Commit Graph

24709 Commits

Author SHA1 Message Date
Andrea Bolognani
0a3f919972 NEWS: Start using the improved format
This entry is meant to both get the ball rolling on the
switch and to provide a blueprint of what NEWS file entries
are supposed to look like.
2016-11-21 15:35:04 +01:00
Peter Krempa
4fa7ba0b32 qemu: process: Set current vcpu count to maximum if it was not specified
Mimic qemu's behavior on the given command line.
2016-11-21 14:35:20 +01:00
Peter Krempa
d3734b7a1d qemu: parse: Assign maximum cpu count from topology if not provided
qemu uses this if 'maxcpus' is not present. Do the same in the parsing
code.
2016-11-21 14:35:20 +01:00
Peter Krempa
0d9a76de6d qemu: parse: Assign topology info earlier
Qemu can also use the topology to calculate the total vcpu count. To
allow parsing this move the assignment earlier.
2016-11-21 14:35:20 +01:00
Peter Krempa
d78a8c26c2 qemu: parse: Allow the 'cpus=' prefix for current cpu number
qemu allows following syntax:

  -smp [cpus=]n[,cores=cores][,threads=threads][,sockets=sockets][,maxcpus=maxcpus]

Allow the "cpus" prefix.
2016-11-21 14:35:20 +01:00
Peter Krempa
4d72d80665 qemu: parse: Validate that the VM has at least one cpu
Libvirt's code relies on this fact so don't allow parsing a command line
which would have none.

Libvirtd would crash in the post parse callback on such config.
2016-11-21 14:35:20 +01:00
Martin Kletzander
d29ba605af docs: Use Overpass Mono as the monospace font
Because this makes the font the same size as the non-monospaced one,
let's change those properties a little bit as well.  <pre/> will be
shrunk a bit and <code/> inside <p/> or <dd/> will have slightly greyer
background.

Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
2016-11-21 14:04:05 +01:00
Martin Kletzander
63f20e6677 docs: Add monospaced Overpass fonts
Overpass 3.0 has monospaced fonts, so why not have the same font for
the monospaced parts of the documentation.

Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
2016-11-21 13:15:12 +01:00
Martin Kletzander
9d936f12db docs: Upgrade Overpass fonts to 3.0
Since we are using Overpass for the web pages, we might be using the
latest version.

Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
2016-11-21 13:15:12 +01:00
Martin Kletzander
b458de9b9e docs: Clean-up font definitions
Some of those were duplicate, so remove those.  In order to better see
such things in the future, sort them so they are "regular, italic,
bold, bold-italic, light, light-italic".

Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
2016-11-21 13:15:12 +01:00
Michal Privoznik
0c1bfd2c8d tests: Adapt to gluster_debug_level in qemu.conf
After a944bd92 we gained support for setting gluster debug level.
However, due to a space we haven't tested whether augeas file
actually works.

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
2016-11-21 10:50:48 +01:00
Martin Kletzander
8158a19fd7 util: Print pid_t as long long
After commit f2bf5fbb04, MinGW strikes again.  Simply print pid as any
other place after commit b7d2d4af2b.

Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
2016-11-20 21:46:21 +01:00
Guido Günther
112b095944 configure: Look for daemons in **/sbin
Unify the logic we use for looking up daemons and admin binaries. Some
lookups prefered $PATH over **/sbin while others left out $PATH
entierly.  We add **/sbin since non-root users might not have these in
their path.

This also unbreaks libvirt when built on Debian systems with usrmerge[0]
and run on systems without it.

[0]: https://packages.debian.org/sid/usrmerge
2016-11-19 17:46:52 +01:00
Martin Kletzander
f2bf5fbb04 Fix scheduler support check
Commit 94cc577807 tried fixing build on systems that did not have
SCHED_BATCH or SCHED_IDLE defined.  But instead of changing it to
conditional support, it rather completely disabled the support for
setting any scheduler.  Since then, such old systems are not
supported, but rather than reverting that commit, let's change that to
the conditional support.  That way any addition to the list of
schedulers can follow the same style so that we're consistent in the
future.

Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
2016-11-18 16:08:52 +01:00
John Ferlan
135e77d32f fs: Add proper switch to create filesystem with overwrite
https://bugzilla.redhat.com/show_bug.cgi?id=1366460

When using the --overwrite switch on a pool-build or pool-create, the
The mkfs.ext{2|3|4} commands use mke2fs which requires using the '-F' switch
in order to force overwriting the current filesystem on the whole disk.

Likewise, the mkfs.vfat command uses mkfs.fat which requires using the '-I'
switch in order to force overwriting the current filesystem on the whole disk.
2016-11-16 06:52:35 -05:00
Jiri Denemark
29e17f65a8 Avoid compiler warnings in virCPUDefStealModel
Old GCC on CentOS 6 thinks vendor and vendor_id might be used
uninitialized in virCPUDefStealModel. The compiler is wrong, though.

Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
2016-11-16 09:04:51 +01:00
Eric Blake
85fdcd5520 build: update to latest gnulib
In particular, this is a quick gnulib fix for a mingw compile error.

* .gnulib: Update to latest.

Signed-off-by: Eric Blake <eblake@redhat.com>
2016-11-15 10:26:58 -06:00
Roman Bogorodskiy
a6b81d5581 bhyve: fix memory leaks in bhyvexml2argvtest
* virNetDevTapCreateInBridgePort() mock: free '*ifname' before
   strdupping a hardoded value to it
 * testCompareXMLToArgvFiles(): unref 'conn' object in cleanup
 * testCompareXMLToArgvHelper(): free 'ldargs' and 'dmargs' in
   cleanup
2016-11-15 19:07:48 +03:00
Pino Toscano
9b39a9dcf4 docs: fix default value for sshauth option of libssh2/libssh
Both transports include "password" in their default authentication
methods.
2016-11-15 15:50:51 +01:00
Pino Toscano
e7a7807bd6 spec: enable libssh transport on Fedora 2016-11-15 15:50:51 +01:00
Pino Toscano
22eaee8e01 remote: expose a new libssh transport
Implement in virtNetClient and VirNetSocket the needed functions to
expose a new libssh transport, providing all the options that the
libssh2 transport supports.
2016-11-15 15:50:51 +01:00
Pino Toscano
6917467c2b libssh_transport: add new libssh-based transport
Implement a new libssh transport, which uses libssh to communicate with
remote hosts, and add all the build system stuff (search of libssh,
private symbols, etc) to built it.

This new transport supports all the common ssh authentication methods,
making use of libvirt's auth callbacks for interaction with the user.
2016-11-15 15:50:51 +01:00
Pino Toscano
24ee5dc907 virnetsocket: improve search for default SSH key
Add a couple of helper functions to check whether one of the default
names of SSH keys (as documented in ssh-keygen(1)) exists, and use them
to specify a key for the libssh2 transport if none was passed.
2016-11-15 15:50:51 +01:00
Pino Toscano
f0e7f90bff virerror: add error for libssh transport
Add a new error domain and number for a new libssh-based transport.
2016-11-15 15:50:51 +01:00
Pino Toscano
0e9fec979d virNetSocket: allow to not close FD
Add an internal variable to mark the FD as "not owned" by the
virNetSocket, in case the internal implementation takes the actual
ownership of the descriptor; this avoids a warning when closing the
socket, as the FD would be invalid.
2016-11-15 15:50:51 +01:00
Jiri Denemark
03fa904c0c cpu: Drop cpuGuestData
The API is not used anywhere in the code.

Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
2016-11-15 15:49:16 +01:00
Jiri Denemark
98b7c37d37 cpu: Avoid adding <vendor> to custom CPUs
Guest CPU definitions with mode='custom' and missing <vendor> are
expected to run on a host CPU from any vendor as long as the required
CPU model can be used as a guest CPU on the host. But even though no CPU
vendor was explicitly requested we would sometimes force it due to a bug
in virCPUUpdate and virCPUTranslate.

