Commit Graph

21269 Commits

Author SHA1 Message Date
Peter Krempa
d773b57d22 qemu: don't iterate vcpus using priv->nvcpupids in qemuProcessSetSchedParams
This should be the last offender.
2016-01-28 09:58:24 +01:00
Peter Krempa
763941749e conf: disallow empty cpuset for emulatorpin
It's disallowed in the API.
2016-01-27 17:27:54 +01:00
Peter Krempa
31b782a147 conf: disallow empty cpusets for vcpu pinning when parsing XML
They are disallowed in the pinning API and as default cpuset.

Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=1293241
2016-01-27 17:27:54 +01:00
Peter Krempa
414b7eeae9 qemu: Don't use priv->ncpus to iterate cgroup setting
Iterate over all cpus skipping inactive ones.
2016-01-27 17:27:54 +01:00
Andrea Bolognani
d87f0c0052 virnetdevopenvswitch: Don't call strlen() twice on the same string
Commit 871e10f fixed a memory corruption error, but called strlen()
twice on the same string to do so. Even though the compiler is
probably smart enough to optimize the second call away, having a
single invocation makes the code slightly cleaner.

Suggested-by: Michal Privoznik <mprivozn@redhat.com>
2016-01-27 13:01:24 +01:00
Michal Privoznik
720bc953f8 virnetdevmacvlan: Provide stubs for build without macvtap
In 370608b4c7 we have introduced two new internal APIs.
However, there are no stubs for build without macvtap. Therefore
build on systems lacking macvtap support (e.g. mingw or freebds)
fails when trying to link.

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2016-01-27 10:07:46 +01:00
Jason J. Herne
871e10fc95 Fix libvirtd free() segfault when migrating guest with deleted open vswitch port
libvirtd crashes on free()ing portData for an open vswitch port if that port
was deleted.  To reproduce:

ovs-vsctl del-port vnet0
virsh migrate --live kvm1 qemu+ssh://dstHost/system

Error message:
libvirtd: *** Error in `/usr/sbin/libvirtd': free(): invalid pointer: 0x000003ff90001e20 ***

The problem is that virCommandRun can return an empty string in the event that
the port being queried does not exist. When this happens then we are
unconditionally overwriting a newline character at position strlen()-1. When
strlen is 0, we overwrite memory that does not belong to the string.

The fix: Only overwrite the newline if the string is not empty.

Reviewed-by: Bjoern Walk <bwalk@linux.vnet.ibm.com>
Signed-off-by: Jason J. Herne <jjherne@linux.vnet.ibm.com>
2016-01-27 10:01:58 +01:00
Laine Stump
370608b4c7 util: keep/use a bitmap of in-use macvtap devices
This patch creates two bitmaps, one for macvlan device names and one
for macvtap. The bitmap position is used to indicate that libvirt is
currently using a device with the name macvtap%d/macvlan%d, where %d
is the position in the bitmap. When requested to create a new
macvtap/macvlan device, libvirt will now look for the first clear bit
in the appropriate bitmap and derive the device name from that rather
than just starting at 0 and counting up until one works.

When libvirtd is restarted, the qemu driver code that reattaches to
active domains calls the appropriate function to "re-reserve" the
device names as it is scanning the status of running domains.

Note that it may seem strange that the retry counter now starts at
8191 instead of 5. This is because we now don't do a "pre-check" for
the existence of a device once we've reserved it in the bitmap - we
move straight to creating it; although very unlikely, it's possible
that someone has a running system where they have a large number of
network devices *created outside libvirt* named "macvtap%d" or
"macvlan%d" - such a setup would still allow creating more devices
with the old code, while a low retry max in the new code would cause a
failure. Since the objective of the retry max is just to prevent an
infinite loop, and it's highly unlikely to do more than 1 iteration
anyway, having a high max is a reasonable concession in order to
prevent lots of new failures.
2016-01-26 12:20:04 -05:00
Leno Hou
8c70d04bab util: increase libnl buffer size
In the following cases nl_recv() was returning the error "No buffer
space available":

* When switching CPUs to offline/online in a system more than 128 cpus
* When using virsh to destroy domain in a system with many interfaces

This patch sets the buffer size for all netlink sockets created by
libnl to 128K and turns on message peeking for nl_recv(). This
eliminates the "No buffer space available" errors seen in the cases
above, and also preempts other future errors the smaller buffers could
have caused.

