Currently, any attempt to change MTU on an interface that is
plugged to a running domain is silently ignored. We should either
do what's asked or error out. Well, we can update the host side
of the interface, but we cannot change 'host_mtu' attribute for
the virtio-net device. Therefore we have to error out.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Laine Stump <laine@laine.org>
While implementing MTU (572eda12ad and friends), I've forgotten
to actually set MTU on the host NIC in case of hotplug. We
correctly tell qemu on the monitor what the MTU should be, but we
are not actually setting it on the host NIC.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Laine Stump <laine@laine.org>
We try to get the last element of the passed path by calling
strrch(path, '/'). However, the pointer that strrchr() returns
points at the slash, We want string that starts right after that.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Currently, we are querying for vhostuser interface name in post
parse callback. At that time interface might not yet exist.
However, it has to exist when starting domain. Therefore it makes
more sense to query its name at that point. This partially
reverts 57b5e27.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
virNetClientStreamQueuePacket: Set st->incomingEOF on the end of stream
While reworking client side of streams, I had to postpone payload
decoding so that stream holes and stream data can be
distinguished in virNetClientStreamRecvPacket. That's merely what 18944b7aea46d does. However, I accidentally removed one important
bit: when server sends us an empty STREAM packet (with no
payload) - meaning end of stream - st->incomingEOF flag needs to
be set. It used to be before I touched the code. After I removed
it, virNetClientStreamRecvPacket will try to fetch more data from
the stream, but it will never come.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com> Reviewed-by: Jim Fehlig <jfehlig@suse.com>
In 9cb891141c we've introduced some logic to clearing suggested
macvtap/macvlan ifnames. The logic consists of comparing ifname
string with strings that libvirt would generate. However, due to
a typo only VIR_NET_GENERATED_MACVTAP_PREFIX was compared. Twice.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Laine Stump [Tue, 6 Jun 2017 16:07:49 +0000 (12:07 -0400)]
docs: correct improper information about domain states in virsh manpage
Commit 24d4a0a1f removed the non-existent "dying" state from the list
of possible domain states given in the virsh manpage, but didn't
correct the count of states from 8 down to 7. This patch fixes that
mismatch by completely removing any reference to the exact number of
states (thus preventing a potential future mismatch), while wording
the sentence in a more readable/truthful manner.
Erik Skultety [Mon, 29 May 2017 10:58:34 +0000 (12:58 +0200)]
qemu: Fix serial stub console allocation
When adding the aliased serial stub console, the structure wasn't
properly allocated (VIR_ALLOC instead of virDomainChrDefNew) which then
resulted in SIGSEGV in virDomainChrSourceIsEqual during a serial device
coldplug.
Jiri Denemark [Wed, 31 May 2017 10:34:10 +0000 (12:34 +0200)]
qemu: Use updated CPU when starting QEMU if possible
If QEMU is new enough and we have the live updated CPU definition in
either save or migration cookie, we can use it to enforce ABI. The
original guest CPU from domain XML will be stored in private data.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com> Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Jiri Denemark [Fri, 2 Jun 2017 20:52:03 +0000 (22:52 +0200)]
qemu: Store updated CPU in save cookie
Since the domain XML saved in a snapshot or saved image uses the
original guest CPU definition but we still want to enforce ABI when
restoring the domain if libvirt and QEMU are new enough, we save the
live updated CPU definition in a save cookie.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com> Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Jiri Denemark [Fri, 26 May 2017 15:06:56 +0000 (17:06 +0200)]
qemu: Send updated CPU in migration cookie
Since the domain XML send during migration uses the original guest CPU
definition but we still want the destination to enforce ABI if it is new
enough, we send the live updated CPU definition in a migration cookie.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com> Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Jiri Denemark [Wed, 7 Jun 2017 07:35:25 +0000 (09:35 +0200)]
qemu: Always send persistent XML during migration
When persistent migration of a transient domain is requested but no
custom XML is passed to the migration API we would just let the
destination daemon make a persistent definition from the live definition
itself. This is not a problem now, but once the destination daemon
starts replacing the original CPU definition with the one from migration
cookie before starting a domain, it would need to add more ugly hacks to
reverse the operation. Let's just always send the persistent definition
