFlag): Update comment.
* src/qemu/qemu_monitor_text.c (qemuMonitorTextMigrateToFile): Use
two invocations of dd to output non-aligned large blocks.
* src/qemu/qemu_monitor_json.c (qemuMonitorJSONMigrateToFile):
Likewise.
Match earlier change for qemu pause support with virDomainCreateXML.
* src/qemu/qemu_driver.c (qemudDomainObjStart): Add parameter; all
callers changed.
(qemudDomainStartWithFlags): Implement flag support.
When a disk is on a root squashed NFS server, it may not be
possible to stat() the disk file in virCgroupAllowDevice.
The virStorageFileGetMeta method may also fail to extract
the parent backing store. Both of these errors have to be
ignored to avoid breaking NFS deployments
* src/qemu/qemu_driver.c: Ignore errors in cgroup setup to
keep root squash NFS happy
https://bugzilla.redhat.com/show_bug.cgi?id=589465
Some guests (eg with badly configured grub, or Windows' installation cd)
require quick response from the console user. That's why we have a
"launchPaused" option in vdsm.
To implement it via libvirt, we need to ask libvirt not to call
qemuMonitorStartCPUs() after starting qemu. Calling virDomainStop
immediately after the domain is up is inherently raceful.
* src/qemu/qemu_driver.c (qemudStartVMDaemon): Add new parameter;
all callers adjusted.
(qemudDomainCreate): Implement support for new flag.
When an attempt to hotplug a PCI device to a guest fails,
the device was left attached to pci-stub. It is neccessary
to reset the device and then attach it to the host driver
again.
* src/qemu/qemu_driver.c: Reattach PCI device to host if
hotadd fails
When SELinux is running in MLS mode, libvirtd will have a
different security level to the VMs. For libvirtd to be
able to connect to the monitor console, the client end of
the UNIX domain socket needs a different label. This adds
infrastructure to set the socket label via the security
driver framework
* src/qemu/qemu_driver.c: Call out to socket label APIs in
security driver
* src/qemu/qemu_security_stacked.c: Wire up socket label
drivers
* src/security/security_driver.h: Define security driver
entry points for socket labelling
* src/security/security_selinux.c: Set socket label based on
VM label
Adjust args to qemudStartVMDaemon() to also specify path to stdin_fd,
so this can be passed to the AppArmor driver via SetSecurityAllLabel().
This updates all calls to qemudStartVMDaemon() as well as setting up
the non-AppArmor security driver *SetSecurityAllLabel() declarations
for the above. This is required for the following
"apparmor-fix-save-restore" patch since AppArmor resolves the passed
file descriptor to the pathname given to open().
See https://bugzilla.redhat.com/show_bug.cgi?id=599091
Saving a paused 512MB domain took 3m47s with the old block size of 512
bytes. Changing the block size to 1024*1024 decreased the time to 56
seconds. (Doubling again to 2048*1024 yielded 0 improvement; lowering
to 512k increased the save time to 1m10s, about 20%)
The pointer to the xml describing the domain is saved into an object
prior to calling VIR_REALLOC_N() to make the size of the memory it
points to a multiple of QEMU_MONITOR_MIGRATE_TO_FILE_BS. If that
operation needs to allocate new memory, the pointer that was saved is
no longer valid.
To avoid this situation, adjust the size *before* saving the pointer.
(This showed up when experimenting with very large values of
QEMU_MONITOR_MIGRATE_TO_FILE_BS).
This patch that adds support for configuring 802.1Qbg and 802.1Qbh
switches. The 802.1Qbh part has been successfully tested with real
hardware. The 802.1Qbg part has only been tested with a (dummy)
server that 'behaves' similarly to how we expect lldpad to 'behave'.
