* src/qemu/qemu_driver.c (qemudDomainAttachSCSIDisk): The ".controller"
member is an index, and *may* be 0. As such, the commit that we're
reverting broke SCSI disk hot-plug on controller 0.
Reported by Wolfgang Mauerer.
We need to call PrepareHostdevs to determine the USB device path before
any security calls. PrepareHostUSBDevices was also incorrectly skipping
all USB devices.
* src/qemu/qemu_conf.c: add the ",readonly=on" for read-only disks
and also parse it back in qemuParseCommandLineDisk()
* tests/qemuxml2argvtest.c
tests/qemuxml2argvdata/qemuxml2argv-disk-drive-readonly-disk.args
tests/qemuxml2argvdata/qemuxml2argv-disk-drive-readonly-disk.xml:
add a specific regression test
The nodeGetInfo code was always assuming that machine had a
single NUMA node, which is not correct. The good news is that
libnuma gives us this information pretty easily, so let's
properly report it.
NOTE: With recent hardware starting to support CPU hot-add
and hot-remove, both this code and the nodeCapsInitNUMA()
code are quickly going to become obsolete. We'll have to
think of a more dynamic solution for dealing with NUMA
nodes and CPUs that can come and go at will.
Signed-off-by: Chris Lalancette <clalance@redhat.com>
Currently if you dump the core of a qemu guest with
qemudDomainCoreDump, subsequent commands will hang
up libvirtd. This is because qemudDomainCoreDump
uses qemuDomainWaitForMigrationComplete, which expects
the qemuDriverLock to be held when it's called. This
patch does the simple thing and moves the qemuDriveUnlock
to the end of the qemudDomainCoreDump so that the driver
lock is held for the entirety of the call (as it is done
in qemudDomainSave). We will probably want to make the
lock more fine-grained than that in the future, but
we can fix both qemudDomainCoreDump and qemudDomainSave
at the same time.
Signed-off-by: Chris Lalancette <clalance@redhat.com>
The code to add job support into libvirtd caused a problem
in qemudDomainSetVcpus. In particular, a qemuDomainObjEndJob()
call was added at the end of the function, but a
corresponding qemuDomainObjBeginJob() was not. Additionally,
a call to qemuDomainObj{Enter,Exit}Monitor() was also missed
in qemudDomainHotplugVcpus(). These missing calls conspired to
cause a hang in the libvirtd process after the command was
finished. Fix this by adding the missing calls.
Signed-off-by: Chris Lalancette <clalance@redhat.com>
As previously discussed[1], this patch removes the
qemudDomainSetMaxMemory() function, since it doesn't
work. This means that instead of getting somewhat
cryptic errors, you will now get:
error: Unable to change MaxMemorySize
error: this function is not supported by the hypervisor: virDomainSetMaxMemory
Which describes the situation perfectly.
[1] https://www.redhat.com/archives/libvir-list/2010-February/msg00928.html
Signed-off-by: Chris Lalancette <clalance@redhat.com>
When using the JSON monitor, qemuMonitorJSONExtractCPUInfo
was returning 0 on success. Unfortunately, higher levels of
the cpuinfo code expect that it returns the number of CPUs
it found on success. This one-line patch fixes it so that
it returns the correct number. This makes "virsh vcpuinfo <domain>"
work when using the JSON monitor.
Signed-off-by: Chris Lalancette <clalance@redhat.com>
As pointed out by eblake, I made a real hash of the
nodeinfo code with commit
aa2f6f96dd. This patch
cleans it up:
1) Do more work at compile time instead of runtime (minor)
2) Properly handle the hex digits that come from
/sys/devices/system/cpu/cpu*/topology/thread_siblings
3) Fix up some error paths that could cause SEGV
4) Used unsigned's for the cpu numbers (cpu -1 doesn't
make any sense)
Along with the recent patch from jdenemar to zero out
the nodeinfo structure, I've re-tested this on the
machines having the problems, and it seems to be good.
Signed-off-by: Chris Lalancette <clalance@redhat.com>
It is a bad idea to call gettext on an already-translated
string. In cases where a string must be translated separately
from where it is exposed to xgettext, the gettext manual
recommends the idiom of N_() wrapping gettext_noop for
marking the string.
* src/internal.h (N_): Fix definition to match gettext manual.
* tools/virsh.c: (cmdHelp, cmdList, cmdDomstate, cmdDominfo)
(cmdVcpuinfo, vshUsage): Replace incorrect use of N_ with _.
(vshCmddefHelp): Likewise. Mark C format strings appropriately.
The nodeinfo structure wasn't initialized in qemu driver and with the
recent change in CPU topology parsing, old value of nodeinfo->sockets
could be used and incremented giving totally bogus results.
Let's just wipe the structure completely.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
A few more non-literal format strings in error log messages have crept
in. Fix them in the standard way - turn the format string into "%s"
with the original string as the arg.
If a special cache strategy for a disk has been specified in a domain
definition, but no driverName has been set, virDomainGetXMLDesc would not
include the <driver> tag at all.
* src/conf/domain_conf.c: make sure any <driver> tag setting is
serialized if set.
The current code for "nodeinfo" is pretty naive
about socket and thread information. To determine the
sockets, it just takes the number of cpus and divides
by the number of cores. For the thread count, it always
sets it to 1. With more recent Intel machines, however,
hyperthreading is again an option, meaning that these
heuristics no longer work and give bogus numbers. This
patch goes through /sys to get the additional
information so we properly report it.
Note that I had to edit the tests not to report on
socket and thread counts, since these are determined
dynamically now.
v2: As pointed out by Eric Blake, gnulib provides
count-one-bits (which is LGPLv2+). Use it instead
of a hand-coded popcnt.