in the cookie to make things a bit cleaner.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com> Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Jiri Denemark [Fri, 19 May 2017 13:20:31 +0000 (15:20 +0200)]
qemu: Report the original CPU in migratable xml
The destination host may not be able to start a domain using the live
updated CPU definition because either libvirt or QEMU may not be new
enough. Thus we need to send the original guest CPU definition.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com> Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Jiri Denemark [Tue, 16 May 2017 11:26:54 +0000 (13:26 +0200)]
qemu: Remember CPU def from domain start
When starting a domain we update the guest CPU definition to match what
QEMU actually provided (since it is allowed to add or removed some
features unless check='full' is specified). Let's store the original CPU
in domain private data so that we can use it to provide a backward
compatible domain XML.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com> Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Jiri Denemark [Thu, 1 Jun 2017 22:44:06 +0000 (00:44 +0200)]
conf: Introduce virSaveCookie
The code will be used by snapshots and domain save/restore code to store
additional data for a saved running domain. It is analogous to migration
cookies, but simple and one way only.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com> Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Jiri Denemark [Tue, 6 Jun 2017 21:41:57 +0000 (23:41 +0200)]
qemu: Refactor qemuDomainSaveHeader
The function is now called virQEMUSaveDataWrite and it is now doing
everything it needs to save both the save image header and domain XML to
a file. Be it a new file or an existing file in which a user wants to
change the domain XML.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com> Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Jiri Denemark [Tue, 23 May 2017 18:32:07 +0000 (20:32 +0200)]
qemu: Rename xml_len in virQEMUSaveHeader as data_len
Since virQEMUSaveHeader will be followed by more than just domain XML,
the old name would be confusing as it was designed to describe the
length of all data following the save image header.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com> Reviewed-by: Pavel Hrdina <phrdina@redhat.com>
Peter Krempa [Tue, 6 Jun 2017 13:13:51 +0000 (15:13 +0200)]
qemu: Conditionally allow block-copy for persistent domains
Allow starting the block-copy job for a persistent domain if a user
declares by using a flag that the job will not be recovered if the VM is
switched off while the job is active.
This allows to use the block-copy job with persistent VMs under the same
conditions as would apply to transient domains.
John Ferlan [Mon, 15 May 2017 18:31:44 +0000 (14:31 -0400)]
interface: Convert virInterfaceObj to use virObjectLockable
Now that we have a bit more control, let's convert our object into
a lockable object and let that magic handle the create and lock/unlock.
This commit also introduces virInterfaceObjEndAPI in order to handle the
lock unlock and object unref in one call for consumers returning a NULL
obj upon return. This removes the need for virInterfaceObj{Lock|Unlock}
external API's.
John Ferlan [Fri, 2 Jun 2017 11:56:59 +0000 (07:56 -0400)]
interface: Consume @def in virInterfaceObjNew
Move the consumption of @def in virInterfaceObjNew and then handle that
in the error path of virInterfaceObjListAssignDef since it's caller expects
to need to free @def when NULL is returned and so would virInterfaceObjFree.
Peter Krempa [Tue, 6 Jun 2017 05:39:25 +0000 (07:39 +0200)]
qemu: process: Save vcpu ordering information on reconnect
vCPU ordering information would not be updated if a vCPU emerged or
disappeared during the time libvirtd is not running. This allowed to
create invalid configuration like:
virFDStreamThread: Make sure we won't exceed @length
There's a problem with current streams after I switched them from
iohelper to thread implementation. Previously, iohelper made sure
not to exceed specified @length resulting in the pipe EOF
appearing at the exact right moment (the pipe was used to tunnel
the data from the iohelper to the daemon). Anyway, when switching
to thread I had to write the I/O code from scratch. Whilst doing
that I took an inspiration from the iohelper code, but since the
usage of pipe switched to slightly different meaning, there was
no 1:1 relationship between the codes.
Moreover, after introducing VIR_FDSTREAM_MSG_TYPE_HOLE, the
condition that should made sure we won't exceed @length was
completely wrong.
The fix is to:
a) account for holes for @length
b) cap not just data sections but holes too (if @length would be
exceeded)
For this purpose, the condition needs to be brought closer to the
code that handles holes and data sections.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
They do the same thing with only one difference. Let's put them
together (like we already do with virFDStreamCloseInt) so that future
changes don't miss one of the implementations. Also to clean up the
code.
Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
The memset() was resetting only 30 bytes in the array (size of the
array), but it is array of pointers. Since it is a static array,
let's just reset it by its size.
Found by gcc-7.1:
testutils.c: In function 'virTestRun':
testutils.c:243:13: error: 'memset' used with length equal to number
of elements without multiplication by element size [-Werror=memset-elt-size]
memset(testAllocStack, 0, ARRAY_CARDINALITY(testAllocStack));
^~~~~~
Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
While checking for ABI stability, drivers might pose additional
checks that are not valid for general case. For instance, qemu
driver might check some memory backing attributes because of how
qemu works. But those attributes may work well in other drivers.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
John Ferlan [Fri, 12 May 2017 17:07:43 +0000 (13:07 -0400)]
nodedev: Remove privateData from virNodeDeviceObj
It was only ever used in node_device_hal.c which really never used it
anyway since the NODE_DEV_UDI was never referenced. Remove free_udi()
and @privData as well as the references to obj->privateData & obj->privateFree.
John Ferlan [Fri, 12 May 2017 16:24:56 +0000 (12:24 -0400)]
nodedev: Create helper for finding by name in driver
Create nodeDeviceObjFindByName which will perform the corresponding
virNodeDeviceObjFindByName call for various node_device_driver callers
rather than having the same repetitive code.
John Ferlan [Fri, 12 May 2017 15:49:18 +0000 (11:49 -0400)]
nodedev: Cleanup driver code and prototypes
Alter the node_device_driver source and prototypes to follow more
recent code style guidelines w/r/t spacing between functions, format
of the function, and the prototype definitions.
While the new names for nodeDeviceUpdateCaps, nodeDeviceUpdateDriverName,
and nodeDeviceGetTime don't follow exactly w/r/t a "vir" prefix, they
do follow other driver nomenclature style.
John Ferlan [Thu, 25 May 2017 19:43:45 +0000 (15:43 -0400)]
nodedev: Use switch for virNodeDeviceObjHasCap and virNodeDeviceCapMatch
In order to ensure that whenever something is added to virNodeDevCapType
that both functions are considered for processing of a new capability,
change the if-then-else construct into a switch statement.
John Ferlan [Tue, 16 May 2017 17:21:32 +0000 (13:21 -0400)]
nodedev: Need to check for vport capable scsi_host for vHBA searches
When searching for an NPIV capable fc_host, not only does there need to
be an "fc_host" capability with the specified wwnn/wwpn or fabric_wwn,
but that scsi_host must be vport capable; otherwise, one could end up
picking an exising vHBA/NPIV which wouldn't be good.
Currently not a problem since scsi_hosts are in an as found forward linked
list and the vport capable scsi_hosts will always appear before a vHBA by
definition. However, in the near term future a hash table will be used to
lookup the devices and that could cause problems for these algorithms.
John Ferlan [Sat, 15 Apr 2017 11:08:01 +0000 (07:08 -0400)]
interface: Clean up virInterfaceObjListFindByMACString
Alter the algorithm to return a list of matching names rather than a
list of match virInterfaceObjPtr which are then just dereferenced
extracting the def->name and def->mac. Since the def->mac would be
the same as the passed @mac, just return a list of names and as long
as there's only one, extract the [0] entry from the passed list.
Also alter the error message on failure to include the mac that wasn't
found.
John Ferlan [Fri, 14 Apr 2017 19:45:27 +0000 (15:45 -0400)]
interface: Make _virInterfaceObjList struct private
Move the structs into virinterfaceobj.c, create necessary accessors, and
initializers.
This also includes reworking virInterfaceObjListClone to handle receiving
a source interfaces list pointer, creating the destination interfaces object,
and copying everything from source into dest.
John Ferlan [Fri, 14 Apr 2017 13:05:21 +0000 (09:05 -0400)]
interface: Remove some unnecessary goto's for Interface tests
Rather than using goto cleanup on object find failure and having cleanup
need to check if the obj was present before unlocking, just return immediately.
