Commit Graph

40798 Commits

Author SHA1 Message Date
Daniel P. Berrangé
c32f172d12 qemu: wire up support for maximum CPU model
The "max" model can be treated the same way as "host" model in general.

Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
2021-02-10 11:44:48 +00:00
Daniel P. Berrangé
9c89cc5d6f qemu: probe for "-cpu max" support
This is a special CPU model similar to "-cpu host", so won't use our
normal CPU model detection logic.

Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
2021-02-10 11:44:48 +00:00
Daniel P. Berrangé
7c1653f63a cpu: wire up support for maximum CPU mode
The logic applied in the ppc64 case isn't quite correct, as the
interpretation of maximum mode depends on whether hardware virt
is used or not. This is information the CPU driver doesn't have.

Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
2021-02-10 11:44:48 +00:00
Daniel P. Berrangé
09cbd460fb conf: add reporting of "maximum" CPU mode in domain caps
The data reported is the same as for "host-passthrough"

Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
2021-02-10 11:44:48 +00:00
Daniel P. Berrangé
d153c101d2 conf: define a new "maximum" CPU mode
For hardware virtualization this is functionally identical to the
existing host-passthrough mode so the same caveats apply.

For emulated guest this exposes the maximum featureset supported by
the emulator. Note that despite being emulated this is not guaranteed
to be migration safe, especially if different emulator software versions
are used on each host.

Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
2021-02-10 11:44:48 +00:00
Michal Privoznik
676e56e515 virStorageSourceClear: Unref @vhostuser
The @vhostuser member of virStorageSource structure is allocated
during parsing in virDomainDiskSourceVHostUserParse() but never
freed leading to a memleak. Since the member is an object it has
to be unrefed instead of g_free()-d.

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
2021-02-10 11:54:38 +01:00
Michal Privoznik
7415e72e68 vsh: Drop unused @text arg from readline generators
After previous patches neither vshReadlineCommandGenerator() nor
vshReadlineOptionsGenerator() use prefix that user wants to
complete. The argument is marked as unused in both functions.
Drop it then.

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Jonathon Jongsma <jjongsma@redhat.com>
2021-02-10 11:51:59 +01:00
Michal Privoznik
72636ed86d vsh: Rework vshReadlineCommandGenerator()
Firstly, move variable declarations into the inner most block
they are used. Secondly, use for() loop instead of while so that
we don't have to advance loop counter explicitly on 'continue'.

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Jonathon Jongsma <jjongsma@redhat.com>
2021-02-10 11:51:59 +01:00
Michal Privoznik
9b005b1967 vsh: Simplify condition for calling completer callback
The way we currently call completer callbacks is that if we've
found --option that user wants to complete value for and it has
callback set then the callback is called.

And just before that, if no --option to have the value completed
is found or is found and is of boolean type then a list of
--option is generated (for given command).

But these two conditions can never be true at the same time
because boolean type of --options do not accept values. Therefore
the calling of completer callback can be promoted onto the same
level as the --option list generation.

This means that merging of two lists can be dropped to and
completer callback can store its retval directly into @list (but
as shown earlier one of the string lists to merge is always
empty).

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Jonathon Jongsma <jjongsma@redhat.com>
2021-02-10 11:51:59 +01:00
Michal Privoznik
b1eab47c2d vsh: Deduplicate filtering in vshReadlineCommandGenerator()
Completer callbacks generate all possible outputs ignoring any partial
input (e.g. prefix of a domain name) and then use vshCompleterFilter() to
filter out those strings which don't fit the partial input (prefix).

In contrast, vshReadlineCommandGenerator() does some internal filtering and
only generates completions that match a given prefix. Rather than treating
these scenarios differently, simply generate all possible options and
filter them all at the end.

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Jonathon Jongsma <jjongsma@redhat.com>
2021-02-10 11:51:59 +01:00
Michal Privoznik
c3ea585e18 vsh: Deduplicate filtering in vshReadlineOptionsGenerator()
Completer callbacks generate all possible outputs ignoring any partial
input (e.g. prefix of a domain name) and then use vshCompleterFilter() to
filter out those strings which don't fit the partial input (prefix).

