virPCIDeviceReattach and virPCIDeviceUnbindFromStub (called by
virPCIDeviceReattach) had previously required the name of the stub
driver as input. This is unnecessary, because the name of the driver
the device is currently bound to can be found by looking at the link:
/sys/bus/pci/dddd:bb:ss.ff/driver
Instead of requiring that the name of the expected stub driver name
and only unbinding if that one name is matched, we no longer take a
driver name in the arglist for either of these
functions. virPCIDeviceUnbindFromStub just compares the name of the
currently bound driver to a list of "well known" stubs (right now
contains "pci-stub" and "vfio-pci" for qemu, and "pciback" for xen),
and only performs the unbind if it's one of those devices.
This allows virsh nodedevice-reattach to work properly across a
libvirtd restart, and fixes a couple of cases where we were
erroneously still hard-coding "pci-stub" as the drive name.
For some unknown reason, virPCIDeviceReattach had been calling
modprobe on the stub driver prior to unbinding the device. This was
problematic because we no longer know the name of the stub driver in
that function. However, it is pointless to probe for the stub driver
at that time anyway - because the device is bound to the stub driver,
we are guaranteed that it is already loaded, and so that call to
modprobe has been removed.
Python code generator "generate_source" section that handles
code generation to "free" inherited objects needs to generate
DISPATCH_FREE calls for all extended_by objects.
For s390 we don't want to have a default USB device generated even
if QEMU is silently tolerating -usb on the command line. This may change
in the future.
Another reason to avoid the USB controller is that it implies a PCI
bus which might cause a regression at some later point in time.
The following change will set the USB controller model to 'none'
unless a model or address has been specified, which can be the case
if a legacy definition is loaded or the XML writer knows what
she/he's doing.
Requiring the user to explicitly disable USB on systems not supporting
it seems cumbersome.
Signed-off-by: Viktor Mihajlovski <mihajlov@linux.vnet.ibm.com>
Commit eca3fdf inadvertantly caused a failure to start for any domain
with the following in its config:
<graphics type='spice' autoport='yes'/>
The problem is that when tlsPort == 0 and defaultMode == "any" (which
is the default for defaultMode), this would be flagged in the code as
"needTLSPort", and if there was then no spice tls config, the new
error+fail would happen.
This patch checks for the case of defaultMode == "any", and in that
case simply doesn't allocate a TLS port (since that's probably not
what the user wanted, and it would have failed later anyway.). It does
leave the error in place for cases when the user specifically asked to
use tls in one way or another, though.
On cygwin, compilation failed because SIOCSIFHWADDR is undefined.
* src/util/virnetdev.c (virNetDevSetMAC): Cygwin can query but not
set mac address.
Signed-off-by: Eric Blake <eblake@redhat.com>
As a result of commit id '19c345f2', 'make -C tests valgrind' has the
following for qemuxml2argvtest:
==22482== 197 (80 direct, 117 indirect) bytes in 1 blocks are definitely lost in loss record 101 of 120
==22482== at 0x4A06B6F: calloc (vg_replace_malloc.c:593)
==22482== by 0x4C6F301: virAlloc (viralloc.c:124)
==22482== by 0x4C840FC: virSaveLastError (virerror.c:308)
==22482== by 0x431882: qemuBuildCommandLine (qemu_command.c:8204)
==22482== by 0x41E8F0: testCompareXMLToArgvHelper (qemuxml2argvtest.c:155)
==22482== by 0x41FE9F: virtTestRun (testutils.c:157)
==22482== by 0x419DEB: mymain (qemuxml2argvtest.c:654)
==22482== by 0x4204DA: virtTestMain (testutils.c:719)
==22482== by 0x39D0821A04: (below main) (libc-start.c:225)
==22482==
FreeBSD (and maybe other BSDs) have different member
names in struct ifreq when compared to Linux, such as:
- uses ifr_data instead of ifr_newname for setting
interface names
- uses ifr_index instead of ifr_ifindex for interface
index
Also, add a check for SIOCGIFHWADDR for virNetDevValidateConfig().
Use AF_LOCAL if AF_PACKET is not available.
Signed-off-by: Eric Blake <eblake@redhat.com>
If virPCIDeviceGetVFIOGroupDev() failed,
virSecurity*(Set|Restore)HostdevLabel() would fail to free a
virPCIDevice that had been allocated.
These leaks were all introduced (by me) very recently, in commit
f0bd70a.
libxlBuildDomainConfig() was disposing the libxl_domain_config object
on error, only to have it disposed again by libxlBuildDomainConfig()'s
caller, which resulted in a segfault. Leave disposing of the config
object to it's owner.
These fixes solve a compilation failure on FreeBSD:
util/virnetdevtap.c: In function 'virNetDevTapGetName':
util/virnetdevtap.c:56: warning: unused parameter 'tapfd' [-Wunused-parameter]
util/virnetdevtap.c:56: warning: unused parameter 'ifname' [-Wunused-parameter]
* src/util/virnetdevtap.c (virNetDevTapGetName): Add attributes
when TUNGETIFF is not present.
Signed-off-by: Eric Blake <eblake@redhat.com>
qemuBuildMemballoonDevStr returns NULL if memballoon doesn't have
the right address type, but it doesn't report an error, leading to:
error: An error occurred, but the cause is unknown
Report a helpful error message instead, e.g.:
error: XML error: memballoon unsupported with address type 'usb'
When a user requests auto-allocation of the spice TLS port but spice TLS
is disabled in qemu.conf, we start the machine and let qemu fail instead
of erroring out sooner.
Add an error message so that this doesn't happen.
On the off-chance that creation of persistent configuration file would
fail when defining a network that is already started as transient, the
code would remove the transient data structure and thus the network.
This patch changes the code so that in such case, the network is again
marked as transient and left behind.
The USB-specific cgroup setup had been inserted inline in
qemuDomainAttachHostUsbDevice and qemuSetupCgroup, but now there is a
common cgroup setup function called for all hostdevs, so it makes sens
to put the usb-specific setup there and just rely on that function
being called.
The one thing I'm uncertain of here (and a reason for not pushing
until after release) is that previously hostdev->missing was checked
only when starting a domain (and cgroup setup for the device skipped
if missing was true), but with this consolidation, it is now checked
in the case of hotplug as well. I don't know if this will have any
practical effect (does it make sense to hotplug a "missing" usb
device?)
PCIO device assignment using VFIO requires read/write access by the
qemu process to /dev/vfio/vfio, and /dev/vfio/nn, where "nn" is the
VFIO group number that the assigned device belongs to (and can be
found with the function virPCIDeviceGetVFIOGroupDev)
/dev/vfio/vfio can be accessible to any guest without danger
(according to vfio developers), so it is added to the static ACL.
The group device must be dynamically added to the cgroup ACL for each
vfio hostdev in two places:
1) for any devices in the persistent config when the domain is started
(done during qemuSetupCgroup())
2) at device attach time for any hotplug devices (done in
qemuDomainAttachHostDevice)
The group device must be removed from the ACL when a device it
"hot-unplugged" (in qemuDomainDetachHostDevice())
Note that USB devices are already doing their own cgroup setup and
teardown in the hostdev-usb specific function. I chose to make the new
functions generic and call them in a common location though. We can
then move the USB-specific code (which is duplicated in two locations)
to this single location. I'll be posting a followup patch to do that.
Don't reserve slot 2 for video if the machine has no PCI buses.
Error out when the user specifies a video device without
a PCI address when there are no PCI buses.
(This wouldn't work on a machine with no PCI bus anyway since
we do add PCI addresses for video devices to the command line)
In the past we automatically added a USB controller and assigned
it a PCI address (0:0:1.2) even on machines without a PCI bus.
This didn't break machines with no PCI bus because the command
line for it is just '-usb', with no mention of the PCI bus.
The implicit IDE controller (reserved address 0:0:1.1) has
no command line at all.
Commit b33eb0dc removed the ability to reserve PCI addresses
on machines without a PCI bus. This made them stop working,
since there would always be the implicit USB controller.
Skip the reservation of addresses for these controllers when
there is no PCI bus, instead of failing.
This isn't strictly speaking a bugfix, but I realized I'd gotten a bit
too verbose when I chose the names for
VIR_DOMAIN_HOSTDEV_PCI_BACKEND_TYPE_*. This shortens them all a bit.
I remembered to document this bit, but somehow forgot to implement it.
