cpu: add 'amd-ssbd' and 'amd-no-ssb' CPU features (CVE-2018-3639)
AMD x86 CPUs have two separate ways to mitigate the Speculative Store
Bypass hardware flaw. In current processors only non-architectural MSRs
are available, and so hypervisors must expose a virtualized MSR and CPU
flag "virt-ssbd" (CPUID Function 8000_0008, EBX[25]=1).
In future processors AMD will provide an architectural MSR, indicated by
existance of the CPUID Function 8000_0008, EBX[24]=1, to which QEMU has
given the name "amd-ssbd".
The "amd-ssbd" flag should be used in preference to "virt-ssbd", if it
is available, since it provides improved performance. For virtual
machine configuration, both should be exposed when available, to allow
for maximal guest OS compatibility as not all guests yet support both.
If future processes are not vulnerable to the flaw, this will be
indicated by the existance of CPUID Function 8000_0008, EBX[26]=1,
to which QEMU has given the name "amd-no-ssb".
See also 124441_AMD64_SpeculativeStoreBypassDisable_Whitepaper_final.pdf
from:
Note that neither amd-ssbd or amd-no-ssb will be reported by the kernel
in /proc/cpuinfo. It knows about these CPUID bits and does the right thing,
but doesn't report their existance as distinct flags in /proc/cpuinfo.
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
qemu: format serial and geometry on frontend disk device
Currently we format the serial, geometry and error policy on the -drive
backend argument.
QEMU added the ability to set serial and geometry on the frontend in
the 1.2 release deprecating use of -drive, with support being deleted
from -drive in 3.0.
We keep formatting error policy on -drive for now, because we don't
ahve support for that with -device for usb-storage just yet.
Note that some disk buses (sd) still don't support -device. Although
QEMU allowed these properties to be set on -drive for if=sd, they
have been ignored so we now report an error in this case.
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Peter Krempa [Tue, 3 Jul 2018 08:45:34 +0000 (10:45 +0200)]
qemu: hotplug: Don't access srcPriv when it's not allocated
The private data of a virStorageSource which is backing an iSCSI hostdev
may be NULL if no authentication is present. The code handling the
hotplug would attempt to extract the authentication info stored in
'secinfo' without checking if it is allocated which resulted in a crash.
Here we opt the easy way to check if srcPriv is not NULL so that we
don't duplicate all the logic which selects whether the disk source has
a secret.
qemu: Remove unused bypassSecurityDriver from qemuOpenFileAs
This argument is not used anymore. The only function that is
passing non-NULL (qemuDomainSaveMemory) does not actually care
for the value (after 23087cfdb) and every other caller just
passes NULL anyway.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: John Ferlan <jferlan@redhat.com>
esx_vi.c: Simplify error handling in MachineByName
The same pattern is used in lots of other places.
Also, reporting error message is not desired because all callers
check the return value and report errors on their own.
Signed-off-by: Marcos Paulo de Souza <marcos.souza.org@gmail.com> Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
qemu: don't use chardev FD passing with standalone args
When using domxml-to-native, we must generate CLI args that can be used
in a standalone scenario. This means no FD passing can be used. To
achieve this we must clear the QEMU_CAPS_CHARDEV_FD_PASS capability bit.
Reviewed-by: John Ferlan <jferlan@redhat.com> Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Peter Krempa [Mon, 18 Jun 2018 08:22:18 +0000 (10:22 +0200)]
tests: qemumonitorjson: Add only required replies for blockstats test
testQemuMonitorJSONqemuMonitorJSONGetBlockStatsInfo added 4 replies but
only one was used. Additionally the comment stated that 7 replies are
going to be added.
Peter Krempa [Wed, 20 Jun 2018 08:39:21 +0000 (10:39 +0200)]
qemu: domain: update only newly detected images in qemuDomainDetermineDiskChain
The processing code which prepares images should be executed really only
for the images which were detected. The code actually tried to update
the last user-specified layer as well. Thankfully we don't do anything
that would be a problem at this point.
The @prefix attribute to <ip/> element for interface type user is
optional. Therefore, if left out it has value of zero in which
case we should not check whether it falls into <4, 27> range.
Otherwise we fail parsing domain XML for no good reason.
Julio Faracco [Wed, 27 Jun 2018 15:06:26 +0000 (12:06 -0300)]
util: moving 'type' argument to avoid issues with mount() syscall.
This commit fixes a mount call inside virgroup.c file. The NULL value
into 'type' argument is causing a valgrind issue. See commit 794b576c
for more details. The best approach to fix it is moving NULL to "none"
filesytem.