In contrast, vshReadlineOptionsGenerator() does some internal filtering and
only generates completions that match a given prefix. Rather than treating
these scenarios differently, simply generate all possible options and
filter them all at the end.

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Jonathon Jongsma <jjongsma@redhat.com>
2021-02-10 11:51:59 +01:00
Michal Privoznik
3cc6572fab vsh: Rewrite opt->type check in vshReadlineParse()
The vshReadlineParse() function is called whenever user hits
<TAB><TAB>. If there is no command (or a partially written one),
then a list of possible commands is printed to the user. But, if
there is a command then its --options are generated. But
obviously, we can not generate --options if there already is an
--option that's expecting a value. For instance, consider:

  virsh # start --domain <TAB><TAB>

In this case we want to call completer for --domain option, but
that's a different story.

Anyway, the way that we currently check whether --options list
should be generated is checking the type of the last --option. If
it isn't DATA, STRING, INT, or ARGV (all these expect a value),
then we can generate --option list. Well, writing the condition
this way is needlessly verbose and also prone to errors (see
d9a320bf97 for example).

We know that boolean type does not require a value. This leaves
us with the only type that was not mentioned yet - VSH_OT_ALIAS.
This is a special type for backwards compatibility and it refers
to another --option which can be just any type.

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Jonathon Jongsma <jjongsma@redhat.com>
2021-02-10 11:51:59 +01:00
Michal Privoznik
9ad2cb6e73 vsh: Use g_auto() for string lists returned in readline command/options generators
There are two functions that are used to generate completion
lists: vshReadlineCommandGenerator() for command names and
vshReadlineOptionsGenerator() for --options for given command.
Both return a string list, but may also fail while constructing
it. For that case, they call g_strfreev() explicitly, which is
needless since we have g_auto(GStrv).

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Jonathon Jongsma <jjongsma@redhat.com>
2021-02-10 11:51:59 +01:00
Michal Privoznik
eac14234c8 vsh: Prefer g_strdup_printf() over g_snprintf() in vshReadlineOptionsGenerator()
The vshReadlineOptionsGenerator() function returns a string list
of all --options for given command. But the way that individual
items on the list are allocated can be written better.

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Jonathon Jongsma <jjongsma@redhat.com>
2021-02-10 11:51:59 +01:00
Michal Privoznik
6ba61373b7 vsh: Accept NULL @list in vshCompleterFilter()
The aim of vshCompleterFilter() is to take a string list and a
prefix and remove all strings from the list that don't have the
desired prefix. The function is used to filter out those strings
returned by a completer callback that don't correspond with
user's (partial) input. For instance, domain name completer
virshDomainNameCompleter() returns all domain names and then
vshCompleterFilter() refines the list so that only domains with
correct prefix of their name are offered to user. This was a
design choice - it allows us to have shorter completers as they
do not have to copy the list filtering over and over.

Having said all of that, it may happen that a completer does not
return anything (e.g. there is no domain in requested state,
virsh is not connected and thus completer exited early, etc.). In
that case, the string list is NULL and vshCompleterFilter() can
simply return early.

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Jonathon Jongsma <jjongsma@redhat.com>
2021-02-10 11:51:59 +01:00
Michal Privoznik
268f16293c vsh: Use g_auto(GStrv) to free string list returned by completer callback
This saves us explicit call of g_strfreev() in error path.

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Jonathon Jongsma <jjongsma@redhat.com>
2021-02-10 11:51:59 +01:00
Michal Privoznik
58aeebe096 vsh: Don't put VSH_OT_ALIAS onto list of completions
We've invented VSH_OT_ALIAS type for --option so that we can
rewrite some --options (e.g. fix spelling). For instance
blkdeviotune command uses this feature heavily:
--options-with-dash are preferred over old
--options_with_underscore. Both versions are supported but only
the new ones (not aliased) are documented and reported in --help.

Except for options completer, which happily put also aliased
versions in front of user's eyes.

Note, there is a second (gross) way we use aliases: to rewrite
options from --oldoption to --newoption=value (for instance
--shareable option of attach-disk is an alias of
--mode=shareable). And just like with the previous group - don't
generate them into the list of possible options.

