Laine Stump [Sun, 17 Jan 2021 19:27:20 +0000 (14:27 -0500)]
rpm: disable netcf for the interface driver in rpm build on new targets
libvirt.spec currently adds a hardcoded -Dnetcf=enabled to the meson
commandline, so just setting the default in the meson.build file won't
have any effect for rpm builds - it will be overridden.
This patch changes the meson commandline in the spec file from
hardcoded -Dnetcf=enabled to %{arg_netcf}, which is itself set
according to the value of %{with_netcf}; and *that* is normally set
according to the distro release of the build target (1 for Fedora >=
34 and RHEL >= 9, 0 otherwise), but can be manually overridden by
adding "-without netcf" to the rpmbuild commandline.
Along with being used to determine what arg to pass to meson,
%{with_netcf} is also checked when deciding on whether or not to add
netcf build time / install time dependencies ("Requires: netcf-libs"
and "BuildRequires: netcf-devel")
Signed-off-by: Laine Stump <laine@redhat.com> Reviewed-by: Neal Gompa <ngompa13@gmail.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Laine Stump [Thu, 21 Jan 2021 21:01:06 +0000 (16:01 -0500)]
build: support explicitly disabling netcf
placing "-Dnetcf=disabled" on the meson commandline was ignored,
meaning that even with that option the build would get WITH_NETCF if
the netcf-devel package was found - the only way to disable it was to
uninstall netcf-devel.
This patch adds the small bit of logic to check the netcf meson
commandline option (in addition to whether netcf-devel is installed)
before defining WITH_NETCF.
Signed-off-by: Laine Stump <laine@redhat.com> Reviewed-by: Neal Gompa <ngompa13@gmail.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
The current virPCIDeviceNew() signature, receiving 4 uints in sequence
(domain, bus, slot, function), is not neat.
We already have a way to represent a PCI address in virPCIDeviceAddress
that is used in the code. Aside from the test files, most of
virPCIDeviceNew() callers have access to a virPCIDeviceAddress reference,
but then we need to retrieve the 4 required uints (addr.domain, addr.bus,
addr.slot, addr.function) to satisfy virPCIDeviceNew(). The result is
that we have extra verbosity/boilerplate to retrieve an information that
is already available in virPCIDeviceAddress.
A better way is presented by virNVMEDeviceNew(), where the caller just
supplies a virPCIDeviceAddress pointer and the function handles the
details internally.
This patch changes virPCIDeviceNew() to receive a virPCIDeviceAddress
pointer instead of 4 uints.
Reviewed-by: Laine Stump <laine@redhat.com> Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
qemu, lxc: move NodeDeviceGetPCIInfo() function to domain_driver.c
libxlNodeDeviceGetPCIInfo() and qemuNodeDeviceGetPCIInfo() are equal.
Let's move the logic to a new virDomainDriverNodeDeviceGetPCIInfo()
info to be used by libxl_driver.c and qemu_driver.c.
Reviewed-by: Laine Stump <laine@redhat.com> Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Boris Fiuczynski [Fri, 29 Jan 2021 11:39:22 +0000 (12:39 +0100)]
conf: rename virDomainCheckVirtioOptions
Rename virDomainCheckVirtioOptions into
virDomainCheckVirtioOptionsAreAbsent since it checks if all
virtio options are absent. The old name was very misleading.
Signed-off-by: Boris Fiuczynski <fiuczy@linux.ibm.com> Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
Signed-off-by: Boris Fiuczynski <fiuczy@linux.ibm.com> Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com>
The virDomainVirtioOptionsCheckABIStability() function is called
from various ABI stability check functions. Every caller checks
if both old and new definitions have virtio options set and only
after that they call the function. This is suboptimal because:
a) this check can be done in the function itself (making all
callers shorter),
b) is inherently wrong, because it doesn't catch case where one
definition has virtio options set and the other doesn't.