Signed-off-by: Julio Faracco <jcfaracco@gmail.com>
Julio Faracco [Wed, 27 Jun 2018 15:06:25 +0000 (12:06 -0300)]
lxc: moving 'type' argument to avoid issues with mount() syscall.
This commit fixes a lots of mount calls inside lxc_container.c file. The
NULL value into 'type' argument is causing a valgrind issue. See commit 794b576c2b for more details. The best approach to fix it is moving NULL
to "none" filesytem.
Signed-off-by: Julio Faracco <jcfaracco@gmail.com>
Laine Stump [Sun, 1 Jul 2018 23:29:03 +0000 (19:29 -0400)]
network: properly check for taps that are connected to an OVS bridge
When libvirtd is restarted, it checks that each guest tap device is
still attached to the bridge device that the configuration info says
it should be connected to. If not, the tap will be disconnected from
[wherever it is] and connected to [wherever it should be].
The previous code that did this did not account for:
1) the IFLA_MASTER attribute in a netdev's ifinfo will be set to
"ovs-system" for any tap device connected to an OVS bridge, *not*
to the name of the bridge it is attached to.
2) virNetDevRemovePort() only works for devices that are attached to a
standard Linux host bridge. If a device is currently attached to an
OVS bridge, then virNetDevOpenvswitchRemovePort() must be called
instead.
This patch remedies those problems, and adds a couple of information
log messages to aid in debugging any future problem.
Laine Stump [Sun, 1 Jul 2018 23:24:19 +0000 (19:24 -0400)]
util: new function virNetDevOpenvswitchInterfaceGetMaster()
This function retrieves the name of the OVS bridge that the given
netdev is attached to. This separate function is necessary because OVS
set the IFLA_MASTER attribute to "ovs-system" for all netdevs that are
attached to an OVS bridge, so the standard method of retrieving the
master can't be used.
Signed-off-by: Laine Stump <laine@laine.org> ACKed-by: Michal Privoznik <mprivozn@redhat.com>
John Ferlan [Tue, 26 Jun 2018 14:14:20 +0000 (10:14 -0400)]
lxc: Rearrange order in lxcDomainUpdateDeviceFlags
Although commit e3497f3f noted that the LIVE option doesn't
matter and removed the call to virDomainDefCompatibleDevice,
it didn't go quite far enough and change the order of the checks
and rework the code to just handle the config change causing
a failure after virDomainObjUpdateModificationImpact updates
the @flags. Since we only support config a lot of previously
conditional code is now just inlined.
Signed-off-by: John Ferlan <jferlan@redhat.com> ACKed-by: Michal Prívozník <mprivozn@redhat.com>
Jiri Denemark [Thu, 28 Jun 2018 09:38:52 +0000 (11:38 +0200)]
qemu_migration: Check for active domain after talking to remote daemon
Once we called qemuDomainObjEnterRemote to talk to the destination
daemon during a peer to peer migration, the vm lock is released and we
only hold an async job. If the source domain dies at this point the
monitor EOF callback is allowed to do its job and (among other things)
clear all private data irrelevant for stopped domain. Thus when we call
qemuDomainObjExitRemote, the domain may already be gone and we should
avoid touching runtime private data (such as current job info).
In other words after acquiring the lock in qemuDomainObjExitRemote, we
need to check the domain is still alive. Unless we're doing offline
migration.
Jiri Denemark [Thu, 28 Jun 2018 12:09:47 +0000 (14:09 +0200)]
qemu_migration: Rename 'offline' variable in SrcPerformPeer2Peer
The variable is used to store the offline migration capability of the
destination daemon. Let's call it 'dstOffline' so that we can later use
'offline' to indicate whether we were asked to do offline migration.
Julio Faracco [Wed, 27 Jun 2018 02:47:53 +0000 (23:47 -0300)]
nwfilter: variable 'obj' must be initialized inside nwfilterBindingCreateXML().
The function nwfilterBindingCreateXML() is failing to compile due to a
conditional branch which leads to an undefined 'obj' variable. So 'obj'
must have an initial value to avoid compilation errors. See the problem:
CC nwfilter/libvirt_driver_nwfilter_impl_la-nwfilter_driver.lo
nwfilter/nwfilter_driver.c:752:9: error: variable 'obj' is used uninitialized whenever 'if' condition is true [-Werror,-Wsometimes-uninitialized]
if (virNWFilterBindingCreateXMLEnsureACL(conn, def) < 0)
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
nwfilter/nwfilter_driver.c:779:10: note: uninitialized use occurs here
if (!obj)
^~~
nwfilter/nwfilter_driver.c:752:5: note: remove the 'if' if its condition is always false
if (virNWFilterBindingCreateXMLEnsureACL(conn, def) < 0)
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
nwfilter/nwfilter_driver.c:742:33: note: initialize the variable 'obj' to silence this warning
virNWFilterBindingObjPtr obj;
^
= NULL
This commit initialized 'obj' with NULL to fix the error properly.
