Peter Krempa [Wed, 28 Jun 2017 08:42:49 +0000 (10:42 +0200)]
qemu: hotplug: Disallow modification of vcpu 0 in inactive config
vcpu 0 must be always enabled and non-hotpluggable, thus you can't
modify it using the vcpu hotplug APIs. Disallow it so that users can't
create invalid configurations.
Lily Zhu [Wed, 28 Jun 2017 06:00:48 +0000 (14:00 +0800)]
man: Fix documentation errors about the paths of the config files
The default conf files, for example libvirtd.conf, virtlockd.conf, and
virtlogd.conf, should be located under the directory "/etc/libvirt" when
root as root, rather than "/etc". When run as non-root, the configuration
files should be located under "$XDG_CONFIG_HOME/libvirt/", rather than
"XDG_CONFIG_HOME".
Signed-off-by: Lily Zhu <lizhu@redhat.com> Signed-off-by: Erik Skultety <eskultet@redhat.com>
docs: Add callback-related info to virStream{Abort,Finish}
When one has a non-blocking stream and aborts or finishes it without
removing the callback, any event loop invocation will trigger that
callback, but it cannot be removed any more. We cannot remove the
callback automatically from virStream{Abort,Finish} functions due to
forward-compatibility. So let's at least document this behaviour,
because it is not easy to find out the reason for.
Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
Daniel Liu [Tue, 27 Jun 2017 15:54:50 +0000 (11:54 -0400)]
virsh: Fix --help problem for domxml-to-native DOMAIN COMMAND
Resolves a bug in domxml-to-native command option, so that the
following command displays the help information correctly:
'virsh domxml-to-native --help'.
Signed-off-by: Daniel Liu <srwx4096@gmail.com> Signed-off-by: Julio Faracco <jcfaracco@gmail.com>
error: Failed to start domain fedora25-aarch64
error: internal error: process exited while connecting to monitor:
2017-06-26T13:55:34.726293Z qemu-system-aarch64: -chardev pty,id=charserial0:
char device redirected to /dev/pts/5 (label charserial0)
2017-06-26T13:55:34.782121Z qemu-system-aarch64: -device
isa-serial,chardev=charserial0,id=serial0: No 'ISA' bus found for device
'isa-serial'
Andrea Bolognani [Mon, 26 Jun 2017 17:13:28 +0000 (19:13 +0200)]
conf: Copy loadparm in virDomainDeviceInfoCopy()
Commit 54fa1b44afc8 added virDomainDeviceInfo::loadparm
and updated virDomainDeviceInfoClear() accordingly, but
omitted the necessary virDomainDeviceInfoCopy() changes.
Signed-off-by: Andrea Bolognani <abologna@redhat.com>
Andrea Bolognani [Mon, 26 Jun 2017 06:45:44 +0000 (08:45 +0200)]
HACKING: Drop from the git repository
Despite being a generated file, HACKING has been tracked in
the git repository along with actual source files. As far as
I'm aware, it's the only generated file for which that happens.
Times and times again, people[1] have committed changes to
the source file without refreshing the generated copy at the
same time.
The rationale for tracking the generated file is to help out
people who just cloned the git repository looking to contribue;
however, README-hacking already contains enough information to
get perspective contributors to a place where they can simply
look at docs/hacking.html instead.
[1] Mostly me, to be honest
Signed-off-by: Andrea Bolognani <abologna@redhat.com>
Andrea Bolognani [Mon, 26 Jun 2017 08:48:20 +0000 (10:48 +0200)]
README: Remove 'git send-email' mention
The use of 'git send-email' is described in detail in the
contributor guidelines (docs/hacking.html), which are
mentioned in docs/contribute.html, the non-local version
of which is in turn mentioned in README.
Signed-off-by: Andrea Bolognani <abologna@redhat.com>
Jiri Denemark [Mon, 19 Jun 2017 15:00:28 +0000 (17:00 +0200)]
qemu: Avoid fd leak on incoming tunneled migration
While qemuProcessIncomingDefNew takes an fd argument and stores it in
qemuProcessIncomingDef structure, the caller is still responsible for
closing the file descriptor.
