Commit Graph

14063 Commits

Author SHA1 Message Date
Laine Stump
48a3f48ac5 qemu: add pcie-root controller
This controller is implicit on q35 machinetypes. It provides 31 PCIe
(*not* PCI) slots as controller 0.

Currently there are no devices that can connect to pcie-root, and no
implicit pci controller on a q35 machine, so q35 is still
unusable. For a usable q35 system, we need to add a
"dmi-to-pci-bridge" pci controller, which can connect to pcie-root,
and provides standard pci slots that can be used to connect other
devices.
2013-08-05 15:13:56 -04:00
Laine Stump
c305783c65 qemu: enable auto-allocate of all PCI addresses
Previous refactoring of the guest PCI address reservation/allocation
code allowed for slot types other than basic PCI (e.g. PCI express,
non-hotpluggable slots, etc) but would not auto-allocate a slot for a
device that required any type other than a basic hot-pluggable
PCI slot.

This patch refactors the code to be aware of different slot types
during auto-allocation of addresses as well - as long as there is an
empty slot of the required type, it will be found and used.

The piece that *wasn't* added is that we don't auto-create a new PCI
bus when needed for anything except basic PCI devices. This is because
there are multiple different types of controllers that can provide,
for example, a PCI express slot (in addition to the pcie-root
controller, these can also be found on a "root-port" or on a
"downstream-switch-port"). Since we currently don't support any PCIe
devices (except pending support for dmi-to-pci-bridge), we can defer
any decision on what to do about this.
2013-08-05 15:11:57 -04:00
Laine Stump
1ecbff4073 build: fix missing max_queued_clients in augeas test file for libvirtd.conf
Broken in commit 1199edb1d4
2013-08-05 15:09:52 -04:00
Jim Fehlig
0e671a1646 xen: fix memory corruption in legacy driver
Commit 632180d1 introduced memory corruption in xenDaemonListDefinedDomains
by starting to populate the names array at index -1, causing all sorts
of havoc in libvirtd such as aborts like the following

*** Error in `/usr/sbin/libvirtd': double free or corruption (out): 0x00007fffe00ccf20 ***
======= Backtrace: =========
/lib64/libc.so.6(+0x7abf6)[0x7ffff3fa0bf6]
/lib64/libc.so.6(+0x7b973)[0x7ffff3fa1973]
/lib64/libc.so.6(xdr_array+0xde)[0x7ffff403cbae]
/usr/sbin/libvirtd(+0x50251)[0x5555555a4251]
/lib64/libc.so.6(xdr_free+0x15)[0x7ffff403ccd5]
/usr/lib64/libvirt.so.0(+0x1fad34)[0x7ffff76b1d34]
/usr/lib64/libvirt.so.0(virNetServerProgramDispatch+0x1fc)[0x7ffff76b16f1]
/usr/lib64/libvirt.so.0(+0x1f214a)[0x7ffff76a914a]
/usr/lib64/libvirt.so.0(+0x1f222d)[0x7ffff76a922d]
/usr/lib64/libvirt.so.0(+0xbcc4f)[0x7ffff7573c4f]
/usr/lib64/libvirt.so.0(+0xbc5e5)[0x7ffff75735e5]
/lib64/libpthread.so.0(+0x7e0f)[0x7ffff48f7e0f]
/lib64/libc.so.6(clone+0x6d)[0x7ffff400e7dd]

Fix by initializing ret to 0 and only setting to error on failure path.
2013-08-05 10:38:52 -06:00
Michal Privoznik
1199edb1d4 Introduce max_queued_clients
This configuration knob lets user to set the length of queue of
connection requests waiting to be accept()-ed by the daemon. IOW, it
just controls the @backlog passed to listen:

  int listen(int sockfd, int backlog);
2013-08-05 11:03:01 +02:00
Michal Privoznik
2737aaafe9 RPC: Don't accept client if it would overcommit max_clients
Currently, even if max_client limit is hit, we accept() incoming
connection request, but close it immediately. This has disadvantage of
not using listen() queue. We should accept() only those clients we
know we can serve and let all other wait in the (limited) queue.
2013-08-05 10:25:01 +02:00
Laine Stump
3bb0125766 qemu: eliminate almost-duplicate code in qemu_command.c
* The functions qemuDomainPCIAddressReserveAddr and
qemuDomainPCIAddressReserveSlot were very similar (and should have
been more similar) and were about to get more code added to them which
would create even more duplicated code, so this patch gives
qemuDomainPCIAddressReserveAddr a "reserveEntireSlot" arg, then
replaces the body of qemuDomainPCIAddressReserveSlot with a call to
qemuDomainPCIAddressReserveAddr.

You will notice that addrs->lastaddr was previously set in
qemuDomainPCIAddressReserveAddr (but *not* set in
qemuDomainPCIAddressReserveSlot). For consistency and cleanliness of
code, that bit was removed and put into the one caller of
qemuDomainPCIAddressReserveAddr (there is a similar place where the
caller of qemuDomainPCIAddressReserveSlot sets lastaddr). This does
guarantee identical functionality to pre-patch code, but in practice
isn't really critical, because lastaddr is just keeping track of where
to start when looking for a free slot - if it isn't updated, we will
just start looking on a slot that's already occupied, then skip up to
one that isn't.

* qemuCollectPCIAddress was essentially doing the same thing as
qemuDomainPCIAddressReserveAddr, but with some extra special case
checking at the beginning. The duplicate code has been replaced with
a call to qemuDomainPCIAddressReserveAddr. This required adding a
"fromConfig" boolean, which is only used to change the log error
code from VIR_ERR_INTERNAL_ERROR (when the address was
auto-generated by libvirt) to VIR_ERR_XML_ERROR (when the address is
coming from the config); without this differentiation, it would be
difficult to tell if an error was caused by something wrong in
libvirt's auto-allocate code or just bad config.

* the bit of code in qemuDomainPCIAddressValidate that checks the
connect type flags is going to be used in a couple more places where
we don't need to also check the slot limits (because we're generating
the slot number ourselves), so that has been pulled out into a
separate qemuDomainPCIAddressFlagsCompatible function.
2013-08-03 15:42:20 -04:00
Laine Stump
29e3a1df47 qemu: rename some functions in qemu_command.c
* qemuDomainPCIAddressSetNextAddr

The name of this function was confusing because 1) other functions in
the file that end in "Addr" are only operating on a single function of
one PCI slot, not the entire slot, while functions that do something
with the entire slot end in "Slot", and 2) it didn't contain a verb
describing what it is doing (the "Set" refers to the set that contains
all PCI buses in the system, used to keep track of which slots in
which buses are already reserved for use).

It is now renamed to qemuDomainPCIAddressReserveNextSlot, which more
clearly describes what it is doing. Arguably, it could have been
changed to qemuDomainPCIAddressSetReserveNextSlot, but 1) the word
"set" is confusing in this context because it could be intended as a
verb or as a noun, and 2) most other functions that operate on a
single slot or address within this set are also named
qemuDomainPCIAddress... rather than qemuDomainPCIAddressSet... Only
the Create, Free, and Grow functions for an address set (which modify the
entire set, not just one element) use "Set" in their name.

* qemuPCIAddressAsString, qemuPCIAddressValidate

All the other functions in this set are named
qemuDomainPCIAddressxxxxx, so I renamed these to be consistent.
2013-08-03 15:40:38 -04:00
Laine Stump
c66da9d224 conf: add default USB controller in qemu post-parse callback
The parser shouldn't be doing arch-specific things like adding in
implicit controllers to the config. This should instead be done in the
hypervisor's post-parse callback.

This patch removes the auto-add of a usb controller from the domain
parser, and puts it into the qemu driver's post-parse callback (just
as is already done with the auto-add of the pci-root controller). In
the future, any machine/arch that shouldn't have a default usb
controller added should just set addDefaultUSB = false in this
function.