Signed-off-by: Chris Lalancette <clalance@redhat.com>
When adding domainMemoryStats API support for the qemu driver, I didn't
follow the locking rules exactly. The job condition must be held when
executing monitor commands. This corrects the segfaults I was seeing
when calling domainMemoryStats in a multi-threaded environment.
* src/qemu/qemu_driver.c: in qemudDomainMemoryStats() add missing
calls to qemuDomainObjBeginJob/qemuDomainObjEndJob
doTunnelSendAll function (used by QEMU migration) uses a 64k buffer on
the stack, which could be problematic. This patch replaces that with a
buffer from the heap.
While in the neighborhood, this patch also improves error reporting in
the case that saferead fails - previously, virStreamAbort() was called
(resetting errno) before reporting the error. It's been changed to
report the error first.
* src/qemu/qemu_driver.c: fix doTunnelSendAll() to use a malloc'ed
buffer
* src/qemu/qemu_driver.c (qemudDomainAttachSCSIDisk): Handle
the (theoretical) case of an empty controller list, so that
clang does not think the subsequent dereference of "cont"
would dereference an undefined variable (due to preceding
loop not iterating even once).
* src/qemu/qemu_driver.c (qemudDomainRestore): A corrupt save file
(in particular, a too-large header.xml_len value) would cause an
unwarranted out-of-memory error. Do not trust the just-read
header.xml_len. Instead, merely use that as a hint, and
read/allocate up to that number of bytes from the file.
Also verify that header.xml_len is positive; if it were negative,
passing it to virFileReadLimFD could cause trouble.
to saferead_lim, which interprets it as a size_t.
* src/util/util.c (virFileReadLimFD): Do not malfunction when
maxlen < -1. Return -1,EINVAL in that case. Handle maxlen==0
in the same manner.
* src/xen/proxy_internal.c (xenProxyDomainDumpXML): An invalid packet
could include a too-large "ans.len" value, which would make us allocate
too much memory and then copy data from beyond the end of "ans",
possibly evoking a segfault. Ensure that the value we use is no
larger than the remaining portion of "ans".
Also, change unnecessary memmove to memcpy (src and dest obviously
do not overlap, so no need to use memmove).
(xenProxyDomainGetOSType): Likewise.
(xenProxyGetCapabilities): Likewise.
The code erroneously searched the entire "reply" for a comma, when
its intent was to search only that portion after "balloon: actual="
* src/qemu/qemu_monitor_text.c (qemuMonitorTextGetMemoryStats):
Search for "," only starting *after* the BALLOON_PREFIX string.
Otherwise, we'd be more prone to false positives.
Changeset
commit 5073aa994a
Author: Cole Robinson <crobinso@redhat.com>
Date: Mon Jan 11 11:40:46 2010 -0500
Added support for product/vendor based passthrough, but it only
worked at the security driver layer. The main guest XML config
was not updated with the resolved bus/device ID. When the QEMU
argv refactoring removed use of product/vendor, this then broke
launching guests.
THe solution is to move the product/vendor resolution up a layer
into the QEMU driver. So the first thing QEMU does is resolve
the product/vendor to a bus/device and updates the XML config
with this info. The rest of the code, including security drivers
and QEMU argv generated can now rely on bus/device always being
set.
* src/util/hostusb.c, src/util/hostusb.h: Split vendor/product
resolution code out of usbGetDevice and into usbFindDevice.
Add accessors for bus/device ID
* src/security/virt-aa-helper.c, src/security/security_selinux.c,
src/qemu/qemu_security_dac.c: Remove vendor/product from the
usbGetDevice() calls
* src/qemu/qemu_driver.c: Use usbFindDevice to resolve vendor/product
into a bus/device ID
The pci_del command is not being ported to QMP. Convert all the
QEMU hotplug code over to use device_del whenever it is available
to avoid the pci_del problem
* src/qemu/qemu_driver.c: Convert unplug code to device_del
Previously hot-unplug could not be supported for USB devices
in QEMU, since usb_del required the guest visible address
which libvirt never knows. With 'device_del' command we can
now unplug based on device alias, so support that.
* src/qemu/qemu_driver.c: Use device_del to remove USB devices
Upstart crashes & burns in a heap if $TERM environment variable
is missing. Presumably the kernel always sets this when booting
init on a real machine, so libvirt should set it for containers
too.
To make a typical inittab / mingetty setup happier, we need to
symlink the primary console /dev/pts/0 to /dev/tty1.
Improve logging in certain scenarios to make troubleshooting
easier
* src/lxc/lxc_container.c: Create /dev/tty1 and set $TERM
When using the 'ns' cgroup controller, the moment a process calls
'unshare(CLONE_NEWNS)', it will be given a private cgroup tree
under its current location. This really messages up the LXC
controller process, because it ends up creating the containers'
cgroup in the wrong place. The fix is fairly easy, just move
the cgroup setup before the code which calls unshare(). The
'ns' controller will still create extra undesired cgroups, but
they at least won't break libvirt's setup now.
The patch also adds a missing cgroups allow rule for /dev/tty
device node
When getting the driver/domain cgroup it is possible to specify
whether it should be auto created. If auto-creation was turned
off, libvirt still mistakenly created its own top level cgroup
* src/util/cgroup.c: Honour autocreate flag for top level cgroup
This allows the config to have a setting that means "leave it alone",
eg when building a pool where the directory already exists the user
may want the current uid/gid of the directory left intact. This
actually gets us back to older behavior - before recent changes to the
pool building code, we weren't as insistent about honoring the uid/gid
settings in the XML, and virt-manager was taking advantage of this
behavior.
As a side benefit, removing calls to getuid/getgid from the XML
parsing functions also seems like a good idea. And having a default
that is different from a common/useful value (0 == root) is a good
thing in general, as it removes ambiguity from decisions (at least one
place in the code was checking for (perms.uid == 0) to see if a
special uid was requested).
