Jiri Denemark [Wed, 22 Jun 2016 13:53:48 +0000 (15:53 +0200)]
qemu: Update guest CPU def in live XML
Storing the updated CPU definition in the live domain definition saves
us from having to update it over and over when we need it. Not to
mention that we will soon further update the CPU definition according to
QEMU once it's started.
A highly wanted side effect of this patch, libvirt will pass all CPU
features explicitly specified in domain XML to QEMU, even those that are
already included in the host model.
This patch should fix the following bugs:
https://bugzilla.redhat.com/show_bug.cgi?id=1207095
https://bugzilla.redhat.com/show_bug.cgi?id=1339680
https://bugzilla.redhat.com/show_bug.cgi?id=1371039
https://bugzilla.redhat.com/show_bug.cgi?id=1373849
https://bugzilla.redhat.com/show_bug.cgi?id=1375524
https://bugzilla.redhat.com/show_bug.cgi?id=1377913
Jiri Denemark [Tue, 9 Aug 2016 11:26:53 +0000 (13:26 +0200)]
cpu: Rework cpuCompare* APIs
Both cpuCompare* APIs are renamed to virCPUCompare*. And they should now
work for any guest CPU definition, i.e., even for host-passthrough
(trivial) and host-model CPUs. The implementation in x86 driver is
enhanced to provide a hint about -noTSX Broadwell and Haswell models
when appropriate.
The function is similar to virCPUDataCheckFeature, but it works directly
on CPU definition rather than requiring it to be transformed into CPU
data first.
Jiri Denemark [Fri, 17 Jun 2016 07:45:48 +0000 (09:45 +0200)]
cpu: Introduce virCPUTranslate
The API is supposed to make sure the provided CPU definition does not
use a CPU model which is not supported by the hypervisor (if at all
possible, of course).
Jiri Denemark [Thu, 4 Aug 2016 22:01:42 +0000 (00:01 +0200)]
cpu: Set nfeatures_max correctly in x86Decode
Keeping nfeatures_max set to 0 while nfeatures > 0 and some features are
already stored in features array is just asking for problems once we
want to add a new feature into the array.
Jiri Denemark [Thu, 23 Jun 2016 13:27:07 +0000 (15:27 +0200)]
cpu: Rework cpuUpdate
The reworked API is now called virCPUUpdate and it should change the
provided CPU definition into a one which can be consumed by the QEMU
command line builder:
- host-passthrough remains unchanged
- host-model is turned into custom CPU with a model and features
copied from host
- custom CPU with minimum match is converted similarly to host-model
- optional features are updated according to host's CPU
Jiri Denemark [Sat, 18 Jun 2016 09:19:17 +0000 (11:19 +0200)]
cpu: Make x86ModelFromCPU a bit smarter
x86ModelFromCPU is used to provide CPUID data for features matching
@policy. This patch allows callers to set @policy to -1 to get combined
CPUID for all CPU features (including those implicitly provided a CPU
model) specified in CPU def.
Jiri Denemark [Wed, 15 Jun 2016 14:45:47 +0000 (16:45 +0200)]
Show host model in domain capabilities
The domain capabilities XML is capable of showing whether each guest CPU
mode is supported or not with a possibility to provide additional
details. This patch enhances host-model capability to advertise the
exact CPU model which will be used as a host-model:
Jiri Denemark [Tue, 21 Jun 2016 14:02:07 +0000 (16:02 +0200)]
cpu: Drop false support for ARM cpu-model
The ARM CPU driver wrongly reported host CPU model as "host", which made
host-model to be just an alias for host-passthrough. Let's drop this
insanity.
Jiri Denemark [Wed, 15 Jun 2016 12:35:18 +0000 (14:35 +0200)]
qemu: Store host-model CPU in qemu capabilities
Host capabilities provide libvirt's view of the host CPU, but for a
useful support for host-model CPUs we really need a hypervisor's view of
the CPU. And since the view can be differ with emulator, qemu
capabilities is the best place to store the host CPU model.
This patch just copies the CPU model from host capabilities, but this
will change in the future.
