Commit Graph

22903 Commits

Author SHA1 Message Date
Michal Privoznik
263a88806c virNetDevBridgeGet: Don't require users to virNetDevSetupControl
So far, this function has just three callers. Two of them call
virNetDevSetupControl to create a socket that we can then
optionally use for ioctl() to fetch data. However, querying sysfs
is preferred. Therefore it doesn't make much sense to require
users to set up the socket if they don't even know it will be
used in favour of sysfs. We can set up the socket iff we need to.

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2016-06-02 11:35:29 +02:00
Laine Stump
93b59fcff6 network: restart dnsmasq after adding/removing txt and srv records
Although dns host records are stored in a separate configuration file
that is reread by dnsmasq when it receives a SIGHUP, the txt and srv
records are directly in the dnsmasq .conf file which can't be reread
after initial dnsmasq startup. This means that if an srv or txt record
is modified in a network config, libvirt needs to restart the dnsmasq
process rather than just sending a SIGHUP.

This was pointed out in a question in
https://bugzilla.redhat.com/show_bug.cgi?id=988718 , but no separate
BZ was filed.
2016-06-01 11:45:25 -04:00
Pavel Hrdina
de0b091ae0 QXL: fix reloading of vram64 attribute
Commit b4a5fd95 introduced vram64 attribute for QXL video device but
there were two issues.  Only function
qemuMonitorJSONUpdateVideoVram64Size should update the vram64 attribute
and also the value is in MiB, not in B.

Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
2016-06-01 14:33:08 +02:00
Michal Privoznik
99809fd482 esxStorageVolGetXMLDesc: Lookup SCSI lun properly
So the idea is as follows: firstly we obtain a list of all the
luns, then iterate over it trying to find the one we want to work
with and after all the iterations we detect whether we have found
something. Now, the last check is broken, because it compares a
value form previous iteration, not the one we've just been
through.

Then, when computing md5 sum of lun's UUID, we use wrong variable
again. Well, @hostScsiDisk which is type of esxVI_HostScsiDisk
extends esxVI_ScsiLun type so they both have the uuid member, but
it just doesn't feel right to access the data via two different
variables in one function call.

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2016-05-31 13:36:25 +02:00
Michal Privoznik
c94720f86a qemuMonitorTextGetAllBlockStatsInfo: Fix line validation
There's a bug in the function. We expect the following format for
the data we are parsing here:

  key: value

So we use strchr() to find ':' and then see if it is followed by
space. But the check that does just that is slightly incorrect.

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2016-05-31 13:06:55 +02:00
Michal Privoznik
2bd61c8448 virSocketAddrIsPrivate: Work on 32bits platforms
Yet another one of those where signed int (or long int) is not
enough. And useless to as we're aiming at unsigned anyway.

../../src/util/virsocketaddr.c: In function 'virSocketAddrIsPrivate':
../../src/util/virsocketaddr.c:289:45: error: result of '192l << 24' requires 33 bits to represent, but 'long int' only has 32 bits [-Werror=shift-overflow=]
        return ((val & 0xFFFF0000) == ((192L << 24) + (168 << 16)) ||
                                             ^~
../../src/util/virsocketaddr.c:290:45: error: result of '172l << 24' requires 33 bits to represent, but 'long int' only has 32 bits [-Werror=shift-overflow=]
                (val & 0xFFF00000) == ((172L << 24) + (16  << 16)) ||
                                             ^~
cc1: all warnings being treated as errors

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2016-05-30 15:45:19 +02:00
Michal Privoznik
3864d86385 apibuild: Substitute only pure number tokens
In 38df47c9af I've tried to prepare our apibuild.py script for
change made in 0628f3498c (1U << 31). What I've done in the
former commit was to replace \d+U in parsed tokens with \d.
Problem was, my regular expression there was not quite right as
it also translated VIR_123U_VAL into VIR_123_VAL.

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2016-05-30 15:38:07 +02:00
Michal Privoznik
b099aa88e9 build: use gnulib's unsetenv
Now that gnulib has lifted it's licensing of unsetenv, we should
use it. Just like we use its counterpart - setenv, already.