Note that this will only affect newly created pools and volumes. Due
to the way that the XML is parsed, then formatted for newly created
volumes, all existing pools/volumes already have an explicit uid and
gid set.
src/conf/storage_conf.c: Remove calls to setuid/setgid for default values
of uid/gid, and set them to -1 instead
src/storage/storage_backend.c:
src/storage/storage_backend_fs.c:
Make account for the new default values of perms.uid
and perms.gid.
With the recent changes to the linking defaults in Fedora 13 (namely
enabling --no-add-needed behaviour by default), we have to pass the
dlopen()-providing libraries directly at the link of the module; use the
same AC_SEARCH_LIBS function as used before to look for it and add it to
the Makefile.
QEMU has a monitor command 'set_cpu' which allows a specific
CPU to be toggled between online& offline state. libvirt CPU
hotplug does not work in terms of individual indexes CPUs.
Thus to support this, we iteratively toggle the online state
when the total number of vCPUs is adjusted via libvirt
NB, currently untested since QEMU segvs when running this!
* src/qemu/qemu_driver.c: Toggle online state for CPUs when
doing hotplug
* src/qemu/qemu_monitor.c, src/qemu/qemu_monitor.h,
src/qemu/qemu_monitor_json.c, src/qemu/qemu_monitor_json.h,
src/qemu/qemu_monitor_text.c, src/qemu/qemu_monitor_text.h: Add
monitor API for toggling a CPU's online status via 'set_cpu
The storage backend implementations all presume that the XML parser
is validating correctness of the source specification. The check for
a source device was lost at some point. This allowed for a potential
crash in the disk backend. Re-introduce the sanity check
* src/conf/storage_conf.c: Re-add check for source device
When trying to delete a pool the error message claimed the volume
could not be deleted.
* src/storage/storage_driver.c: Error message referred to
volumes instead of pools
The QMP code was running query-migration instead of query-migrate.
This doesn't work so well
* src/qemu/qemu_monitor_json.c: s/query-migration/query-migrate/
The code to remove the cgroup after QEMU failed to startup could
be obscuring a real error from earlier on. It is not neccessary
to raise an error in this case, so tell cgroups to keep quiet
* src/qemu/qemu_driver.c: Don't raise cgroups error in QEMU start
cleanup code.
The QEMU hotunplug code for PCI devices was looking at host
devices in the guest config without first filtering non
PCI devices. This means it was reading garbage
* src/qemu/qemu_driver.c: Filter out non-PCI devices
Commit 3c12a67b76 added
a dependency on the NFS_SUPER_MAGIC macro, which is
defined in linux/magic.h. Unfortunately linux/magic.h
is not available in RHEL-5, and causes a compile error.
Just define it locally, since this is something that
can't change.
Signed-off-by: Chris Lalancette <clalance@redhat.com>
Move *all* file operations related to creation and writing of libvirt
header to the domain save file into a hook function that is called by
virFileOperation. First try to call virFileOperation as root. If that
fails with EACCESS, and (in the case of Linux) statfs says that we're
trying to save the file on an NFS share, rerun virFileOperation,
telling it to fork a child process and setuid to the qemu user. This
is the only way we can successfully create a file on a root-squashed
NFS server.
This patch (along with setting dynamic_ownership=0 in qemu.conf)
makes qemudDomainSave work on root-squashed NFS.
* src/qemu/qemu_driver.c: provide new qemudDomainSaveFileOpHook()
utility, use it in qemudDomainSave() if normal creation of the
file as root failed, and after checking the filesystem type for
the storage is NFS. In that case we also bypass the security
driver, as this would fail on NFS.
If qemudDomainRestore fails to open the domain save file, create a
pipe, then fork a process that does setuid(qemu_user) and opens the
file, then reads this file and stuffs it into the pipe. the parent
libvirtd process will use the other end of the pipe as its fd, then
reap the child process after it's done reading.
This makes domain restore work on a root-squash NFS share that is only
visible to the qemu user.
* src/qemu/qemu_driver.c: add new qemudOpenAsUID() helper function,
and use it in qemudDomainRestore() if reading as root directly failed.
The USB/PCI device hotplug code for the QEMU driver was forgetting
to allocate a unique device alias.
* src/qemu/qemu_driver.c: Fill in device alias for USB/PCI devices
The code assumed that 'device_add' returned an empty string upon
success. This is not true, it sometimes prints random debug info.
THus we need to check for an explicit fail string
* src/qemu/qemu_monitor_text.c: Fix error checking of the device_add
monitor command
Another warning caught by coverity. Continue to perform best-effort
closing and resource release, but warn the caller about the failure.
* src/esx/esx_driver.c (esxClose): Return an error on failure to close.
Various safezero() implementations used either -1, errno or -errno
return values. This patch fixes them all to return -1 and set errno
appropriately.
There was also a bug in size parameter passed to safewrite() which could
result in an attempt to write gigabytes out of a megabyte buffer.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
When a VM save attempt failed, the VM would be left in a paused
state. It is neccessary to resume CPU execution upon failure
if it was running originally
* src/qemu/qemu_driver.c: Resume CPUs upon save failure
This supports cancellation of jobs for the QEMU driver against
the virDomainMigrate, virDomainSave and virDomainCoreDump APIs.
It is not yet supported for the virDomainRestore API, although
it is desirable.