Jiri Denemark [Thu, 23 Jun 2016 10:54:19 +0000 (12:54 +0200)]
conf: Introduce virCPUDefStealModel
The function moves CPU model related parts from one CPU definition to
another. It can be used to avoid unnecessary copies from a temporary CPU
definitions which will be freed anyway.
Jiri Denemark [Wed, 15 Jun 2016 14:15:44 +0000 (16:15 +0200)]
domcaps: Add CPU usable flag
In case a hypervisor is able to tell us a list of supported CPU models
and whether each CPU models can be used on the current host, we can
propagate this to domain capabilities. This is a better alternative
to calling virConnectCompareCPU for each supported CPU model.
Jiri Denemark [Tue, 14 Jun 2016 09:12:49 +0000 (11:12 +0200)]
cpu: Special case models == NULL in cpuGetModels
Some CPU drivers (such as arm) do not provide list of CPUs libvirt
supports and just pass any CPU model from domain XML directly to QEMU.
Such driver need to return models == NULL and success from cpuGetModels.
Jiri Denemark [Fri, 5 Aug 2016 12:42:15 +0000 (14:42 +0200)]
qemuxml2argvtest: Reorder CPU features
The x86 CPU driver translated each CPU definition from domain XML into
CPUID data and then back to CPU definition. This effectively sorted the
list of CPU features according to their CPUID values. Since this is
going to change, we need to reorder CPU features in a few test files to
make sure the generated QEMU command lines will not change.
Jiri Denemark [Thu, 4 Aug 2016 11:19:39 +0000 (13:19 +0200)]
qemuxml2argvtest: Set correct architecture for KVM guests
Testing PPC64/AArch64 KVM domains on x86_64 host only works because we
have a lot of bugs in our code. Since this series is going to fix them,
we need to make sure the host architecture matches guest for KVM
domains.
Jiri Denemark [Thu, 4 Aug 2016 11:25:02 +0000 (13:25 +0200)]
qemuxml2argvtest: Properly setup CPU models in qemuCaps
Adding x86 CPU models into a list of supported CPUs for non-x86
architectures is not a very good idea. Each architecture we test needs
to maintain its own list of supported CPU models.
Jiri Denemark [Thu, 4 Aug 2016 11:16:55 +0000 (13:16 +0200)]
testutilsqemu: Helpers for changing host CPU and arch
Changing a host architecture or a CPU is not as easy as assigning a new
value to the appropriate element in virCaps since there is a relation
between the CPU and host architecture (we don't really want to test
anything on an AArch64 host with core2duo CPU). This patch introduces
qemuTestSetHostArch and qemuTestSetHostCPU helpers which will make sure
the host architecture matches the host CPU.
Jiri Denemark [Wed, 3 Aug 2016 10:21:19 +0000 (12:21 +0200)]
qemuxml2argvtest: Update qemuCaps after parsing domain XML
Some parts of qemuCaps depend on guest architecture, machine type, and
possibly other things that we know only once the domain XML has been
parsed. Let's move all these updates into a dedicated function.
qemu: Use virDomainCapsCPUModels for cpuDefinitions
The list of supported CPU models in domain capabilities is stored in
virDomainCapsCPUModels. Let's use the same object for storing CPU models
in QEMU capabilities.
Applications can use it to inspect what CPU configuration modes are
supported for a specific combination of domain type, emulator binary,
guest architecture and machine type.
Jim Fehlig [Mon, 19 Sep 2016 17:34:34 +0000 (11:34 -0600)]
libxl: support VIR_MIGRATE_PERSIST_DEST migration flag
By default, virt-manager (and likely other libvirt-based apps) sets
the VIR_MIGRATE_PERSIST_DEST flag when invoking the migrate API, which
fails in a Xen setup since the libxl driver does not support the flag.
Persisting a domain is a trivial task in the grand scheme of migration,
so be nice to libvirt apps and add support for VIR_MIGRATE_PERSIST_DEST
in the libxl driver.
qemu_process: show shutoff reasons when debug log disabled
We have a few of senarios that libvirtd would invoke qemuProcessStop
and leave a "shutting down" in /var/log/libvirt/qemu/$DOMAIN.log.