Do proper checks at the beginning of the function and simplify
its calls.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Michal Privoznik [Thu, 28 Jan 2021 13:06:55 +0000 (14:06 +0100)]
conf: Move virDomainCheckVirtioOptions() into domain_validate.c
The aim of virDomainCheckVirtioOptions() function is to check
whether no virtio options are set, i.e. no @iommu no @ats and no
@packed attributes were present in given device's XML (yeah, the
function has very misleading name). Nevertheless, this kind of
check belongs to validation phase, but now is done in post parse
phase. Move the function and its calls to domain_validate.c so
that future code is not tempted to repeat this mistake.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
While the PCI docs are linked from formatdomain.html, finding those
links is not straightforward. It is good for users to highlight them in
the kbase pages. The PCI docs are intentionally not moved to the kbase/
sub-directory in order to avoid breaking hyperlinks.
Reviewed-by: Andrea Bolognani <abologna@redhat.com> Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Notable changes detected by libvirt are the new machine types and
'intel-pt-lip', 'avx512-fp16', 'kvm-msi-ext-dest-id' cpu features
reported by qemu.
Other qemu changes not detected by libvirt include removal of the
'change' command, addition of 'sev-inject-launch-secret', 'yank',
'query-yank' commands and other device properties.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Erik Skultety [Tue, 26 Jan 2021 16:15:32 +0000 (17:15 +0100)]
spec: Drop gettext-devel from the SPEC file
With meson, we don't need the gettext headers anymore, meson takes care
of that and we only need to have xgettext installed.
Without this patch RPM build in Fedora containers fails.
Signed-off-by: Erik Skultety <eskultet@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
qemu: fix release of virDomainObjPtr in SSH key APIs
The qemuDomainObjFromDomain() API must be paired with
the virDomainObjEndAPI API. The qemuDomainAuthorizedSSHKeysGet
method simply did 'return -1' leaking a reference and lock
in two paths.
The qemuDomainAuthorizedSSHKeysSet method marked the object
as an autoptr while also have some code paths that will call
virDomainObjEndAPI. As a result the object will be released
but not unlocked in error paths.
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com> Reviewed-by: Michal Privoznik <mprivozn@redhat.com> Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Michal Privoznik [Fri, 22 Jan 2021 09:25:45 +0000 (10:25 +0100)]
qemu: Avoid crash in qemuStateShutdownPrepare() and qemuStateShutdownWait()
If QEMU driver fails to initialize for whatever reason (it can be
as trivial as a typo on qemu.conf), the control jumps to error
label in qemuStateInitialize() where qemuStateCleanup() is called
which frees the driver. But the daemon then asks drivers to
prepare for shutdown, which in case of QEMU driver is implemented
in qemuStateShutdownPrepare(). In here, the driver is
dereferenced but since it was freed earlier, the pointer is NULL
which leads to instant crash.
Solution is simple - just check if qemu_driver is not NULL. But
doing so only in qemuStateShutdownPrepare() would push the
problem down to virStateShutdownWait(), well
qemuStateShutdownWait(). Therefore, duplicate the trick there
too.
Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=1895359#c14 Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Jiri Denemark <jdenemar@redhat.com>
Peter Krempa [Mon, 25 Jan 2021 14:02:26 +0000 (15:02 +0100)]
virStorageSourceChainLookup: Handle names like 'vda[4]' internally
All callers of this function called virStorageFileParseChainIndex
before. Internalize the logic of that function to prevent multiple calls
and passing around unnecessary temporary variables.
This is achieved by calling virStorageFileParseBackingStoreStr and using
it to fill the values internally.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Peter Krempa [Mon, 25 Jan 2021 13:57:35 +0000 (14:57 +0100)]
test: storage: Remove double testing in testStorageLookup
The function attempts two calls to virStorageSourceChainLookup to see
whether the function handles NULL correctly. This isn't very useful and
additionally upcoming patch will remove the 'idx' parameter thus the
test becomes obsolete. Remove it.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Peter Krempa [Fri, 22 Jan 2021 16:16:38 +0000 (17:16 +0100)]
virt-aa-helper: Don't probe image metadata for terminated chains
A terminated chain has a virStorageSource with type ==
VIR_STORAGE_TYPE_NONE at the end. Since virStorageSourceHasBacking
is explicitly returning false in that case we'd probe the chain
needlessly. Just check whether src->backingStore is non-NULL.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Peter Krempa [Fri, 22 Jan 2021 15:03:53 +0000 (16:03 +0100)]
storage_source: Move backing store parsers into new file
The parsers for the backing store strings are relatively self-contained
and rather massive piece of code. Move them to a new module called
storage_source_backingstore.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Peter Krempa [Fri, 22 Jan 2021 14:08:23 +0000 (15:08 +0100)]
virStorageSourceGetBackingStoreStr: Return relative paths only
Rename the function to virStorageSourceFetchRelativeBackingPath and
return relative paths only. The function is only used to restore the
relative relationship between images so there's no need for it to be
universal.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Peter Krempa [Fri, 22 Jan 2021 14:15:01 +0000 (15:15 +0100)]
virStorageSourceGetBackingStoreStr: Move the function earlier
Move it together with virStorageSourceGetRelativeBackingPath which is
the main reason why it exists. Upcoming patch will modify the comment
and arguments refering to virStorageSourceGetRelativeBackingPath so it's
better if they are together.