This adds <driver name='kvm|vfio'/> as a subelement to the <forward>
element of a network (this puts it parallel to the match between
mode='hostdev' attribute in a network and type='hostdev' in an
<interface>).
Since it's already documented, only the parser, formatter, backend
driver recognition (it just translates/moves the flag into the
<interface> at the appropriate time), and a test case were needed.
(I used a separate enum for the values both because the original is
defined in domain_conf.h, which is unavailable from network_conf.h,
and because in the future it's possible that we may want to support
other non-hostdev oriented driver names in the network parser; this
makes sure that one can be expanded without the other).
<source type='bridge'> uses a helper application to do the necessary
TUN/TAP setup to use an existing network bridge, thus letting
unprivileged users use TUN/TAP interfaces.
However, libvirt should be preventing QEMU from running any setuid
programs at all, which would include this helper program. From
a security POV, any setuid helper needs to be run by libvirtd itself,
not QEMU.
This is what this patch does. libvirt now invokes the setuid helper,
gets the TAP fd and then passes it to QEMU in the normal manner.
The path to the helper is specified in qemu.conf.
As a small advantage, this adds a <target dev='tap0'/> element to the
XML of an active domain using <interface type='bridge'>.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
This will be used on a tap file descriptor returned by the bridge helper
to populate the <target> element, because the helper does not provide
the interface name.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
this patch fix the wrong sequence for fd and timeout register. the sequence
was right in dfa1e1dd for fd register, but it changed in e0622ca2.
in this patch, set priv, xl_priv in info and increase info->priv ref count
before virEventAddHandle. if do this after virEventAddHandle, the fd
callback or fd deregister maybe got the empty priv, xl_priv or wrong ref
count.
after apply this patch, test more than 100 rounds passed compare to fail
within 3 rounds without this patch. each round includes define -> start ->
destroy -> create -> suspend -> resume -> reboot -> shutdown -> save ->
resotre -> dump -> destroy -> create -> setmem -> setvcpus -> destroy.
Signed-off-by: Bamvor Jian Zhang <bjzhang@suse.com>
VFIO requires all of the guest's memory and IO space to be lockable in
RAM. The domain's max_balloon is the maximum amount of memory the
domain can have (in KiB). We add a generous 1GiB to that for IO space
(still much better than KVM device assignment, where the KVM module
actually *ignores* the process limits and locks everything anyway),
and convert from KiB to bytes.
In the case of hotplug, we are changing the limit for the already
existing qemu process (prlimit() is used under the hood), and for
regular commandline additions of vfio devices, we schedule a call to
setrlimit() that will happen after the qemu process is forked.
These were previously being set in a custom hook function, but now
that virCommand directly supports setting them, we can eliminate that
part of the hook and call the APIs directly.
This patch adds two sets of functions:
1) lower level virProcessSet*() functions that will immediately set
the RLIMIT_MEMLOCK. RLIMIT_NPROC, or RLIMIT_NOFILE of either the
current process (using setrlimit()) or any other process (using
prlimit()). "current process" is indicated by passing a 0 for pid.
2) functions for virCommand* that will setup a virCommand object to
set those limits at a later time just after it has forked a new
process, but before it execs the new program.
configure.ac has prlimit and setrlimit added to the list of functions
to check for, and the low level functions log an "unsupported" error)
on platforms that don't support those functions.
If a user cgroup name begins with "cgroup.", "_" or with any of
the controllers from /proc/cgroups followed by a dot, then they
need to be prefixed with a single underscore. eg if there is
an object "cpu.service", then this would end up as "_cpu.service"
in the cgroup filesystem tree, however, "waldo.service" would
stay "waldo.service", at least as long as nobody comes up with
a cgroup controller called "waldo".
Since we require a '.XXXX' suffix on all partitions, there is
no scope for clashing with the kernel 'tasks' and 'release_agent'
files.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
If the partition named passed in the XML does not already have
a suffix, ensure it gets a '.partition' added to each component.
The exceptions are /machine, /user and /system which do not need
to have a suffix, since they are fixed partitions at the top
level.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Recently we changed to create VM cgroups with the naming pattern
$VMNAME.$DRIVER.libvirt. Following discussions with the systemd
community it was decided that only having a single '.' in the
names is preferrable. So this changes the naming scheme to be
$VMNAME.libvirt-$DRIVER. eg for LXC 'mycontainer.libvirt-lxc' or
for KVM 'myvm.libvirt-qemu'.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Legacy kvm style pci device assignment requires changes to the
labelling of several sysfs files for each device, but for vfio device
assignment, the only thing that needs to be relabelled/chowned is the
"group" device for the group that contains the device to be assigned.
This was the only hypervisor driver other than qemu that implemented
virNodeDeviceDettach. It doesn't currently support multiple pci device
assignment driver backends, but it is simple to plug in this new API,
which will make it easier for Xen people to fill it in later when they
decide to support VFIO (or whatever other) device assignment. Also it
means that management applications will have the same API available to
them for both hypervisors on any given version of libvirt.
The only acceptable value for driverName in this case is NULL, since
there is no alternate, and I'm not willing to pick a name for the
default driver used by Xen.
The differences from virNodeDeviceDettach are very minor:
1) Check that the flags are 0.
2) Set the virPCIDevice's stubDriver according to the driverName that
is passed in.
3) Call virPCIDeviceDetach with a NULL stubDriver, indicating it
should get the name of the stub driver from the virPCIDevice
object.
This requires a custom function for remoteNodeDeviceDetachFlags,
because it is named *NodeDevice, but it goes through the hypervisor
driver rather than nodedevice driver, and so it uses privateData
instead of nodeDevicePrivateData. (It has to go through the hypervisor
driver, because that is the driver that knows about the backend drivers
that will perform the pci device assignment).
The existing virNodeDeviceDettach() assumes that there is only a
single PCI device assignment backend driver appropriate for any
hypervisor. This is no longer true, as the qemu driver is getting
support for PCI device assignment via VFIO. The new API
virNodeDeviceDetachFlags adds a driverName arg that should be set to
the exact same string set in a domain <hostdev>'s <driver name='x'/>
element (i.e. "vfio", "kvm", or NULL for default). It also adds a
flags arg for good measure (and because it's possible we may need it
when we start dealing with VFIO's "device groups").
If the config for a device has specified <driver name='vfio'/>,
"backend" in the pci part of the hostdev object will be set to
..._VFIO. In this case, when creating a virPCIDevice set the
stubDriver to "vfio-pci", otherwise set it to "pci-stub". We will rely
on the lower levels to report an error if the vfio driver isn't
loaded.
The detach/attach functions in virpci.c will pay attention to the
stubDriver setting in the device, and bind/unbind the appropriate
driver when preparing hostdevs for the domain.
Note that we don't yet attempt to do anything to mark active any other
devices in the same vfio "group" as a single device that is being
marked active. We do need to do that, but in order to get basic VFIO
functionality testing sooner rather than later, initially we'll just
live with more cryptic errors when someone tries to do that.
This can be set when the virPCIDevice is created and placed on a list,
then used later when traversing the list to determine which stub
driver to bind/unbind for managed devices.
The existing Detach and Attach functions' signatures haven't been
changed (they still accept a stub driver name in the arg list), but if
the arg list has NULL for stub driver and one is available in the
device's object, that will be used. (we may later deprecate and remove
the arg from those functions).
The device option for vfio-pci is nearly identical to that for
pci-assign - only the configfd parameter isn't supported (or needed).
Checking for presence of the bootindex parameter is done separately
from constructing the commandline, similar to how it is done for
pci-assign.
This patch contains tests to check for proper commandline
construction. It also includes tests for parser-formatter-parser
roundtrips (xml2xml), because those tests use the same data files, and
would have failed had they been included before now.
qemu: xml/args tests for VFIO hostdev and <interface type='hostdev'/>
These should be squashed in with the patch that adds commandline
handling of vfio (they would fail at any earlier time).
A domain's <interface> or <hostdev>, as well as a <network>'s
<forward>, can now have an optional <driver name='kvm|vfio'/>
element. As of this patch, there is no functionality behind this new
knob - this patch adds support to the domain and network
formatter/parser, and to the RNG and documentation.
When the backend is added, legacy KVM PCI device assignment will
continue to be used when no driver name is specified (or if <driver
name='kvm'/> is specified), but if driver name is 'vfio', the new UEFI
Secure Boot compatible VFIO device assignment will be used.