Signed-off-by: Julio Faracco <jcfaracco@gmail.com> Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
When updating a live device users might pass different alias than
the one the device has. Currently, this is silently ignored which
goes against our behaviour for other parts of the device where we
explicitly allow only certain changes and error out loudly on
anything else.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: John Ferlan <jferlan@redhat.com>
Michal Privoznik [Tue, 26 Jun 2018 08:37:27 +0000 (10:37 +0200)]
conf: Reintroduce action to virDomainDefCompatibleDevice
This was lost in c57f3fd2f8999d17e01. But now we are going to
need it again (except the DETACH action where checking for device
compatibility does not make much sense anyway).
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: John Ferlan <jferlan@redhat.com>
John Ferlan [Wed, 20 Jun 2018 22:37:15 +0000 (18:37 -0400)]
docs: Add news article for volume encryption modifications
Include both the domain and storage modifications in a "Removed
features" section as well as describing the improvement to allow
using a raw input volume to create the luks encrypted volume.
Signed-off-by: John Ferlan <jferlan@redhat.com> ACKed-by: Michal Privoznik <mprivozn@redhat.com>
John Ferlan [Wed, 20 Jun 2018 19:51:47 +0000 (15:51 -0400)]
storage: Add support for using inputvol for encryption
Starting with QEMU 2.9, encryption convert processing requires
a multi-step process in order to generate an encrypted image from
some non encrypted raw image.
Processing requires to first create an encrypted image using the
sizing parameters from the input source and second to use the
--image-opts, -n, and --target-image-opts options along with inline
driver options to describe the input and output files, generating
two commands such as:
This patch handles the convert processing by running the processing
in a do..while loop essentially reusing the existing create logic and
arguments to create the target vol from the inputvol and then converting
the inputvol using new arguments.
This then allows the following virsh command to work properly:
John Ferlan [Tue, 19 Jun 2018 15:56:21 +0000 (11:56 -0400)]
storage: Remove storageBackendGenerateSecretData
Since we no longer support creating qcow2 encryption format
volumes, we no longer have to possibly create some secret and
have no real need for the function, so move the remaining
functionality to build the secret path back into the caller
storageBackendCreateQemuImg.
Signed-off-by: John Ferlan <jferlan@redhat.com> ACKed-by: Michal Privoznik <mprivozn@redhat.com>
Since commit c4eedd793 disallowed qcow2 encrypted images to be
used for domains, it no longer makes sense to allow a qcow2
encrypted volume to be created or resized.
Add a test that will exhibit the failure of creation as well
as the xml2xml validation of the format still being correct.
Update the documentation to note the removal of the capability
to create and use qcow/default encrypted volumes.
Signed-off-by: John Ferlan <jferlan@redhat.com> ACKed-by: Michal Privoznik <mprivozn@redhat.com>
John Ferlan [Wed, 20 Jun 2018 19:52:30 +0000 (15:52 -0400)]
tests: Remove qcow2 encryption from storagevol tests
We're about to disallow creation of a qcow2 encrypted storage
volume, so let's remove the qcow encryption element from the
tests which are testing whether other format='qcow2' related
features work properly.
Signed-off-by: John Ferlan <jferlan@redhat.com> ACKed-by: Michal Privoznik <mprivozn@redhat.com>
John Ferlan [Tue, 19 Jun 2018 16:05:31 +0000 (12:05 -0400)]
storage: Don't allow encryption secretPath to be NULL
Allowing a NULL @secretPath for virStorageBackendCreateQemuImgCmdFromVol
would result in a generated command line with a dangling "file=" output.
So let's make sure the @secretPath exists before processing.
This means we should pass a dummy path from the storage test.
Signed-off-by: John Ferlan <jferlan@redhat.com> ACKed-by: Michal Privoznik <mprivozn@redhat.com>
nwfilter: convert virt drivers to use public API for nwfilter bindings
Remove the callbacks that the nwfilter driver registers with the domain
object config layer. Instead make the current helper methods call into
the public API for creating/deleting nwfilter bindings.