We've recently seen that q35 and ARMV7L domains shouldn't get a default USB
controller, so I've set addDefaultUSB to false for both of those.
2013-08-03 15:37:41 -04:00
Jiri Denemark
010065d702 spec: Explicitly claim ownership of channel subdir
As both /var/lib/libvirt/qemu and /var/lib/libvirt/qemu/channel/target
are owned by us, the intermediate /var/lib/libvirt/qemu/channel should
be owned by us too.
2013-08-02 17:22:33 +02:00
Daniel P. Berrange
1166eeba61 Fix crashing upgrading from older libvirts with running guests
If upgrading from a libvirt that is older than 1.0.5, we can
not assume that vm->def->resource is non-NULL. This bogus
assumption caused libvirtd to crash

Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-08-02 15:32:26 +01:00
Daniel P. Berrange
b4ca299902 Avoid crash if NULL is passed for filename/funcname in logging
The journald code would crash if a NULL was passed for the
filename / funcname in the logging code. This shouldn't
happen in general, but it is better to be safe, since there
have been bugs triggering this.

Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-08-02 15:32:26 +01:00
Daniel P. Berrange
cb3868f701 Ensure LXC/QEMU APIs set the filename for errors
The virLibConnError macros in libvirt-lxc.c and
libvirt-qemu.c were passing NULL for the filename.
This causes a crash if the logging code is configured
to use journald.

Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-08-02 15:32:26 +01:00
Daniel P. Berrange
311eae8257 Remove reference to python/tests from RPM %doc
The python/tests files were recently deleted, but a reference was
left in the RPM %doc entry

Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-08-02 16:29:43 +02:00
Peter Krempa
8f95d5d84c qemumonitortestutils: Don't skip va_end() on error path
When allocation of memory failed the error path didn't call va_end()
causing a coverity failure.

Reported by John Ferlan.
2013-08-02 12:28:58 +02:00
Peter Krempa
ae9ef97b9b qemuagenttest: Check invalid response in shutdown test
The shutdown test utilizes waiting for condition to exit the test. This
addition will return an error for the shutdown command to see if the
condition waiting code will not hang.
2013-08-02 12:28:58 +02:00
Peter Krempa
baebb34120 qemuagenttest: Fix checking of shutdown mode
Coverity complained about unused variable that contains the shutdown
mode. The original intention was to check it against the requested mode.

Also the fixed check revealed a mistake in the expected shutdown mode.

Reported by John Ferlan.
2013-08-02 12:28:50 +02:00
Roman Bogorodskiy
4ac708f250 bridge driver: extract platform specifics
* Move platform specific things (e.g. firewalling and route
  collision checks) into bridge_driver_platform
* Create two platform specific implementations:
    - bridge_driver_linux: Linux implementation using iptables,
      it's actually the code moved from bridge_driver.c
    - bridge_driver_nop: dumb implementation that does nothing

Signed-off-by: Eric Blake <eblake@redhat.com>
2013-08-01 15:47:02 -06:00
John Ferlan
6e69166ef1 valgrind: Adjust filter for _dl_allocate_tls
More tests are now using the path - adjust the filter to include any
path from a test through pthread_create to _dl_allocate_tls
2013-08-01 17:04:28 -04:00
John Ferlan
3b4b121423 tests: Coverity found new NULL_RETURNS
Coverity reported the existing missing check of the return value and
subsequent use from a call to virJSONValueFromString() in testJSONAddRemove().
2013-08-01 17:04:28 -04:00
Jincheng Miao
a320730a88 build: fix qemuagenttest build with -O0 in fedora 19.
When building libvirt with -O0 flag in fedora 19, it will fail to
generate qemuagenttest, a link error occurs like:

./.libs/libqemumonitortestutils.a(qemumonitortestutils.o): In function `qemuMonitorTestFree':
libvirt/tests/qemumonitortestutils.c:346: undefined reference to `qemuMonitorClose'
./.libs/libqemumonitortestutils.a(qemumonitortestutils.o): In function `qemuMonitorTestNew':
libvirt/tests/qemumonitortestutils.c:870: undefined reference to `qemuMonitorOpen'
collect2: error: ld returned 1 exit status

Fix it by listing libraries in the correct order to avoid lazy linkage.