Note that the parser doesn't automatically insert the current default
value of this setting. This is done on purpose because the two
possibilities are functionally equivalent from the guest's point of
view, and we want to be able to automatically start using vfio as the
default (even for existing domains) at some time in the future. This
is similar to what was done with the "vhost" driver option in
<interface>.
There will soon be other items related to pci hostdevs that need to be
in the same part of the hostdevsubsys union as the pci address (which
is currently a single member called "pci". This patch replaces the
single member named pci with a struct named pci that contains a single
member named "addr".
QEMU_CAPS_DEVICE_VFIO_PCI is set if the device named "vfio-pci" is
supported in the qemu binary.
QEMU_CAPS_VFIO_PCI_BOOTINDEX is set if the vfio-pci device supports
the "bootindex" parameter; for some reason, the bootindex parameter
wasn't included in early versions of vfio support (qemu 1.4) so we
have to check for it separately from vfio itself.
POSIX says that both basename() and dirname() may return static
storage (aka they need not be thread-safe); and that they may but
not must modify their input argument. Furthermore, <libgen.h>
is not available on all platforms. For these reasons, you should
never use these functions in a multi-threaded library.
Gnulib instead recommends a way to avoid the portability nightmare:
gnulib's "dirname.h" provides useful thread-safe counterparts. The
obvious dir_name() and base_name() are GPL (because they malloc(),
but call exit() on failure) so we can't use them; but the LGPL
variants mdir_name() (malloc's or returns NULL) and last_component
(always points into the incoming string without modifying it,
differing from basename semantics only on corner cases like the
empty string that we shouldn't be hitting in the first place) are
already in use in libvirt. This finishes the swap over to the safe
functions.
* cfg.mk (sc_prohibit_libgen): New rule.
* src/util/vircgroup.c: Fix offenders.
* src/parallels/parallels_storage.c (parallelsPoolAddByDomain):
Likewise.
* src/parallels/parallels_network.c (parallelsGetBridgedNetInfo):
Likewise.
* src/node_device/node_device_udev.c (udevProcessSCSIHost)
(udevProcessSCSIDevice): Likewise.
* src/storage/storage_backend_disk.c
(virStorageBackendDiskDeleteVol): Likewise.
* src/util/virpci.c (virPCIGetDeviceAddressFromSysfsLink):
Likewise.
* src/util/virstoragefile.h (_virStorageFileMetadata): Avoid false
positive.
Signed-off-by: Eric Blake <eblake@redhat.com>
Jim Fehlig reported on IRC that older gcc/glibc triggers this warning:
cc1: warnings being treated as errors
qemu/qemu_domain.c: In function 'qemuDomainDefFormatBuf':
qemu/qemu_domain.c:1297: error: declaration of 'remove' shadows a global declaration [-Wshadow]
/usr/include/stdio.h:157: error: shadowed declaration is here [-Wshadow]
make[3]: *** [libvirt_driver_qemu_impl_la-qemu_domain.lo] Error 1
Fix it like we have done in the past (such as commit 2e6322a).
* src/qemu/qemu_domain.c (qemuDomainDefFormatBuf): Avoid shadowing
a function name.
Signed-off-by: Eric Blake <eblake@redhat.com>
When all usb controllers connected to the same bus have <master
startport='x'/> specified, none of them have 'id=usb' assigned and
thus qemu fails due to invalid masterport specification (we use 'usb'
for that purpose). Adding a check that at least one of the
controllers is specified without <master startport='x'/> and in case
this happens, error out due to invalid configuration.
After 9d6e56db the syntax-check was unhappy due to wrong whitespacing:
src/qemu/qemu_command.c:1637: for ( ; a.slot < QEMU_PCI_ADDRESS_SLOT_LAST; a.slot++) {
maint.mk: incorrect whitespace around brackets, see HACKING for rules
make: *** [bracket-spacing-check] Error 1
After 78d7c3c5 we are strdup()-ing path to qemu-bridge-helper.
However, the check for its return value is missing. So it is
possible we've ignored the OOM error silently.
Add a "dry run" address allocation to figure out how many bridges
will be needed for all the devices without explicit addresses.
Auto-add just enough bridges to put all the devices on, or up to the
bridge with the largest specified index.
<controller type='pci' index='0' model='pci-root'/>
is auto-added to pc* machine types.
Without this controller PCI bus 0 is not available and
no PCI addresses are assigned by default.
Since older libvirt supported PCI bus 0 even without
this controller, it is removed from the XML when migrating.
Now we set the default disk driver name when parsing
the qemu command line too, hence all the test changes.
Assume format type is 'auto' when none is specified on
qemu command line.
For pSeries guest in QEMU, NVRAM is one kind of spapr-vio device.
Users are allowed to specify spapr-vio devices'address.
But NVRAM is not supported in libvirt. So this patch is to
add NVRAM device to allow users to specify its address.
In QEMU, NVRAM device's address is specified by
"-global spapr-nvram.reg=xxxxx".
In libvirt, XML file is defined as the following:
<nvram>
<address type='spapr-vio' reg='0x3000'/>
</nvram>
Signed-off-by: Li Zhang <zhlcindy@linux.vnet.ibm.com>
Currently, if there has been an error in building command line
process after virtual interfaces has been created, the flow jumps
to 'error' label, where virDomainConfNWFilterTeardown() is
called. This may report an error as well, but should not
overwrite the original cause why we jumped to 'error' label.
Instead of making a choice between the underscore and camelCase, this
simply changes "num_queues" into "queues", which is also consistent
with Michal's multiple queue support for interface.
Improve error reporting and generating of SPICE command line arguments
according to the need to enable TLS. If TLS is disabled, there's no need
to pass the certificate dir to qemu.
This patch resolves:
https://bugzilla.redhat.com/show_bug.cgi?id=953126
Ensure that all drivers implementing public APIs use a
naming convention for their implementation that matches
the public API name.
eg for the public API virDomainCreate make sure QEMU
uses qemuDomainCreate and not qemuDomainStart
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
It will simplify later work if the sub-drivers have dedicated
APIs / field names. ie virNetworkDriver should have
virDrvNetworkOpen and virDrvNetworkClose methods
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
The driver.h struct for node devices used an inconsistent
naming scheme 'DeviceMonitor' instead of the more usual
'NodeDeviceDriver'. Fix this everywhere it has leaked
out to.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
The driver.h file has no consistent indentation usage across
all the typedefs. Attempts to vertically align struct field
members have also been inconsistently applied. Sanitize the
whitespace used for typedefs & remove all vertical alignment
from structs
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Ensure that the driver struct field names match the public
API names. For an API virXXXX we must have a driver struct
field xXXXX. ie strip the leading 'vir' and lowercase any
leading uppercase letters.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Ensure that the virDrvXXX method names exactly match
the public APIs virYYY method names. ie XXX == YYY.
Add a test case to prevent any regressions.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
A number of the remote procedure names did not match the
corresponding API names. For example, many lacked the
word 'CONNECT', others re-arranged the names. Update the
procedures so their names exactly match the API names.
Then remove the special case handling of these APIs in
the generator
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
There are many declared options in gendispatch.pl that were
no longer used. Those which were used were obscure '-b', '-k'
and '-d'. Switch to use --mode={debug|client|server}.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Switch the function from a bunch of ifs to a switch statement with
correct type and reflow some code.
Also fix comment in enum describing possible graphics types
Decrease size of qemuBuildGraphicsCommandLine() by splitting out
spice-related code into qemuBuildGraphicsVNCCommandLine().
This patch also fixes 2 possible memory leaks on error path in the code
that was split-out. The buffer containing the already generated options
and a listen address string could be leaked.
Also break a few very long lines and reflow code that fits now.
Decrease size of qemuBuildGraphicsCommandLine() by splitting out
spice-related code into qemuBuildGraphicsSPICECommandLine().
This patch also fixes 2 possible memory leaks on error path in the code
that was split-out. The buffer containing the already generated options
and a listen address string could be leaked.
Also break a few very long lines.
Currently the RPC protocol files can contain annotations after
the protocol enum eg
REMOTE_PROC_DOMAIN_SNAPSHOT_LIST_CHILDREN_NAMES = 247, /* autogen autogen priority:high */
This is not very extensible as the number of annotations grows.
Change it to use
/**
* @generate: both
* @priority: high
*/
REMOTE_PROC_DOMAIN_SNAPSHOT_LIST_CHILDREN_NAMES = 247,
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Refactoring done in 19c6ad9ac7 didn't
correctly take into account the order cgroup limit modification needs to
be done in. This resulted into errors when decreasing the limits.