The following changes were made during the development of this patch:
- Merging Scott's v13-pre1 patch
- Fixing endptr related bug while using virStrToLong_ui() pointed out
by Jim Meyering
- Addressing Jim Meyering's comments to v11
- requiring mac address to the vpDisassociateProfileId() function to
pass it further to the 802.1Qbg disassociate part (802.1Qbh untouched)
- determining pid of lldpad daemon by reading it from /var/run/libvirt.pid
(hardcode as is hardcode alson in lldpad sources)
- merging netlink send code for kernel target and user space target
(lldpad) using one function nlComm() to send the messages
- adding a select() after the sending and before the reading of the
netlink response in case lldpad doesn't respond and so we don't hang
- when reading the port status, in case of 802.1Qbg, no status may be
received while things are 'in progress' and only at the end a status
will be there.
- when reading the port status, use the given instanceId and vf to pick
the right IFLA_VF_PORT among those nested under IFLA_VF_PORTS.
- never sending nor parsing IFLA_PORT_SELF type of messages in the
802.1Qbg case
- iterating over the elements in a IFLA_VF_PORTS to pick the right
IFLA_VF_PORT by either IFLA_PORT_PROFILE and given profileId
(802.1Qbh) or IFLA_PORT_INSTANCE_UUID and given instanceId (802.1Qbg)
and reading the current status in IFLA_PORT_RESPONSE.
- recycling a previous patch that adds functionality to interface.c to
- get the vlan identifier on an interface
- get the flags of an interface and some convenience function to
check whether an interface is 'up' or not (not currently used here)
- adding function to determine the root physical interface of an
interface. For example if a macvtap is linked to eth0.100, it will
find eth0. Also adding a function that finds the vlan on the 'way to
the root physical interface'
- conveying the root physical interface name and index in case of 802.1Qbg
- conveying mac address of macvlan device and vlan identifier in
IFLA_VFINFO_LIST[ IFLA_VF_INFO[ IFLA_VF_MAC(mac), IFLA_VF_VLAN(vlan) ] ]
to (future) lldpad via netlink
- To enable build with --without-macvtap rename the
[dis|]associatePortProfileId functions, prepend 'vp' before their
name and make them non-static functions.
- Renaming variable multicast to nltarget_kernel and inverting
the logic
- Addressing Jim Meyering's comments; this also touches existing
code for example for correcting indentation of break statements or
simplification of switch statements.
- Renamed occurrencvirVirtualPortProfileDef to virVirtualPortProfileParamses
- 802.1Qbg part prepared for sending a RTM_SETLINK and getting
processing status back plus a subsequent RTM_GETLINK to
get IFLA_PORT_RESPONSE.
Note: This interface for 802.1Qbg may still change
- [David Allan] move getPhysfn inside IFLA_VF_PORT_MAX to avoid
compiler
warning when latest if_link.h isn't available
- move from Stefan's 802.1Qb{g|h} XML v8 to v9
- move hostuuid and vf index calcs to inside doPortProfileOp8021Qbh
- remove debug fprintfs
- use virGetHostUUID (thanks Stefan!)
- fix compile issue when latest if_link.h isn't available
- change poll timeout to 10s, at 1/8 intervals
- if polling times out, log msg and return -ETIMEDOUT
- Add Stefan's code for getPortProfileStatus
- Poll for up to 2 secs for port-profile status, at 1/8 sec intervals:
- if status indicates error, abort openMacvtapTap
- if status indicates success, exit polling
- if status is "in-progress" after 2 secs of polling, exit
polling loop silently, without error
My patch finishes out the 802.1Qbh parts, which Stefan had mostly complete.
I've tested using the recent kernel updates for VF_PORT netlink msgs and
enic for Cisco's 10G Ethernet NIC. I tested many VMs, each with several
direct interfaces, each configured with a port-profile per the XML. VM-to-VM,
and VM-to-external work as expected. VM-to-VM on same host (using same NIC)
works same as VM-to-VM where VMs are on diff hosts. I'm able to change
settings on the port-profile while the VM is running to change the virtual
port behaviour. For example, adjusting a QoS setting like rate limit. All
VMs with interfaces using that port-profile immediatly see the effect of the
change to the port-profile.
I don't have a SR-IOV device to test so source dev is a non-SR-IOV device,
but most of the code paths include support for specifing the source dev and
VF index. We'll need to complete this by discovering the PF given the VF
linkdev. Once we have the PF, we'll also have the VF index. All this info-
mation is available from sysfs.
