Moving tasks to cgroups implied sched_setaffinity. Changing the cpus in
a set implies the same for all tasks in the group.
The old code put the the thread into the cpuset inherited from the
machine cgroup, which allowed it to run outside of vcpupin for a short
while.
Signed-off-by: Henning Schild <henning.schild@siemens.com>
The machine cgroup is a superset, a parent to the emulator and vcpuX
cgroups. The parent cgroup should never have any tasks directly in it.
In fact the parent cpuset might contain way more cpus than the sum of
emulatorpin and vcpupins. So putting tasks in the superset will allow
them to run outside of <cputune>.
Signed-off-by: Henning Schild <henning.schild@siemens.com>
virCgroupNewMachine used to add the pidleader to the newly created
machine cgroup. Do not do this implicit anymore.
Signed-off-by: Henning Schild <henning.schild@siemens.com>
Firstly, there's a bug (or typo) in the only place where we call
this function: @multiqueue is set whenever @tapfdSize is greater
than zero, while in fact the condition should have been 'greater
than one'.
Then, secondly, since the condition depends on just one
variable, that we are even passing down to the function, we can
move the condition into the function and drop useless argument.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
When user configures vhost-user interface and forgets to also configure
any shared memory, the search for the root cause of non-operational
interface might take unpleasantly long time. Let's enhance user
experience by emitting a warning in the logs.
Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=1266982
Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
Some older systems, e.g. RHEL-6 do not have IFF_MULTI_QUEUE flag
which we use to enable multiqueue feature. Therefore one gets the
following compile error there:
CC util/libvirt_util_la-virnetdevmacvlan.lo
util/virnetdevmacvlan.c: In function 'virNetDevMacVLanTapSetup':
util/virnetdevmacvlan.c:338: error: 'IFF_MULTI_QUEUE' undeclared (first use in this function)
util/virnetdevmacvlan.c:338: error: (Each undeclared identifier is reported only once
util/virnetdevmacvlan.c:338: error: for each function it appears in.)
make[3]: *** [util/libvirt_util_la-virnetdevmacvlan.lo] Error 1
So, whenever user wants us to enable the feature on such systems,
we will just throw a runtime error instead.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
The libvirt file system storage driver determines what file to
act on by concatenating the pool location with the volume name.
If a user is able to pick names like "../../../etc/passwd", then
they can escape the bounds of the pool. For that matter,
virStoragePoolListVolumes() doesn't descend into subdirectories,
so a user really shouldn't use a name with a slash.
Normally, only privileged users can coerce libvirt into creating
or opening existing files using the virStorageVol APIs; and such
users already have full privilege to create any domain XML (so it
is not an escalation of privilege). But in the case of
fine-grained ACLs, it is feasible that a user can be granted
storage_vol:create but not domain:write, and it violates
assumptions if such a user can abuse libvirt to access files
outside of the storage pool.
Therefore, prevent all use of volume names that contain "/",
whether or not such a name is actually attempting to escape the
pool.
This changes things from:
$ virsh vol-create-as default ../../../../../../etc/haha --capacity 128
Vol ../../../../../../etc/haha created
$ rm /etc/haha
to:
$ virsh vol-create-as default ../../../../../../etc/haha --capacity 128
error: Failed to create vol ../../../../../../etc/haha
error: Requested operation is not valid: volume name '../../../../../../etc/haha' cannot contain '/'
Signed-off-by: Eric Blake <eblake@redhat.com>
Commit id '56e2171c6' removed a variable from the argument list, but
neglected to update the ATTRIBUTE_NONNULL values, so when commit id
'08da97bfb' added a couple of arguments, the values were off.
Always return LLONG_MAX even on 32 bit systems. The limitation
originates from our use of "unsigned long" in several APIs. The internal
data type is unsigned long long. Make the test suite deterministic by
removing the architecture difference.
Flaw was introduced in 645881139b where
I've added a test that uses too large numbers.
https://bugzilla.redhat.com/show_bug.cgi?id=1240439
Ta-da! Now that we know how to open a macvtap device multiple
times, we can finally enable the multiqueue feature. Everything
else is already prepared (e.g. command line generation) from the
previous iteration where the feature was implemented for
TUN/TAP devices.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
For the multiqueue on macvtaps we are going to need to open
the device multiple times. Currently, this is not supported.
Rework the function, so that upper layers can be reworked too.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Like we are doing for TUN/TAP devices, we should do the same for
macvtaps. Although, it's not as critical as in that case, we
should do it for the consistency.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
For the multiqueue on macvtaps we are going to need to open
the device multiple times. Currently, this is not supported.
Rework the function, so that upper layers can be reworked too.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
For the multiqueue on macvtaps we are going to need to open
the device multiple times. Currently, this is not supported.
Rework the function, so that upper layers can be reworked too.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
There are few outdated things. Firstly, we don't need to undergo
the torture of fopen, fscanf and fclose just to get the interface
index when we have nice wrapper over that: virNetDevGetIndex.
Secondly, we don't need to have statically allocated buffer for
the path we are opening.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
So yet again one of integer arguments that we use as a boolean.