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2016-05-28 23:13:26 +02:00
Michal Privoznik
0628f3498c Turn 1<<31 into 1U<<31
Apparently, 1 << 31 is signed which in turn does not fit into
a signed integer variable:

../../include/libvirt/libvirt-domain.h:1881:57: error: result of '1 << 31' requires 33 bits to represent, but 'int' only has 32 bits [-Werror=shift-overflow=]
     VIR_CONNECT_GET_ALL_DOMAINS_STATS_ENFORCE_STATS = 1 << 31, /* enforce requested stats */
                                                         ^~
cc1: all warnings being treated as errors

The solution is to make it an unsigned value. I've found only two
such occurrences in our code base.

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2016-05-28 13:49:14 +02:00
Michal Privoznik
38df47c9af docs: Teach apibuild to deal with (1U << 31) too
The apibuild script is a terrifying beast that parses some source
files of ours and produces an XML representation of them. When it
comes to parsing enums we have in some header files, it tries to
be clever and detect a value that an enum member has (or if it is
an alias for a different member). Whilst doing that it has to
deal with values we give to the members in many formats. At some
places we just pass the value in decimal:

    VIR_DOMAIN_BLOCK_JOB_TYPE_PULL = 1,

in other places, we use the aliasing:

    VIR_CONNECT_GET_ALL_DOMAINS_STATS_ACTIVE = VIR_CONNECT_LIST_DOMAINS_ACTIVE,

and in other places bitwise shifts are used:

    VIR_CONNECT_GET_ALL_DOMAINS_STATS_ENFORCE_STATS = 1 << 31, /* enforce requested stats */

The script tries to parse all of these resulting in the following
tokens: "1", "VIR_CONNECT_LIST_DOMAINS_ACTIVE", "1<<31"; Then, the
script tries to turn these into integers using python's eval()
function. This function succeeds on the first and the last
tokens. But, if we were to modify the last example so that it's
of the following form:

    VIR_CONNECT_GET_ALL_DOMAINS_STATS_ENFORCE_STATS = 1U << 31, /* enforce requested stats */

the token representing enum's member value will then be "1U<<31".
So our parsing is good. Unfortunately, python is not aware of the
difference between signed and unsigned C types, therefore eval()
fails over this token and the parser falls back thinking it's an
alias to another enum member. Well it's not.

The solution is to transform [0-9]U into [0-9] as for our
purposes here it's the same thing.

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2016-05-28 13:21:43 +02:00
Eric Blake
0ebd0b19d3 maint: update to latest gnulib
Fix a regression in checking for realpath (which caused link
failures regarding duplicate rpl_canonicalize_file_name), and
fix the mingw build regarding unsetenv.

* .gnulib: Update to latest.

Signed-off-by: Eric Blake <eblake@redhat.com>
2016-05-27 14:06:45 -06:00
Cole Robinson
e9ef4dfac8 spec: Advertise nvram paths of official fedora edk2 builds
Fedora now ships edk2 firmware in its official repos, so adapt
the nvram path list to match. Eventually we can remove the nightly
links as well once some integration kinks have been worked out,
and documentation updated.

Move the macro building into the %build target, which lets us
build up a shell variable and make things a bit more readable

https://bugzilla.redhat.com/show_bug.cgi?id=1335395
2016-05-27 12:53:50 -04:00
Katerina Koukiou
9b9d0f13d3 lxc: Fix virLXCDomainObjBeginJob position in lxcDomainSetMemoryParameters
Adjust the code to perform the virLXCDomainObjBeginJob first
and then the call virDomainLiveConfigHelperMethod.
As Ján Tomko pointed out, in virDomainLiveConfigHelperMethod,
there is a check to see if the domain is active when AFFECT_LIVE is set.
Since virLXCDomainObjBeginJob unlocks the virDomainObjPtr lock,
the domain could possibly be destroyed while we wait for the job
and the check results would no longer be valid.

Signed-off-by: Katerina Koukiou <k.koukiou@gmail.com>
2016-05-27 15:15:44 +02:00
Eric Blake
0cd64883dc maint: update to latest gnulib
Pulls in several portability fixes, including the fact that gnulib
now only works on platforms with two's complement signed integers.
Also makes for a smaller delta on the next update (we are waiting
on a license change to unsetenv for the sake of mingw).

* .gnulib: Update to latest.
* bootstrap: Resync from upstream.
* tests/virstringtest.c: Drop use of obsolete probes of integer
properties.