The operations need to take place in this order:
decrease hard limit
change swap hard limit
or
change swap hard limit
increase hard limit
This patch also fixes the check if the hard_limit is less than
swap_hard_limit to print better error messages. For this purpose I
introduced a helper function virCompareLimitUlong to compare limit
values where value of 0 is equal to unlimited. Additionally the check is
now applied also when the user does not provide all of the tunables
through the API and in that case the currently set values are used.
This patch resolves:
https://bugzilla.redhat.com/show_bug.cgi?id=950478
1. Handle invalid ULong prefix specified.
When parsing for @prefix as a ULong, a -2 can be returned
if the specification is not a valid ULong.
2. Error out if address= is not specified.
3. Merge netmask process/tests under family tests.
4. Max sure that prefix does not exceed maximum.
.
Signed-off-by: Gene Czarcinski <gene@czarc.net>
Create the utility function virSocketAddrGetIpPrefix() to
determine the prefix for this network. The code in this
function was adapted from virNetworkIpDefPrefix().
Update virNetworkIpDefPrefix() in src/conf/network_conf.c
to use the new utility function.
Signed-off-by: Gene Czarcinski <gene@czarc.net>
The change in commit aed4986322
was incomplete, missing a couple of cases of /system. This
caused failure to start VMs.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
libvirt/HACKING suggests omitting braces with a
single-line body; this patch fixes the coding style
problem for the Sheepdog storage backend driver.
Signed-off-by: Harry Wei <harryxiyou@gmail.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
After discussions with systemd developers it was decided that
a better default policy for resource partitions is to have
3 default partitions at the top level
/system - system services
/machine - virtual machines / containers
/user - user login session
This ensures that the default policy isolates guest from
user login sessions & system services, so a mis-behaving
guest can't consume 100% of CPU usage if other things are
contending for it.
Thus we change the default partition from /system to
/machine
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Wrong use of the parentheses causes "rc" always having a boolean value,
either "1" or "0", and thus we can't get the detailed error message
when it fails:
Before (I only have 1 node):
% virsh numatune f18 --nodeset 12
error: Unable to change numa parameters
error: unable to set numa tunable: Unknown error -1
After:
virsh numatune f18 --nodeset 12
error: Unable to change numa parameters
error: unable to set numa tunable: Invalid argument
http://www.uhv.edu/ac/newsletters/writing/grammartip2009.07.01.htm
(and several other sites) give hints that 'onto' is best used if
you can also add 'up' just before it and still make sense. In many
cases in the code base, we really want the two-word form, or even
a simplification to just 'on' or 'to'.
* docs/hacking.html.in: Use correct 'on to'.
* python/libvirt-override.c: Likewise.
* src/lxc/lxc_controller.c: Likewise.
* src/util/virpci.c: Likewise.
* daemon/THREADS.txt: Use simpler 'on'.
* docs/formatdomain.html.in: Better usage.
* docs/internals/rpc.html.in: Likewise.
* src/conf/domain_event.c: Likewise.
* src/rpc/virnetclient.c: Likewise.
* tests/qemumonitortestutils.c: Likewise.
* HACKING: Regenerate.
Signed-off-by: Eric Blake <eblake@redhat.com>
https://bugzilla.redhat.com/show_bug.cgi?id=922186
Commit d04916fa introduced a regression in audit quality - even
though the code was computing the proper escaped name for a
path, it wasn't feeding that escaped name on to the audit message.
As a result, /var/log/audit/audit.log would mention a pair of
fields class=path path=/dev/hpet instead of the intended
class=path path="/dev/hpet", which in turn caused ausearch to
format the audit log with path=(null).
* src/conf/domain_audit.c (virDomainAuditCgroupPath): Use
constructed encoding.
Signed-off-by: Eric Blake <eblake@redhat.com>
Each bus is represented as an array of 32 8-bit integers
where each bit represents a PCI function and each byte represents
a PCI slot.
Uses just one bus so far.
Create a new function qemuPCIAddressValidate and call it everywhere
the user might supply an incorrect address:
* qemuCollectPCIAddress for domain definition
* qemuDomainPCIAddressEnsureAddr and ReleaseSlot for hotplug
Slot and function shouldn't be wrong at this point, since values
out of range should be rejected by the XML parser.
Change QEMU_PCI_ADDRESS_LAST_SLOT to the number of slots in the bus,
not the maximum slot value, to match QEMU_PCI_ADDRESS_LAST_FUNCTION
and rename them both to have _LAST at the end.
The error message reported when attempting to change/get persistent
configuration of a transient domain suggests that changes are being
made. Reword it to suit getter APIs too.
Before:
$ virsh vcpucount transient-domain --config
error: Requested operation is not valid: cannot change persistent config of a transient domain
After:
$ virsh vcpucount transient-domain --config
error: Requested operation is not valid: transient domains do not have any persistent config
Until now tranisent networks weren't really useful as libvirtd wasn't
able to remember them across restarts. This patch adds support for
loading status files of transient networks (that already were generated)
so that the status isn't lost.
This patch chops up virNetworkObjUpdateParseFile and turns it into
virNetworkLoadState and a few friends that will help us to load status
XMLs and refactors the functions that are loading the configs to use
them.
The cpu_map.xml file is there to separate CPU model definitions from the
code. Having the only interesting data for PowerPC models only in the
source code. This patch moves this data to the XML file and removes the
hardcoded list completely.
PowerPC CPUs are either identical or incompatible and thus we just need
to look up the right model for given PVR without pretending we have
several candidates which we may choose from.
The function is also renamed as ppcDecode to match other functions in
PowerPC CPU driver.
Baseline API is supposed to return guest CPU definition that can be used
on any of the provided host CPUs. Since PowerPC CPUs are either
identical or incompatible, the API just needs to check that all provided
CPUs are identical. Previous implementation was completely bogus.
The function is also renamed as ppcBaseline to match other functions in
PowerPC CPU driver.
When ppcVendorLoad fails to parse the vendor element for whatever
reason, it is supposed to ignore it and return 0 rather than -1. The
patch also removes PowerPC vendor string from the XML as it is not
actually used for anything.
Make getting node CPU data for PowerPC unsupported on other
architectures. The function is also renamed as ppcNodeData to match
other functions in PowerPC CPU driver.
Currently, -device xxx still doesn't work well for ppc64 platform.
It's better use legacy USB option with default for ppc64.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
When running unprivileged, virSetUIDGIDWithCaps will fail because it
tries to add the requested capabilities to the permitted and effective
sets.
Detect this case, and invoke the child with cleared permitted and
effective sets. If it is a setuid program, it will get them.
Some care is needed also because you cannot drop capabilities from the
bounding set without CAP_SETPCAP. Because of that, ignore errors from
setting the bounding set.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
The need_prctl variable is not really needed. If it is false,
capng_apply will be called twice with the same set, causing
a little extra work but no problem. This keeps the code a bit
simpler.
It is also clearer to invoke capng_apply(CAPNG_SELECT_BOUNDS)
separately, to make sure it is done while we have CAP_SETPCAP.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Reusing the result of virArchFromHost instead of calling it multiple times
Signed-off-by: Tal Kain <tal.kain@ravellosystems.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
Directories python/tools/examples should include them in <> form,
though this patch allows "" form in these directories by excluding
them, a later patch will do the cleanup.
The recent qemu requires "0x" prefix for the disk wwn, this patch
changes virValidateWWN to allow the prefix, and prepend "0x" if
it's not specified. E.g.
qemu-kvm: -device scsi-hd,bus=scsi0.0,channel=0,scsi-id=0,lun=0,\
drive=drive-scsi0-0-0-0,id=scsi0-0-0-0,wwn=6000c60016ea71ad:
Property 'scsi-hd.wwn' doesn't take value '6000c60016ea71ad'
Though it's a qemu regression, but it's nice to allow the prefix,
and doesn't hurt for us to always output "0x".
Detected by a simple Shell script:
for i in $(git ls-files -- '*.[ch]'); do
awk 'BEGIN {
fail=0
}
/# *include.*\.h/{
match($0, /["<][^">]*[">]/)
arr[substr($0, RSTART+1, RLENGTH-2)]++
}
END {
for (key in arr) {
if (arr[key] > 1) {
fail=1
printf("%d %s\n", arr[key], key)
}
}
if (fail == 1)
exit 1
}' $i
if test $? != 0; then
echo "Duplicate header(s) in $i"
fi
done;
A later patch will add the syntax-check to avoid duplicate
headers.