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Jonathon Jongsma <jjongsma@redhat.com>
2021-02-10 11:51:59 +01:00
Michal Privoznik
d3b2d5158a lib: Substitute some STREQLEN with STRPREFIX
There are few cases where STREQLEN() is called like this:

  STREQLEN(var, string, strlen(string))

which is the same as STRPREFIX(var, string). Use STRPREFIX()
because it is more obvious what the check is doing.

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Jonathon Jongsma <jjongsma@redhat.com>
2021-02-10 11:51:59 +01:00
Andrea Bolognani
0ad938adc0 tests: Only mock $INODE64 symbols on x86_64 macOS
The version of macOS running on Apple Silicon doesn't need to
concern itself with backwards compatibility with 32-bit
applications, and so it could jettison all the symbol aliasing
shenanigans involved.

https://gitlab.com/libvirt/libvirt/-/issues/121

Signed-off-by: Andrea Bolognani <abologna@redhat.com>
Reviewed-by: Roman Bolshakov <r.bolshakov@yadro.com>
Tested-by: Roman Bolshakov <r.bolshakov@yadro.com>
2021-02-10 11:03:39 +01:00
Michal Privoznik
aebb2b6edf qemu_capabilities: Parse "deprecated" in virQEMUCapsLoadMachines() properly
A <machine/> element can have "deprecated" attribute that
corresponds to 'deprecated' member of _virQEMUCapsMachineType
struct. But the member is of boolean type. Therefore, the string
returned by virXMLPropString() must be freed.

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
2021-02-10 09:38:59 +01:00
Michal Privoznik
71609477a2 qemu_capabilities: Don't leak @str in virQEMUCapsLoadMachines()
If parsing "maxCpus" attribute of <machine/> element fails an
error is printed but the corresponding string is not freed. While
it is very unlikely to happen (parsed XML is not user provided
and we are the ones generating it), it is possible. Instead of
freeing the variable in the error path explicitly, let's declare
it as g_autofree. And while I'm at it, let's bring it into the
loop where it's used.

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
2021-02-10 09:38:49 +01:00
Jim Fehlig
e3d60f761c build: Fix generation of virtproxyd socket files
The various virtproxyd socket files are generated with invalid syntax,
e.g. from virtproxyd.socket

[Unit]
Description=Libvirt proxy local socket
Before=virtproxyd.service
libvirtd.socket libvirtd-ro.socket libvirtd-admin.socket libvirtd-tcp.socket libvirtd-tls.socket

Note the missing 'Conflicts=' in the last line. Fix it by prepending
'Conflicts=' to libvirtd_socket_conflicts when adding virtproxyd
to virt_daemon_units.

Signed-off-by: Jim Fehlig <jfehlig@suse.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
2021-02-09 11:39:40 -07:00
Peter Krempa
a619e28dba qemucapabilitiesdata: Update 6.0.0 x86_64 capability test data
Update to qemu commit v5.2.0-1684-gd0dddab40e which includes the removal
of pc-1.0/pc-1.1/pc-1.2 machine types, adds the new QMP commands for
internal snapshots as well as includes the background-snapshot
capability.

Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
2021-02-09 11:30:19 +01:00
Peter Krempa
af53024523 qemuxml2argvdata: luks-disks-source-qcow2: Remove specific machine type
The test doesn't depend on any specific machine type.

Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
2021-02-09 11:30:19 +01:00
Peter Krempa
5949b8d714 qemuxml2argvdata: disk-cache: Remove specific machine type
The test doesn't depend on any specific machine type.

Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
2021-02-09 11:30:19 +01:00
Peter Krempa
f990ddaf3a qemuxml2argvdata: disk-cdrom-network: Remove specific machine type
The test doesn't depend on a specific machine type.

The test uses a machine type which is becoming deprecated so it would
break the _LATEST version of the test once we update the qemu data.

Signed-off-by: Peter Krempa <pkrempa@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
2021-02-09 11:30:19 +01:00
Andrea Bolognani
65edb00539 gitlab: Fix build container dependencies for build jobs
A couple of these were wrong, resulting in the build job not
waiting for the correct container to be built before starting.