Signed-off-by: Eric Blake <eblake@redhat.com>
2016-05-26 10:47:03 -06:00
Dawid Zamirski
3ef5e218b1 esx: do not store escaped password in esxVI_Context.
This patch fixes an issue where screenshot API call was failing when
the esx/vcenter password contains special characters such as
apostrophee. The reason for failures was that passwords were escaped
for XML and stored in esxVI_Context which was then passed to raw CURL
API calls where the password must be passed in original form to
authenticate successfully. So this patch addresses this by storing
original passwords in the esxVI_Context struct and escape only for
esxVI_Login call.
2016-05-26 18:27:55 +02:00
Andrea Bolognani
c7289cf3b5 qemu: Fix error message when PCI bridge has index <= bus
Commit ff2126225d changed the error message to be more
detailed about the failure at hand; however, while the new
error message claims that "bus must be <= index", the error
message is displayed if "idx <= addr->bus", ie. when bus
is larger than or *equal to* index.

Change the error message to report the correct constraint,
and format it in a way that mirrors the check exactly to
make it clearer to people reading the code. The new error
message reads "index must be larger than bus".

Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=1339900
2016-05-26 16:24:04 +02:00
Nikolay Shirokovskiy
3ba93c754b daemon: cleanup state drivers in order reverse to init order
This patch aims to fix observed crash on daemon shutdown. Main thread is in
the process of state drivers cleanup, network driver is cleaned up and
qemu driver is not yet. Meanwhile eof event from qemu process triggers
qemuProcessStop -> networkReleaseActualDevice and crash happens as
network driver is already cleaned up.

Signed-off-by: Nikolay Shirokovskiy <nshirokovskiy@virtuozzo.com>
2016-05-26 08:25:40 -04:00
Dawid Zamirski
f8f7440870 esx: use newer virtualHW version for 5.1+ hosts
This is because there's a known issue where ESX will refuse to attach
drives bigger than 4TB when virtualHW < 9. Therefore, to avoid that
use the higher virtualHW for hosts that support it.

https://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=2054952
2016-05-26 11:47:38 +02:00
Dawid Zamirski
5b36410f37 esx: Add VMCI device for virtualHW >= 7
This patch fixes an issue where vMotion fails when VMCI device is not
present in the vmx file.
2016-05-26 11:17:06 +02:00
Dawid Zamirski
2b89f1d8f5 esx: add pciBridge devices when SCSI is used
When a SCSI controller is present, ESX adds several pciBridge devices
to vmx file. This fixes an error message where it refuses to create VM
due to not enough PCI devices available. This applies only to virtualHW
version >= 7.
2016-05-26 10:55:09 +02:00
Laine Stump
4d100c7a41 conf: permit auto-assignment of controller indexes
Hand-entering indexes for 20 PCI controllers is not as tedious as
manually determining and entering their PCI addresses, but it's still
annoying, and the algorithm for determining the proper index is
incredibly simple (in all cases except one) - just pick the lowest
unused index.

The one exception is USB2 controllers because multiple controllers in
the same group have the same index. For these we look to see if 1) the
most recently added USB controller is also a USB2 controller, and 2)
the group *that* controller belongs to doesn't yet have a controller
of the exact model we're just now adding - if both are true, the new
controller gets the same index, but in all other cases we just assign
the lowest unused index.

With this patch in place and combined with the automatic PCI address
assignment, we can define a PCIe switch with several ports like this:

  <controller type='pci' model='pcie-root-port'/>
  <controller type='pci' model='pcie-switch-upstream-port'/>
  <controller type='pci' model='pcie-switch-downstream-port'/>
  <controller type='pci' model='pcie-switch-downstream-port'/>
  <controller type='pci' model='pcie-switch-downstream-port'/>
  <controller type='pci' model='pcie-switch-downstream-port'/>
  <controller type='pci' model='pcie-switch-downstream-port'/>
  ...

These will each get a unique index, and PCI addresses that connect
them together appropriately with no pesky numbers required.
2016-05-25 15:00:25 -04:00
Laine Stump
808e16ff13 conf: make virDomainControllerFindUnusedIndex() more generally usable
Make virDomainControllerFindUnusedIndex() a global function so that it
can be used outside domain_conf.c (as well as higher up in
domain_conf.c itself)/ Also make its DomainDef arg a const* so that
functions which only have a const* to the domain can use it.
2016-05-25 15:00:25 -04:00
Laine Stump
1140b31f25 conf/qemu: make IS_USB2_CONTROLLER globally available
IS_USB2_CONTROLLER() is useful in more places aside from just when
assigning PCI addresses in QEMU, and is checking for enum values that
are all defined in conf/domain_conf.h anyway, so define it there
instead.
2016-05-25 15:00:25 -04:00
Chunyan Liu
ba56642885 libxl: add .domainInterfaceAddresses
Add .domainInterfaceAddresses so that user can have a way to
get domain interface address by 'virsh domifaddr'. Currently
it only supports '--source lease'.