Since the macvtap device needs active tear-down and the teardown logic
is based on the interface name, it can happen that if for example 1 out
of 3 interfaces was successfully created, that during the failure path
the macvtap's target device name is used to tear down an interface that
is doesn't own (owned by another VM).
So, in this patch, the target interface name is reset so that there is
no target interface name and the interface name is always cleared after
a tear down.
The hotplug methods still had the qemuCmdFlags variable declared
as an int, instead of unsigned long long. This caused flag checks
to be incorrect for flags > 31
* src/qemu/qemu_driver.c: Fix integer overflow in hotplug
This allows libvirt to open the PCI device sysfs config file prior
to dropping privileges so qemu can access the full config space.
Without this, a de-privileged qemu can only access the first 64
bytes of config space.
* src/qemu/qemu_conf.c, src/qemu/qemu_conf.h: Detect support
for pci-assign.configfd option. Use this option when formatting
PCI device string if possible
* src/qemu/qemu_driver.c: Pre-open PCI sysfs config file and pass
to QEMU
We've been running into a lot of situations where
virGetHostname() is returning "localhost", where a plain
gethostname() would have returned the correct thing. This
is because virGetHostname() is *always* trying to canonicalize
the name returned from gethostname(), even when it doesn't
have to.
This patch changes virGetHostname so that if the value returned
from gethostname() is already FQDN or localhost, it returns
that string directly. If the value returned from gethostname()
is a shortened hostname, then we try to canonicalize it. If
that succeeds, we returned the canonicalized hostname. If
that fails, and/or returns "localhost", then we just return
the original string we got from gethostname() and hope for
the best.
Note that after this patch it is up to clients to check whether
"localhost" is an allowed return value. The only place
where it's currently not is in qemu migration.
Signed-off-by: Chris Lalancette <clalance@redhat.com>
This patch parses the following two XML descriptions, one for
802.1Qbg and one for 802.1Qbh, and stores the data internally.
The actual triggering of the switch setup protocol has not been
implemented here but the relevant code to do that should go into
the functions associatePortProfileId() and disassociatePortProfileId().
<interface type='direct'>
<source dev='eth0.100' mode='vepa'/>
<model type='virtio'/>
<virtualport type='802.1Qbg'>
<parameters managerid='12' typeid='0x123456' typeidversion='1'
instanceid='fa9b7fff-b0a0-4893-8e0e-beef4ff18f8f'/>
</virtualport>
<filterref filter='clean-traffic'/>
</interface>
<interface type='direct'>
<source dev='eth0.100' mode='vepa'/>
<model type='virtio'/>
<virtualport type='802.1Qbh'>
<parameters profileid='my_profile'/>
</virtualport>
</interface>
I'd suggest to use this patch as a base for triggering the setup
protocol with the 802.1Qb{g|h} switch.
Several rounds of changes were made to this patch. The
following is a list of these changes.
- Renamed structure virVirtualPortProfileDef to virVirtualPortProfileParams
as per Daniel Berrange's request
- Addressing Daniel Berrange's comments:
- removing macvtap.h's dependency on domain_conf.h by
moving the virVirtualPortProfileDef structure into macvtap.h
and not passing virtDomainNetDefPtr to any functions in
macvtap.c
- Addressed most of Chris Wright's comments:
- indicating error in case virtualport XML node cannot be parsed
properly
- parsing hex and decimal numbers using virStrToLong_ui() with
parameter '0' for base
- tgifname (target interface name) variable wasn't necessary
to pass to openMacvtapTap function anymore
- assigning the virtual port data structure to the virDomainNetDef
only if it was previously parsed
- make sure that the error code returned by openMacvtapTap() is a negative n
in case the associatePortProfileId() function failed.
- renaming vsi in the XML to virtualport
- replace all occurrences of vsi in the source as well
- removing mode and MAC address parameters from the functions that
will communicate with the hareware diretctly or indirectly
- moving the associate and disassociate functions to the end of the
file for subsequent patches to easier make them generally available
for export
- passing the macvtap interface name rather than the link device since
this otherwise gives funny side effects when using netlink messages
where IFLA_IFNAME and IFLA_ADDRESS are specified and the link dev
all of a sudden gets the MAC address of the macvtap interface.