Andrea Bolognani [Thu, 15 Jun 2017 03:25:31 +0000 (11:25 +0800)]
hacking: Improve 'git send-email' documentation
For the benefit of first time contributors, we point out that 'git
send-email' might have to be installed separately; however, we omit
the fact that some configuration will likely be needed before it
can successfully deliver patches to the mailing list.
Some minor tweaks to the existing contents are included as well.
Signed-off-by: Andrea Bolognani <abologna@redhat.com> Reviewed-by: John Ferlan <jferlan@redhat.com>
John Ferlan [Wed, 14 Jun 2017 11:32:15 +0000 (07:32 -0400)]
events: Avoid double free possibility on remote call failure
If a remote call fails during event registration (more than likely from
a network failure or remote libvirtd restart timed just right), then when
calling the virObjectEventStateDeregisterID we don't want to call the
registered @freecb function because that breaks our contract that we
would only call it after succesfully returning. If the @freecb routine
were called, it could result in a double free from properly coded
applications that free their opaque data on failure to register, as seen
in the following details:
Program terminated with signal 6, Aborted.
#0 0x00007fc45cba15d7 in raise
#1 0x00007fc45cba2cc8 in abort
#2 0x00007fc45cbe12f7 in __libc_message
#3 0x00007fc45cbe86d3 in _int_free
#4 0x00007fc45d8d292c in PyDict_Fini
#5 0x00007fc45d94f46a in Py_Finalize
#6 0x00007fc45d960735 in Py_Main
#7 0x00007fc45cb8daf5 in __libc_start_main
#8 0x0000000000400721 in _start
The double dereference of 'pyobj_cbData' is triggered in the following way:
(1) libvirt_virConnectDomainEventRegisterAny is invoked.
(2) the event is successfully added to the event callback list
(virDomainEventStateRegisterClient in
remoteConnectDomainEventRegisterAny returns 1 which means ok).
(3) when function remoteConnectDomainEventRegisterAny is hit,
network connection disconnected coincidently (or libvirtd is
restarted) in the context of function 'call' then the connection
is lost and the function 'call' failed, the branch
virObjectEventStateDeregisterID is therefore taken.
(4) 'pyobj_conn' is dereferenced the 1st time in
libvirt_virConnectDomainEventFreeFunc.
(5) 'pyobj_cbData' (refered to pyobj_conn) is dereferenced the
2nd time in libvirt_virConnectDomainEventRegisterAny.
(6) the double free error is triggered.
Resolve this by adding a @doFreeCb boolean in order to avoid calling the
freeCb in virObjectEventStateDeregisterID for any remote call failure in
a remoteConnect*EventRegister* API. For remoteConnect*EventDeregister* calls,
the passed value would be true indicating they should run the freecb if it
exists; whereas, it's false for the remote call failure path.
Patch based on the investigation and initial patch posted by
fangying <fangying1@huawei.com>.
The function to check if -chardev is supported by QEMU was written a
long time ago, where adding chardevs did not make sense on the fixed ARM
platforms. Since then, we now have a general purpose virt platform,
which should support plugging in any device over PCIe which is supported
in a similar fashion on x86.
Signed-off-by: Christoffer Dall <cdall@linaro.org> Reviewed-by: Andrea Bolognani <abologna@redhat.com>
Daniel Liu [Fri, 2 Jun 2017 15:04:52 +0000 (11:04 -0400)]
virsh: add [--domain DOMAIN] option to domxml-to-native DOMAIN COMMAND
The option allows someone to run domain-to-native on already existing
domain without the need of supplying their XML. It is basically
wrapper around 'virsh dumpxml | virsh domxml-to-native /dev/stdin'.
Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=835476 Signed-off-by: Daniel Liu <srwx4096@gmail.com> Reviewed-by: Martin Kletzander <mkletzan@redhat.com> Reviewed-by: John Ferlan <jferlan@redhat.com>
As it turns out the volume create, build, and refresh path was not peeking
at the meta data, so immediately after a create operation the value displayed
for capacity was still incorrect. However, if a pool refresh was done the
correct value was fetched as a result of a meta data peek.
The reason is it seems historically if the file type is RAW then peeking
at the file just took the physical value for the capacity. However, since
we know if it's an encrypted file, then peeking at the meta data will be
required in order to get a true capacity value.
So check for encryption in the source and if present, use the meta data
in order to fill in the capacity value and set the payload_offset.
Peter Krempa [Wed, 21 Jun 2017 09:44:59 +0000 (11:44 +0200)]
Revert "qemu: Check duplicate WWNs also for hotplugged disks"
Similarly to commit 5da28cc3069b573f54f0bcaf8eb75476bcfdc6e9 this check
actually does not make sense since duplicate WWNs are used e.g. when
multipathing disks.
When fetching stats for a vhost-user type of interface, we run
couple of ovs-vsctl commands and parse their output. However, not
all stats exist at all times, for instance "rx_dropped" or
"tx_errors" can be missing. Thing is, we ask for a bulk of
statistics and if one of them is missing an error is reported
instead of returning the rest. Since we ignore errors, we fail to
set statistics. Fix this by asking for each piece alone.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Michal Privoznik [Thu, 22 Jun 2017 11:51:58 +0000 (13:51 +0200)]
security: Don't skip relabel for all chardevs
Our commit e13e8808f9 was way too generic. Currently, virtlogd is
used only for chardevs type of file and nothing else. True, we
must not relabel the path in this case, but we have to in all
other cases. For instance, if you want to have a physical console
attached to your guest:
Julio Faracco [Wed, 21 Jun 2017 17:08:29 +0000 (14:08 -0300)]
util: fix locale problem with virStrToDouble().
This commit fixes a locale problem with locales that use comma as a mantissa
separator. Example: 12.34 en_US = 12,34 pt_BR. Since strtod() is a non-safe
function, virStrToDouble() will have problems to parse double numbers from
kernel settings and other double numbers from static files (XMLs, JSONs, etc).
Julio Faracco [Wed, 21 Jun 2017 17:08:28 +0000 (14:08 -0300)]
util: moving virDoubleToStr() from virutil to virstring.
The function virDoubleToStr() is defined in virutil.* and virStrToDouble() is
defined in virstring.*. Joining both functions into the same file makes more
sense.
Signed-off-by: Julio Faracco <jcfaracco@gmail.com>
Jiri Denemark [Wed, 21 Jun 2017 13:31:38 +0000 (15:31 +0200)]
qemu: Do not skip virCPUUpdateLive if priv->origCPU is set
Even though we got both the original CPU (used for starting a domain)
and the updated version (the CPU really provided by QEMU) during
incoming migration, restore, or snapshot revert, we still need to update
the CPU according to the data we got from the freshly started QEMU.
Otherwise we don't know whether the CPU we got from QEMU matches the one
before migration. We just need to keep the original CPU in
priv->origCPU.
Michal Privoznik [Tue, 20 Jun 2017 15:55:21 +0000 (17:55 +0200)]
qemuProcessBuildDestroyHugepagesPath: Don't warn on destroying non-existent path
This function is called unconditionally from qemuProcessStop to
make sure we leave no dangling dirs behind. However, whenever the
directory we want to rmdir() is not there (e.g. because it hasn't
been created in the first place because domain doesn't use
hugepages at all), we produce a warning like this:
Peter Krempa [Tue, 20 Jun 2017 06:19:02 +0000 (08:19 +0200)]
qemu: snapshot: Load data necessary for relative block commit to work
Commit 7456c4f5f introduced a regression by not reloading the backing
chain of a disk after snapshot. The regression was caused as
src->relPath was not set and thus the block commit code could not
determine the relative path.
This patch adds code that will load the backing store string if
VIR_DOMAIN_SNAPSHOT_CREATE_REUSE_EXT and store it in the correct place
when a snapshot is successfully completed.