* src/qemu/qemu_driver.c: Issue 'migrate_cancel' command if
virDomainAbortJob is issued during a migration operation
* tools/virsh.c: Add a domjobabort command
This defines the wire protocol for the new API
* src/remote/remote_protocol.x: Wire protocol definition
* src/remote/remote_driver.c,daemon/remote.c: Client and server
side implementation
* daemon/remote_dispatch_args.h, daemon/remote_dispatch_prototypes.h,
daemon/remote_dispatch_table.h, src/remote/remote_protocol.c,
src/remote/remote_protocol.h: Re-generate from remote_protocol.x
This provides the internal glue for the driver API
* src/driver.h: Internal API contract
* src/libvirt.c, src/libvirt_public.syms: Connect public API
to driver API
* src/esx/esx_driver.c, src/lxc/lxc_driver.c, src/opennebula/one_driver.c,
src/openvz/openvz_driver.c, src/phyp/phyp_driver.c,
src/qemu/qemu_driver.c, src/remote/remote_driver.c,
src/test/test_driver.c src/uml/uml_driver.c, src/vbox/vbox_tmpl.c,
src/xen/xen_driver.c: Stub out entry points
Introduce support for virDomainGetJobInfo in the QEMU driver. This
allows for monitoring of any API that uses the 'info migrate' monitor
command. ie virDomainMigrate, virDomainSave and virDomainCoreDump
Unfortunately QEMU does not provide a way to monitor incoming migration
so we can't wire up virDomainRestore yet.
The virsh tool gets a new command 'domjobinfo' to query status
* src/qemu/qemu_driver.c: Record virDomainJobInfo and start time
in qemuDomainObjPrivatePtr objects. Add generic shared handler
for calling 'info migrate' with all migration based APIs.
* src/qemu/qemu_monitor_text.c: Fix parsing of 'info migration' reply
* tools/virsh.c: add new 'domjobinfo' command to query progress
* src/remote/remote_protocol.x: Define wire protocol format
for virDomainGetJobInfo API
* src/remote/remote_driver.c, daemon/remote.c: Implement client
and server marshalling code for virDomainGetJobInfo()
* daemon/remote_dispatch_args.h, daemon/remote_dispatch_prototypes.h
daemon/remote_dispatch_ret.h, daemon/remote_dispatch_table.h,
src/remote/remote_protocol.c, src/remote/remote_protocol.h: Rebuild
files from src/remote/remote_protocol.x
The internal glue layer for the new pubic API
* src/driver.h: Define internal driver API contract
* src/libvirt.c, src/libvirt_public.syms: Wire up public
API to internal driver API
* src/esx/esx_driver.c, src/lxc/lxc_driver.c, src/opennebula/one_driver.c,
src/openvz/openvz_driver.c, src/phyp/phyp_driver.c,
src/qemu/qemu_driver.c, src/remote/remote_driver.c,
src/test/test_driver.c, src/uml/uml_driver.c, src/vbox/vbox_tmpl.c,
src/xen/xen_driver.c: Stub new entry point
All other uses of get_str_prop in this file that ignored
failure explicitly cast to void.
* src/node_device/node_device_hal.c (dev_create): Silence coverity
warning.
A number of the error messages raised when parsing USB devices
refered to PCI devices by mistake
* src/qemu/qemu_conf.c: s/PCI/USB/ in qemuParseCommandLineUSB()
when the underlying qemu supports the drive/device model and the
controller has been added this way.
* src/qemu/qemu_driver.c: use qemuMonitorDelDevice() when detaching
PCI controller and if supported
* src/qemu/qemu_monitor.[ch]: add new qemuMonitorDelDevice() function
* src/qemu/qemu_monitor_json.[ch]: JSON backend for DelDevice command
* src/qemu/qemu_monitor_text.[ch]: Text backend for DelDevice command
* src/qemu/qemu_driver.c: in qemudDomainDetachPciControllerDevice()
when a controller is not present in the system anymore, the PCI
address must be deleted from libvirt's hashtable because it can
be re-used for other purposes.
* src/qemu/qemu_driver.c: in qemudDomainAttachDevice(), one must not
delete the data part when the operation succeeds because it is
required later on. The correct pattern to handlethe parsed
representation of the device information on success
is dev->data.controller = NULL; virDomainDeviceDefFree(dev);,
which leaves the structure pointed at by data in memory.
* src/cpu/cpu_x86.c (x86Decode): Don't dereference NULL when passed
a NULL "models" pointer, or when passed a nonzero "nmodels" value
and a corresponding NULL models[i].
* src/phyp/phyp_driver.c (phypUUIDTable_Push): Move incr/decr
of ptr/nread into the loop where those variables are used.
Also, remove "exit" label and just-preceding "goto".
Allow an arbitrary timezone with QEMU by setting the $TZ environment
variable when launching QEMU
* src/qemu/qemu_conf.c: Set TZ environment variable if a timezone
is requested
* tests/qemuxml2argvtest.c: Add test case for timezones
* tests/qemuxml2argvdata/qemuxml2argv-clock-france.xml,
tests/qemuxml2argvdata/qemuxml2argv-clock-france.args: Data
for timezone tests
This extends the XML to allow for
<clock offset='timezone' timezone='Europe/Paris'/>
This is useful if the admin has not configured any timezone on the
host OS, but still wants to synchronize a guest to a specific one.
* src/conf/domain_conf.h, src/conf/domain_conf.c: Support extra
'timezone' attribute on clock configuration
* docs/schemas/domain.rng: Add 'timezone' attribute
* src/xen/xend_internal.c, src/xen/xm_internal.c: Reject configs
with a configurable timezone
This allows QEMU guests to be started with an arbitrary clock
offset
The test case can't actually be enabled, since QEMU argv expects
an absolute timestring, and this will obviously change every
time the test runs :-( Hopefully QEMU will allow a relative
time offset in the future.
* src/qemu/qemu_conf.c, src/qemu/qemu_conf.h: Use the -rtc arg
if available to support variable clock offset mode
* tests/qemuhelptest.c: Add QEMUD_CMD_FLAG_RTC for qemu 0.12.1
* qemuxml2argvdata/qemuxml2argv-clock-variable.args,
qemuxml2argvdata/qemuxml2argv-clock-variable.xml,
qemuxml2argvtest.c: Test case, except we can't actually enable
it yet.