The shutoff reason showing in debug log is also very important
for us to know why VM shutting down in domain log,
as we seldom enable debug log of libvirtd.
During offline migration, no storage is copied. Nor disks, nor
NVRAM file, nor anything. We use qemu for that and because domain
is not running there's nobody to copy that for us.
We should document this to avoid confusing users.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Peter Krempa [Wed, 14 Sep 2016 05:37:16 +0000 (07:37 +0200)]
qemu: driver: Don't return automatic NUMA emulator pinning data for persistentDef
Calling virDomainGetEmulatorPinInfo on a live VM with automatic NUMA
pinning and VIR_DOMAIN_AFFECT_CONFIG would return the automatic pinning
data in some cases which is bogus. Use the autoCpuset property only when
called on a live definition.
Peter Krempa [Wed, 14 Sep 2016 05:37:16 +0000 (07:37 +0200)]
qemu: driver: Don't return automatic NUMA vCPU pinning data for persistentDef
Calling virDomainGetVcpuPinInfo on a live VM with automatic NUMA pinning
and VIR_DOMAIN_AFFECT_CONFIG would return the automatic pinning data
in some cases which is bogus. Use the autoCpuset property only when
called on a live definition.
Peter Krempa [Wed, 14 Sep 2016 05:28:18 +0000 (07:28 +0200)]
qemu: domain: Add macro to simplify access to vm private data
Sometimes adding a separate variable to access vm->privateData is not
necessary. Add a macro that will do the typecasting rather than having
to add a temp variable to force the compiler to typecast it.
In this mode, QEMU doesn't listen for SPICE connection anywhere and
clients have to use virDomainOpenGraphics* APIs to attach to the domain.
That is, the client has to run on the same host where the domains runs
and it's impossible to tell the client to reconnect to the destination
QEMU during migration (unless there is some kind of proxy on the host).
While current libvirt correctly ignores such graphics devices when
creating graphics migration cookie, old libvirt just sends
in the cookie. After seeing this cookie, we happily would call
client_migrate_info QMP command and wait for SPICE_MIGRATE_COMPLETED
event, which is quite pointless since the doesn't know where to connecti
anyway. We should just ignore such cookies.
qemu: agent: give better error messages whe agent monitor is down
We can receive NULL as sync reply in two situations. First
is garbage sync reply and this situation is handled by
resending sync message. Second is different cases
of rebooting guest, destroing domain etc and we can
give more meaningful error message. Actually we have
this error message in qemuAgentCommand already which checks
for the same sitatuion. AFAIK case with mon->running
is just to be safe on adding some future(?) cases of
returning NULL reply.
We can easily handle receiving garbage on sync. We don't
have to make client deal with this situation. We just
need to resend sync command but this time garbage is
not be possible.
When we wait for sync reply we can receive delayed
reply to syncs or commands that were sent erlier. We can
safely skip them until we receive sync reply with correct id.
There is no much sense report this situation to client.
Actually with a bit of "luck" if we involve client into
this the play can go on forever: send sync 0, receive
sync reply -1, send sync 1, receive reply 0 ...
qemu: agent: ignore garbage while waiting for sync reply
After sync is sent we can receive garbare and this is not error.
Consider next regular case:
1. libvirtd sends sync
2. qga sends partial sync reply and die
3. libvirtd sends sync
4. qga sends sync reply
5. libvirtd receives garbage
(half of first reply and second reply together)
We should handle this situation as it is recoverable.
Next sync can succeed. Let's report reply is NULL,
it will be converted to the VIR_ERR_AGENT_UNSYNCED
which signals client to retry.
Errors in qemuAgentIOProcessLine stop agent IO processing just
like any regular IO error, however some of current errors
that this functions spawns are false positives. Consider
next case for example:
Last IO triggers error because current code ignores
only delayed syncs when unsynced
We should not treat any delayed reply as error in unsynced
state. Until client and qga are not in sync delayed reply to any
command is possible. msg == NULL is the exact criterion
that we are not in sync.