Peter Krempa [Mon, 19 Jun 2017 16:16:30 +0000 (18:16 +0200)]
storage: Add helper to retrieve the backing store string of a storage volume
It is necessary for some parts of the code to refresh just data
based on the based on the backing store string. Add a convenience
function that will retrieve this data.
Peter Krempa [Tue, 20 Jun 2017 07:06:35 +0000 (09:06 +0200)]
qemu: block commit: Determine relative path of images before initializing
Changing labelling of the images does not need to happen after setting
the labeling and lock manager access. This saves the cleanup of the
labeling if the relative path can't be determined.
Farhan Ali [Thu, 1 Jun 2017 16:36:25 +0000 (12:36 -0400)]
qemu: Add loadparm to qemu command line string
Check for the LOADPARM capabilility and potentially add a loadparm=x to
the "-machine" string for the QEMU command line.
Also add xml2argv test cases for loadparm.
Signed-off-by: Farhan Ali <alifm@linux.vnet.ibm.com> Reviewed-by: Bjoern Walk <bwalk@linux.vnet.ibm.com> Reviewed-by: Boris Fiuczynski <fiuczy@linux.vnet.ibm.com> Reviewed-by: Marc Hartmayer <mhartmay@linux.vnet.ibm.com>
Farhan Ali [Thu, 1 Jun 2017 16:36:24 +0000 (12:36 -0400)]
conf: Add loadparm boot option for a boot device
Update the per device boot schema to add an optional loadparm parameter.
eg: <boot order='1' loadparm='2'/>
Extend the virDomainDeviceInfo to support loadparm option.
Modify the appropriate functions to parse loadparm from boot device xml.
Add the xml2xml test to validate the field.
Signed-off-by: Farhan Ali <alifm@linux.vnet.ibm.com> Reviewed-by: Bjoern Walk <bwalk@linux.vnet.ibm.com> Reviewed-by: Boris Fiuczynski <fiuczy@linux.vnet.ibm.com> Reviewed-by: Marc Hartmayer <mhartmay@linux.vnet.ibm.com>
Peter Krempa [Mon, 19 Jun 2017 12:37:47 +0000 (14:37 +0200)]
util: storage: Add support for type 'inet' in virStorageSourceParseBackingJSONSocketAddress
'SocketAddress' structure was changed to contain 'inet' instead of
'tcp' since qemu commit c5f1ae3ae7b. Existing entries have a backward
compatibility layer.
Libvirt will parse 'inet' and 'tcp' as equivalents.
It was added in commit 6c2e4c3856c8ed48c378bf1bf357cab46271a47a
so that Coverity would not complain about passing -1 to
qemuDomainDetachThisHostDevice(), but the function in question
has changed since and so the annotation doesn't apply anymore.
Signed-off-by: Andrea Bolognani <abologna@redhat.com> Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Peter Krempa [Thu, 15 Jun 2017 15:23:15 +0000 (17:23 +0200)]
util: storage: make virStorageSourceParseBackingJSONGlusterHost universal
The same json strucutre is used for NBD and sheepdog volumes for
specifying of the host. Rename the function and fix up error messages to
be more universal.
When added in multiple previous commits, it was used only with -device
qxl(-vga), but for some QEMUs (< 1.6) we need to add this
functionality when using -vga qxl as well.
Pavel Hrdina [Thu, 15 Jun 2017 06:34:55 +0000 (08:34 +0200)]
qemu: introduce chardevStdioLogd to qemu private data
In QEMU driver we can use virtlogd as stdio handler for source backend
of char devices if current QEMU is new enough and it's enabled in
qemu.conf. We should store this information while starting a guest
because the config option may change while the guest is running.