- Removing rc = -1 error indications in the case of 802.1Qbg|h setup in case
we wanted to use hook scripts for the setup and so the setup doesn't fail
here.
- if instance ID UUID is not supplied it will automatically be generated
- adapted schema to make instance ID UUID optional
- added test case
- parser and XML generator have been separated into their own
functions so they can be re-used elsewhere (passthrough case
for example)
- Adapted XML parser and generator support the above shown type
(802.1Qbg, 802.1Qbh).
- Adapted schema to above XML
- Adapted test XML to above XML
- Passing through the VM's UUID which seems to be necessary for
802.1Qbh -- sorry no host UUID
- adding virtual function ID to association function, in case it's
necessary to use (for SR-IOV)
Allow for a host UUID in the capabilities XML. Local drivers
will initialize this from the SMBIOS data. If a sanity check
shows SMBIOS uuid is invalid, allow an override from the
libvirtd.conf configuration file
* daemon/libvirtd.c, daemon/libvirtd.conf: Support a host_uuid
configuration option
* docs/schemas/capability.rng: Add optional host uuid field
* src/conf/capabilities.c, src/conf/capabilities.h: Include
host UUID in XML
* src/libvirt_private.syms: Export new uuid.h functions
* src/lxc/lxc_conf.c, src/qemu/qemu_driver.c,
src/uml/uml_conf.c: Set host UUID in capabilities
* src/util/uuid.c, src/util/uuid.h: Support for host UUIDs
* src/node_device/node_device_udev.c: Use the host UUID functions
* tests/confdata/libvirtd.conf, tests/confdata/libvirtd.out: Add
new host_uuid config option to test
The cgroups ACL code was only allowing the primary disk image.
It is possible to chain images together, so we need to search
for backing stores and add them to the ACL too. Since the ACL
only handles block devices, we ignore the EINVAL we get from
plain files. In addition it was missing code to teardown the
cgroup when hot-unplugging a disk
* src/qemu/qemu_driver.c: Allow backing stores in cgroup ACLs
and add missing teardown code in unplug path
Basic live migration was broken by the commit that added
non-shared block support in two ways:
1) It added a virCheckFlags() to doNativeMigrate(). Besides
the fact that typical usage of virCheckFlags() is in driver
entry points, and doNativeMigrate() is not an entry point,
it was missing important flags like VIR_MIGRATE_LIVE. Move
the virCheckFlags to the top-level qemuDomainMigratePrepare2
and friends.
2) It also added a memory leak in qemuMonitorTextMigrate()
by not freeing the memory used by virBufferContentAndReset().
This is fixed by storing the pointer in a temporary variable
and freeing it at the end.
With this patch in place, normal live migration works again.
v3: Instead of the churn for virCheckFlagsUI and UL, instead
always promote flags to an unsigned long and always use %lx
for the fprintf.
v2: Add back flags check, which required adding virCheckFlagsUI
and virCheckFlagsUL
Signed-off-by: Chris Lalancette <clalance@redhat.com>
The device path doesn't make use of guestAddr, so the memcpy corrupts
the guest info struct.
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
The virDomainGetBlockInfo API allows query physical block
extent and allocated block extent. These are normally the
same value unless storing a special format like qcow2
inside a block device. In this scenario we can query QEMU
to get the actual allocated extent.
Since last time:
- Return fatal error in text monitor
- Only invoke monitor command for block devices
- Fix error handling JSON code
* src/qemu/qemu_driver.c: Fill in block aloction extent when VM
is running
* src/qemu/qemu_monitor.c, src/qemu/qemu_monitor.h,
src/qemu/qemu_monitor_json.c, src/qemu/qemu_monitor_json.h,
src/qemu/qemu_monitor_text.c, src/qemu/qemu_monitor_text.h: Add
API to query the highest block extent via info blockstats
The qemu driver contains a subtle race in the logic to find next
available vnc port. Currently it iterates through all available ports
and returns the first for which bind(2) succeeds. However it is possible
that a previously issued port has not yet been bound by qemu, resulting
in the same port used for a subsequent domain.