Fixes: 77296c807b

Signed-off-by: Andrea Bolognani <abologna@redhat.com>
2021-02-09 10:28:50 +01:00
Andrea Bolognani
f834c341fb cpu: Fix build on some aarch64 platforms
On platforms that lack both getauxval() and elf_aux_info(),
such as OpenBSD and macOS, host CPU detection can't work.

https://gitlab.com/libvirt/libvirt/-/issues/121

Signed-off-by: Andrea Bolognani <abologna@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
2021-02-09 09:14:44 +01:00
Andrea Bolognani
03af15c024 cpu: Only retrieve AT_HWCAP once
No need to fetch the same information twice.

As a side effect, this solves a bug where, on platforms where
elf_aux_info() is used instead of getauxval(), we would not
make sure the CPUID feature is available before attempting to
use it.

Signed-off-by: Andrea Bolognani <abologna@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
2021-02-09 09:14:42 +01:00
Andrea Bolognani
82ffb81c9c cpu: Only include <sys/auxv.h> if available
This header is not present on several non-Linux targets that
nonetheless support aarch64.

Signed-off-by: Andrea Bolognani <abologna@redhat.com>
Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
2021-02-09 09:14:05 +01:00
Michal Privoznik
f8f7bc254f qemu: Release <memory/> device address on failed hotplug
A few commits back I've introduced new 'virtio-pmem' <memory/>
device. Since it's virtio it goes onto PCI bus. Therefore, on
hotplug new PCI address is generated (or provided one is
reserved). However, if hotplug fails (for whatever reason) the
address needs to be released. This is different to 'dimm' type of
address because for that type we don't keep a map of used slots
rather generate one on each address assign request. The map is
then thrown away. But for PCI addresses we keep internal state
and thus has to keep it updated. Therefore, this new
qemuDomainReleaseMemoryDeviceSlot() function is NOP for those
models which use 'dimm' address type ('dimm' and 'nvdimm').

While I'm at it, let's release the address in case of hot unplug.
Not that is supported (any such attempt fails with the following
error:

  "virtio based memory devices cannot be unplugged"

But if QEMU ever implements hot unplug then we don't have to
remember to fix our code.

Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Reviewed-by: Jiri Denemark <jdenemar@redhat.com>
2021-02-08 20:05:08 +01:00
Roman Bogorodskiy
b1e6324ca2 bhyve: auto allocate nmdm console paths
Currently, nmdm console device requires user to specify master and slave
path attributes (such as /dev/nmdm0A and /dev/nmdm0B respectively).
However, making user find a non-occupied device name might be not
convenient, especially for the remote connections.

Update the logic to make these attributes optional. In case if not
specified, use /dev/nmdm$UUID[AB], where $UUID is a domain's UUID.
With this schema it's unlikely nmdm device will clash with other domains
or even other non-bhyve nmdm devices.

Signed-off-by: Roman Bogorodskiy <bogorodskiy@gmail.com>
Reviewed-by: Ján Tomko <jtomko@redhat.com>
2021-02-08 18:50:47 +04:00
Jan Kuparinen
1fc23c885e Translated using Weblate (Finnish)
Currently translated at 13.5% (1415 of 10451 strings)

Translation: libvirt/libvirt
Translate-URL: https://translate.fedoraproject.org/projects/libvirt/libvirt/fi/

Co-authored-by: Jan Kuparinen <copper_fin@hotmail.com>
Signed-off-by: Jan Kuparinen <copper_fin@hotmail.com>
2021-02-08 12:42:57 +01:00
Pavel Hrdina
d3f4f01fa7 conf: allow virtio driver attributes for vhostuser disk
All of these options are actually supported by vhostuser disk so
we should allow them to be usable.

Signed-off-by: Pavel Hrdina <phrdina@redhat.com>
Reviewed-by: Peter Krempa <pkrempa@redhat.com>
2021-02-08 10:49:59 +01:00
Roman Bogorodskiy
a58edc602e bhyve: drop emulator binary value check
Currently, requesting domain capabilities fails when the specified
emulator binary does not equal to "/usr/sbin/bhyve". As we're
not using user-specified emulator anyway, drop this check to avoid
showing errors for values like "bhyve" (without absolute path).

Signed-off-by: Roman Bogorodskiy <bogorodskiy@gmail.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
2021-02-06 12:54:50 +04:00
Jim Fehlig
3068294e77 docs: Remove broken link to Xen channel doc
Many of Xen's text documents have been converted to man pages over
the years, the channel doc being one of them. Replace the broken
channel.txt link with the name of the man page providing the same
information.