Signed-off-by: Pavel Hrdina <phrdina@redhat.com> Reviewed-by: John Ferlan <jferlan@redhat.com>
Imagine a FS mounted on /dev/blah/blah2. Our process of creating
suffix for temporary location where all the mounted filesystems
are moved is very simplistic. We want:
/var/run/libvirt/qemu/$domName.$suffix\
were $suffix is just the mount point path stripped of the "/dev/"
prefix. For instance:
/var/run/libvirt/qemu/fedora.mqueue for /dev/mqueue
/var/run/libvirt/qemu/fedora.pts for /dev/pts
and so on. Now if we plug /dev/blah/blah2 into the example we see
some misbehaviour:
/var/run/libvirt/qemu/fedora.blah/blah2
Well, misbehaviour if /dev/blah/blah2 is a file, because in that
case we call virFileTouch() instead of virFileMakePath().
The solution is to replace all the slashes in the suffix with say
dots. That way we don't have to care about nested directories.
IOW, the result we want for given example is:
/var/run/libvirt/qemu/fedora.blah.blah2
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: John Ferlan <jferlan@redhat.com>
There can be nested mount points. For instance /dev/shm/blah can
be a mount point and /dev/shm too. It doesn't make much sense to
return the former path because callers preserve the latter (and
with that the former too). Therefore prune nested mount points.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: John Ferlan <jferlan@redhat.com>
After 290a00e41d we know how to deal with file mount points.
However, when cleaning up the temporary location for preserved
mount points we are still calling rmdir(). This won't fly for
files. We need to call unlink(). Now, since we don't really care
if the cleanup succeeded or not (it's the best effort anyway), we
can call both rmdir() and unlink() without need for
differentiation between files and directories.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: John Ferlan <jferlan@redhat.com>
On some platforms the number of bits in the cbm_mask might not be
divisible by 4 (and not even by 2), so we need to properly count the
bits. Similar file, min_cbm_bits, is properly parsed and used, but if
the number is greater than one, we lose the information about
granularity when reporting the data in capabilities. For that matter
always report granularity, but if it is not the same as the minimum,
add that information in there as well.
Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
Serge Hallyn [Tue, 23 May 2017 16:22:48 +0000 (18:22 +0200)]
apparmor, libvirt-qemu: Allow access to certificates used by libvirt-vnc
When setting up VncTLS according to the official Libvirt documentation,
only one certificate for libvirt/libvirt-vnc is used. The document
indicates to use the following directories :
Simon McVittie [Tue, 23 May 2017 16:22:39 +0000 (18:22 +0200)]
virt-aa-helper, apparmor: allow /usr/share/OVMF/ too
The split firmware and variables files introduced by
https://bugs.debian.org/764918 are in a different directory for
some reason. Let the virtual machine read both.
Signed-off-by: Christian Ehrhardt <christian.ehrhardt@canonical.com> Signed-off-by: Stefan Bader <stefan.bader@canonical.com>
GCC 7.1 introduces a new -Wformat-truncation warning
flag that reports if it thinks the maximum possible
size of the formatted output will exceed the provided
fixed buffer. This is enabled automatically by the
-Wformat warning flag. There are quite a few places
hit by this in libvirt which need rewriting. This is
non-trivial work in some places, so temporarily
disable the new warning until those fixes can be
implemented.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Depending on the platform/architecture, a number of conditionals
in libvirt code expand the same on both branches. This is expected
behaviour and harmless, so disable the warning to avoid creating
unexpected build failures
Two examples, mingw32:
../../src/util/vircommand.c: In function 'virCommandWait':
../../src/util/vircommand.c:2562:51: error: this condition has identical branches [-Werror=duplicated-branches]
*exitstatus = cmd->rawStatus ? status : WEXITSTATUS(status);
^
and gcc7.1
In file included from util/virobject.c:28:0:
util/virobject.c: In function 'virClassNew':
util/viratomic.h:176:46: error: this condition has identical branches [-Werror=duplicated-branches]
(void)(0 ? *(atomic) ^ *(atomic) : 0); \
^
util/virobject.c:144:20: note: in expansion of macro 'virAtomicIntInc'
klass->magic = virAtomicIntInc(&magicCounter);
^~~~~~~~~~~~~~~
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Jiri Denemark [Wed, 14 Jun 2017 12:56:21 +0000 (14:56 +0200)]
qemu: Use qemuDomainCheckABIStability where needed
Most places which want to check ABI stability for an active domain need
to call this API rather than the original
qemuDomainDefCheckABIStability. The only exception is in snapshots where
we need to decide what to do depending on the saved image data.