Signed-off-by: Peter Krempa <pkrempa@redhat.com> Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Michal Privoznik [Mon, 25 Jan 2021 12:48:31 +0000 (13:48 +0100)]
tools: Set IFS for bash completion script
The way our bash completion string is that is gets user's input
and lets virsh completion code do all the work by calling 'virsh
complete -- $INPUT". The 'complete' command is a "secret",
unlisted command that exists solely for this purpose. After it
has done it's part, it prints candidates onto stdout, each
candidate on its own line, e.g. like this:
Bash interprets that as another array item because by default,
Internal Field Separator (IFS) = set of characters that bash uses
to split words at, is: space, TAB, newline. We don't want space
nor TAB. Therefore, we have to set $IFS when storing 'virsh
complete' output into the array.
Thanks to Peter who suggested it.
Resolves: https://gitlab.com/libvirt/libvirt/-/issues/116 Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
vsh: Allow double quotes imbalance for auto completion in vshCommandStringGetArg()
If user is trying to auto complete a value that contains a space,
they have two options: use backslash to escape space or use
quotes, like this:
virsh # start --domain "domain with space<TAB>
However, in this case our tokenizer sees imbalance in (double)
quotes: there is a starting one that's missing its companion.
Well, that's obvious - user is still in process of writing the
command. What we need to do in this case is to ignore the
imbalance and return success (from the tokenizer) - readline will
handle closing the quote properly.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Michal Privoznik [Tue, 26 Jan 2021 08:51:27 +0000 (09:51 +0100)]
vsh: Rework how option to complete is found
The way that auto completion works currently is that user's input
is parsed, and then we try to find the first --option (in the
parsed structure) that has the same value as user's input around
where <TAB> was pressed. For instance, for the following input:
virsh # command --arg1 hello --arg2 world<TAB>
we will see "world" as text that user is trying to autocomplete
(this is affected by rl_basic_word_break_characters which
readline uses internally to break user's input into individual
words) and find that it is --arg2 that user is trying to
autocomplete. So far so good, for this naive approach. But
consider the following example:
virsh # command --arg1 world --arg2 world<TAB>
Here, both arguments have the same value and because we see
"world" as text that user is trying to autocomplete we would
think that it is --arg1 that user wants to autocomplete. This is
obviously wrong.
Fortunately, readline stores the current position of cursor (into
rl_point) and we can use that when parsing user's input: whenever
we reach a position that matches the cursor then we know that
that is the place where <TAB> was pressed and hence that is the
--option that user wants to autocomplete. Readline stores the
cursor position as offset (numbered from 1) from the beginning of
user's input. We store this input into @parser->pos initially,
but then advance it as we tokenize it. Therefore, what we need is
to store the original position too.
Thanks to Martin who helped me with this.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Michal Privoznik [Tue, 26 Jan 2021 08:23:24 +0000 (09:23 +0100)]
vshReadlineParse: Escape list of candidates earlier
The way our completer callbacks work is that they return all
possible candidates and then vshCompleterFilter() is called to
prune the list of all candidates removing those which don't match
user's input. This allows us to have simpler completer callbacks
as their only job is to fetch all possible candidates.
Anyway, if the completion candidate we're returning contains a
space, it has to be escaped (shell like escaping), unless there
is already a quote character (single quote or double quote).