This patch addresses the race by using a simple bitmap to "reserve" the
ports allocated by libvirt.
V2:
- Put port bitmap in struct qemud_driver
- Initialize bitmap in qemudStartup
V3:
- Check for failure of virBitmapGetBit
- Additional check for port != -1 before calling virbitmapClearBit
V4:
- Check for failure of virBitmap{Set,Clear}Bit
We need a common internal function for starting managed domains to be
used during autostart. This patch factors out relevant code from
qemudDomainStart into qemudDomainObjStart and makes it use the
refactored code for domain restore instead of calling qemudDomainRestore
API directly.
We need to be able to restore a domain which we already locked and
started a job for it without undoing these steps. This patch factors
out internals of qemudDomainRestore into separate functions which work
for locked objects.
The current cleanup: in StartVMDaemon path is a poor duplication.
qemuShutdownVMDaemon can handle teardown for inactive VMs, so let's use it.
v2: Remove old abort: label, only use cleanup:
There doesn't seem to be anything specific to tap devices for this
array of file descriptors which need to stay open of the guest to use.
Rename then for others to make use of.
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
Signed-off-by: Chris Lalancette <clalance@redhat.com>
Do not require each caller of virStorageFileGetMetadata and
virStorageFileGetMetadataFromFD to first clear the storage of the
"meta" buffer. Instead, initialize that storage in
virStorageFileGetMetadataFromFD.
* src/util/storage_file.c (virStorageFileGetMetadataFromFD): Clear
"meta" here, not before each of the following callers.
* src/qemu/qemu_driver.c (qemuSetupDiskCgroup): Don't clear "meta" here.
(qemuTeardownDiskCgroup): Likewise.
* src/qemu/qemu_security_dac.c (qemuSecurityDACSetSecurityImageLabel):
Likewise.
* src/security/security_selinux.c (SELinuxSetSecurityImageLabel):
Likewise.
* src/security/virt-aa-helper.c (get_files): Likewise.
Approximately 60 messages were marked. Since these diagnostics are
intended solely for developers and maintainers, encouraging translation
is deemed to be counterproductive:
http://thread.gmane.org/gmane.comp.emulators.libvirt/25050/focus=25052
Run this command:
git grep -l VIR_WARN|xargs perl -pi -e \
's/(VIR_WARN0?)\s*\(_\((".*?")\)/$1($2/'
There were three very similar uses of qemuMonitorAddDrive.
This change makes the three 17-line sequences identical.
* src/qemu/qemu_driver.c (qemudDomainAttachPciDiskDevice): Detect
failure. Add VIR_WARN and braces.
(qemudDomainAttachSCSIDisk): Add VIR_WARN and braces.
(qemudDomainAttachUsbMassstorageDevice): Likewise.
Any method which intends to invoke a monitor command must have
a check for virDomainObjIsActive() before using the monitor to
ensure that priv->mon != NULL.
There is one subtle edge case in this though. If a method invokes
multiple monitor commands, and calls qemuDomainObjExitMonitor()
in between two of these commands then there is no guarentee that
priv->mon != NULL anymore. This is because the QEMU process may
exit or die at any time, and because qemuDomainObjEnterMonitor()
releases the lock on virDomainObj, it is possible for the background
thread to close the monitor handle and thus qemuDomainObjExitMonitor
will release the last reference allowing priv->mon to become NULL.
This affects several methods, most notably migration but also some
hotplug methods. This patch takes a variety of approaches to solve
the problem, depending on the particular usage scenario. Generally
though it suffices to add an extra virDomainObjIsActive() check
if qemuDomainObjExitMonitor() was called during the method.