This introduces a third option for clock offset synchronization,
that allows an arbitrary / variable adjustment to be set. In
essence the XML contains the time delta in seconds, relative to
UTC.
<clock offset='variable' adjustment='123465'/>
The difference from 'utc' mode, is that management apps should
track adjustments and preserve them at next reboot.
* docs/schemas/domain.rng: Schema for new clock mode
* src/conf/domain_conf.c, src/conf/domain_conf.h: Parse
new clock time delta
* src/libvirt_private.syms, src/util/xml.c, src/util/xml.h: Add
virXPathLongLong() method
The XML will soon be extended to allow more than just a simple
localtime/utc boolean flag. This change replaces the plain
'int localtime' with a separate struct to prepare for future
extension
* src/conf/domain_conf.c, src/conf/domain_conf.h: Add a new
virDomainClockDef structure
* src/libvirt_private.syms: Export virDomainClockOffsetTypeToString
and virDomainClockOffsetTypeFromString
* src/qemu/qemu_conf.c, src/vbox/vbox_tmpl.c, src/xen/xend_internal.c,
src/xen/xm_internal.c: Updated to use new structure for localtime
While building under RHEL-5, I got a compile warning because
virDomainObjFormat was defined but not used. That came about
because in RHEL-5 we build with "#define PROXY", and
virDomainObjFormat is only used with !PROXY. Move the
define.
Signed-off-by: Chris Lalancette <clalance@redhat.com>
* src/openvz/openvz_conf.c (openvzGetVEID): Always call fclose.
Diagnose parse failure also when vzlist output is empty.
If somehow we read a -1, diagnose that (albeit as a parse failure).
Kind of minor, but it annoys me that the default auth callback
doesn't put a space between the prompt and the input, like a typical
terminal, ssh, etc. This patch changes the current prompt:
Please enter your authentication name:myuser
to
Please enter your authentication name: myuser
In bug 567931 we found that virt-top would exit occasionally
when the terminal window was resized. Tracking this down it
turned out that SIGWINCH was being delivered to the process at
exactly the point where the libvirt remote driver was calling
poll(2) waiting for a reply from libvirtd.
This caused the poll(2) call to be interrupted (returning errno
EINTR). However handling EINTR the same way as EAGAIN was not
the solution to this problem since we found previously that this
would break Ctrl-C handling (commit 47fec8eac2).
The correct solution is to mask out SIGWINCH for the duration
of the poll(2) system call. The per-thread mask is changed and
restored immediately after the call. Since we are using
pthread_sigmask, this should not affect other threads, and
since we restore the signal mask immediately afterwards it should
not affect the current thread visibly either. Other possibly
problematic signals are SIGCHLD and SIGPIPE and these are
masked too.
Note use of ignore_value: It's not fatal if we cannot mask out
SIGWINCH, and in any case pthread_sigmask never fails on Linux
as long as you supply the correct arguments.
I tested this patch and it cures the original problem with
virt-top.
Create the filesystem on the partition used by the pool
* configure.ac: check for mkfs availability
* libvirt.spec.in: add extra require on util-linux for mkfs
* src/storage/storage_backend_fs.c: run mkfs with the expected
fs type when creating a filesystem pool
We were using 'Y' to mean exabyte, when the correct abbreviation would be
'E' ('Y' is yettabyte, which is exabyte * 1024 * 1024). While it isn't
strictly backwards compatible, I highly doubt anyone was actually using
this broken behavior, so I don't see any harm in in dropping 'Y' handling.
Recently we introduced O_DSYNC flag when creating raw storage files to
avoid filling all disk cache with dirty pages. However, the patch got
lost when virStorageBackendCreateRaw was reworked using
virFileOperation. Let's use O_DSYNC again.
* src/xen/xend_internal.c (xenDaemonDomainSetAutostart): Avoid a NULL
dereference upon non-SEXPR_VALUE'd on_xend_start. This bug was
introduced by commit 37ce5600c0.
There were a few operations on the storage volume file that were still
being done as root, which will fail if the file is on a root-squashed
NFS share. The result was that attempts to create a storage volume of
type "raw" on a root-squashed NFS share would fail.
This patch uses the newly introduced "hook" function in
virFileOperation to execute all those file operations in the child
process that's run under the uid that owns the file (and, presumably,
has permission to write to the NFS share)
* src/storage/storage_backend.c: use virFileOperation() in
virStorageBackendCreateRaw, turning virStorageBackendCreateRaw()
into a new createRawFileOpHook() hook
It turns out it is also useful to be able to perform other operations
on a file created while running as a different uid (eg, write things
to that file), and possibly to do this to a file that already
exists. This patch adds an optional hook function to the renamed (for
more accuracy of purpose) virFileOperation; the hook will be called
after the file has been opened (possibly created) and gid/mode
checked/set, before closing it.
As with the other operations on the file, if the VIR_FILE_OP_AS_UID
flag is set, this hook function will be called in the context of a
child process forked from the process that called virFileOperation.
The implication here is that, while all data in memory is available to
this hook function, any modification to that data will not be seen by
the caller - the only indication in memory of what happened in the
hook will be the return value (which the hook should set to 0 on
success, or one of the standard errno values on failure).
Another piece of making the function more flexible was to add an
"openflags" argument. This arg should contain exactly the flags to be
passed to open(2), eg O_RDWR | O_EXCL, etc.
In the process of adding the hook to virFileOperation, I also realized
that the bits to fix up file owner/group/mode settings after creation
were being done in the parent process, which could fail, so I moved
them to the child process where they should be.
* src/util/util.[ch]: rename and rework virFileCreate-->virFileOperation,
and redo flags in virDirCreate
* storage/storage_backend.c, storage/storage_backend_fs.c: update the
calls to virFileOperation/virDirCreate to reflect changes in the API,
but don't yet take advantage of the hook.