But ordering is critical. Completer callback returns string
without any escaping, but the filter function sees the user input
escaped. For instance, if user's input is "domain with
space<TAB>" then the filtering function gets "domain\ with\
space" as user's input but completer returns "domain with space".
Since these two strings don't match the filtering function
removes this candidate from the list. What we need to do is to
escape strings before calling the filtering function. This way,
the filtering function will see two same strings.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Michal Privoznik [Tue, 26 Jan 2021 08:23:10 +0000 (09:23 +0100)]
vshReadlineParse: Rename @buf to @line
In next commit the block that does escaping of returned string
will be brought into this block. But both contain variable @buf
and use it in different contexts. Rename @buf from @state == 0
block to @line which reflects its purpose better.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Michal Privoznik [Tue, 26 Jan 2021 08:18:21 +0000 (09:18 +0100)]
vshReadlineParse: Bring some variables into !state block
On readline completion vshReadlineCompletion() is called which
does nothing more than calling rl_completion_matches() with
vshReadlineParse() as a callback. This means, that
vshReadlineParse() is called repeatedly, each time returning next
completion candidate, until it returns NULL which is interpreted
as the end of the list of candidates.
The function takes two parameters: @text which is a portion of
input line around cursor when TAB was pressed, and @state. The
@state is an integer that is zero on the very first call and
non-zero on each subsequent call (in fact, readline does @state++
on each call).
Anyway, the idea is that the callback gets the whole list of
candidates on @state == 0 and returns one candidate at each call.
And this is what vshReadlineParse() is doing but some variables
(@partial, @cmd and @opt) are really used only in the @state == 0
case but declared for whole function. We can limit their scope by
declaring them inside the @state == 0 body which also means that
they don't have to be static anymore.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
A backslash is the way we escape characters in virsh. For
instance:
virsh # start domain\ with\ long\ name
For readline completion, we do not want to get four separate
words ("domain", "with", "long", "name"). This means, that we
can't use virBufferEscapeShell() because it doesn't escape spaces
the way we want.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Ján Tomko <jtomko@redhat.com>
Jiri Denemark [Fri, 22 Jan 2021 14:04:42 +0000 (15:04 +0100)]
cpu_map: Remove intel-pt from x86 CPU models
As explained in QEMU commit 4c257911dcc7c4189768e9651755c849ce9db4e8
intel-pt features should never be included in the CPU models as it was
not supported by KVM back then and even once it started to be supported,
users have to enable it by passing pt_mode=1 parameter to kvm_intel
module. The Icelake-* CPU models with intel-pt included were added to
QEMU 3.1.0 and removed right in the following 4.0.0 release (and even in
3.1.1 maintenance release).
In libvirt 6.10.0 I introduced 'removed' attribute for features included
in our CPU model definitions which we can use to drop intel-pt from
Icelake-* CPU models. Back then I explained we can safely do so only for
features which could never be enabled, which is not the case of intel-pt.
Theoretically, it could be possible to create an environment in which
QEMU would enable intel-pt without asking for it explicitly: it would
need to use a new enough kernel (not available at the time of QEMU
3.1.0) and pt_mode KVM parameter in combination with QEMU 3.1.0 running
a domain with q35 machine type and all that on a CPU which didn't really
exist at that time.
Migrating such domain to a host with newer SW stack including libvirt
with this patch applied would result in incompatible guest ABI (the
virtual CPU would lose intel-pt). However, QEMU changed its CPU models
unconditionally and thus migration would not work even without this
patch. That said, it is safe to follow QEMU and remove the feature from
Icelake-* CPU models in our cpu_map.
Cédric Bosdonnat [Mon, 25 Jan 2021 13:56:01 +0000 (14:56 +0100)]
Fix format network dns doc
The code block on the srv name in the formatnetwork page is confusing
since the actual parameter is service. Moving the code block to the
service work makes it better.
Reviewed-by: Andrea Bolognani <abologna@redhat.com> Signed-off-by: Cédric Bosdonnat <cbosdonnat@suse.com>
gitlab: force dwarf4 format for debuginfo in Fedora rawhide
Fedora 34 rawhide has pulled in a new GCC 11 build which now
defaults to dwarf5 format. This format is not compatible with
the pdwtags program used in our test suite to validate the
RPC files.