The bug would effectively forbid cross vendor migrations even if they
were previously working just fine.

Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
2016-11-15 15:49:16 +01:00
Jiri Denemark
d73422c186 cpu: Introduce virCPUConvertLegacy API
PPC driver needs to convert POWERx_v* legacy CPU model names into POWERx
to maintain backward compatibility with existing domains. This patch
adds a new step into the guest CPU configuration work flow which CPU
drivers can use to convert legacy CPU definitions.

Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
2016-11-15 15:49:16 +01:00
Jiri Denemark
509a4a40f9 cputest: Don't test cpuGuestData
The API is no longer used anywhere else since it was replaced by a much
saner work flow utilizing new APIs that work on virCPUDefPtr directly:
virCPUCompare, virCPUUpdate, and virCPUTranslate.

Not testing the new work flow caused some bugs to be hidden. This patch
reveals them, but doesn't attempt to fix them. To make sure all test
still pass after this patch, all affected test results are modified to
pretend the tests succeeded. All of the bugs will be fixed in the
following commits and the artificial modifications will be reverted.

The following is the list of bugs in the new CPU model work flow:

- a guest CPU with mode='custom' and missing <vendor/> gets the vendor
  copied from host's CPU (the vendor should only be copied to host-model
  CPUs):
    DO_TEST_UPDATE("x86", "host", "min", VIR_CPU_COMPARE_IDENTICAL)
    DO_TEST_UPDATE("x86", "host", "pentium3", VIR_CPU_COMPARE_IDENTICAL)
    DO_TEST_GUESTCPU("x86", "host-better", "pentium3", NULL, 0)

- when a guest CPU with mode='custom' needs to be translated into
  another model because the original model is not supported by a
  hypervisor, the result will have its vendor set to the vendor of the
  original CPU model as specified in cpu_map.xml even if the original
  guest CPU XML didn't contain <vendor/>:
    DO_TEST_GUESTCPU("x86", "host", "guest", model486, 0)
    DO_TEST_GUESTCPU("x86", "host", "guest", models, 0)
    DO_TEST_GUESTCPU("x86", "host-Haswell-noTSX", "Haswell-noTSX",
                     haswell, 0)

- legacy POWERx_v* model names are not recognized:
    DO_TEST_GUESTCPU("ppc64", "host", "guest-legacy", ppc_models, 0)

Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
2016-11-15 15:49:16 +01:00
Jiri Denemark
2a2ce08a6d cpu: Make models array in virCPUTranslate constant
The API doesn't change the array so let's make it constant.

Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
2016-11-15 15:49:16 +01:00
Jiri Denemark
adf44c7bae cputest: Don't use preferred CPU model at all
Now that all tests pass NULL as the preferred model, we can just drop
that test parameter.

Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
2016-11-15 15:49:16 +01:00
Jiri Denemark
6384277611 cputest: Don't use superfluous preferred model
In some cases preferred model doesn't really do anything since the
result remains the same even if it is removed.

Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
2016-11-15 15:49:16 +01:00
Jiri Denemark
797bdcedc9 cputest: Don't use preferred model with forbidden fallback
Using a preferred model for guest CPUs with forbidden fallback masks a
bug in the code. It would just happily use another CPU model supported
by a hypervisor even though it is explicitly forbidden in the CPU XML.

This patch temporarily changes the expected result to -2, which is used
when the result XML file cannot be found (but it was supposed not to be
found since the tested API should have failed). The result will be
switched back to -1 few commits later when the original bug gets fixed.

Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
2016-11-15 15:49:16 +01:00
Jiri Denemark
f60c5e4eb6 cputest: Don't use unsupported preferred model
Using a preferred CPU model which is not in the list of CPU models
supported by a hypervisor does not make sense.

Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
2016-11-15 15:49:16 +01:00
Jiri Denemark
43a9427046 cputest: Don't use preferred model for minimum match CPUs
Guest CPUs with match='minimum' should always be updated to match host
CPU model. Trying to get different results by supplying preferred models
does not make sense.

Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
2016-11-15 15:49:16 +01:00
Jiri Denemark
53a5986ad6 cpu: Rename cpuDataFormat
The new name is virCPUDataFormat.

Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
2016-11-15 15:49:15 +01:00
Jiri Denemark
be57e68954 cpu: Rename cpuDataParse
The new name is virCPUDataParse.

Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
2016-11-15 15:49:15 +01:00
Jiri Denemark
938ec1620a cpu: Rename and document cpuModelIsAllowed
The new name is virCPUModelIsAllowed.

Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
2016-11-15 15:49:15 +01:00
Jiri Denemark
b7011dfe44 cpu: Rename cpuGetModels
The new name is virCPUGetModels.

Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
2016-11-15 15:49:15 +01:00
Maxim Nestratov
007fb4388f qemu: fix libvirtd crash when querying halted cpus info
It was introduced by commit 7a51d9ebb, which started to use
monitor commands without job acquiring, which is unsafe and leads
to simultaneous access to vm->mon structure by different threads.

Crash backtrace is the following (shortened):

Program received signal SIGSEGV, Segmentation fault.
qemuMonitorSend (mon=mon@entry=0x7f4ef4000d20, msg=msg@entry=0x7f4f18e78640) at qemu/qemu_monitor.c:1011
1011        while (!mon->msg->finished) {

0  qemuMonitorSend () at qemu/qemu_monitor.c:1011
1  0x00007f691abdc720 in qemuMonitorJSONCommandWithFd () at qemu/qemu_monitor_json.c:298
2  0x00007f691abde64a in qemuMonitorJSONCommand at qemu/qemu_monitor_json.c:328
3  qemuMonitorJSONQueryCPUs at qemu/qemu_monitor_json.c:1408
4  0x00007f691abcaebd in qemuMonitorGetCPUInfo g@entry=false) at qemu/qemu_monitor.c:1931
5  0x00007f691ab96863 in qemuDomainRefreshVcpuHalted at qemu/qemu_domain.c:6309
6  0x00007f691ac0af99 in qemuDomainGetStatsVcpu at qemu/qemu_driver.c:18945
7  0x00007f691abef921 in qemuDomainGetStats  at qemu/qemu_driver.c:19469
8  qemuConnectGetAllDomainStats at qemu/qemu_driver.c:19559
9  0x00007f693382e806 in virConnectGetAllDomainStats at libvirt-domain.c:11546
10 0x00007f6934470c40 in remoteDispatchConnectGetAllDomainStats at remote.c:6267

(gdb) p mon->msg
$1 = (qemuMonitorMessagePtr) 0x0

This change fixes it by calling qemuDomainRefreshVcpuHalted only when job is acquired.

Signed-off-by: Maxim Nestratov <mnestratov@virtuozzo.com>
2016-11-15 17:39:24 +03:00
Laine Stump
70d15c9ac6 qemu: initially reserve one open pcie-root-port for hotplug
For machinetypes with a pci-root bus (all legacy PCI), libvirt will
make a "fake" reservation for one extra slot prior to assigning
addresses to unaddressed PCI endpoint devices in the domain. This will
trigger auto-adding of a pci-bridge for the final device to be
assigned an address *if that device would have otherwise instead been
the last device on the last available pci-bridge*; thus it assures
that there will always be at least one slot left open in the domain's
bus topology for expansion (which is important both for hotplug (since
a new pci-bridge can't be added while the guest is running) as well as
for offline additions to the config (since adding a new device might
otherwise in some cases require re-addressing existing devices, which
we want to avoid)).