Fix the error
util/vircgroup.c: In function 'virCgroupNewDomainPartition':
util/vircgroup.c:1299:11: error: declaration of 'dirname' shadows a global declaration [-Werror=shadow]
Signed-off-by: Stefan Berger <stefanb@linux.vnet.ibm.com>
Check for an unsupported QMP command when using the query-tpm-models
and query-tpm-types commands before checking for general errors
in order to avoid error messages in the log.
Signed-off-by: Stefan Berger <stefanb@linux.vnet.ibm.com>
Check for QMP query-tpm-models and set a capability flag. Do not use
this QMP command if it is not supported.
Signed-off-by: Stefan Berger <stefanb@linux.vnet.ibm.com>
The LXC driver currently has code to detect cgroups mounts
and then re-mount them inside the new root filesystem. Replace
this fragile code with a call to virCgroupIsolateMount.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Add a virCgroupIsolateMount method which looks at where the
current process is place in the cgroups (eg /system/demo.lxc.libvirt)
and then remounts the cgroups such that this sub-directory
becomes the root directory from the current process' POV.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
If a cgroup controller is co-mounted with another, eg
/sys/fs/cgroup/cpu,cpuacct
Then it is a requirement that there exist symlinks at
/sys/fs/cgroup/cpu
/sys/fs/cgroup/cpuacct
pointing to the real mount point. Add support to virCgroupPtr
to detect and track these symlinks
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
The virCgroupNewDriver method had a 'bool privileged' param.
If a false value was ever passed in, it would simply not
work, since non-root users don't have any privileges to create
new cgroups. Just delete this broken code entirely and make
the QEMU driver skip cgroup setup in non-privileged mode
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Historically QEMU/LXC guests have been placed in a cgroup layout
that is
$LOCATION-OF-LIBVIRTD/libvirt/{qemu,lxc}/$VMNAME
This is bad for a number of reasons
- The cgroup hierarchy gets very deep which seriously
impacts kernel performance due to cgroups scalability
limitations.
- It is hard to setup cgroup policies which apply across
services and virtual machines, since all VMs are underneath
the libvirtd service.
To address this the default cgroup location is changed to
be
/system/$VMNAME.{lxc,qemu}.libvirt
This puts virtual machines at the same level in the hierarchy
as system services, allowing consistent policy to be setup
across all of them.
This also honours the new resource partition location from the
XML configuration, for example
<resource>
<partition>/virtualmachines/production</partitions>
</resource>
will result in the VM being placed at
/virtualmachines/production/$VMNAME.{lxc,qemu}.libvirt
NB, with the exception of the default, /system, path which
is intended to always exist, libvirt will not attempt to
auto-create the partitions in the XML. It is the responsibility
of the admin/app to configure the partitions. Later libvirt
APIs will provide a way todo this.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Allow VMs to be placed into resource groups using the
following syntax
<resource>
<partition>/virtualmachines/production</partition>
</resource>
A resource cgroup will be backed by some hypervisor specific
functionality, such as cgroups with KVM/LXC.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
A resource partition is an absolute cgroup path, ignoring the
current process placement. Expose a virCgroupNewPartition API
for constructing such cgroups
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Currently if virCgroupMakeGroup fails, we can get in a situation
where some controllers have been setup, but others not. Ensure
we call virCgroupRemove to remove what we've done upon failure
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Currently the virCgroupPtr struct contains 3 pieces of
information
- path - path of the cgroup, relative to current process'
cgroup placement
- placement - current process' placement in each controller
- mounts - mount point of each controller
When reading/writing cgroup settings, the path & placement
strings are combined to form the file path. This approach
only works if we assume all cgroups will be relative to
the current process' cgroup placement.
To allow support for managing cgroups at any place in the
heirarchy a change is needed. The 'placement' data should
reflect the absolute path to the cgroup, and the 'path'
value should no longer be used to form the paths to the
cgroup attribute files.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Rename all the virCgroupForXXX methods to use the form
virCgroupNewXXX since they are all constructors. Also
make sure the output parameter is the last one in the
list, and annotate all pointers as non-null. Fix up
all callers, and make sure they use true/false not 0/1
for the boolean parameters
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
The definition of structs for cgroups are kept in vircgroup.c since
they are intended to be private from users of the API. To enable
effective testing, however, they need to be accessible. To address
the latter issue, without compronmising the former, this introduces
a new vircgrouppriv.h file to hold the struct definitions.
To prevent other files including this private header, it requires
that __VIR_CGROUP_ALLOW_INCLUDE_PRIV_H__ be defined before inclusion
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Instead of calling virCgroupForDomain every time we need
the virCgrouPtr instance, just do it once at Vm startup
and cache a reference to the object in virLXCDomainObjPrivatePtr
until shutdown of the VM. Removing the virCgroupPtr from
the LXC driver state also means we don't have stale mount
info, if someone mounts the cgroups filesystem after libvirtd
has been started
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Instead of calling virCgroupForDomain every time we need
the virCgrouPtr instance, just do it once at Vm startup
and cache a reference to the object in qemuDomainObjPrivatePtr
until shutdown of the VM. Removing the virCgroupPtr from
the QEMU driver state also means we don't have stale mount
info, if someone mounts the cgroups filesystem after libvirtd
has been started
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
The virCgroupForDriver method recently gained an 'int controllers'
parameter, but the stub impl did not
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Though they are the same thing, mixed use of them is uncomfortable.
"unsigned" is used a lot in old codes, this just tries to change the
ones in utils.
If libvirt makes any gcry_control() calls, then this
prevents gnutls for doing any initialization. As such
we must take care to do full initialization of libcrypt
on a par with what gnutls would have done. In particular
we must disable "sec mem" for cases where the user does
not have mlock() permission. We also skip our init of
libgcrypt if something else (ie the app using libvirt)
has beaten us to it.
https://bugzilla.redhat.com/show_bug.cgi?id=951630
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Report the errors as:
Domain not found: no domain with matching uuid '41414141-4141-4141-4141-414141414141' (crashtest)
instead of:
Domain not found: no domain with matching uuid '41414141-4141-4141-4141-414141414141'
Use the helper to lookup the domain object in the remaining places.
This patch also fixes error reporting when the domain was not found in several
functions that were printing the raw UUID buffer instead of the formatted
string. The offending functions were:
qemuDomainGetInterfaceParameters
qemuDomainSetInterfaceParameters
qemuGetSchedulerParametersFlags
qemuSetSchedulerParametersFlags
qemuDomainGetNumaParameters
qemuDomainSetNumaParameters
qemuDomainGetMemoryParameters
qemuDomainSetMemoryParameters
qemuDomainGetBlkioParameters
qemuDomainSetBlkioParameters
qemuDomainGetCPUStats
Some refactoring for virDomainChrSourceDef type of devices so
we can use common code.
Signed-off-by: Stefan Berger <stefanb@linux.vnet.ibm.com>
Reviewed-by: Corey Bryant <coreyb@linux.vnet.ibm.com>
Tested-by: Corey Bryant <coreyb@linux.vnet.ibm.com>
When a VM with a TPM passthrough device is started, the audit daemon
logs the following type of message:
type=VIRT_RESOURCE msg=audit(1365170222.460:3378): pid=16382 uid=0 auid=4294967295 ses=4294967295 subj=system_u:system_r:virtd_t:s0-s0:c0.c1023 msg='virt=kvm resrc=dev reason=start vm="TPM-PT" uuid=a4d7cd22-da89-3094-6212-079a48a309a1 device="/dev/tpm0" exe="/usr/sbin/libvirtd" hostname=? addr=? terminal=? res=success'
Signed-off-by: Stefan Berger <stefanb@linux.vnet.ibm.com>
Reviewed-by: Corey Bryant <coreyb@linux.vnet.ibm.com>
Tested-by: Corey Bryant <coreyb@linux.vnet.ibm.com>
Parse the domain XML with TPM passthrough support.
The TPM passthrough XML may look like this:
<tpm model='tpm-tis'>
<backend type='passthrough'>
<device path='/dev/tpm0'/>
</backend>
</tpm>
Signed-off-by: Stefan Berger <stefanb@linux.vnet.ibm.com>
Reviewed-by: Corey Bryant <coreyb@linux.vnet.ibm.com>
Tested-by: Corey Bryant <coreyb@linux.vnet.ibm.com>
Probe for QEMU's QMP TPM support by querying the lists of
supported TPM models (query-tpm-models) and backend types
(query-tpm-types).