Signed-off-by: Leno Hou <houqy@linux.vnet.ibm.com>
Signed-off-by: Laine Stump <laine@laine.org>
2016-01-26 12:20:04 -05:00
Pavel Hrdina
36785c7e77 device: cleanup input device code
The current code was a little bit odd.  At first we've removed all
possible implicit input devices from domain definition to add them later
back if there was any graphics device defined while parsing XML
description.  That's not all, while formating domain definition to XML
description we at first ignore any input devices with bus different to
USB and VIRTIO and few lines later we add implicit input devices to XML.

This seems to me as a lot of code for nothing.  This patch may look
to be more complicated than original approach, but this is a preferred
way to modify/add driver specific stuff only in those drivers and not
deal with them in common parsing/formating functions.

The update is to add those implicit input devices into config XML to
follow the real HW configuration visible by guest OS.

There was also inconsistence between our behavior and QEMU's in the way,
that in QEMU there is no way how to disable those implicit input devices
for x86 architecture and they are available always, even without graphics
device.  This applies also to XEN hypervisor.  VZ driver already does its
part by putting correct implicit devices into live XML.

Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
2016-01-26 17:53:33 +01:00
Pavel Hrdina
2686e44e05 tests: add some missing tests to qemuxml2xmltest
Those tests are in qemuargv2xmltest and it makes sense to include them
also in qemuxml2xmltest and qemuxml2argvtest.

Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
2016-01-26 17:53:33 +01:00
Pavel Hrdina
2d446b6eeb tests: use virtTestDifferenceFull in tests where we have output file
This will enable regenerate functionality for those tests to make
developer lives easier while updating tests.

Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
2016-01-26 17:53:33 +01:00
Michal Privoznik
c7f5e26b5f vircgroup: Finish renaming of virCgroupIsolateMount
In dc576025c3 we renamed virCgroupIsolateMount function to
virCgroupBindMount. However, we forgot about one occurrence in
section of the code which provides stubs for platforms without
support for CGroups like *BSD for instance.

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2016-01-26 17:39:47 +01:00
Daniel P. Berrange
dc576025c3 lxc: don't try to hide parent cgroups inside container
On the host when we start a container, it will be
placed in a cgroup path of

   /machine.slice/machine-lxc\x2ddemo.scope

under /sys/fs/cgroup/*

Inside the containers' namespace we need to setup
/sys/fs/cgroup mounts, and currently will bind
mount /machine.slice/machine-lxc\x2ddemo.scope on
the host to appear as / in the container.

While this may sound nice, it confuses applications
dealing with cgroups, because /proc/$PID/cgroup
now does not match the directory in /sys/fs/cgroup

This particularly causes problems for systems and
will make it create repeated path components in
the cgroup for apps run in the container eg

  /machine.slice/machine-lxc\x2ddemo.scope/machine.slice/machine-lxc\x2ddemo.scope/user.slice/user-0.slice/session-61.scope

This also causes any systemd service that uses
sd-notify to fail to start, because when systemd
receives the notification it won't be able to
identify the corresponding unit it came from.
In particular this break rabbitmq-server startup

Future kernels will provide proper cgroup namespacing
which will handle this problem, but until that time
we should not try to play games with hiding parent
cgroups.

Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2016-01-26 16:11:32 +00:00
Daniel P. Berrange
511e7c5bba qemu: add reporting of vCPU wait time
The VIR_DOMAIN_STATS_VCPU flag to virDomainListGetStats
enables reporting of stats about vCPUs. Currently we
only report the cumulative CPU running time and the
execution state.

This adds reporting of the wait time - time the vCPU
wants to run, but the host scheduler has something else
running ahead of it.

The data is reported per-vCPU eg

$ virsh domstats --vcpu demo
 Domain: 'demo'
   vcpu.current=4
   vcpu.maximum=4
   vcpu.0.state=1
   vcpu.0.time=1420000000
   vcpu.0.wait=18403928
   vcpu.1.state=1
   vcpu.1.time=130000000
   vcpu.1.wait=10612111
   vcpu.2.state=1
   vcpu.2.time=110000000
   vcpu.2.wait=12759501
   vcpu.3.state=1
   vcpu.3.time=90000000
   vcpu.3.wait=21825087

In implementing this I notice our reporting of CPU execute
time has very poor granularity, since we are getting it
from /proc/$PID/stat. As a future enhancement we should
prefer to get CPU execute time from /proc/$PID/schedstat
or /proc/$PID/sched (if either exist on the running kernel)

Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2016-01-26 14:34:23 +00:00
Luyao Huang
985f01a65f virsh: fix cpu-stats command output format issue
After commit 57177f1, the cpu-stats command format change to:

CPU0:
    cpu_time         14401.507878990 seconds
    vcpu_time    14378732785511

vcpu_time is not user friendly. After this patch, it will
change back:
CPU0:
    cpu_time         14401.507878990 seconds
    vcpu_time        14378.732785511 seconds

https://bugzilla.redhat.com/show_bug.cgi?id=1301807

Signed-off-by: Luyao Huang <lhuang@redhat.com>
2016-01-26 09:23:49 +01:00
Peter Krempa
356e28b35e util: buffer: Sanitize comment for virBufferAddBuffer
Idioms are usually weird and obscure when translated literally.
2016-01-25 17:53:08 +01:00
Peter Krempa
7141fc7a27 test: Touch up error message when attempting to pin invalid vCPU
Report
error: invalid argument: requested vcpu '100' is not present in the domain
instead of
error: invalid argument: requested vcpu is higher than allocated vcpus
2016-01-25 17:53:08 +01:00
Peter Krempa
f82a8014c0 tests: qemuxml2xml: Order pinning information numerically
A future patch will refactor the storage of the pinning information in a
way where the ordering will be lost. Order them numerically to avoid
changing the tests later.
2016-01-25 17:53:07 +01:00
Peter Krempa
a2e80549a2 virsh: cpu-stats: Remove unneeded flags
virDomainGetCPUStats doesn't support flags so there's no need to carry
the 'flags' variable around. Additionally since the API is poorly
designed I doubt that it will be extended.
2016-01-25 17:45:09 +01:00
Peter Krempa
57177f1abd virsh: cpu-stats: Extract common printing code into a function
Simplify the code by extracting a common code path.
2016-01-25 17:45:09 +01:00
Peter Krempa
51f07d8f0f (qemu|lxc)DomainGetCPUStats: Clean up
Remove unnecessary condition and variable.
2016-01-25 17:45:09 +01:00
Peter Krempa
68ee703bfe vz: Fix invalid iteration of def->cputune.vcpupin
The array doesn't necessarily have the same cardinality as the count of
vCPUs for a domain. Iterating it can cause access beyond the end of the
array.
2016-01-25 17:45:09 +01:00
Peter Krempa
b3c91b8a50 qemu: process: Disallow VMs with 0 vcpus
Counterintuitively the user would end up with a VM with maximum number
of vCPUs available.

Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=1290324
2016-01-25 17:45:09 +01:00
Peter Krempa
adca15cf15 qemu: process: refactor and rename qemuValidateCpuMax to qemuValidateCpuCount
Next patch will add minimum checking, so use a more generic name.
Refactor return values to the commonly used semantics.
2016-01-25 17:45:09 +01:00
Michal Privoznik
99f8fb4c55 virt-host-validate: Fix error level for user namespace check
From the code it seems to me that we need user namespace if
configured in domain XML. Otherwise we don't use it at all.
However our tool is more strict about that. Fix this discrepancy.

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2016-01-25 16:53:23 +01:00
Michal Privoznik
d55e11a302 virt-host-validate: Check those CGroups that we actually use
Since the introduction of virt-host-validate tool the set of
cgroup controllers we use has changed so the tool is checking for
some cgroups that we don't need (e.g. net_cls, although I doubt
we have ever used that one) and is not checking for those we
actually use (e.g. cpuset).

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2016-01-25 16:53:18 +01:00
Michal Privoznik
9cbd1ecc3e virsh: Correctly detect inserted media in change-media command
https://bugzilla.redhat.com/show_bug.cgi?id=1250331

It all works like this. The change-media command dumps domain
XML, finds the corresponding cdrom device we want to change media
in and returns it in the xmlNodePtr form. This way we don't have
to bother with keeping all the subelements or attributes that we
don't care about in the XML that is fed back to libvirt for the
update API.

Now, the problem is we try to be clever here and detect if disk
already has a source (indicated by <source/> subelement).
However, bare fact that the element is there does not mean disk
has source. Make our clever check better.

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2016-01-25 15:57:23 +01:00
Michal Privoznik
35c3aab44d vmx: Adapt to emptyBackingString for cdrom-image
https://bugzilla.redhat.com/show_bug.cgi?id=1266088

We are missing this value for cdrom-image device. It seems like
there's no added value to extend this to other types of disk
devices [1].

1: https://www.redhat.com/archives/libvir-list/2016-January/msg01038.html

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2016-01-25 08:34:23 +01:00
Peter Krempa
4ac14cde9a qemu: snapshot: Correctly report qemu error on 'savevm'
Since 'savevm' was not converted to QMP libvirt has to parse for error
strings in the text monitor output. One of the unhandled errors is
produced when qemu treats a device as unmigratable.