It's important to note that for the above case (legacy PCI), we must
check for the special case of all slots on all buses being occupied
*prior to assigning any addresses*, and avoid attempting to reserve
the extra address in that case, because there is no free address in
the existing topology, so no place to auto-add a pci-bridge for
expansion (i.e. it would always fail anyway). Since that condition can
only be reached by manual intervention, this is acceptable.

For machinetypes with pcie-root (Q35, aarch64 virt), libvirt's
methodology for automatically expanding the bus topology is different
- pcie-root-ports are plugged into slots (soon to be functions) of
pcie-root as needed, and the new endpoint devices are assigned to the
single slot in each pcie-root-port. This is done so that the devices
are, by default, hotpluggable (the slots of pcie-root don't support
hotplug, but the single slot of the pcie-root-port does). Since
pcie-root-ports can only be plugged into pcie-root, and we don't
auto-assign endpoint devices to the pcie-root slots, this means
topology expansion doesn't compete with endpoint devices for slots, so
we don't need to worry about checking for all "useful" slots being
free *prior* to assigning addresses to new endpoint devices - as a
matter of fact, if we attempt to reserve the open slots before the
used slots, it can lead to errors.

Instead this patch just reserves one slot for a "future potential"
PCIe device after doing the assignment for actual devices, but only
if the only PCI controller defined prior to starting address
assignment was pcie-root, and only if we auto-added at least one PCI
controller during address assignment. This assures two things:

1) that reserving the open slots will only be done when the domain is
   initially defined, never at any time after, and

2) that if the user understands enough about PCI controllers that they
   are adding them manually, that we don't mess up their plan by
   adding extras - if they know enough to add one pcie-root-port, or
   to manually assign addresses such that no pcie-root-ports are
   needed, they know enough to add extra pcie-root-ports if they want
   them (this could be called the "libguestfs clause", since
   libguestfs needs to be able to create domains with as few
   devices/controllers as possible).

This is set to reserve a single free port for now, but could be
increased in the future if public sentiment goes in that direction
(it's easy to increase later, but essentially impossible to decrease)
2016-11-14 14:23:48 -05:00
Laine Stump
8d873a5a47 qemu: try to put ich9 sound device at 00:1B.0
Real Q35 hardware has an ICH9 chip that includes several integrated
devices at particular addresses (see the file docs/q35-chipset.cfg in
the qemu source). libvirt already attempts to put the first two sets
of ich9 USB2 controllers it finds at 00:1D.* and 00:1A.* to match the
real hardware. This patch does the same for the ich9 "HD audio"
device.

The main inspiration for this patch is that currently the *only*
device in a reasonable "workstation" type virtual machine config that
requires a legacy PCI slot is the audio device, Without this patch,
the standard Q35 machine created by virt-manager will have a
dmi-to-pci-bridge and a pci-bridge just for the sound device; with the
patch (and if you change the sound device model from the default
"ich6" to "ich9"), the machine definition constructed by virt-manager
has absolutely no legacy PCI controllers - any legacy PCI devices
(e.g. video and sound) are on pcie-root as integrated devices.
2016-11-14 14:23:01 -05:00
Laine Stump
d8bd837669 qemu: add a USB3 controller to Q35 domains by default
Previously we added a set of EHCI+UHCI controllers to Q35 machines to
mimic real hardware as closely as possible, but recent discussions
have pointed out that the nec-usb-xhci (USB3) controller is much more
virtualization-friendly (uses less CPU), so this patch switches the
default for Q35 machinetypes to add an XHCI instead (if it's
supported, which it of course *will* be).