The setting of the capability flags following the strings
returned from the commands above is only provided in the
patch where domain_conf.c gets TPM support due to dependencies
on functions only introduced there.
Signed-off-by: Stefan Berger <stefanb@linux.vnet.ibm.com>
Reviewed-by: Corey Bryant <coreyb@linux.vnet.ibm.com>
Tested-by: Corey Bryant <coreyb@linux.vnet.ibm.com>
This patch adds the ability to configure non-contiguous boot orders on boot
devices. This allows unplugging devices that have boot order specified without
breaking migration.
The new code now uses a slightly less memory efficient approach to store the
boot order fields in a hashtable instead of a bitmap.
To avoid the collision for creating USB controllers in machine->init()
and -device xx command line, it needs to set usb=off to avoid one USB
controller created in machine->init(). So that libvirt can use -device
or -usb to create USB controller sucessfully.
So QEMU_CAPS_MACHINE_USB_OPT capability is added, and it is for QEMU
v1.3.0 onwards which supports USB option.
Signed-off-by: Li Zhang <zhlcindy@linux.vnet.ibm.com>
When migrating a domain with disk images stored locally (and using
storage migration), we should not complain about unsafe migration no
matter what cache policy is used for that disk.
https://bugzilla.redhat.com/show_bug.cgi?id=920441
Currently, we are discarding listen attribute from qemu cookie even though
we strive to gather it. This result in not so cool bug: if user have
different networks, one for management/migration, and one for VNC/SPICE we
pass incorrect host to the qemu in client_migrate_info. What we actually
pass is remote hostname, while we should be passing remote listen address.
It doesn't matter as long as these two are the same, but they don't need
necessary to be like that.
==5306== 8 bytes in 1 blocks are definitely lost in loss record 24 of 277
==5306== at 0x4C28B2F: calloc (vg_replace_malloc.c:593)
==5306== by 0x5293CAF: virAllocN (viralloc.c:152)
==5306== by 0x52DFEAE: virXPathNodeSet (virxml.c:611)
==5306== by 0x5313DD9: virNetworkDefParseXML (network_conf.c:1408)
==5306== by 0x53170F6: virNetworkObjUpdateParseFile (network_conf.c:2031)
==5306== by 0x131DA63C: networkStartup (bridge_driver.c:279)
==5306== by 0x53481DF: virStateInitialize (libvirt.c:822)
==5306== by 0x40DF44: daemonRunStateInit (libvirtd.c:877)
==5306== by 0x52D2FF5: virThreadHelper (virthreadpthread.c:161)
==5306== by 0x5D00C52: start_thread (in /usr/lib64/libpthread-2.17.so)
==5306== by 0x6410ECC: clone (in /usr/lib64/libc-2.17.so)
This patch fixes crash of the daemon that happens due to the following race
condition:
Let's have two threads in the libvirtd daemon's qemu driver:
A - thread executing undefine on the same domain
B - thread executing a API call to get information about a domain
Assume following serialization of operations done by the threads:
1) A has the lock on the domain object and is executing some code prior to
virDomainObjListRemove()
2) B takes the lock on the domain object list, looks up the domain object
pointer and blocks in the attempt to lock the domain object as A is holding the
lock
3) A reaches virDomainObjListRemove() and unlocks the lock on the domain object
4) A blocks on the attempt to get the domain list lock
5) B is able to lock the domain object now and unlocks the domain list
6) A is now able to lock the domain list, and sheds the last reference on the
domain object, this triggers the freeing function.
6) B starts executing the code on the pointer that is being freed
7) The libvirtd daemon crashes while attempting to access invalid pointer in
thread B.
This patch fixes the race by acquiring a reference on the domain object before
unlocking it in virDomainObjListRemove() and re-locks the object prior to
removing and freeing it. This ensures that no thread holds a lock on the domain
object at the time it is removed from the list, and that doing a list lookup
will never find a domain that is about to vanish.
This is a minimal fix of the problem, but a better solution will be to switch to
full reference counting for domain objects.
Commit 9a3ff01d7f (which was ACKed at
the end of January, but for some reason didn't get pushed until during
the 1.0.4 freeze) fixed the logic in virPCIGetVirtualFunctions().
Unfortunately, a typo in the fix (replacing VIR_REALLOC_N with
VIR_ALLOC_N during code movement) caused not only a memory leak, but
also resulted in most of the elements of the result array being
replaced with NULL. virNetDevGetVirtualFunctions() assumed (and I think
rightly so) that virPCIGetVirtualFunctions() wouldn't return any NULL
elements in the array, so it ended up segfaulting.
This was found when attempting to use a virtual network with an
auto-created pool of SRIOV VFs, e.g.:
<forward mode='hostdev' managed='yes'>
<pf dev='eth4'/>
</forward>
(the pool of PCI addresses is discovered by calling
virNetDevGetVirtualFunctions() on the PF dev).
The helper function to look up disk controller model may be used by scsi
hostdev. But it should be changed to use device info.
Signed-off-by: Han Cheng <hanc.fnst@cn.fujitsu.com>
Even though http://libvirt.org/formatdomain.html#elementsMetadata
states that it requires RFC4122 compliance UUIDs that are generated
by virUUIDGenerate() are not. Following patch modifies generated
UUIDs to conform to rules described in RFC.
Signed-off-by: Milos Vyletel <milos.vyletel@sde.cz>
If the user requests a mount for /run, this may hide any existing
mounts that are lower down in /run. The result is that the
container still sees the mounts in /proc/mounts, but cannot
access them
sh-4.2# df
df: '/run/user/501/gvfs': No such file or directory
df: '/run/media/berrange/LIVE': No such file or directory
df: '/run/media/berrange/SecureDiskA1': No such file or directory
df: '/run/libvirt/lxc/sandbox': No such file or directory
Filesystem 1K-blocks Used Available Use% Mounted on
/dev/mapper/vg_t500wlan-lv_root 151476396 135390200 8384900 95% /
tmpfs 1970888 3204 1967684 1% /run
/dev/sda1 194241 155940 28061 85% /boot
devfs 64 0 64 0% /dev
tmpfs 64 0 64 0% /sys/fs/cgroup
tmpfs 1970888 1200 1969688 1% /etc/libvirt-sandbox/scratch
Before mounting any filesystem at a particular location, we
must recursively unmount anything at or below the target mount
point
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Ensure lxcContainerUnmountSubtree is at the top of the
lxc_container.c file so it is easily referenced from
any other method. No functional change
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
This allows a container-type domain to have exclusive access to one of
the host's NICs.
Wire <hostdev caps=net> with the lxc_controller - when moving the newly
created veth devices into a new namespace, also look for any hostdev
devices that should be moved. Note: once the container domain has been
destroyed, there is no code that moves the interfaces back to the
original namespace. This does happen, though, probably due to default
cleanup on namespace destruction.
Signed-off-by: Bogdan Purcareata <bogdan.purcareata@freescale.com>
This updates the definitions and supporting structures in the XML
schema and domain configuration files.
Signed-off-by: Bogdan Purcareata <bogdan.purcareata@freescale.com>
The virCgroupMounted method is badly named, since a controller can be
mounted, but disabled in the current object. Rename the method to be
virCgroupHasController. Also make it tolerant to a NULL virCgroupPtr
and out-of-range controller index, to avoid duplication of these
checks in all callers
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
To support "shareable" for volume type disk, we have to translate
the source before trying to add the shared disk entry. To achieve
the goal, this moves the helper qemuTranslateDiskSourcePool into
src/qemu/qemu_conf.c, and introduce an internal only member (voltype)
for struct _virDomainDiskSourcePoolDef, to record the underlying
volume type for use when building the drive string.
Later patch will support "shareable" volume type disk.
This adds a new helper qemuTranslateDiskSourcePool which uses the
storage pool/vol APIs to translate the disk source before building
the drive string. Network volume is not supported yet. Disk chain
for volume type disk may be supported later, but before I'm confident
it doesn't break anything, it's just disabled now.