As current qemu actually does support AHCI migration this bug is
applicable only to older versions of qemu.

Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=1293899
2016-01-25 07:21:25 +01:00
Peter Krempa
0c1b0d83bb qemu: monitor: Refactor error handling for 'savevm'
Unify few error conditions into a single error reporting case.
2016-01-25 07:21:25 +01:00
Roman Bogorodskiy
ef01addb38 bhyve: bhyveload: respect boot dev and boot order
Make bhyveload respect boot order as specified by os.boot section of the
domain XML or by "boot order" for specific devices. As bhyve does not
support a real boot order specification right now, it's just about
choosing a single device to boot from.
2016-01-25 04:19:33 +03:00
Roman Bogorodskiy
318ae9f3be conf: expose virDomainBootType(From|To)String
These functions are going to be used by the Bhyve driver.
2016-01-25 03:54:07 +03:00
Laine Stump
29cc45cb79 util: reset MAC address of macvtap passthrough physdev after disassociate
libvirt always resets the MAC address of the physdev used for macvtap
passthrough when the guest is finished with it. This was happening
prior to the 802.1Qb[gh] DISASSOCIATE command, and was quite often
failing, presumably because the driver wouldn't allow the MAC address
to be reset while the association was still active, with a log message
like this:

virNetDevSetMAC:168 : Cannot set interface MAC to 00:00:00:00:00:00 on 'eth13': Cannot assign requested address

This patch changes the order - we now do the 802.1Qb[gh] disassociate
and delete the macvtap interface first, then and reset the MAC
address.
2016-01-22 13:16:24 -05:00
Cole Robinson
81da8bc73b lxc: fuse: Stub out Slab bits in /proc/meminfo
'free' on fedora23 wants to use the Slab field for calculated used
memory. The equation is:

used = MemTotal - MemFree - (Cached + Slab) - Buffers

We already set Cached and Buffers to 0, do the same for Slab and its
related values

https://bugzilla.redhat.com/show_bug.cgi?id=1300781
2016-01-22 08:32:00 -05:00
Cole Robinson
c7be484d11 lxc: fuse: Fill in MemAvailable for /proc/meminfo
'free' on Fedora 23 will use MemAvailable to calculate its 'available'
field, but we are passing through the host's value. Set it to match
MemFree, which is what 'free' will do for older linux that don't have
MemAvailable

https://bugzilla.redhat.com/show_bug.cgi?id=1300781
2016-01-22 08:32:00 -05:00
Cole Robinson
8418245a7e lxc: fuse: Fix /proc/meminfo size calculation
We virtualize bits of /proc/meminfo by replacing host values with
values specific to the container.

However for calculating the final size of the returned data, we are
using the size of the original file and not the altered copy, which
could give garbelled output.
2016-01-22 08:32:00 -05:00
Cole Robinson
f65dcfcd14 lxc: fuse: Unindent meminfo logic
Reverse the conditional at the start so we aren't stuffing all the logic
in an 'if' block
2016-01-22 08:32:00 -05:00
Ian Campbell
daeace5c5d libxl: Support cmdline= in xl config files
... and consolidate the cmdline/extra/root parsing to facilitate doing
so.

The logic is the same as xl's parse_cmdline from the current xen.git master
branch (e6f0e099d2c17de47fd86e817b1998db903cab61).

On the formatting side switch to producing cmdline= instead of extra=.

Update a few tests and add serveral more.
  - test-cmdline is added to test the exclusive use of cmdline.
  - test-fullvirt-direct-kernel-boot.cfg is updated due to the switch
    on the formatting side and now tests the exclusive use of cmdline=.
  - Tests are added for both paravirt and fullvirt where the .cfg uses
    extra= and (paravirt only) root=. These are format (xl->xml) only
    since the inverse will generate cmdline= hence is not a round trip
    (which was already true if using root=, which used to generate
    extra= on the way back).
  - Tests are added for both paravirt and fullvirt where the .cfg
    declares cmdline= as well as bogus extra= and (paravirt only) root=
    entries which should be ignored. Again these are format only tests
    since the inverse won't include the bogus lines.

The last two bullets here required splitting the DO_TEST macro into
two halves, as is done in the xmconfigtest.c case.

In order to introduce a use of VIR_WARN for logging I had to add
virerror.h and VIR_LOG_INIT.

Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
2016-01-21 10:48:44 -07:00
Joao Martins
d18d6a85f9 libxl: dispose libxl_dominfo after libxl_domain_info()
As suggested in a previous thread [0] this patch adds some missing calls
to libxl_dominfo_{init,dispose} when doing some of the libxl_domain_info
operations which would otherwise lead to memory leaks.