Since none of the existing test cases left out USB controllers in the
input XML, a new Q35 test case was added which has *no* devices, so
ends up with only the defaults always put in by qemu, plus those added
by libvirt.
2016-11-14 14:22:23 -05:00
Laine Stump
807232203a qemu: don't force-add a dmi-to-pci-bridge just on principle
Now the a dmi-to-pci-bridge is automatically added just as it's needed
(when a pci-bridge is being added), we no longer have any need to
force-add one to every single Q35 domain.
2016-11-14 14:21:43 -05:00
Laine Stump
815b51d97a qemu: update tests to not assume dmi-to-pci-bridge is always added
A few of the qemu test cases assume that a dmi-to-pci-bridge will
always be added at index 1, and so they omit it from the input data
even though a pci-bridge is present at index 2, e.g.:

   <controller type='pci' index='0' model='pcie-root'/>
   <controller type='pci' index='2' model='pci-bridge'/>

Support for this odd practice was discussed on libvir-list and we
decided that the complex code required to make this continue was not
worth the headache of maintaining. So instead, this patch modifies the
test cases to manually add a dmi-to-pci-bridge at index 1 (since an
upcoming patch is going to eliminate the unconditional adding of
dmi-to-pci-bridge).

Because the auto-add was placing the dmi-to-pci-bridge later in the
list (even though it has a lower index) the test output is also
updated to take account for the new order (which puts the pci
controllers in index-order)
2016-11-14 14:21:15 -05:00
Laine Stump
0702f48ef4 qemu: auto-add pcie-root-port/dmi-to-pci-bridge controllers as needed
Previously libvirt would only add pci-bridge devices automatically
when an address was requested for a device that required a legacy PCI
slot and none was available. This patch expands that support to
dmi-to-pci-bridge (which is needed in order to add a pci-bridge on a
machine with a pcie-root), and pcie-root-port (which is needed to add
a hotpluggable PCIe device). It does *not* automatically add
pcie-switch-upstream-ports or pcie-switch-downstream-ports (and
currently there are no plans for that).

Given the existing code to auto-add pci-bridge devices, automatically
adding pcie-root-ports is fairly straightforward. The
dmi-to-pci-bridge support is a bit tricky though, for a few reasons:

1) Although the only reason to add a dmi-to-pci-bridge is so that
   there is a reasonable place to plug in a pci-bridge controller,
   most of the time it's not the presence of a pci-bridge *in the
   config* that triggers the requirement to add a dmi-to-pci-bridge.
   Rather, it is the presence of a legacy-PCI device in the config,
   which triggers auto-add of a pci-bridge, which triggers auto-add of
   a dmi-to-pci-bridge (this is handled in
   virDomainPCIAddressSetGrow() - if there's a request to add a
   pci-bridge we'll check if there is a suitable bus to plug it into;
   if not, we first add a dmi-to-pci-bridge).

2) Once there is already a single dmi-to-pci-bridge on the system,
   there won't be a need for any more, even if it's full, as long as
   there is a pci-bridge with an open slot - you can also plug
   pci-bridges into existing pci-bridges. So we have to make sure we
   don't add a dmi-to-pci-bridge unless there aren't any
   dmi-to-pci-bridges *or* any pci-bridges.

3) Although it is strongly discouraged, it is legal for a pci-bridge
   to be directly plugged into pcie-root, and we don't want to
   auto-add a dmi-to-pci-bridge if there is already a pci-bridge
   that's been forced directly into pcie-root.

Although libvirt will now automatically create a dmi-to-pci-bridge
when it's needed, the code still remains for now that forces a
dmi-to-pci-bridge on all domains with pcie-root (in
qemuDomainDefAddDefaultDevices()). That will be removed in a future
patch.

For now, the pcie-root-ports are added one to a slot, which is a bit
wasteful and means it will fail after 31 total PCIe devices (30 if
there are also some PCI devices), but helps keep the changeset down
for this patch. A future patch will have 8 pcie-root-ports sharing the
functions on a single slot.
2016-11-14 14:19:36 -05:00
Laine Stump
b2c887844f qemu: only force an available legacy-PCI slot on domains with pci-root
Andrea had the right idea when he disabled the "reserve an extra
unused slot" bit for aarch64/virt. For *any* PCI Express-based
machine, it is pointless since 1) an extra legacy-PCI slot can't be
used for hotplug, since hotplug into legacy PCI slots doesn't work on
PCI Express machinetypes, and 2) even for "coldplug" expansion,
everybody will want to expand using Express controllers, not legacy
PCI.