With this patch, one can specify the disk source using libvirt
storage like:
<disk type='volume' device='disk'>
<driver name='qemu' type='raw' cache='none'/>
<source pool='default' volume='fc18.img'/>
<target dev='vdb' bus='virtio'/>
</disk>
"seclabels" and "startupPolicy" are not supported for this new
disk type ("volume"). They will be supported in later patches.
docs/formatdomain.html.in:
* Add documents for new XMLs
docs/schemas/domaincommon.rng:
* Add rng for new XMLs;
src/conf/domain_conf.h:
* New struct for 'volume' type disk source (virDomainDiskSourcePoolDef)
* Add VIR_DOMAIN_DISK_TYPE_VOLUME for enum virDomainDiskType
src/conf/domain_conf.c:
* New helper virDomainDiskSourcePoolDefParse to parse the 'volume'
type disk source.
* New helper virDomainDiskSourcePoolDefFree to free the source def
if 'volume' type disk.
tests/qemuxml2argvdata/qemuxml2argv-disk-source-pool.xml:
tests/qemuxml2xmltest.c:
* New test
This finds the parent for vHBA by iterating over all the HBA
which supports vport_ops capability on the host, and return
the first one which is online, not saturated (vports in use
is less than max_vports).
startPool creates the vHBA if it's not existed yet, stopPool destroys
the vHBA. Also to support autostart, checkPool will creates the vHBA
if it's not existed yet.
The helper iterates over sysfs, to find out the matched scsi host
name by comparing the wwnn,wwpn pair. It will be used by checkPool
and refreshPool of storage scsi backend. New helper getAdapterName
is introduced in storage_backend_scsi.c, which uses the new util
helper virGetFCHostNameByWWN to get the fc_host adapter name.
node device driver names the HBA like "scsi_host5", but storage
driver uses "host5", which could make the user confused. This
changes them to be consistent. However, for back-compat reason,
adapter name like "host5" is still supported.
This introduces 4 new attributes for storage pool source adapter.
E.g.
<adapter type='fc_host' parent='scsi_host5' wwnn='20000000c9831b4b' wwpn='10000000c9831b4b'/>
Attribute 'type' can be either 'scsi_host' or 'fc_host', and defaults
to 'scsi_host' if attribute 'name' is specified. I.e. It's optional
for 'scsi_host' adapter, for back-compat reason. However, mandatory
for 'fc_host' adapter and any new future adapter types. Attribute
'parent' is to specify the parent for the fc_host adapter.
* docs/formatstorage.html.in:
- Add documents for the 4 new attrs
* docs/schemas/storagepool.rng:
- Add RNG schema
* src/conf/storage_conf.c:
- Parse and format the new XMLs
* src/conf/storage_conf.h:
- New struct virStoragePoolSourceAdapter, replace "char *adapter" with it;
- New enum virStoragePoolSourceAdapterType
* src/libvirt_private.syms:
- Export TypeToString and TypeFromString
* src/phyp/phyp_driver.c:
- Replace "adapter" with "adapter.data.name", which is member of the union
of the new struct virStoragePoolSourceAdapter now. Later patch will
add the checking, as "adapter.data.name" is only valid for "scsi_host"
adapter.
* src/storage/storage_backend_scsi.c:
- Like above
* tests/storagepoolxml2xmlin/pool-scsi-type-scsi-host.xml:
* tests/storagepoolxml2xmlin/pool-scsi-type-fc-host.xml:
- New test for 'fc_host' and "scsi_host" adapter
* tests/storagepoolxml2xmlout/pool-scsi.xml:
- Change the expected output, as the 'type' defaults to 'scsi_host' if 'name"
specified now
* tests/storagepoolxml2xmlout/pool-scsi-type-scsi-host.xml:
* tests/storagepoolxml2xmlout/pool-scsi-type-fc-host.xml:
- New test
* tests/storagepoolxml2xmltest.c:
- Include the test
There are a number of places which generate cast alignment
warnings, which are difficult or impossible to address. Use
pragmas to disable the warnings in these few places
conf/nwfilter_conf.c: In function 'virNWFilterRuleDetailsParse':
conf/nwfilter_conf.c:1806:16: warning: cast increases required alignment of target type [-Wcast-align]
item = (nwItemDesc *)((char *)nwf + att[idx].dataIdx);
conf/nwfilter_conf.c: In function 'virNWFilterRuleDefDetailsFormat':
conf/nwfilter_conf.c:3238:16: warning: cast increases required alignment of target type [-Wcast-align]
item = (nwItemDesc *)((char *)def + att[i].dataIdx);
storage/storage_backend_mpath.c: In function 'virStorageBackendCreateVols':
storage/storage_backend_mpath.c:247:17: warning: cast increases required alignment of target type [-Wcast-align]
names = (struct dm_names *)(((char *)names) + next);
nwfilter/nwfilter_dhcpsnoop.c: In function 'virNWFilterSnoopDHCPDecode':
nwfilter/nwfilter_dhcpsnoop.c:994:15: warning: cast increases required alignment of target type [-Wcast-align]
pip = (struct iphdr *) pep->eh_data;
nwfilter/nwfilter_dhcpsnoop.c:1004:11: warning: cast increases required alignment of target type [-Wcast-align]
pup = (struct udphdr *) ((char *) pip + (pip->ihl << 2));
nwfilter/nwfilter_learnipaddr.c: In function 'procDHCPOpts':
nwfilter/nwfilter_learnipaddr.c:327:33: warning: cast increases required alignment of target type [-Wcast-align]
uint32_t *tmp = (uint32_t *)&dhcpopt->value;
nwfilter/nwfilter_learnipaddr.c: In function 'learnIPAddressThread':
nwfilter/nwfilter_learnipaddr.c:501:43: warning: cast increases required alignment of target type [-Wcast-align]
struct iphdr *iphdr = (struct iphdr*)(packet +
nwfilter/nwfilter_learnipaddr.c:538:43: warning: cast increases required alignment of target type [-Wcast-align]
struct iphdr *iphdr = (struct iphdr*)(packet +
nwfilter/nwfilter_learnipaddr.c:544:48: warning: cast increases required alignment of target type [-Wcast-align]
struct udphdr *udphdr= (struct udphdr *)
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
When reading the inotify FD, we get back a sequence of
struct inotify_event, each with variable length data following.
It is not safe to simply cast from the char *buf to the
struct inotify_event struct since this may violate data
alignment rules. Thus we must copy from the char *buf
into the struct inotify_event instance before accessing
the data.
uml/uml_driver.c: In function 'umlInotifyEvent':
uml/uml_driver.c:327:13: warning: cast increases required alignment of target type [-Wcast-align]
e = (struct inotify_event *)tmp;
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
The current way virObject instances are allocated using
VIR_ALLOC_N causes alignment warnings
util/virobject.c: In function 'virObjectNew':
util/virobject.c:195:11: error: cast increases required alignment of target type [-Werror=cast-align]
Changing to use VIR_ALLOC_VAR will avoid the need todo
the casts entirely.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
The virNetlinkCommand() method takes an 'unsigned char **'
parameter to be filled with the received netlink message.
The callers then immediately cast this to 'struct nlmsghdr',
triggering (bogus) warnings about increasing alignment
requirements
util/virnetdev.c: In function 'virNetDevLinkDump':
util/virnetdev.c:1300:12: warning: cast increases required alignment of target type [-Wcast-align]
resp = (struct nlmsghdr *)*recvbuf;
^
util/virnetdev.c: In function 'virNetDevSetVfConfig':
util/virnetdev.c:1429:12: warning: cast increases required alignment of target type [-Wcast-align]
resp = (struct nlmsghdr *)recvbuf;
Since all callers cast to 'struct nlmsghdr' we can avoid
the warning problem entirely by simply changing the
signature of virNetlinkCommand to return a 'struct nlmsghdr **'
instead of 'unsigned char **'. The way we do the cast inside
virNetlinkCommand does not have any alignment issues.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Playing games with field offsets in a struct causes all sorts
of alignment warnings on ARM platforms
util/virkeycode.c: In function '__virKeycodeValueFromString':
util/virkeycode.c:26:7: warning: cast increases required alignment of target type [-Wcast-align]
(*(typeof(field_type) *)((char *)(object) + field_offset))
^
util/virkeycode.c:91:28: note: in expansion of macro 'getfield'
const char *name = getfield(virKeycodes + i, const char *, name_offset);
^
util/virkeycode.c:26:7: warning: cast increases required alignment of target type [-Wcast-align]
(*(typeof(field_type) *)((char *)(object) + field_offset))
^
util/virkeycode.c:94:20: note: in expansion of macro 'getfield'
return getfield(virKeycodes + i, unsigned short, code_offset);
^
util/virkeycode.c: In function '__virKeycodeValueTranslate':
util/virkeycode.c:26:7: warning: cast increases required alignment of target type [-Wcast-align]
(*(typeof(field_type) *)((char *)(object) + field_offset))
^
util/virkeycode.c:127:13: note: in expansion of macro 'getfield'
if (getfield(virKeycodes + i, unsigned short, from_offset) == key_value)
^
util/virkeycode.c:26:7: warning: cast increases required alignment of target type [-Wcast-align]
(*(typeof(field_type) *)((char *)(object) + field_offset))
^
util/virkeycode.c:128:20: note: in expansion of macro 'getfield'
return getfield(virKeycodes + i, unsigned short, to_offset);
There is no compelling reason to use a struct for the keycode
tables. It can easily just use an array of arrays instead,
avoiding all alignment problems
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
For both "live" and "config" changes of vcpupin and emulatorpin, an
all clear bitmap doesn't make sense, and it can just cause corruptions.