* src/qemu/qemu_driver.c: Fix multiple potential NULL pointer flaws
in usage of the monitor
* src/qemu/qemu_driver.c (qemudDomainSetVcpus): Upon look-up failure,
i.e., vm==NULL, goto cleanup, rather than to "endjob", superficially
since the latter would dereference vm, but more fundamentally because
we certainly don't want to call qemuDomainObjEndJob before we've
even attempted qemuDomainObjBeginJob.
A fedora translator filed:
https://bugzilla.redhat.com/show_bug.cgi?id=580816
Pointing out these two error messages as unclear: "write save" sounds
like a typo without context, and lack of a colon made the second message
difficult to parse.
qemuReadLogOutput early VM death detection is racy and won't always work.
Startup then errors when connecting to the VM monitor. This won't report
the emulator cmdline output which is typically the most useful diagnostic.
Check if the VM has died at the very end of the monitor connection step,
and if so, report the cmdline output.
See also: https://bugzilla.redhat.com/show_bug.cgi?id=581381
* src/qemu/qemu_driver.c (qemudDomainSetVcpus): Avoid NULL-deref
upon unknown UUID. Call qemuDomainObjBeginJob(vm) only after
ensuring that vm != NULL, not before. This potential NULL-deref
was introduced by commit 2c555d87b0.
The code specifies driver->cacheDir as the format string,
but it usually doesn't contain '%s', so the subsequent
argument, "/qemu.mem.XXXXXX", is always ignored.
The patch fixes the misuse.
When QEMU runs with its disk on NFS, and as a non-root user, the
disk is chownd to that non-root user. When migration completes
the last step is shutting down the QEMU on the source host. THis
normally resets user/group/security label. This is bad when the
VM was just migrated because the file is still in use on the dest
host. It is thus neccessary to skip the reset step for any files
found to be on a shared filesystem
* src/libvirt_private.syms: Export virStorageFileIsSharedFS
* src/util/storage_file.c, src/util/storage_file.h: Add a new
method virStorageFileIsSharedFS() to determine if a file is
on a shared filesystem (NFS, GFS, OCFS2, etc)
* src/qemu/qemu_driver.c: Tell security driver not to reset
disk labels on migration completion
* src/qemu/qemu_security_dac.c, src/qemu/qemu_security_stacked.c,
src/security/security_selinux.c, src/security/security_driver.h,
src/security/security_apparmor.c: Add ability to skip disk
restore step for files on shared filesystems.
The cgroups ACL code was only allowing the primary disk image.
It is possible to chain images together, so we need to search
for backing stores and add them to the ACL too. Since the ACL
only handles block devices, we ignore the EINVAL we get from
plain files. In addition it was missing code to teardown the
cgroup when hot-unplugging a disk
* src/qemu/qemu_driver.c: Allow backing stores in cgroup ACLs
and add missing teardown code in unplug path
QEMU is gaining a new monitor command netdev_add for hotplugging
NICs using the netdev backend code. We already support this on
the command this, though it is disabled. This adds support for
hotplug too, also to remain disabled until 0.13 QEMU is released
* src/qemu/qemu_driver.c: Support netdev hotplug for NICs
* src/qemu/qemu_monitor.c, src/qemu/qemu_monitor.h,
src/qemu/qemu_monitor_json.c, src/qemu/qemu_monitor_json.h,
src/qemu/qemu_monitor_text.c, src/qemu/qemu_monitor_text.h: Add
support for netdev_add and netdev_remove commands
With the introduction of the generic qemu device model, unplugging
SCSI disks works like a charm, so support it in libvirt.
* src/qemu/qemu_driver.c: Add qemudDomainDetachSCSIDiskDevice() to do the
unplugging, extend qemudDomainDetachDeviceAdd().
Signed-off-by: Wolfgang Mauerer <wolfgang.mauerer@siemens.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
Support for live migration between hosts that do not share storage was
added to qemu-kvm release 0.12.1.
It supports two flags:
-b migration without shared storage with full disk copy
-i migration without shared storage with incremental copy (same base image
shared between source and destination).
I tested the live migration without shared storage (both flags) for native
and p2p with and without tunnelling. I also verified that the fix doesn't
affect normal migration with shared storage.
