July 2009 - Devel - libvirt List Archives

[libvirt] [PATCH 1/3] Move ldconfig calls to libvirt-client %post/%postun
by Mark McLoughlin 29 Jul '09

29 Jul '09

Now that the library has moved to libvirt-client --- libvirt.spec.in | 5 +++-- 1 files changed, 3 insertions(+), 2 deletions(-) diff --git a/libvirt.spec.in b/libvirt.spec.in index 6228d0b..b79e766 100644 --- a/libvirt.spec.in +++ b/libvirt.spec.in @@ -411,7 +411,6 @@ chmod 0644 $RPM_BUILD_ROOT%{_sysconfdir}/sysconfig/libvirtd rm -fr %{buildroot} %post -/sbin/ldconfig %if %{with_libvirtd} %if %{with_qemu} @@ -440,7 +439,9 @@ if [ $1 = 0 ]; then fi %endif -%postun -p /sbin/ldconfig +%post client -p /sbin/ldconfig + +%postun client -p /sbin/ldconfig %files %defattr(-, root, root) -- 1.6.2.5

2 5

[libvirt] [PATCH] libvirt-glib: add missing details argument to callbacks
by Guido Günther 29 Jul '09

29 Jul '09

Hi, the python example is lacking the details argument in the callback. This makes the example fail due to the wrong number of arguments. Cheers, -- Guido

2 5

[libvirt] Debug the username/passwords
by Bryan Kearney 29 Jul '09

29 Jul '09

Hopefully this is the last question I have on the Java drivers. I am trying to debug the callback from openAuth(). I am passing back what I think is valid information, but I see in /var/log/messages the following: Jul 28 16:21:24 localhost libvirtd: sasl step failed -20 (SASL(-13): user not found: no secret in database) Is there a way to output what I have passed back in the credential responses? Setting the following in the /etc/libvirt/libvirtd.conf file does not seem to increase the logging: log_level=1 log_outputs="1:stderr 1:syslog:bk 1:file:virsh.log" and setting LIBVIRT_DEBUG=1,and launching libvirtd from the command line does not spit out the info. -- bk

2 2

[libvirt] ANNOUNCE: New release virt-manager 0.8.0
by Cole Robinson 29 Jul '09

29 Jul '09

I'm happy to announce a new virt-manager release, version 0.8.0. The release can be downloaded from: http://virt-manager.org/download.html The direct download link is: http://virt-manager.org/download/sources/virt-manager/virt-manager-0.8.0.ta… This release includes: - New 'Clone VM' Wizard - Improved UI, including an overhaul of the main 'manager' view - System tray icon for easy VM access (start, stop, view console/details) - Wizard for adding serial, parallel, and video devices to existing VMs. - CPU pinning support (Michal Novotny) - Ability to view and change VM security (sVirt) settings (Dan Walsh) - Many bug fixes and improvements Thanks to everyone who has contributed to this release through testing, bug reporting, submitting patches, and otherwise sending in feedback! Thanks, Cole

4 3

[libvirt] PATCH: Fix remote domain events deadlock
by Daniel P. Berrange 29 Jul '09