E.g (similar for emulatorpin).
% virsh vcpupin hame 0 8,^8 --config
% virsh vcpupin hame
VCPU: CPU Affinity
----------------------------------
0:
1: 0-63
2: 0-63
3: 0-63
% virsh dumpxml hame | grep cpuset
<vcpupin vcpu='0' cpuset=''/>
% virsh start hame
error: Failed to start domain hame
error: An error occurred, but the cause is unknown
This introduce a new attribute "num_queues" (same with the good name
QEMU uses) for virtio-scsi controller. An example of the XML:
<controller type='scsi' index='0' model='virtio-scsi' num_queues='8'/>
The corresponding QEMU command line:
-device virtio-scsi-pci,id=scsi0,num_queues=8,bus=pci.0,addr=0x3 \
By default, libtool builds two .o files for every .lo rule:
src/foo.o - static builds
src/.libs/foo.o - shared library builds
But since commit ad42b34b disabled static builds, src/foo.o is
no longer built by default. On a fresh checkout, this means our
protocol check rules using pdwtags were testing a missing file,
and thanks to a lousy behavior of pdwtags happily giving no output
and 0 exit status (http://bugzilla.redhat.com/949034), we were
merely claiming that "dwarves is too old" and skipping the test.
However, if you swap between branches and do incremental builds,
such as building v0.10.2-maint and then switching back to master,
you end up with src/foo.o being leftover from its 0.10.2 state,
and then 'make check' fails because the .o file does not match
the protocol-structs file due to API additions in the meantime.
A simpler fix would be to always look in .libs for the .o to
be parsed; but since it is possible to pass ./configure options
to tell libtool to do a static-only build with no shared .o,
I went with the approach of finding the newest of the two files,
whenever both exist.
* src/Makefile.am (PDWTAGS): Ensure we test just-built file.
When setting processor count for a domain using the API libvirt enforced
a maximum processor count, while it isn't enforced when taking the XML path.
This patch removes the check to match the XML.
Currently when getting an instance of virCgroupPtr we will
create the path in all cgroup controllers. Only at the virt
driver layer are we attempting to filter controllers. This
is bad because the mere act of creating the dirs in the
controllers can have a functional impact on the kernel,
particularly for performance.
Update the virCgroupForDriver() method to accept a bitmask
of controllers to use. Only create dirs in the controllers
that are requested. When creating cgroups for domains,
respect the active controller list from the parent cgroup
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
The virCgroupGetAppRoot is not clear in its meaning. Change
to virCgroupForSelf to highlight that this returns the
cgroup config for the caller's process
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
The last Viktor's effort to fix the race and memory corruption unfortunately
wasn't complete in the case the close callback was not registered in an
connection. At that time, the trail of event's that I'll describe later could
still happen and corrupt the memory or cause a crash of the client (including
the daemon in case of a p2p migration).
Consider the following prerequisities and trail of events:
Let's have a remote connection to a hypervisor that doesn't have a close
callback registered and the client is using the event loop. The crash happens in
cooperation of 2 threads. Thread E is the event loop and thread W is the worker
that does some stuff. R denotes the remote client.
1.) W - The client finishes everything and sheds the last reference on the client
2.) W - The virObject stuff invokes virConnectDispose that invokes doRemoteClose
3.) W - the remote close method invokes the REMOTE_PROC_CLOSE RPC method.
4.) W - The thread is preempted at this point.
5.) R - The remote side receives the close and closes the socket.
6.) E - poll() wakes up due to the closed socket and invokes the close callback
7.) E - The event loop is preempted right before remoteClientCloseFunc is called
8.) W - The worker now finishes, and frees the conn object.
9.) E - The remoteClientCloseFunc accesses the now-freed conn object in the
attempt to retrieve pointer for the real close callback.
10.) Kaboom, corrupted memory/segfault.
This patch tries to fix this by introducing a new object that survives the
freeing of the connection object. We can't increase the reference count on the
connection object itself or the connection would never be closed, as the
connection is closed only when the reference count reaches zero.
The new object - virConnectCloseCallbackData - is a lockable object that keeps
the pointers to the real user registered callback and ensures that the
connection callback is either not called if the connection was already freed or
that the connection isn't freed while this is being called.
By adjusting the reference count of the connection object we
prevent races between callback function and virConnectClose.
Signed-off-by: Viktor Mihajlovski <mihajlov@linux.vnet.ibm.com>
This patch refactors various places to allow removing of the
defaultConsoleTargetType callback from the virCaps structure.
A new console character device target type is introduced -
VIR_DOMAIN_CHR_CONSOLE_TARGET_TYPE_NONE - to mark that no type was
specified in the XML. This type is at the end converted to the standard
VIR_DOMAIN_CHR_CONSOLE_TARGET_TYPE_SERIAL. Other types that are
different from this default have to be processed separately in the
device post parse callback.
Use the virDomainXMLConf structure to hold this data and tweak the code
to avoid semantic change.
Without configuration the KVM mac prefix is used by default. I chose it
as it's in the privately administered segment so it should be usable for
any purposes.
Use the qemu specific callback to fill this data in the qemu driver as
it's the only place where it was used and fix tests as the qemu test
capability object didn't configure the defaults for the tests.
This patch removes the emulatorRequired field and associated
infrastructure from the virCaps object. Instead the driver specific
callbacks are used as this field isn't enforced by all drivers.
This patch implements the appropriate callbacks in the qemu and lxc
driver and moves to check to that location.
This patch removes the defaultDiskDriverName from the virCaps
structure. This particular default value is used only in the qemu driver
so this patch uses the recently added callback to fill the driver name
if it's needed instead of propagating it through virCaps.
This gets rid of the parameter in favor of using the new callback
infrastructure to do the same stuff.
This patch implements the domain adjustment callback in the openVZ
driver and moves the check from the parser to a new validation method in
the callback infrastructure.
This patch implements the devices post parse callback and uses it to fill
the default qemu network card model into the XML if none is specified.
Libvirt assumes that the network card model for qemu is the "rtl8139".
Record this in the XML using the new callback to avoid user
confusion.
This patch adds instrumentation that will allow hypervisor drivers to
fill and validate domain and device definitions after parsed by the XML
parser.
With this patch, after the XML is parsed, a callback to the driver is
issued requesting to fill and validate driver specific details of the
configuration. This allows to use sensible defaults and checks on a per
driver basis at the time the XML is parsed.
Two callback pointers are stored in the new virDomainXMLConf object:
* virDomainDeviceDefPostParseCallback (devicesPostParseCallback)
- called for a single device parsed and for every single device in a
domain config. A virDomainDeviceDefPtr is passed along with the
domain definition and virCaps.
* virDomainDefPostParseCallback, (domainPostParseCallback)
- A callback that is meant to process the domain config after it's
parsed. A virDomainDefPtr is passed along with virCaps.
Both types of callbacks support arbitrary opaque data passed for the
callback functions.
Errors may be reported in those callbacks resulting in a XML parsing
failure.
This patch is the result of running:
for i in $(git ls-files | grep -v html | grep -v \.po$ ); do
sed -i -e "s/virDomainXMLConf/virDomainXMLOption/g" -e "s/xmlconf/xmlopt/g" $i
done
and a few manual tweaks.
If libnuma is not compiled in, or numa_available() returns an
error, stub out fake NUMA info consisting of one NUMA cell
containing all CPUs and memory.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Not all kernel builds have any entries under the location
/sys/devices/system/cpu/cpu0/topology. We already cope with
that being missing in some cases, but not all. Update the
code which looks for thread_siblings to cope with the missing
file
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