WIN32 is always defined when __MINGW32__ is defined, but the
converse is not true. WIN32 is more generic, if someone were
to ever attempt porting to a microsoft compiler. This does
not affect Cygwin, which intentionally does not define WIN32.
* src/qemu/qemu_driver.c (qemuDomainGetBlockInfo): Use more
generic flag macro.
* src/storage/storage_backend.c
(virStorageBackendUpdateVolTargetInfoFD)
(virStorageBackendRunProgRegex): Likewise.
* tools/console.h (vshRunConsole): Likewise.
This introduces a new event type
VIR_DOMAIN_EVENT_ID_IO_ERROR_REASON
This event is the same as the previous VIR_DOMAIN_ID_IO_ERROR
event, but also includes a string describing the cause of
the event.
Thus there is a new callback definition for this event type
typedef void (*virConnectDomainEventIOErrorReasonCallback)(virConnectPtr conn,
virDomainPtr dom,
const char *srcPath,
const char *devAlias,
int action,
const char *reason,
void *opaque);
This is currently wired up to the QEMU block IO error events
* daemon/remote.c: Dispatch IO error events to client
* examples/domain-events/events-c/event-test.c: Watch for
IO error events
* include/libvirt/libvirt.h.in: Define new IO error event ID
and callback signature
* src/conf/domain_event.c, src/conf/domain_event.h,
src/libvirt_private.syms: Extend API to handle IO error events
* src/qemu/qemu_driver.c: Connect to the QEMU monitor event
for block IO errors and emit a libvirt IO error event
* src/remote/remote_driver.c: Receive and dispatch IO error
events to application
* src/remote/remote_protocol.x: Wire protocol definition for
IO error events
* src/qemu/qemu_monitor.c, src/qemu/qemu_monitor.h,
src/qemu/qemu_monitor_json.c: Watch for BLOCK_IO_ERROR event
from QEMU monitor
When using -device syntax, the IO event will have a different
prefix, 'drive-' that needs to be skipped over before matching
against the libvirt disk alias
* src/qemu/qemu_driver.c: Skip QEMU_DRIVE_HOST_PREFIX in IO event
This defines the internal driver API and stubs out each driver
* src/driver.h: Define virDrvDomainGetBlockInfo signature
* src/libvirt.c, src/libvirt_public.syms: Glue public API to drivers
* src/esx/esx_driver.c, src/lxc/lxc_driver.c, src/opennebula/one_driver.c,
src/openvz/openvz_driver.c, src/phyp/phyp_driver.c,
src/test/test_driver.c, src/uml/uml_driver.c, src/vbox/vbox_tmpl.c,
src/xen/xen_driver.c, src/xenapi/xenapi_driver.c: Stub out driver
The previous commit changes a goto from 'endjob' to 'cleanup',
leaving the endjob label unused. Remove it to avoid compile
warning.
* src/qemu/qemu_driver.c: Remove 'endjob' label
* src/qemu/qemu_driver.c (qemuDomainSnapshotCreateXML): When setting
"vm" to NULL, jump over vm-dereferencing code to "cleanup".
(qemuDomainRevertToSnapshot): Likewise.
In cases where the security driver failed to restore a label after a
guest has saved, we mistakenly jumped to the error cleanup paths.
This is not good, because the operation has in fact completed and
cannot be rolled back completely. Label restore is non-critical, so
just log the problem instead. Also add a missing restore call in
the error cleanup path
* src/qemu/qemu_driver.c: Fix handling of security driver
restore failures in QEMU domain save
When cgroups is enabled, access to block devices is likely to be
restricted to a whitelist. Prior to saving a guest to a block device,
it is necessary to add the block device to the whitelist. This is
not required upon restore, since QEMU reads from stdin
* src/qemu/qemu_driver.c: Add block device to cgroups whitelist
if neccessary during domain save.
The save process was relying on use of the shell >> append
operator to ensure the save data was placed after the libvirt
header + XML. This doesn't work for block devices though.
Replace this code with use of 'dd' and its 'seek' parameter.
This means that we need to pad the header + XML out to a
multiple of dd block size (in this case we choose 512).