Signed-off: Chunyan Liu <cyliu@suse.com>
2016-05-25 09:41:03 -06:00
Ján Tomko
5da23bbedf security: label the slic_table
Add support for the slic_table to the security drivers.
2016-05-25 17:15:21 +02:00
Ján Tomko
ea04d1a659 qemu: format SLIC ACPI table command line
<os>
  <acpi>
    <table type="slic">/path/to/acpi/table/file</table>
  </acpi>
</os>

will result in:

-acpitable sig=SLIC,file=/path/to/acpi/table/file

This option was introduced by QEMU commit 8a92ea2 in 2009.

https://bugzilla.redhat.com/show_bug.cgi?id=1327537
2016-05-25 17:15:21 +02:00
Ján Tomko
72f652da63 conf: add <acpi><table> to <os>
Add a new element to <domain> XML:
<os>
  <acpi>
    <table type="slic">/path/to/acpi/table/file</table>
  </acpi>
</os>

To supply a path to a SLIC (Software Licensing) ACPI
table blob.

https://bugzilla.redhat.com/show_bug.cgi?id=1327537
2016-05-25 17:15:21 +02:00
Pavel Hrdina
c81415065e qemucapstest: replace caps-1.6.50 with updated caps-1.7.0
The qemu-1.6.50 is a beta before the new minor version, let's replace it with
the release qemu-1.7.0.

Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
2016-05-25 17:04:22 +02:00
Peter Krempa
a5c70e58f2 qemu: Remove virDomainLiveConfigHelperMethod from qemuDomainSetSchedulerParametersFlags
This refactor also makes a distinction between the pointer to the
original definition and copied one to prevent mixups.
2016-05-25 16:59:58 +02:00
Peter Krempa
48dc930aab qemu: Remove virDomainLiveConfigHelperMethod from qemuDomainSetBlockIoTune 2016-05-25 16:59:58 +02:00
Peter Krempa
d314410fb8 qemu: Refactor qemuDomainGetSchedulerParametersFlags
Use virDomainCputune struct to store the data rather than exploding the
fields and use macros to fill the typed params.
2016-05-25 16:59:58 +02:00
Peter Krempa
e6e144689d conf: Change virDomainCputune member 'shares' to unsigned long long
cgroup functions set and get the longer type so use it everywhere
2016-05-25 16:59:58 +02:00
Peter Krempa
8bf8838fb4 qemu: Remove virDomainLiveConfigHelperMethod from qemuDomainGetSchedulerParametersFlags 2016-05-25 16:59:58 +02:00
Peter Krempa
9f50f6e288 qemu: Refactor qemuDomainGetBlkioParameters
Get rid of lots of duplicated code.
2016-05-25 16:59:58 +02:00
Peter Krempa
31b4d75877 qemu: Remove virDomainLiveConfigHelperMethod from qemuDomainGetBlkioParameters 2016-05-25 16:59:58 +02:00
Peter Krempa
b30387732a qemu: Remove virDomainLiveConfigHelperMethod from qemuDomainSetMemoryParameters 2016-05-25 16:59:58 +02:00
Peter Krempa
62a73bf631 qemu: Refactor typed params assignment in qemuDomainGetBlockIoTune
Introduce a macro to assign the parameters to avoid the for loop and
shuffle around various checks for a simpler and saner function.
2016-05-25 16:59:58 +02:00
Peter Krempa
6448356f27 qemu: Replace virDomainLiveConfigHelperMethod in qemuDomainGetBlockIoTune
Use virDomainObjGetDefs since the API guarantees that both live and
config are never set together.
2016-05-25 16:59:58 +02:00
Peter Krempa
2fde4e724e qemu: monitor: Remove 'supportMaxOptions' argument from qemuMonitorGetBlockIoThrottle
The caller is already aware that the params are missing and the
extractor is ignoring the missing ones so the parameter isn't necessary.
2016-05-25 16:59:58 +02:00
Dawid Zamirski
77298458d0 esx: use lsilogic adapter type in vol create.
ESX will refuse to attach VMDKS that have buslogic adatper type to 64bit
VMs whereas lsilogic works fine both 32bit and 64bit VMs.
2016-05-25 16:49:49 +02:00
Jim Fehlig
a1c9a81a31 libxl: default to qemu driver for network disks
Xen only supports network-based disks with the qemu (aka qdisk) driver.
Set the driverName to 'qemu' in libxlDomainDeviceDefPostParse() if
not already set. When starting a domain with network-based disks,
ensure the driverName is 'qemu'.