Peter Krempa [Wed, 31 May 2017 07:12:09 +0000 (09:12 +0200)]
daemon: Don't initialize SASL context if not necessary
SASL context would be initialized even if the corresponding TCP or TLS
sockets are not enabled.
fe772f24a68 attempted to fix the symptom by commenting out the settings,
but that did not fix the root cause. 3c647ee4bbb later reverted those
changes so that the more secure algorithm is used.
Peter Krempa [Mon, 22 May 2017 15:44:00 +0000 (17:44 +0200)]
qemu: Don't error out if allocation info can't be queried
qemuDomainGetBlockInfo would error out if qemu did not report
'wr_highest_offset'. This usually does not happen, but can happen
briefly during active layer block commit. There's no need to report the
error, we can simply report that the disk is fully alocated at that
point.
Julio Faracco [Fri, 2 Jun 2017 01:56:09 +0000 (22:56 -0300)]
util: remove dead code inside virstoragefile
The host address or the socket path have already been checked at the
begining of the function virStorageSourceParseNBDColonString(). So,
when the parameter is not a unix socket, there is no reason to check
the address again because if it does not exists, the logic will fail
in the first IF conditional.
Signed-off-by: Julio Faracco <jcfaracco@gmail.com>
VIR_STRDUP returns -1 if the string copy was not successful. So, the
current comparison/logic is throwing an error when VIR_STRDUP() returns
1. Only when source is NULL, it is considering as a success which is
not right.
Signed-off-by: Julio Faracco <jcfaracco@gmail.com>
Michal Privoznik [Sat, 27 May 2017 10:24:30 +0000 (12:24 +0200)]
qemu: mkdir memory_backing_dir on startup
In 48d9e6cdcc and friends we've allowed users to back guest
memory by a file inside the host. And in order to keep things
manageable the memory_backing_dir variable was introduced to
qemu.conf to specify the directory where the files are kept.
However, libvirt's policy is that directories are created on
domain startup if they don't exist. We've missed this one.
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Erik Skultety [Tue, 30 May 2017 08:28:18 +0000 (10:28 +0200)]
qemu: json: Fix daemon crash on handling domain shutdown event
commit a8eba5036 added further checking of the guest shutdown cause, but
this enhancement is available since qemu 2.10, causing a crash because
of a NULL pointer dereference on older qemus.
Thread 1 "libvirtd" received signal SIGSEGV, Segmentation fault.
0x00007ffff72441af in virJSONValueObjectGet (object=0x0,
key=0x7fffd5ef11bf "guest")
at util/virjson.c:769
769 if (object->type != VIR_JSON_TYPE_OBJECT)
(gdb) bt
0 in virJSONValueObjectGet
1 in virJSONValueObjectGetBoolean
2 in qemuMonitorJSONHandleShutdown
3 in qemuMonitorJSONIOProcessEvent
4 in qemuMonitorJSONIOProcessLine
5 in qemuMonitorJSONIOProcess
6 in qemuMonitorIOProcess
Signed-off-by: Erik Skultety <eskultet@redhat.com>
Nitesh Konkar [Thu, 25 May 2017 10:17:43 +0000 (15:47 +0530)]
util: hostcpu: Correctly report total number of vcpus in virHostCPUGetMap
Callers expect the return value to be the total number of vcpus in the
host (including offline vcpus). The refactor in c67e04e25fa58104e0fae41
broke this assumption by using virHostCPUGetOnlineBitmap which only
creates a bitmap long enough to hold the last online vcpu.
Report the full number of host vcpus by returning value from
virHostCPUGetCount().
Signed-off-by: Nitesh Konkar <nitkon12@linux.vnet.ibm.com> Signed-off-by: Peter Krempa <pkrempa@redhat.com>
ning.bo [Tue, 9 May 2017 02:09:07 +0000 (10:09 +0800)]
nodedev: Increase the netlink socket buffer size to the one used by udev
When a number of SRIOV VFs (up to 128 on Intel XL710) is created:
for i in `seq 0 1`; do
echo 63 > /sys/class/net/<interface>/device/sriov_numvfs
done
libvirtd will then report "udev_monitor_receive_device returned NULL"
error because the netlink socket buffer is not big enough (using GDB on
libudev confirmed this with ENOBUFFS) and thus some udev events were
dropped. This results in some devices being missing in the nodedev-list
output. This patch overrides the system's rmem_max limit but for that,
we need to make sure we've got root privileges.