This patch eliminates the extra slot reserve unless the system has a
pci-root (i.e. legacy PCI)
2016-11-14 14:18:49 -05:00
Laine Stump
5266426b21 qemu: assign nec-xhci (USB3) controller to a PCIe address when appropriate
The nec-usb-xhci device (which is a USB3 controller) has always
presented itself as a PCI device when plugged into a legacy PCI slot,
and a PCIe device when plugged into a PCIe slot, but libvirt has
always auto-assigned it to a legacy PCI slot.

This patch changes that behavior to auto-assign to a PCIe slot on
systems that have pcie-root (e.g. Q35 and aarch64/virt).

Since we don't yet auto-create pcie-*-port controllers on demand, this
means a config with an nec-xhci USB controller that has no PCI address
assigned will also need to have an otherwise-unused pcie-*-port
controller specified:

   <controller type='pci' model='pcie-root-port'/>
   <controller type='usb' model='nec-xhci'/>

(this assumes there is an otherwise-unused slot on pcie-root to accept
the pcie-root-port)
2016-11-14 14:18:06 -05:00
Laine Stump
9dfe733e99 qemu: assign e1000e network devices to PCIe slots when appropriate
The e1000e is an emulated network device based on the Intel 82574,
present in qemu 2.7.0 and later. Among other differences from the
e1000, it presents itself as a PCIe device rather than legacy PCI. In
order to get it assigned to a PCIe controller, this patch updates the
flags setting for network devices when the model name is "e1000e".

(Note that for some reason libvirt has never validated the network
device model names other than to check that there are no dangerous
characters in them. That should probably change, but is the subject of
another patch.)

Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=1343094
2016-11-14 14:17:14 -05:00
Laine Stump
c7fc151eec qemu: assign virtio devices to PCIe slot when appropriate
libvirt previously assigned nearly all devices to a "hotpluggable"
legacy PCI slot even on machines with a PCIe root bus (and even though
most such machines don't even support hotplug on legacy PCI slots!)
Forcing all devices onto legacy PCI slots means that the domain will
need a dmi-to-pci-bridge (to convert from PCIe to legacy PCI) and a
pci-bridge (to provide hotpluggable legacy PCI slots which, again,
usually aren't hotpluggable anyway).

To help reduce the need for these legacy controllers, this patch tries
to assign virtio-1.0-capable devices to PCIe slots whenever possible,
by setting appropriate connectFlags in
virDomainCalculateDevicePCIConnectFlags(). Happily, when that function
was written (just a few commits ago) it was created with a
"virtioFlags" argument, set by both of its callers, which is the
proper connectFlags to set for any virtio-*-pci device - depending on
the arch/machinetype of the domain, and whether or not the qemu binary
supports virtio-1.0, that flag will have either been set to PCI or
PCIe. This patch merely enables the functionality by setting the flags
for the device to whatever is in virtioFlags if the device is a
virtio-*-pci device.

NB: the first virtio video device will be placed directly on bus 0
slot 1 rather than on a pcie-root-port due to the override for primary
video devices in qemuDomainValidateDevicePCISlotsQ35(). Whether or not
to change that is a topic of discussion, but this patch doesn't change
that particular behavior.

NB2: since the slot must be hotpluggable, and pcie-root (the PCIe root
complex) does *not* support hotplug, this means that suitable
controllers must also be in the config (i.e. either pcie-root-port, or
pcie-downstream-port). For now, libvirt doesn't add those
automatically, so if you put virtio devices in a config for a qemu
that has PCIe-capable virtio devices, you'll need to add extra
pcie-root-ports yourself. That requirement will be eliminated in a
future patch, but for now, it's simple to do this:

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

Partially Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=1330024
2016-11-14 14:16:12 -05:00