Reviewed-by: John Ferlan <jferlan@redhat.com> Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
nwfilter: wire up new APIs for creating and deleting nwfilter bindings
This allows the virsh commands nwfilter-binding-create and
nwfilter-binding-delete to be used.
Note using these commands lets you delete filters that were
previously created automatically by the virt drivers, or add
filters for VM nics that were not there before. Generally it
is expected these new APIs will only be used by virt drivers.
It is the admin's responsibility to not shoot themselves in
the foot.
Reviewed-by: John Ferlan <jferlan@redhat.com> Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
nwfilter: remove virt driver callback layer for rebuilding filters
Now that the nwfilter driver keeps a list of bindings that it has
created, there is no need for the complex virt driver callbacks. It is
possible to simply iterate of the list of recorded filter bindings.
This means that rebuilding filters no longer has to acquire any locks on
the virDomainObj objects, as they're never touched.
Reviewed-by: John Ferlan <jferlan@redhat.com> Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Currently the nwfilter driver does not keep any record of what filter
bindings it has active. This means that when it needs to recreate
filters, it has to rely on triggering callbacks provided by the virt
drivers. This introduces a hash table recording the virNWFilterBinding
objects so the driver has a record of all active filters.
Reviewed-by: John Ferlan <jferlan@redhat.com> Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Anya Harter [Wed, 20 Jun 2018 13:17:00 +0000 (09:17 -0400)]
qemu: Escape commas for qemuBuildSCSIiSCSIHostdevDrvStr
Add comma escaping for netsource. This is done here because
qemuBuildNetworkDriveStr has other external callers which
may not expect an escaped comma; however, this particular
command building path needs to perform the escaping for the
hostdev command line, so we do it now to ensure src->path
and src->host->name are covered.
Michal Privoznik [Wed, 20 Jun 2018 12:17:45 +0000 (14:17 +0200)]
qemuDomainObjBeginJobInternal: Report agent job in error message
If a thread is unable to acquire a job (e.g. because of timeout)
an error is reported and the error message contains reference to
the other thread holding the job. Well, the error message should
report agent job too as it is yet another source of possible
failure.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: John Ferlan <jferlan@redhat.com>
Chen Hanxiao [Thu, 21 Jun 2018 11:28:55 +0000 (19:28 +0800)]
set-lifecycle-action: add description of type and action
In [1], <type> are described as "on_poweroff", "on_reboot",
"on_crash". but we accept "poweroff", "reboot" and "crash".
This patch adds documentation for them.
conf: report an error if nic needs filtering by no driver is present
If a <interface> includes a filter name but the nwfilter driver is not
present we silently do nothing. This is very bad, because an application
that thinks it is protected by malicious guest traffic will in fact be
vulnerable. Reporting an error gives the administrator the ability to
know there is a problem and fix it.
Reviewed-by: John Ferlan <jferlan@redhat.com> Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
nwfilter: export port binding concept in the public API
When the daemons are split there will need to be a way for the virt
drivers and/or network driver to create and delete bindings between
network ports and network filters. This defines a set of public APIs
that are suitable for managing this facility.
Reviewed-by: John Ferlan <jferlan@redhat.com> Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
conf: change virNWFilterBindingPtr to virNWFilterBindingDefPtr
We are going to want to expose the NWFilter binding concept in the
public API, so the virNWFilterBindingPtr type needs to be used there.
Our internal type will shortly gain an XML representation, so rename
it to virNWFilterBindingDefPtr which follows our normal conventions.
Reviewed-by: John Ferlan <jferlan@redhat.com> Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Andrea Bolognani [Thu, 21 Jun 2018 14:14:48 +0000 (16:14 +0200)]
spec: Move SASL configuration file from -libs to -daemon
SASL authentication is configured server-side, so the sample
configuration file should be shipped along with the daemon
rather than with the libraries.
Signed-off-by: Andrea Bolognani <abologna@redhat.com>
Julio Faracco [Tue, 26 Jun 2018 03:18:21 +0000 (00:18 -0300)]
util: fix mount issue by moving NULL value to "none" in syscall.