We have no need for debuginfo in CI except for pdwtags,
so the simplest short term fix is to force the older dwarf
version in the hope that a fixed dwarves release will
arrive before Fedora 34 is released, or GCC 11 becomes more
widespread. Eventually we might need to figure out a way to
probe for compatibility but for now, we'll hope that any
distro with GCC 11 will be able to have a fixed dwarves too.
https://bugzilla.redhat.com/show_bug.cgi?id=1919965 Reviewed-by: Erik Skultety <eskultet@redhat.com> Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Helmut Grohne [Mon, 18 Jan 2021 23:08:23 +0000 (00:08 +0100)]
meson: Fix cross-building of dtrace probes
dtrace invokes the C compiler, so when cross-building we need
to make sure that $CC is set in the environment and that it
points to the cross-compiler rather than the native one.
Until https://github.com/mesonbuild/meson/issues/266
is addressed, the workaround is to call dtrace via env(1).
Signed-off-by: Helmut Grohne <helmut@subdivi.de> Reviewed-by: Andrea Bolognani <abologna@redhat.com> Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Dmytro Linkin [Thu, 21 Jan 2021 12:15:22 +0000 (14:15 +0200)]
util: Add phys_port_name support on virPCIGetNetName
virPCIGetNetName is used to get the name of the netdev associated with
a particular PCI device. This is used when we have a VF name, but need
the PF name in order to send a netlink command (e.g. in order to
get/set the MAC address of the VF).
In simple cases there is a single netdev associated with any PCI
device, so it is easy to figure out the PF netdev for a VF - just look
for the PCI device that has the VF listed in its "virtfns" directory;
the only name in the "net" subdirectory of that PCI device's sysfs
directory is the PF netdev that is upstream of the VF in question.
In some cases there can be more than one netdev in a PCI device's net
directory though. In the past, the only case of this was for SR-IOV
NICs that could have multiple PF's per PCI device. In this case, all
PF netdevs associated with a PCI address would be listed in the "net"
subdirectory of the PCI device's directory in sysfs. At the same time,
all VF netdevs and all PF netdevs have a phys_port_id in their sysfs,
so the way to learn the correct PF netdev for a particular VF netdev
is to search through the list of devices in the net subdirectory of
the PF's PCI device, looking for the one netdev with a "phys_port_id"
matching that of the VF netdev.
But starting in kernel 5.8, the NVIDIA Mellanox driver began linking
the VFs' representor netdevs to the PF PCI address [1], and so the VF
representor netdevs would also show up in the net
subdirectory. However, all of the devices that do so also only have a
single PF netdev for any given PCI address.
This means that the net directory of the PCI device can still hold
multiple net devices, but only one of them will be the PF netdev (the
others are VF representors):
$ ls '/sys/bus/pci/devices/0000:82:00.0/net'
ens1f0 eth0 eth1
In this case the way to find the PF device is to look at the
"phys_port_name" attribute of each netdev in sysfs. All PF devices
have a phys_port_name matching a particular regex
(p[0-9]+$)|(p[0-9]+s[0-9]+$)
Since there can only be one PF in the entire list of devices, once we
match that regex, we've found the PF netdev.
Co-Authored-by: Moshe Levi <moshele@nvidia.com> Signed-off-by: Dmytro Linkin <dlinkin@nvidia.com> Reviewed-by: Adrian Chiris <adrianc@nvidia.com> Reviewed-by: Laine Stump <laine@redhat.com>
Moshe Levi [Thu, 21 Jan 2021 12:15:21 +0000 (14:15 +0200)]
util: add virNetDevGetPhysPortName
This commit add virNetDevGetPhysPortName to read netdevice
phys_port_name from sysfs. It also refactor the code so
virNetDevGetPhysPortName and virNetDevGetPhysPortID will use
same method to read the netdevice sysfs.
Signed-off-by: Moshe Levi <moshele@nvidia.com> Reviewed-by: Laine Stump <laine@redhat.com>
projects / thirdparty / libvirt.git / log