The qemuMonitorMigateToCommand() monitor API is used for both
save/coredump, and migration via UNIX socket. We can't simply
switch this to use 'dd' since this causes problems with the
migration usage. Thus, create a dedicated qemuMonitorMigateToFile
which can accept an filename + offset, and remove the filename
from the current qemuMonitorMigateToCommand() API
* src/qemu/qemu_driver.c: Switch to qemuMonitorMigateToFile
for save and core dump
* src/qemu/qemu_monitor.c, src/qemu/qemu_monitor.h,
src/qemu/qemu_monitor_json.c, src/qemu/qemu_monitor_json.h,
src/qemu/qemu_monitor_text.c, src/qemu/qemu_monitor_text.h: Create
a new qemuMonitorMigateToFile, separate from the existing
qemuMonitorMigateToCommand to allow handling file offsets
It is possible to use block devices with domain save/restore. Upon
failure QEMU unlinks the path being saved to. This isn't good when
it is a block device !
* src/qemu/qemu_driver.c: Don't unlink block devices if save fails
If a transient QEMU crashes during save attempt, then the virDomainPtr
object may be freed. If a persistent QEMU crashes during save, then
the 'priv->mon' field is no longer valid since it will be inactive.
* src/qemu/qemu_driver.c: Fix two crashes when QEMU exits
during a save attempt
In particular I was forgetting to take the qemuMonitorPrivatePtr
lock (via qemuDomainObjBeginJob), which would cause problems
if two users tried to access the same domain at the same time.
This patch also fixes a problem where I was forgetting to remove
a transient domain from the list of domains.
Thanks to Stephen Shaw for pointing out the problem and testing
out the initial patch.
Signed-off-by: Chris Lalancette <clalance@redhat.com>
With JSON qemu monitor, we get a STOP event from qemu whenever qemu
stops guests CPUs. The downside of it is that vm->state is changed to
PAUSED and a new generic paused event is send to applications. However,
when we ask qemu to stop the CPUs we are not really interested in qemu
event and we usually want to issue a more specific event.
By setting vm->status to PAUSED before actually sending the request to
qemu (and resetting it back if the request fails) we can ignore the
event since the event handler does nothing when the guest is already
paused. This solution is quite hacky but unfortunately it's the best
solution which I was able to come up with and it doesn't introduce a
race condition.
While doing some testing of the snapshot code I noticed that
if qemuDomainSnapshotLoad failed, it would print a NULL as
part of the error. That's not desirable, so leave the
full_path variable around until after we are done printing
errors.
Signed-off-by: Chris Lalancette <clalance@redhat.com>
The hang fix in d376b7d63e was incomplete
since it left quite a few {Enter,Exit}Monitor calls which require driver
to be unlocked. Since the driver is locked throughout the whole
function, {Enter,Exit}MonitorWithDriver need to be used instead to
ensure driver is not locked when issuing monitor commands.
The comment in qemuDomainWaitForMigrationComplete says we are polling
every 50ms but the code sleeps only for 50us. This was already discussed
during review but apparently forgotten when the series was pushed.
First, inital VCPU pinning is set correctly but then it is reset by
assigning qemu process to a new cgroup (which contains all CPUs). It's
easily fixed by swapping these two actions.
The initial boot of VMs uses -device for NICs where available. The
corresponding monitor command is device_add, but the network hotplug
code was still using device_del by mistake.
* src/qemu/qemu_driver.c: Use device_add for NIC hotplug where
available
Disk devices in QEMU have two parts, the guest device and the host
backend driver. Historically these two parts have had the same
"unique" name. With the switch to using -device though, they now
have separate names. Thus when changing CDROM media, for guests
using -device syntax, we need to prepend the QEMU_DRIVE_HOST_PREFIX
constant
* src/qemu/qemu_conf.c, src/qemu/qemu_conf.h: Add helper function
qemuDeviceDriveHostAlias() for building a host backend alias
* src/qemu/qemu_driver.c: Use qemuDeviceDriveHostAlias() to determine
the host backend alias for performing eject/change commands in the
monitor