Fix multiple veth problem.
NETIF setting was overwritten after first CT because any CT could not be
found by name.
* src/openvz/openvz_conf.c src/openvz/openvz_conf.h: add the
openvzGetVEID lookup function
* src/openvz/openvz_driver.c: use it in openvzDomainSetNetwork()
If the hostname as returned by "gethostname" resolves
to "localhost" (as it does with the broken Fedora-12
installer), then live migration will fail because the
source will try to migrate to itself. Detect this
situation up-front and abort the live migration before
we do any real work.
* src/util/util.h src/util/util.c: add a new virGetHostnameLocalhost
with an optional localhost check, and rewire virGetHostname() to use
it
* src/libvirt_private.syms: expose the new function
* src/qemu/qemu_driver.c: use it in qemudDomainMigratePrepare2()
* src/xen/xend_internal.c (xenDaemonDomainSetAutostart): Rewrite to
avoid dereferencing the result of sexpr_lookup. While in this
particular case, it was guaranteed never to be NULL, due to the
preceding "if sexpr_node(...)" guard, it's cleaner to skip the
sexpr_node call altogether, and also saves a lookup.
This patch sets or unsets the IFF_VNET_HDR flag depending on what device
is used in the VM. The manipulation of the flag is done in the open
function and is only fatal if the IFF_VNET_HDR flag could not be cleared
although it has to be (or if an ioctl generally fails). In that case the
macvtap tap is closed again and the macvtap interface torn.
* src/qemu/qemu_conf.c src/qemu/qemu_conf.h: pass qemuCmdFlags to
qemudPhysIfaceConnect()
* src/util/macvtap.c src/util/macvtap.h: add vnet_hdr boolean to
openMacvtapTap(), and private function configMacvtapTap()
* src/qemu/qemu_driver.c: add extra qemuCmdFlags when calling
qemudPhysIfaceConnect()
For __virExec() this is a semantic NOP except for when fork()
fails. __virExec() would previously forget to restore the signal mask
in this case; virFork() corrects this behavior.
virFileCreate() and virDirCreate() gain the code to reset the logging
and properly deal with the signal handling race condition.
This also removes a log message that had a typo ("cannot fork o create
file '%s'") - this error is now logged in a more generic manner in
virFork() (more generic, but really just as informative, since the
fact that it's forking to create a file is immaterial to the fact that
it simply can't fork)
* src/util/util.c: use the generic virFork() in the 3 functions
virFork() contains bookkeeping that must be done any time a process
forks. Currently this includes:
1) Call virLogLock() prior to fork() and virLogUnlock() just after,
to avoid a deadlock if some other thread happens to hold that lock
during the fork.
2) Reset the logging hooks and send all child process log messages to
stderr.
3) Block all signals prior to fork(), then either a) reset the signal
mask for the parent process, or b) clear the signal mask for the
child process.
Note that the signal mask handling in __virExec erroneously fails to
restore the signal mask when fork() fails. virFork() fixes this
problem.
Other than this, it attempts to behave as closely to fork() as
possible (including preserving errno for the caller), with a couple
exceptions:
1) The return value is 0 (success) or -1 (failure), while the pid is
returned via the pid_t* argument. Like fork(), if pid < 0 there is
no child process, otherwise both the child and the parent will
return to the caller, and both should look at the return value,
which will indicate if some of the extra processing outlined above
encountered an error.
2) If virFork() returns with pid < 0 or with a return value < 0
indicating an error condition, the error has already been
reported. You can log an additional message if you like, but it
isn't necessary, and may be awkwardly extraneous.
Note that virFork()'s child process will *never* call _exit() - if a
child process is created, it will return to the caller.
* util.c util.h: add virFork() function, based on what is currently
done in __virExec().
Support virtio-serial controller and virtio channel in QEMU backend.
Will output
the following for virtio-serial controller:
-device
virtio-serial-pci,id=virtio-serial0,bus=pci.0,addr=0x4,max_ports=16,vectors=4
and the following for a virtio channel:
-chardev pty,id=channel0 \
-device
virtserialport,bus=virtio-serial0.0,chardev=channel0,name=org.linux-kvm.port.0
* src/qemu/qemu_conf.c: Add argument output for virtio
* tests/qemuxml2argvtest.c
tests/qemuxml2argvdata/qemuxml2argv-channel-virtio.args: Add test for
QEMU command line generation
Add support for virtio-serial by defining a new 'virtio' channel target type
and a virtio-serial controller. Allows the following to be specified in a
domain:
<controller type='virtio-serial' index='0' ports='16' vectors='4'/>
<channel type='pty'>
<target type='virtio' name='org.linux-kvm.port.0'/>
<address type='virtio-serial' controller='0' bus='0'/>
</channel>
* docs/schemas/domain.rng: Add virtio-serial controller and virtio
channel type.
* src/conf/domain_conf.[ch]: Domain parsing/serialization for
virtio-serial controller and virtio channel.
* tests/qemuxml2xmltest.c
tests/qemuxml2argvdata/qemuxml2argv-channel-virtio.xml: add domain xml
parsing test
* src/libvirt_private.syms src/qemu/qemu_conf.c:
virDomainDefAddDiskControllers() renamed to
virDomainDefAddImplicitControllers()
Remove virDomainDevicePCIAddressEqual and virDomainDeviceDriveAddressEqual,
which are defined but not used anywhere.
* src/conf/domain_conf.[ch] src/libvirt_private.syms: Remove
virDomainDevicePCIAddressEqual and virDomainDeviceDriveAddressEqual.
Currently we just error with ex. 'virbr0: No such device'.