Andrea Bolognani [Tue, 20 Sep 2016 13:22:04 +0000 (15:22 +0200)]
virtlogd: Don't stop or restart along with libvirtd
Commit 839a060 tied the lifecycle of virtlogd more
closely to that of libvirtd. Unfortunately, while starting
virtlogd when libvirtd is started is definitely a good idea,
restarting virtlogd or shutting it down at any time outside
of system poweroff is not.
Revert part of that commit by removing the PartOf= lines,
meaning that only startup requests will be propagated from
libvirtd to virtlogd.
Erik Skultety [Wed, 7 Sep 2016 15:30:48 +0000 (17:30 +0200)]
virt-admin: Tweak command parsing logic so that aliases point to new commands
Change the logic in a way, so that VSH_CMD_FLAG_ALIAS behaves similarly to
how VSH_OT_ALIAS for command options, i.e. there is no need for code duplication
for the alias and the aliased command structures. Along with that change,
switch any existing VSH_CMD_FLAG_ALIAS occurrences to this new format. Also,
since this patch introduces a new command structure element, adjust the
virsh-self-test test to make sure we won't ever miss to specify the '.alias'
member for an aliased command because doing that would lead to an internal
error.
Signed-off-by: Erik Skultety <eskultet@redhat.com>
Erik Skultety [Fri, 16 Sep 2016 14:06:47 +0000 (16:06 +0200)]
virsh: man: Do not document our old deprecated nodedev-dettach command
The command is deprecated due to being grammatically incorrect, but for
backwards compatibility reasons cannot be removed. However, we should not
document such commands.
Signed-off-by: Erik Skultety <eskultet@redhat.com>
Erik Skultety [Fri, 16 Sep 2016 08:17:46 +0000 (10:17 +0200)]
vsh: discard vshCmddefOptFill and move its body to vshCmddefOptParse
Recent changes extracted the command internals validation routine from
vshCmddefOptParse method which now just calls vshCmddefOptFill. Therefore, make
vshCmddefOptFill the new vshCmddefOptParse and drop the unnecessary name.
Signed-off-by: Erik Skultety <eskultet@redhat.com>
Erik Skultety [Fri, 16 Sep 2016 07:58:24 +0000 (09:58 +0200)]
vsh: Extract vshCmddefCheckInternals from vshCmddefOptParse
Originally introduced by commit 2432521e which correctly split
vshCmddefOptParse into command's options validation and options parsing.
However, command's 'internals' are not tied solely to .options, rather it
should be about the overall structure, therefore the validation should be
extracted from vshCmddefOptParse and performed only within our test suite, i.e.
in vshSelfTest.
Signed-off-by: Erik Skultety <eskultet@redhat.com>
Erik Skultety [Thu, 15 Sep 2016 15:52:11 +0000 (17:52 +0200)]
vsh: vshCmddefHelp: Drop the unnecessary 'else' branch
If the initial check is true the function immediately returns so there's no
need to enclose the code following the check within an 'else' block.
Also, by removing the 'else' block, the declarations need to be moved to
beginning of the function block to conform with our guidelines.
Signed-off-by: Erik Skultety <eskultet@redhat.com>
Erik Skultety [Mon, 19 Sep 2016 11:10:38 +0000 (13:10 +0200)]
vsh: Enforce checking for missing help string
The intention is to move vshCmddefCheckInternals out of vshCmddefOptParse to
our test suite. First step to do that is to enforce checking for an existing
help string (that also means it's non-empty) in a command because a command
without a help is not much of a use.