Signed-off-by: Eric Blake <eblake@redhat.com>
2013-08-01 14:07:03 -06:00
Eric Blake
13c573740b maint: use modern autoconf idioms
Autoconf 2.59 says that AC_OUTPUT with arguments is obsolete,
and we are already using the replacement for some, but not all,
of our output files.

* configure.ac (AC_OUTPUT): Rewrite to use AC_CONFIG_FILES.

Signed-off-by: Eric Blake <eblake@redhat.com>
2013-08-01 07:31:53 -06:00
Guannan Ren
dbca841457 qemu: check presence of each disk and its backing file as well
For disk with startupPolicy support, such as cdrom and floppy
when its chain is broken, the startup policy will apply,
otherwise, report an error.
2013-08-01 13:26:47 +08:00
Guannan Ren
d7b7aa2c20 qemu: add helper functions for diskchain checking
*src/util/virstoragefile.c: Add a helper function to get
the first name of missing backing files, if the name is NULL,
it means the diskchain is not broken.
*src/qemu/qemu_domain.c: qemuDiskChainCheckBroken(disk) to
check if its chain is broken
2013-08-01 13:26:27 +08:00
Guannan Ren
e2ccc96cf0 qemu: refactor qemuDomainCheckDiskPresence for only disk presence check
Refactor this function to make it focus on disk presence checking,
including diskchain checking, and not only for CDROM and Floppy.
This change is good for the following patches.
2013-08-01 13:26:20 +08:00
Mooli Tayer
6f4a0ebe8c Configuring systemd to restart libvirt on failure
This will create a respawn behaviour in case libvirt
process exits due to nonzero exit code, is terminated
by a signal, an operation times out or the configured
watchdog timeout is triggered.
see http://www.freedesktop.org/software/systemd/man/systemd.service.html
and https://bugzilla.redhat.com/show_bug.cgi?id=981974

Signed-off-by: Eric Blake <eblake@redhat.com>
2013-07-31 16:55:27 -06:00
Stefan Bader
9d0557b965 xen: Avoid double free of virDomainDef in xenDaemonCreateXML
The virDomainDef is allocated by the caller and also used after
calling to xenDaemonCreateXML. So it must not get freed by the
callee.

Signed-off-by: Stefan Bader <stefan.bader@canonical.com>
2013-07-31 16:15:56 -06:00
Daniel P. Berrange
2fe2470181 Enable support for systemd-machined in cgroups creation
Make the virCgroupNewMachine method try to use systemd-machined
first. If that fails, then fallback to using the traditional
cgroup setup code path.

Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-07-31 19:29:19 +01:00
Daniel P. Berrange
75304eaa1a Cope with races while killing processes
When systemd is involved in managing processes, it may start
killing off & tearing down croups associated with the process
while we're still doing virCgroupKillPainfully. We must
explicitly check for ENOENT and treat it as if we had finished
killing processes

Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-07-31 19:27:28 +01:00
Daniel P. Berrange
aedd46e7e3 Add support for systemd cgroup mount
Systemd uses a named cgroup mount for tracking processes. Add
it as another type of controller, albeit one which we have to
special case in a number of places. In particular we must
never create/delete directories there, nor add tasks. Essentially
the systemd mount is to be considered read-only for libvirt.

With this change both the virCgroupDetectPlacement and
virCgroupCopyPlacement methods must be invoked. The copy
placement method will copy setup for resource controllers
only. The detect placement method will probe for any
named controllers, or resource controllers not already
setup.

Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-07-31 19:27:19 +01:00
Daniel P. Berrange
4574b475df Add APIs for formatting systemd slice/scope names
There are some interesting escaping rules to consider when dealing
with systemd slice/scope names. Thus it is helpful to have APIs
for formatting names

Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
2013-07-31 19:24:53 +01:00
Peter Krempa
2d95e8b100 qemuagenttest: Add tests for CPU plug functions and helpers 2013-07-31 14:45:46 +02:00
Peter Krempa
fea2590a02 qemuagenttest: Introduce testing of shutdown commands
This patch exports a few utility functions and adds testing of shutdown
commands of the guest agent.
2013-07-31 14:25:43 +02:00
Peter Krempa
f3a9391bec qemuagenttest: Add testing of agent suspend modes 2013-07-31 14:25:43 +02:00
Peter Krempa
725ee3b1d5 qemuagenttest: Test the filesystem trimming 2013-07-31 14:25:43 +02:00
Peter Krempa
5c94dfdd76 tests: Add qemuagenttest
Add a basic test framework with two simple tests to test guest agent
interaction.
2013-07-31 14:25:43 +02:00
Peter Krempa
e32f6e778b qemumonitortestutils: Add the ability to check arguments of commands
This patch adds helpers that allow to check for argument values in
commands sent to the monitor.
2013-07-31 14:25:43 +02:00
Peter Krempa
69441df4d0 qemumonitortestutils: Improve error reporting from mock qemu monitor
Use the JSON error messages to report errors back to the caller in
addition to erroring out. The error reported from the event loop from
the mock function of the monitor was later overwritten by the call to
the monitor/agent interaction API. This will also allow testing of error
reporting.
2013-07-31 14:25:43 +02:00
Peter Krempa
659fe6d2db qemumonitortestutils: Add instrumentation for guest agent testing
Add helper functions to open guest agent connections and a handler for
replying to the "guest-sync" command.
2013-07-31 14:25:43 +02:00
Peter Krempa
f6bb917e6a qemumonitortestutils: Split lines on \n instead of \r\n
The normal monitor uses windows line endings, where the agent monitor
uses only newlines. Change this to tolerate both approaches and allow to
use the utilities for guest agent tests.
2013-07-31 14:25:43 +02:00
Peter Krempa
5467cbef4b qemumonitortestutils: Refactor the test helpers to allow reuse
Refactor the test helpers to allow adding callbacks to verify the
monitor responses instead of simple command name checking and clean up
various parts to prepare for adding guest agent tests.
2013-07-31 14:25:43 +02:00
Peter Krempa
3383480430 qemumonitortestutils: Split up creation of the test to allow reuse
The instrumentation for the monitor test can be hacked for qemu agent
testing. Split out the monitor specific stuff to allow using the code in
guest agent tests in the future.
2013-07-31 14:25:43 +02:00
Peter Krempa
a63a7a5af9 qemumonitortestutils: Don't crash on non fully initialized test
The qemumonitorjsontest crashed when one of the initialization steps
done before starting the worker thread failed. This patch fixes this by
trying to pthread_join() the thread only after it was created.
2013-07-31 14:25:43 +02:00
Peter Krempa
5bf61f3818 qemumonitortestutils: remove multiline function calls 2013-07-31 14:25:43 +02:00
Peter Krempa
65f3f9dd78 qemumonitortestutils: Use VIR_DELETE_ELEMENT and VIR_APPEND_ELEMENT
Simplify the code using the existing helpers instead of open coding the
same functionality.
2013-07-31 14:25:43 +02:00
Peter Krempa
d24dbdd0a2 qemumonitortestutils: Use consistent header style and line spacing 2013-07-31 14:25:43 +02:00
Peter Krempa
56874f0191 qemu_agent: Remove obvious comments
Most APIs in libvirt report errors, thus no need to state that
explicitly.
2013-07-31 14:25:43 +02:00
Peter Krempa
18a629167c qemu_agent: Move updater function for VCPU hotplug into qemu_agent.c
To allow testing of the cpu updater function, this function needs to be
available separately. Export it from qemu_agent.c where it should
belong.
2013-07-31 14:25:43 +02:00
Peter Krempa
2e38753fb0 qemu_agent: Output newline at the end of the sync JSON message
Although this isn't apparently needed for the guest agent itself, the
test I will be adding later depends on the newline as a separator of
messages to process.
2013-07-31 14:25:43 +02:00
Peter Krempa
cc4df6fd5b conf: Export virDomainChrSourceDefClear() 2013-07-31 14:25:43 +02:00