Signed-off-by: Jim Fehlig <jfehlig@suse.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
2021-02-05 08:54:32 -07:00
Jakob Meng
87320046c0 docs: Add 'known_hosts_verify' parameter for libssh(2) connection uris
Parameter 'known_hosts_verify' is supported for some time now,
but it is not yet documented.

Signed-off-by: Jakob Meng <jakobmeng@web.de>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
2021-02-05 16:02:22 +01:00
Yi Li
f209d40a7e qemuDomainAttachRedirdevDevice: Remove need_release variable
Get rid of the 'need_release' variable. The code can be rewritten
so that it is not needed.

Signed-off-by: Yi Li <yili@winhong.com>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
2021-02-05 12:42:15 +01:00
Andrea Bolognani
ab4834a786 Revert "tests: Avoid gnulib replacements in mocks"
Now that we're no longer using gnulib, we can treat macOS the
same as all other targets.

This reverts commit 0ae6f5cea5

Signed-off-by: Andrea Bolognani <abologna@redhat.com>
Reviewed-by: Roman Bolshakov <r.bolshakov@yadro.com>
Tested-by: Roman Bolshakov <r.bolshakov@yadro.com>
2021-02-05 11:51:53 +01:00
Yalei Li
b29c86ae10 util: Remove '\n' from vhostuser ifname
When deleting the vhostuserclient interface, OVS prompts that the interface does not exist,
Through the XML file, I found that the "target dev" has a '\n', results in an XML parsing error.

XML file:

<target dev='vm-20ac9c030a47
'/>

That is because 'ovs-vsctl' returns a newline result, always come with a '\n',
and the vircommandrun function puts it in ifname.

So virNetDevOpenvswitchGetVhostuserIfname should remove '\n' from ifname.

Signed-off-by: Yalei Li <liyl43@chinatelecom.cn>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
2021-02-05 09:57:14 +01:00
Laine Stump
c0ae2ca081 datatypes: replace VIR_FREE with g_free in all *Dispose() functions
Signed-off-by: Laine Stump <laine@redhat.com>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
2021-02-05 00:22:09 -05:00
Laine Stump
9709c3418e tests: replace VIR_FREE with g_free in all *Dispose() functions
Signed-off-by: Laine Stump <laine@redhat.com>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
2021-02-05 00:22:09 -05:00
Laine Stump
8334440941 conf: replace VIR_FREE with g_free in all *Dispose() functions
Signed-off-by: Laine Stump <laine@redhat.com>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
2021-02-05 00:22:09 -05:00
Laine Stump
2ca7234d7d util: replace VIR_FREE with g_free in all *Dispose() functions
Signed-off-by: Laine Stump <laine@redhat.com>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
2021-02-05 00:22:09 -05:00
Laine Stump
124d8726e8 security: replace VIR_FREE with g_free in all *Dispose() functions
Signed-off-by: Laine Stump <laine@redhat.com>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
2021-02-05 00:22:09 -05:00
Laine Stump
a2182cf871 rpc: replace VIR_FREE with g_free in all *Dispose() functions
Signed-off-by: Laine Stump <laine@redhat.com>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
2021-02-05 00:22:09 -05:00
Laine Stump
85785b7b87 logging: replace VIR_FREE with g_free in all *Dispose() functions
Signed-off-by: Laine Stump <laine@redhat.com>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
2021-02-05 00:22:09 -05:00
Laine Stump
3aea862b59 hypervisor: replace VIR_FREE with g_free in all *Dispose() functions
Signed-off-by: Laine Stump <laine@redhat.com>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
2021-02-05 00:22:09 -05:00
Laine Stump
22a3f0fa97 access: replace VIR_FREE with g_free in all *Dispose() functions
Signed-off-by: Laine Stump <laine@redhat.com>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
2021-02-05 00:22:09 -05:00
Laine Stump
29293c0869 interface: replace VIR_FREE with g_free in all *Dispose() functions
Signed-off-by: Laine Stump <laine@redhat.com>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
2021-02-05 00:22:09 -05:00