Since the argument count of the function is unbearably long
enough, lets turn those booleans into flags.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
On the very first log message we send to any output, we include
the libvirt version number and package string. In some bug reports
we have been given libvirtd.log files that came from a different
host than the corresponding /var/log/libvirt/qemu log files. So
extend the initial log message to include the hostname too.
eg on first log message we would now see:
$ libvirtd
2015-12-04 17:35:36.610+0000: 20917: info : libvirt version: 1.3.0
2015-12-04 17:35:36.610+0000: 20917: info : hostname: dhcp-1-180.lcy.redhat.com
2015-12-04 17:35:36.610+0000: 20917: error : qemuMonitorIO:687 : internal error: End of file from monitor
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
https://bugzilla.redhat.com/show_bug.cgi?id=1276198
Prior to commit id '98322052' failure to saferead the block device would
cause an error to be logged and the device to be skipped while attempting
to discover/create a stable target path for a new LUN (NPIV).
This was because virStorageBackendSCSIFindLUs ignored errors from
processLU and virStorageBackendSCSINewLun.
Ignoring the failure allowed a multipath device with an "active" and
"ghost" to be present on the host with the "ghost" block device being
ignored. This patch will return a -2 to the caller indicating the desire
to ignore the block device since it cannot be used directly rather than
fail the pool startup.
I found this useful while processing a volume that wouldn't end up
showing up in the resulting list of block volumes. In this case, the
partition type wasn't found in the disk_types table.
Similar to the openflags VIR_STORAGE_VOL_OPEN_NOERROR processing, if some
read processing operation fails, check the readflags for the corresponding
error flag being set. If so, rather then causing an error - use VIR_WARN
to flag the error, but return -2 which some callers can use to perform
specific actions. Use a new VIR_STORAGE_VOL_READ_NOERROR flag in a new
VolReadErrorMode enum.
While processing the volume for lseek, virFileReadHeaderFD, and
virStorageFileGetMetadataFromBuf - failure would cause an error,
but ret would not be set. That would result in an error message being
sent, but successful status being returned.
Just so it's clearer what to expect upon input and what types of return
values could be generated. These were loosely copied from existing
virStorageBackendUpdateVolTargetInfoFD.
Similar to the openflags which allow VIR_STORAGE_VOL_OPEN_NOERROR to be
passed to avoid open errors, add a 'readflags' variable so that in the
future read failures could also be ignored.
This updates the test program to make it consistent with recent changes
to the mock libraries, and also opens up the possibility of mocking more
than just /sys in the future.
Instead of fakesysfsdir, which is very generic, use fakesysfspcidir and
fakesysfscgroupdir. This makes it explicit what part of the fake sysfs
filesystem they're referring to, and also leaves open the possibility of
handling files in two unrelated parts of the fake sysfs filesystem.
No functional changes.
We might need to mock files living outside SYSFS_PREFIX later on,
so it's better to treat the temporary directory we are passed via
the environment as the root of the fake filesystem and create
SYSFS_PREFIX inside it.
The environment variable name will be changed to reflect the new use
we're making of it in a later commit.
We might need to mock files living outside PCI_SYSFS_PREFIX later on,
so it's better to treat the temporary directory we are passed via
the environment as the root of the fake filesystem and create
PCI_SYSFS_PREFIX inside it.
The environment variable name will be changed to reflect the new use
we're making of it in a later commit.
Add qemuDomainHasVCpuPids to do the checking and replace in place checks
with it.
We no longer need checking whether the thread contains fake data
(vcpupids[0] == vm->pid) as in b07f3d821d
and 65686e5a81 this was removed.
The vCPU threads make sense in the counterparts that set the vCPU
bandwidth/quota, not in the emulator one. The emulator tunables are set
all the time anyways.
Drop the extra check and remove the now unneeded vm argument.
Since commit 0c04906fa the check for priv->cgroup doesn't make sense as
the calls to virCgroupHasController return the same information. Remove
it and move it's comment partially to the new check.
The already spurious check was also later copied to the iothreads code.
Once more stuff will be moved into the vCPU data structure it will be
necessary to get a specific one in some ocasions. Add a helper that will
simplify this task.
Refactor the code flow so that 'exit_monitor:' can be removed.
This patch moves the auditing functions into places where it's certain
that hotunplug was or was not successful and reports errors from
qemuMonitorGetCPUInfo properly.
Refactor the code flow so that 'exit_monitor:' can be removed.
This patch also moves the auditing and setting of the new vCPU count
right to the place where the hotplug happens, since it's possible that
the hotplug succeeds and adds a cpu while other stuff fails.
Lastly, failures of qemuMonitorGetCPUInfo are now reported rather than
ignored. The function retuns 0 if it "successfully" detected 0 threads.
qemuDomainHotplugVcpus/qemuDomainHotunplugVcpus are complex enough in
regards of adding one CPU. Additionally it will be desired to reuse
those functions later with specific vCPU hotplug.
Move the loops for adding vCPUs into qemuDomainSetVcpusFlags so that the
helpers can be made simpler and more straightforward.
The cpu hotplug helper functions used negative error handling in a part
of them, although some code that was added later didn't properly set the
error codes in some cases. This would cause improper error messages in
cases where we couldn't modify the numa cpu mask and a few other cases.
Fix the logic by converting it to the regularly used pattern.
With a very unfortunate timing, the agent might vanish before we do the
second call while the locks were down. Re-check that the agent is
available before attempting it again.