Signed-off-by: Erik Skultety <eskultet@redhat.com>
This caused libvirt Mingw build to break with the compiler
reporting 100's of definitions of virConnectAuthPtrDefault
./.libs/libvirt_util.a(libvirt_util_la-virarch.o):virarch.c:(.bss+0x0): multiple definition of `virConnectAuthPtrDefault'
./.libs/libvirt_util.a(libvirt_util_la-viralloc.o):viralloc.c:(.bss+0x0): first defined here
./.libs/libvirt_util.a(libvirt_util_la-viratomic.o):viratomic.c:(.bss+0x0): multiple definition of `virConnectAuthPtrDefault'
./.libs/libvirt_util.a(libvirt_util_la-viralloc.o):/home/berrange/src/virt/libvirt/src/util/viralloc.c:87: first defined here
./.libs/libvirt_util.a(libvirt_util_la-viraudit.o):viraudit.c:(.bss+0x0): multiple definition of `virConnectAuthPtrDefault'
./.libs/libvirt_util.a(libvirt_util_la-viralloc.o):/home/berrange/src/virt/libvirt/src/util/viralloc.c:87: first defined here
./.libs/libvirt_util.a(libvirt_util_la-virauth.o):virauth.c:(.bss+0x0): multiple definition of `virConnectAuthPtrDefault'
./.libs/libvirt_util.a(libvirt_util_la-viralloc.o):/home/berrange/src/virt/libvirt/src/util/viralloc.c:87: first defined here
./.libs/libvirt_util.a(libvirt_util_la-virauthconfig.o):virauthconfig.c:(.bss+0x0): multiple definition of `virConnectAuthPtrDefault'
...snip...
The cause is our VIR_EXPORT_VAR macro which has some
magic on win to add dllexport/dllimport to the variable
declaration. Unfortunately the dllexport branch missed
off the 'extern' keyword, so the header file was in
fact declaring an instance of the variable in every
source file.
Previously the linker would merge all these definitions
into one, but that no longer happens due to -fno-common
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Laine Stump [Mon, 19 Sep 2016 17:44:21 +0000 (13:44 -0400)]
tests: fix use of fixedcontent variable
Commit 8563560026d192c2cf047b550ffd468692245ed6 switched from
hardcoded use of strcontent to hardcoded use of fixedcontent
(fixedcontent is *sometimes* a copy of strcontent with a \n
appended). This was a problem because sometimes fixedcontent is *not*
a copy of strcontent, but is instead NULL, leading to the regenerated
test case output being a 0 length file.
This patch creates a new const char *cmpcontent initialized to
strcontent, but changed to fixedcontent if/when fixedcontent is
created, then always uses cmpcontent instead of (str|fixed)content.
come with a linear framebuffer (sometimes called "VGA compatibility
framebuffer"). This linear framebuffer lives in one of the PCI device's
MMIO BARs, and allows guest code (primarily: firmware drivers, and
non-accelerated OS drivers) to display graphics with direct memory access.
Due to architectural reasons on aarch64/KVM hosts, this kind of
framebuffer doesn't / can't work in
qemu-system-(arm|aarch64) -M virt
machines. Cache coherency issues guarantee a corrupted / unusable display.
The problem has been researched by several people, including kvm-arm
maintainers, and it's been decided that the best way (practically the only
way) to have boot time graphics for such guests is to consolidate on
QEMU's "virtio-gpu-pci" device.
but libvirt unconditionally maps @type='virtio' to QEMU's "virtio-vga"
device model. (See the qemuBuildDeviceVideoStr() function and the
"qemuDeviceVideo" enum impl.)
According to the above, this is not right for the "virt" machine type; the
qemu-system-(arm|aarch64) binaries don't even recognize the "virtio-vga"
device model (justifiedly). Whereas "virtio-gpu-pci", which is a pure
virtio device without a compatibility framebuffer, is available, and works
fine.
(The ArmVirtQemu ("AAVMF") platform of edk2 -- that is, the UEFI firmware
for "virt" -- supports "virtio-gpu-pci", as of upstream commit 3ef3209d3028. See
<https://tianocore.acgmultimedia.com/show_bug.cgi?id=66>.)
Override the default mapping of "virtio", from "virtio-vga" to
"virtio-gpu-pci", if qemuDomainMachineIsVirt() evaluates to true.
Cc: Andrea Bolognani <abologna@redhat.com> Cc: Drew Jones <drjones@redhat.com> Cc: Marc-André Lureau <marcandre.lureau@redhat.com> Cc: Martin Kletzander <mkletzan@redhat.com> Suggested-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=1372901 Signed-off-by: Laszlo Ersek <lersek@redhat.com> Acked-by: Martin Kletzander <mkletzan@redhat.com>
There is nothing Linux-specific in that function. Also since commit 8c3b5bf48123783b812b97360db7ac51f1889e17 mingw build is broken due to
the fact that this function is not compiled in the library.
Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
projects / thirdparty / libvirt.git / log