After running libvirt daemon with valgrind tools, some errors are
appearing when you try to start a domain. One example:
==18012== Syscall param mount(type) points to unaddressable byte(s)
==18012== at 0x6FEE3CA: mount (syscall-template.S:78)
==18012== by 0x531344D: virFileMoveMount (virfile.c:3828)
==18012== by 0x27FE7675: qemuDomainBuildNamespace (qemu_domain.c:11501)
==18012== by 0x2800C44E: qemuProcessHook (qemu_process.c:2870)
==18012== by 0x52F7E1D: virExec (vircommand.c:726)
==18012== by 0x52F7E1D: virCommandRunAsync (vircommand.c:2477)
==18012== by 0x52F4EDD: virCommandRun (vircommand.c:2309)
==18012== by 0x2800A731: qemuProcessLaunch (qemu_process.c:6235)
==18012== by 0x2800D6B4: qemuProcessStart (qemu_process.c:6569)
==18012== by 0x28074876: qemuDomainObjStart (qemu_driver.c:7314)
==18012== by 0x280522EB: qemuDomainCreateWithFlags (qemu_driver.c:7367)
==18012== by 0x55484BF: virDomainCreate (libvirt-domain.c:6531)
==18012== by 0x12CDBD: remoteDispatchDomainCreate (remote_daemon_dispatch_stubs.h:4350)
==18012== by 0x12CDBD: remoteDispatchDomainCreateHelper (remote_daemon_dispatch_stubs.h:4326)
==18012== Address 0x0 is not stack'd, malloc'd or (recently) free'd
Some documentation recommends to use "none" when you don't have a
filesystem type to use. Specially, for bind and move actions.
Signed-off-by: Julio Faracco <jcfaracco@gmail.com>
qemu: ensure FDs passed to QEMU for chardevs have correct SELinux labels
The UNIX socket FDs were we passing to QEMU inherited a label based on
libvirtd's context. QEMU is thus denied ability to access the UNIX
socket. We need to use the security manager to change our current
context temporarily when creating the UNIX socket FD.
Reviewed-by: Laine Stump <laine@laine.org> Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
docs: formatdomain: Mention that 'urandom' is the recommended RNG backend
Since libvirt 1.3.4, any RNG source is accepted for the 'random'
backend. However, '/dev/urandom' is the _recommended_ source of
entropy. Therefore we should mention that in the docs.
Suggested-by: Daniel P. Berrangé <berrange@redhat.com> Signed-off-by: Kashyap Chamarthy <kchamart@redhat.com> Reviewed-by: Erik Skultety <eskultet@redhat.com>
Michal Privoznik [Wed, 20 Jun 2018 12:15:42 +0000 (14:15 +0200)]
qemuDomainObjBeginJobInternal: Log agent job too
If a thread is unable to start a job (e.g. because of timeout)
a warning is printed into the logs. So far, the message does not
contain agent job info. Add it as it might help future debugging.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Jiri Denemark <jdenemar@redhat.com>
Weilun Zhu [Wed, 20 Jun 2018 08:45:27 +0000 (16:45 +0800)]
qemu: fix msg could be a wild pointer in qemuMonitorIOProcess()
As qemuMonitorJSONIOProcess will call qemuMonitorJSONIOProcessEvent
which unlocks the monitor mutex, there is some extreme situation,
eg qemu send message to monitor twice in a short time, where the
local viriable 'msg' of qemuMonitorIOProcess could be a wild point:
1. qemuMonitorSend() assign mon->msg to parameter 'msg', which is alse a
local variable of its caller qemuMonitorJSONCommandWithFd(), cause
eventloop to send message to monitor, then wait condition.
2. qemu send message to monitor for the first time immediately.
3. qemuMonitorIOProcess() is called, then wake up the qemuMonitorSend()
thread, but the qemuMonitorSend() thread stuck for a while as cpu pressure
or some other reasons, which means the qemu monitor is still unlocked.
4. qemu send event message to monitor for the second time,
such as RTC_CHANGE event
5. qemuMonitorIOProcess() is called again, the local viriable 'msg' is
assigned to mon->msg.
6. qemuMonitorIOProcess() call qemuMonitorJSONIOProcess() to deal with
the qemu event.
7. qemuMonitorJSONIOProcess() unlock the qemu monitor in the macro
'QEMU_MONITOR_CALLBACK', then qemuMonitorSend() thread get the mutex
and free the mon->msg, assign mon->msg to NULL.
Pino Toscano [Thu, 14 Jun 2018 13:34:25 +0000 (15:34 +0200)]
vmx: allow an odd number of vCPUs
Most probably this was a limitation in older ESX versions, and it seems
it does not exist anymore in more recent versions; see the following
thread:
https://www.redhat.com/archives/libvir-list/2018-May/msg02159.html
https://www.redhat.com/archives/libvir-list/2018-June/msg00043.html
Hence, allow an odd number (greater than 1) of vCPUs, since most
probably older versions of ESXi will error out anyway.
projects / libvirt.git / log