Jiri Denemark [Wed, 14 Jun 2017 11:43:37 +0000 (13:43 +0200)]
qemu: Add qemuDomainCheckABIStability
When making ABI stability checks for an active domain, we need to make
sure we use the same migratable definition which virDomainGetXMLDesc
with the MIGRATABLE flag provides, otherwise the ABI check will fail.
This is implemented in the new qemuDomainCheckABIStability which takes a
domain object and generates the right migratable definition from it.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com> Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
This patch separates the actual ABI checks from getting migratable defs
in qemuDomainDefCheckABIStability so that we can create another wrapper
which will use different methods to get the migratable defs.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com> Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
My fix 671d18594f4 was incomplete. If domain doesn't have
hugepages enabled, because of missing condition we would still be
putting hugepages path onto qemu cmd line. Clean up the
conditions so that it's more visible next time.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
With glibc >= 2.25.90 writev() is only available if you explicitly
include sys/uio.h. This matches the documented requirements, but
older glibc and other *NIX pulled in writev indirectly so the bug
wasn't noticed previously.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Erik Skultety [Tue, 13 Jun 2017 15:55:36 +0000 (17:55 +0200)]
qemu: monitor: Fix a memory leak in qemuMonitorJSONAttachCharDevCommand
With the current logic, we only free @tlsalias as part of the error
label and would have to free it explicitly earlier in the code. Convert
the error label to cleanup, so that we have only one sink, where we
handle all frees. Since JSON object append operation consumes pointers,
make sure @backend is cleared before we hit the cleanup label.
Signed-off-by: Erik Skultety <eskultet@redhat.com> Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Historically, we've always supported memAccess for domains backed
by hugepages. However, somewhere along the way we've regressed
and stopped allowing such configuration. Fix it.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
We have couple of hugepage enabled domains for qemuxml2argvtest.
Unfortunately, often when adding a test case there I forget to
add it to xml2xml test too.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
It may happen that a domain is started without any huge pages.
However, user might try to attach a DIMM module later. DIMM
backed by huge pages (why would somebody want to mix regular and
huge pages is beyond me). Therefore we have to create the dir if
we haven't done so far.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: John Ferlan <jferlan@redhat.com>
Currently, the per-domain path for huge pages mmap() for qemu is
created iff domain has memoryBacking and hugepages in it
configured. However, this alone is not enough because there can
be a DIMM module with hugepages configured too.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: John Ferlan <jferlan@redhat.com>
Right now, there is a lot of exit points from the function.
Depending on their position they need to copy the same free
calls. This goes against our style where we usually have just one
exit point from the function which also does the necessary free.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Jiri Denemark [Tue, 13 Jun 2017 11:25:07 +0000 (13:25 +0200)]
qemu: Fix starting a domain with corrupted managed save file
Commit v3.4.0-44-gac793bd71 fixed a memory leak, but failed to return
the special -3 value. Thus an attempt to start a domain with corrupted
managed save file would removed the corrupted file and report
"An error occurred, but the cause is unknown" instead of starting the
domain from scratch.
Marc Hartmayer [Wed, 7 Jun 2017 08:46:41 +0000 (10:46 +0200)]
Use ATTRIBUTE_FALLTHROUGH
Use ATTRIBUTE_FALLTHROUGH, introduced by commit 5d84f5961b8e28e802f600bb2d2c6903e219092e, instead of comments to
indicate that the fall through is an intentional behavior.
Signed-off-by: Marc Hartmayer <mhartmay@linux.vnet.ibm.com> Reviewed-by: Boris Fiuczynski <fiuczy@linux.vnet.ibm.com> Reviewed-by: Bjoern Walk <bwalk@linux.vnet.ibm.com>
projects / libvirt.git / log