Since we are using public API calls here, we need to ensure that any
raised error is properly saved and restored, since API entry points
always reset messages.
virGetLastError returns NULL if no error has been set, not on
allocation error like virSetError assumed. Use virLastErrorObject
instead. This fixes virSetError when no error is currently stored.
Rework and simplification of teardown of the macvtap device.
Basically all devices with the same MAC address and link device are kept
alive and not attempted to be torn down. If a macvtap device linked to a
physical interface with a certain MAC address 'M' is to be created it
will automatically fail if the interface is 'up'ed and another macvtap
with the same properties (MAC addr 'M', link dev) happens to be 'up'.
This will prevent the VM from starting or the device from being attached
to a running VM. Stale interfaces are assumed to be there for some
reason and not stem from libvirt.
In the VM shutdown path, it's assuming that an interface name is always
available so that if the device type is DIRECT it can be torn down
using its name.
* src/util/macvtap.h src/libvirt_macvtap.syms: change of deleting routine
* src/util/macvtap.c: cleanups and change of deleting routine
* src/qemu/qemu_driver.c: change cleanup on shutdown
* src/qemu/qemu_conf.c: don't delete Macvtap in qemudPhysIfaceConnect()
The QEMU JSON monitor changed balloon commands to return/accept
bytes instead of kilobytes. Update libvirt to cope with this
* src/qemu/qemu_monitor_json.c: Expect/use bytes for ballooning
* src/uml/uml_driver.c (umlMonitorCommand): This function would
sometimes return -1, yet fail to free the "reply" it had allocated.
Hence, no caller would know to free the corresponding argument.
When returning -1, be sure to free all allocated resources.
* src/storage/storage_backend_mpath.c (virStorageBackendIsMultipath):
The result of dm_get_next_target was never used (and isn't needed),
so don't store it.
Similar to the Set*Mem commands, this implementation was bogus and
misleading. Make it clear this is a hotplug only operation, and that the
hotplug piece isn't even implemented.
Also drop the overkill maxvcpus validation: we don't perform this check
at XML define time so clearly no one is missing it, and there is
always the risk that our info will be out of date, possibly preventing
legitimate CPU values.
Signed-off-by: Cole Robinson <crobinso@redhat.com>
SetMem and SetMaxMem are hotplug only APIs, any persistent config
changes are supposed to go via XML definition. The original implementation
of these calls were incorrect and had the nasty side effect of making
a psuedo persistent change that would be lost after libvirtd restart
(I didn't know any better).
Fix these APIs to rightly reject non running domains.
Signed-off-by: Cole Robinson <crobinso@redhat.com>
The plain QEMU tree does not include 'thread_id' in the JSON
output. Thus we need to treat it as non-fatal if missing.
* src/qemu/qemu_monitor_json.c: Treat missing thread_id as non-fatal
A typo in the check for the primary IDE controller could cause
a crash on restore depending on the exact guest config.
* src/qemu/qemu_conf.c: Fix s/video/controller/ typo & slot
number typo
Current error reporting for JSON mode returns the full JSON
command string and full JSON error string. This is not very
user friendly, so this change makes the error report only
contain the basic command name, and friendly error message
description string. The full JSON data is logged instead.
* src/qemu/qemu_monitor_json.c: Always return the 'desc' field from
the JSON error message to users.
When in JSON mode, QEMU requires that 'qmp_capabilities' is run as
the first command in the monitor. This is a no-op when run in the
text mode monitor
* src/qemu/qemu_driver.c: Run capabilities negotiation when
connecting to the monitor
* src/qemu/qemu_monitor.c, src/qemu/qemu_monitor.h,
src/qemu/qemu_monitor_json.c, src/qemu/qemu_monitor_json.h: Add
support for the 'qmp_capabilities' command, no-op in text mode.
This part adds support for qemu making a macvtap tap device available
via file descriptor passed to qemu command line. This also attempts to
tear down the macvtap device when a VM terminates. This includes support
for attachment and detachment to/from running VM.
* src/qemu/qemu_conf.[ch] src/qemu/qemu_driver.c: add support in the
QEmu driver
This part adds the helper code to setup and tear down macvtap devices
using direct communication with the device driver via netlink sockets.
The rather short messages received from the netlink layer are now
written into a dynamically allocated buffer
* src/util/macvtap.h src/util/macvtap.c: provides the new module
* po/POTFILES.in: the module contains translated strings
This part adds support to domain_conf.{c|h} for parsing the new
interface XML of type 'direct'. The parsed mode is now stored as
an int.
* src/conf/domain_conf.c src/conf/domain_conf.h: extend parsing code
* src/util/macvtap.h: empty header to not break compilation
This patch adds build support for libvirt checking for certain contents
of /usr/include/linux/if_link.h to see whether macvtap support is
compilable on that system. One can disable macvtap support in libvirt
via --without-macvtap passed to configure.
* configure.ac src/Makefile.am: new build support
* src/libvirt_macvtap.syms: list of exported symbols
* src/util/macvtap.c: empty module to not break compilation
The virRaiseError macro inside of virSecurityReportError expands to
virRaiseErrorFull and includes the __FILE__, __FUNCTION__ and __LINE__
information. But this three values are always the same for every call
to virSecurityReportError and do not reflect the actual error context.
Converting virSecurityReportError into a macro results in getting the
correct __FILE__, __FUNCTION__ and __LINE__ information.
Current PCI addresses are allocated at time of VM startup.
To make them truely persistent, it is neccessary to do this
at time of virDomainDefine/virDomainCreate. The code in
qemuStartVMDaemon still remains in order to cope with upgrades
from older libvirt releases
* src/qemu/qemu_driver.c: Rename existing qemuAssignPCIAddresses
to qemuDetectPCIAddresses. Add new qemuAssignPCIAddresses which
does auto-allocation upfront. Call qemuAssignPCIAddresses from
qemuDomainDefine and qemuDomainCreate to assign PCI addresses that
can then be persisted. Don't clear PCI addresses at shutdown if
they are intended to be persistent
The old text mode monitor prompts for a password when disks are
encrypted. This interactive approach doesn't work for JSON mode
monitor. Thus there is a new 'block_passwd' command that can be
used.