Resolves: https://bugzilla.opensuse.org/show_bug.cgi?id=981094
2016-05-25 08:28:22 -06:00
Peter Krempa
a09d9f5b7a qemu: driver: Allow disk update of startupPolicy/snapshot for all disks
The libvirt internal bits can be changed for disks that don't otherwise
support changing media. Remove the switch statement and allow changes of
non-source data for all disks.
2016-05-25 13:28:34 +02:00
Peter Krempa
e78794c95f qemu: driver: Move around code to avoid need to rollback
qemuDomainChangeDiskLive rolled back few changes to the disk definition
if changing of the media failed. This can be avoided by moving some code
around.
2016-05-25 13:22:36 +02:00
Shivaprasad G Bhat
cab28101ee Call qemuDomainObjEndJob when qemuCaps is null during hotplug
Signed-off-by: Shivaprasad G Bhat <sbhat@linux.vnet.ibm.com>
2016-05-25 13:15:21 +02:00
Shivaprasad G Bhat
a24cdf6cf7 Unref the cfg in qemuDomainAttachHostPCIDevice()
Signed-off-by: Shivaprasad G Bhat <sbhat@linux.vnet.ibm.com>
2016-05-25 12:24:28 +02:00
John Ferlan
cb0159df9f lxc: Fix lxcDomainDestroyFlags endjob processing
Commit id '15ccb0dbf' added job functions for the lxc driver; however,
for shutdown and nonpersistent path, the vm was removed from the domain
object list and the vm pointer cleared before the endjob.

Adjust the code to perform the endjob first and then perform the
ObjListRemove as long as the vm wasn't NULL. This follows more closely
models from qemu and libxl

Found by Coverity (FORWARD_NULL)
2016-05-25 06:02:42 -04:00
John Ferlan
fb06350021 qemu: Remove unused persistentAddrs
Based on some digital archaeology performed by jtomko, it's been determined
that the persistentAddrs variable is no longer necessary...

The variable was added by:
commit 141dea6bc7
CommitDate: 2010-02-12 17:25:52 +0000
    Add persistence of PCI addresses to QEMU

Where it was set to 0 on domain startup if qemu did not support the
QEMUD_CMD_FLAG_DEVICE capability, to clear the addresses at shutdown,
because QEMU might make up different ones next time.

As of commit f5dd58a608
CommitDate: 2012-07-11 11:19:05 +0200
    qemu: Extended qemuDomainAssignAddresses to be callable from
    everywhere.

this was broken, when the persistentAddrs = 0 assignment was moved
inside qemuDomainAssignPCIAddresses and while it pretends to check
for !QEMU_CAPS_DEVICE, its parent qemuDomainAssignAddresses is only
called if QEMU_CAPS_DEVICE is present.
2016-05-25 06:02:42 -04:00
John Ferlan
f30672d5ee qemu: Remove dead code
Since commit id '20a0fa8e' removed the QEMU_CAPS_DEVICE, Coverity notes
that it's no longer possible to have 'addrs' be NULL when checking for
a live domain since qemuDomainPCIAddressSetCreate would have jumped to
cleanup if addrs was NULL.
2016-05-25 06:02:42 -04:00
Andrea Bolognani
54f325e925 conf: nodedev: Set PCI_PHYSICAL_FUNCTION flag more carefully
Instead of setting the flag before parsing the PCI address, set
it afterwards. This ensure we can never end up in a situation
where the flag has been set but pci_dev.physical_function has
not been filled in.
2016-05-25 10:38:01 +02:00
Andrea Bolognani
fc9ba9d574 pci: Fix virPCIGetPhysicalFunction()'s callers
Commit c8b1a83605 changed the function, making it
impossible for callers to be able to tell whether a
non-negative return value means "physical function
address found and parsed correctly" or "couldn't find
corresponding physical function".

The important difference between the two being that,
in the latter case, the returned pointer is NULL and
should never, ever be dereferenced.

In order to cope with these changes, the callers
have to be updated.
2016-05-25 10:38:01 +02:00