[0]
https://www.redhat.com/archives/libvir-list/2015-September/msg00519.html

Signed-off-by: Joao Martins <joao.m.martins@oracle.com>
2016-01-21 09:49:57 -07:00
Jim Fehlig
8c3c32f16a Xen: add XENXL to virErrorDomain enum
Add "Xen XL Config" to the virErrorDomain enum and use it in
src/xenconfig/xen_xl.c.
2016-01-21 09:31:39 -07:00
Jim Fehlig
7d3698b47c Xen: VIR_FROM_THIS cleanup
The virErrorDomain enum has VIR_FROM_XEN, VIR_FROM_XEND,
VIR_FROM_XENSTORE, VIR_FROM_SEXPR, and VIR_FROM_XENXM. Use
these elements in the corresponding .c files. While at it,
remove the VIR_FROM_THIS define in src/xenconfig/xenxs_private.h.
2016-01-21 09:31:39 -07:00
Jiri Denemark
56635345ad qemu: Add support for migration iteration event
The corresponding event in QEMU is called MIGRATION_PASS.

Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
2016-01-21 16:36:08 +01:00
Jiri Denemark
0b50f4a025 Introduce migration iteration event
The VIR_DOMAIN_EVENT_ID_MIGRATION_ITERATION event will be triggered
whenever VIR_DOMAIN_JOB_MEMORY_ITERATION changes its value, i.e.,
whenever a new iteration over guest memory pages is started during
migration.

Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
2016-01-21 16:36:08 +01:00
Dmitry Andreev
e2b86f580c qemuDomainReboot: use fakeReboot=true only for acpi mode
When acpi is used to reboot/shutdown qemu domain, qemu emits
SHUTDOWN event. Libvirt uses fakeReboot variable in order to
differentiate reboot or shutdown. fakeReboot value is reseted
to false after domain restart/reset.

When mode=agent is used to reboot qemu domain, qemu doesn't emit
SHUTDOWN event and libvirt doesn't reset fakeReboot value to false.
In this case next 'shutdown -h now' performs reboot. That's why
we don't need to set fakeReboot=true for mode=agent.

Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
2016-01-21 15:03:56 +01:00
Michal Privoznik
3bd2ee5d55 virsh: Don't fetch status for all domains in cmdList
We are getting the list of domains and after that we iterate over
the list and try to get status for each domain hoping it will
skip over domains that disappeared meanwhile. However, this
solution to race is bogus - domain may disappear right after we
have checked its state and before we exec another API over it
(e.g. virDomainHasManagedSaveImage()). Also, when printing just
names or uuids (list --name / --uuid) we issue APIs to obtain the
values, however these require no RPC call as all requested info
is in virDomain object that client already has.
Therefore move the status obtaining only to the place that really
needs it.

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2016-01-21 08:13:34 +01:00
Cole Robinson
a6cfd22eba docs: fix syntax-check long line error 2016-01-20 10:27:24 -05:00
Cole Robinson
a1edb05c60 build: predictably generate systemtap tapsets (bz 1173641)
The generated output is dependent on perl hashtable ordering, which
gives different results for i686 and x86_64. Fix this by sorting
the hash keys before iterating over them

https://bugzilla.redhat.com/show_bug.cgi?id=1173641
2016-01-20 10:26:02 -05:00
Daniel P. Berrange
7659bd9221 docs: fix generation of docs from VPATH build
When generating docs in a VPATH build we get a failure to
create a file due to the 'internals' subdir not existing:

  Generating internals/locking.html.tmp
  /bin/sh: line 3: internals/locking.html.tmp: No such file or directory
  rm: cannot remove ‘internals/locking.html.tmp’: No such file or directory
  Makefile:2229: recipe for target 'internals/locking.html.tmp' failed
  make: *** [internals/locking.html.tmp] Error 1

For some reason, make has decided to run the target

  %.html.tmp: %.html.in site.xsl page.xsl sitemap.html.in $(acl_generated)

instead of the target

  internals/%.html.tmp: internals/%.html.in subsite.xsl page.xsl sitemap.html.in

Removing '$(acl_generated)' from the first target, inexplicably
causes make to now run the correct target for the internals/
files.

Rather than figure this out, lets just combine the two targets
into one.

Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2016-01-20 12:22:19 +00:00
Ján Tomko
ce9085eba1 leaseshelper: reduce indentation level in virLeaseReadCustomLeaseFile
Instead of nested ifs, jump out early.

Mostly whitespace changes.
2016-01-20 10:01:52 +01:00