* src/qemu/qemu_driver.c: Split out code for looking up a disk
secret from findVolumeQcowPassphrase, into a new method
getVolumeQcowPassphrase. Enhance qemuInitPasswords() to also
set the disk encryption password via the monitor
* src/qemu/qemu_monitor.c, src/qemu/qemu_monitor.h,
src/qemu/qemu_monitor_json.c, src/qemu/qemu_monitor_json.h,
src/qemu/qemu_monitor_text.c, src/qemu/qemu_monitor_text.h: Add
support for the 'block_passwd' monitor command.
Since c26cb9234f, the dname
parameter has been ignored by these two functions. Use it.
* src/qemu/qemu_driver.c (qemudDomainMigratePrepareTunnel): Honor dname
parameter once again.
(qemudDomainMigratePrepare2): Likewise.
All other libvirt functions use array first and then number of elements
in that array. Let's make cpuDecode follow this rule.
Signed-off-by: Jiri Denemark <jdenemar@redhat.com>
With QEMU >= 0.12 the host and guest side of disks no longer have
the same naming convention. Specifically the host side will now
get a 'drive-' prefix added to its name. The 'info blockstats'
monitor command returns the host side name, so it is neccessary
to strip this off when looking up stats since libvirt stores the
guest side name !
* src/qemu/qemu_conf.c, src/qemu/qemu_conf.h: Move 'drive-' prefix
string to a defined constant
* src/qemu/qemu_monitor_json.c, src/qemu/qemu_monitor_text.c: Strip
off 'drive-' prefix (if found) when looking up disk stats
Currently the timeout for reading startup output is 3 seconds. If the
host is under any sort of load, we can easily trigger this. Lets bump
it to 30 seconds.
Since the polling loop checks to see if the process has died, we shouldn't
erroneously hit this timeout if qemu bombs (only if it is stuck in some
infinite loop).
Use the ATTRIBUTE_NONNULL annotation to mark some virConnectPtr
args as mandatory non-null so the compiler can warn of mistakes
* src/conf/domain_event.h: All virConnectPtr args must be non-null
* src/qemu/qemu_conf.h: qemudBuildCommandLine and
qemudNetworkIfaceConnect() must be given non-null connection
* tests/qemuxml2argvtest.c: Provide a non-null (dummy) connection to
qemudBuildCommandLine()
The virConnectPtr is no longer required for error reporting since
that is recorded in a thread local. Remove use of virConnectPtr
from all APIs in secret_conf.{h,c} and update all callers to
match
The virConnectPtr is no longer required for error reporting since
that is recorded in a thread local. Remove use of virConnectPtr
from all APIs in interface_conf.{h,c} and update all callers to
match
The virConnectPtr is no longer required for error reporting since
that is recorded in a thread local. Remove use of virConnectPtr
from all APIs in cpu_conf.{h,c} and update all callers to
match
The virConnectPtr is no longer required for error reporting since
that is recorded in a thread local. Remove use of virConnectPtr
from all APIs in storage_conf.{h,c} and storage_encryption_conf.{h,c}
and update all callers to match
The virConnectPtr is no longer required for error reporting since
that is recorded in a thread local. Remove use of virConnectPtr
from all APIs in node_device_conf.{h,c} and update all callers to
match
The virConnectPtr is no longer required for error reporting since
that is recorded in a thread local. Remove use of virConnectPtr
from all APIs in network_conf.{h,c} and update all callers to
match
All callers now pass a NULL virConnectPtr into the USB/PCi device
iterator functions. Therefore the virConnectPtr arg can now be
removed from these functions
* src/util/hostusb.h, src/util/hostusb.c: Remove virConnectPtr
from usbDeviceFileIterate
* src/util/pci.c, src/util/pci.h: Remove virConnectPtr arg from
pciDeviceFileIterate
* src/qemu/qemu_security_dac.c, src/security/security_selinux.c: Update
to drop redundant virConnectPtr arg
The QEMU flags are commonly stored as a signed or unsigned int,
allowing only 31 flags. This limit is rather close, so to aid
future patches, change it to a 64-bit int
* src/qemu/qemu_conf.c, src/qemu/qemu_conf.h, src/qemu/qemu_driver.c,
tests/qemuargv2xmltest.c, tests/qemuhelptest.c, tests/qemuxml2argvtest.c:
Use 'unsigned long long' for QEMU flags
The virConnectPtr is no longer required for error reporting since
that is recorded in a thread local. Remove use of virConnectPtr
from all APIs in security_driver.{h,c} and update all callers to
match
The security driver was mistakenly initialized before the QEMU
config file was loaded. This prevents it being turned off again.
The capabilities XML was also getting the wrong security driver
name, due to the stacked driver arrangement.
* src/qemu/qemu_driver.c: Fix initialization order and capabilities
model name
* src/util/util.h (virAsprintf): Remove ATTRIBUTE_RETURN_CHECK, since
it is perfectly fine to ignore the return value, now that the pointer
is guaranteed to be set to NULL upon failure.
* src/util/storage_file.c (absolutePathFromBaseFile): Remove now-
unnecessary use of ignore_value.
* src/util/storage_file.c (absolutePathFromBaseFile): While this use
of virAsprintf is slightly cleaner than using stpncpy(stpcpy(...,
it does impose an artificial limitation on the length of the base_file
name. Rather than asserting that it does not exceed INT_MAX, return
NULL when it does.
When creating preallocated large raw files opening them with O_DSYNC
prevents long delays in reading because cache pages can be immediately
reused without writing them on a disk first.