Michal Privoznik [Mon, 29 May 2017 08:04:36 +0000 (10:04 +0200)]
virCapabilitiesInitCaches: Don't leak @cpus
The @cpus is allocated by virFileReadValueBitmap() but never
freed:
==21274== 40 (32 direct, 8 indirect) bytes in 1 blocks are definitely lost in loss record 808 of 1,004
==21274== at 0x4C2E080: calloc (vg_replace_malloc.c:711)
==21274== by 0x54BA561: virAlloc (viralloc.c:144)
==21274== by 0x54BC604: virBitmapNewEmpty (virbitmap.c:126)
==21274== by 0x54BD059: virBitmapParseUnlimited (virbitmap.c:570)
==21274== by 0x54EECE9: virFileReadValueBitmap (virfile.c:4113)
==21274== by 0x5563132: virCapabilitiesInitCaches (capabilities.c:1548)
==21274== by 0x2BB86E59: virQEMUCapsInit (qemu_capabilities.c:1132)
==21274== by 0x2BBEC067: virQEMUDriverCreateCapabilities (qemu_conf.c:928)
==21274== by 0x2BC3DEAA: qemuStateInitialize (qemu_driver.c:845)
==21274== by 0x5625AAC: virStateInitialize (libvirt.c:770)
==21274== by 0x124519: daemonRunStateInit (libvirtd.c:881)
==21274== by 0x554C927: virThreadHelper (virthread.c:206)
Signed-off-by: Michal Privoznik <mprivozn@redhat.com>
Bjoern Walk [Mon, 22 May 2017 06:38:25 +0000 (08:38 +0200)]
node_device: introduce new capability FC_RPORT
Similar to scsi_host and fc_host, there is a relation between a
scsi_target and its transport specific fc_remote_port. Let's expose this
relation and relevant information behind it.
Bjoern Walk [Mon, 22 May 2017 06:38:24 +0000 (08:38 +0200)]
util: helper functions for fibre channel devices
We will need some convenient helper functions for managing sysfs-entries
for fibre channel-backed devices. Let's implement them and make them
available in the private API.
Signed-off-by: Bjoern Walk <bwalk@linux.vnet.ibm.com>
Bjoern Walk [Mon, 22 May 2017 06:38:23 +0000 (08:38 +0200)]
virsh: nodedev: ability to filter CCW capabilities
Now that the node_device driver is aware of CCW devices, let's hook up
virsh so that we can filter them properly.
Reviewed-by: Boris Fiuczynski <fiuczy@linux.vnet.ibm.com> Reviewed-by: Marc Hartmayer <mhartmay@linux.vnet.ibm.com> Signed-off-by: Bjoern Walk <bwalk@linux.vnet.ibm.com>
Bjoern Walk [Mon, 22 May 2017 06:38:22 +0000 (08:38 +0200)]
node_device: detect CCW devices
Make CCW devices available to the node_device driver. The devices are
already seen by udev so let's implement necessary code for detecting
them properly.
Topologically, CCW devices are similar to PCI devices, e.g.:
Marc Hartmayer [Mon, 22 May 2017 06:38:21 +0000 (08:38 +0200)]
node_device: Unlock obj in case of an error too
Unlock @obj in case of an error too.
Reviewed-by: Bjoern Walk <bwalk@linux.vnet.ibm.com> Reviewed-by: Boris Fiuczynski <fiuczy@linux.vnet.ibm.com> Signed-off-by: Marc Hartmayer <mhartmay@linux.vnet.ibm.com>
Marc Hartmayer [Mon, 22 May 2017 06:38:20 +0000 (08:38 +0200)]
node_device: Use the @cap iterator variable
Since the switch statement is already using the deref'd @cap variable
and the VIR_NODE_DEV_CAP_NET case uses it, the SCSI_HOST and PCI_DEV
cases may as well use it too.
Suggested-by: Bjoern Walk <bwalk@linux.vnet.ibm.com> Signed-off-by: Marc Hartmayer <mhartmay@linux.vnet.ibm.com>
projects / libvirt.git / log