29 Jul '09

There is a deadlock in the remote driver's dispatch code for domain events. If your callback triggers a API call that does an RPC call, then the driver will deadlock. The fix is the same as we did in the QEMU driver. We releae the lock & re-aquire it, with some extreme cleverness to protect against people deleting callbacks during the course of dispatch Daniel commit e2e6f795dd1832227c7d5c7832596be3642fe6e4 Author: Daniel P. Berrange <berrange(a)redhat.com> Date: Tue Jul 28 17:45:19 2009 +0100 Fix deadlock in remote driver domain events * src/remote_internal.c: Release driver lock when dispatching events to callbacks diff --git a/src/remote_internal.c b/src/remote_internal.c index f20ed6e..a58b768 100644 --- a/src/remote_internal.c +++ b/src/remote_internal.c @@ -167,6 +167,8 @@ struct private_data { virDomainEventQueuePtr domainEvents; /* Timer for flushing domainEvents queue */ int eventFlushTimer; + /* Flag if we're in process of dispatching */ + int domainEventDispatching; /* Self-pipe to wakeup threads waiting in poll() */ int wakeupSendFD; @@ -6288,18 +6290,26 @@ static int remoteDomainEventDeregister (virConnectPtr conn, remoteDriverLock(priv); - if (virDomainEventCallbackListRemove(conn, priv->callbackList, - callback) < 0) { - error (conn, VIR_ERR_RPC, _("removing cb fron list")); - goto done; - } - - if ( priv->callbackList->count == 0 ) { - /* Tell the server when we are the last callback deregistering */ - if (call (conn, priv, 0, REMOTE_PROC_DOMAIN_EVENTS_DEREGISTER, - (xdrproc_t) xdr_void, (char *) NULL, - (xdrproc_t) xdr_void, (char *) NULL) == -1) + if (priv->domainEventDispatching) { + if (virDomainEventCallbackListMarkDelete(conn, priv->callbackList, + callback) < 0) { + error (conn, VIR_ERR_RPC, _("marking cb for deletion")); goto done; + } + } else { + if (virDomainEventCallbackListRemove(conn, priv->callbackList, + callback) < 0) { + error (conn, VIR_ERR_RPC, _("removing cb from list")); + goto done; + } + + if ( priv->callbackList->count == 0 ) { + /* Tell the server when we are the last callback deregistering */ + if (call (conn, priv, 0, REMOTE_PROC_DOMAIN_EVENTS_DEREGISTER, + (xdrproc_t) xdr_void, (char *) NULL, + (xdrproc_t) xdr_void, (char *) NULL) == -1) + goto done; + } } rv = 0; @@ -7309,18 +7319,54 @@ done: remoteDriverUnlock(priv); } +static void remoteDomainEventDispatchFunc(virConnectPtr conn, + virDomainEventPtr event, + virConnectDomainEventCallback cb, + void *cbopaque, + void *opaque) +{ + struct private_data *priv = opaque; + + /* Drop the lock whle dispatching, for sake of re-entrancy */ + remoteDriverUnlock(priv); + virDomainEventDispatchDefaultFunc(conn, event, cb, cbopaque, NULL); + remoteDriverLock(priv); +} + void remoteDomainEventQueueFlush(int timer ATTRIBUTE_UNUSED, void *opaque) { virConnectPtr conn = opaque; struct private_data *priv = conn->privateData; + virDomainEventQueue tempQueue; remoteDriverLock(priv); - virDomainEventQueueDispatch(priv->domainEvents, priv->callbackList, - virDomainEventDispatchDefaultFunc, NULL); + priv->domainEventDispatching = 1; + + /* Copy the queue, so we're reentrant safe */ + tempQueue.count = priv->domainEvents->count; + tempQueue.events = priv->domainEvents->events; + priv->domainEvents->count = 0; + priv->domainEvents->events = NULL; + + virDomainEventQueueDispatch(&tempQueue, priv->callbackList, + remoteDomainEventDispatchFunc, priv); virEventUpdateTimeout(priv->eventFlushTimer, -1); + /* Purge any deleted callbacks */ + virDomainEventCallbackListPurgeMarked(priv->callbackList); + + if ( priv->callbackList->count == 0 ) { + /* Tell the server when we are the last callback deregistering */ + if (call (conn, priv, 0, REMOTE_PROC_DOMAIN_EVENTS_DEREGISTER, + (xdrproc_t) xdr_void, (char *) NULL, + (xdrproc_t) xdr_void, (char *) NULL) == -1) + VIR_WARN0("Failed to de-register events"); + } + + priv->domainEventDispatching = 0; + remoteDriverUnlock(priv); } -- |: Red Hat, Engineering, London -o- http://people.redhat.com/berrange/ :| |: http://libvirt.org -o- http://virt-manager.org -o- http://ovirt.org :| |: http://autobuild.org -o- http://search.cpan.org/~danberr/ :| |: GnuPG: 7D3B9505 -o- F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 :|

2 1

[libvirt] [PATCH] Fix trailing whitespace in NEWS
by Mark McLoughlin 29 Jul '09

29 Jul '09

Hey, I had to commit this before I could push Cheers, Mark.

1 0

[libvirt] [PATCH 00/14] Clean up spec files mess
by Mark McLoughlin 29 Jul '09

29 Jul '09

Hey, I get treated like a war criminal when I make a change in the Fedora spec file and don't send it upstream, yet there are a whole heap of changes which haven't been pushed upstream ... only one of which is mine ... two changes less than DV himself :-) The following series of patches pulls in the changes from Fedora which make sense. Some of them I'm not 100% sure of: - defaulting to policy kit - is there some reason we don't want that upstream? - The python cruft thing, not sure its needed - Converting NEWS to UTF-8 could probably be done upstream - Requiring libselinux is dubious, doesn't autorequires handle that? My final contention is that maintaining these spec files in the upstream repo actually makes it harder to maintain the Fedora spec file and helps nobody. Because of half-assed attempts to keep the two in sync, I ended up having to fix a bunch of stuff and re-instate another bunch of stuff in Fedora: - Enable netcf support - Pass --with-qemu-user=qemu etc. to configure - Move various requires to the libvirt-client sub-package - Sync some trivial cleanups from upstream spec file - Remove explicit libxml2 requires, again - Build with --without-capng if capng support is disabled - Remove explicit dir creating in makeinstall, replaced by attr in files - Set perms on /var/{run,lib,cache}/libvirt/qemu I really think we should kill these off and just point people at the Fedora spec file if they're looking for a reference spec file. I need some very strong rationale before I ever go wasting my time (2 hours) on this again. Cheers, Mark.

4 23

[libvirt] [PATCH] Remove MAX_TAP_ID
by Aron Griffis 29 Jul '09

29 Jul '09

As far as I can tell, there's no reason to format the device string in brAddTap(). Delegate the job to TUNSETIFF, thereby removing the loop and the MAX_TAP_ID artificial limit. This patch allows me to get 421 guests running before hitting other limits. Signed-off-by: Aron Griffis <aron.griffis(a)hp.com> --- src/bridge.c | 94 +++++++++++++++++++++------------------------------------- 1 files changed, 34 insertions(+), 60 deletions(-) diff --git a/.gnulib b/.gnulib deleted file mode 160000 index b653eda..0000000 --- a/.gnulib +++ /dev/null @@ -1 +0,0 @@ -Subproject commit b653eda3ac4864de205419d9f41eec267cb89eeb diff --git a/src/bridge.c b/src/bridge.c index 0509afd..ec37192 100644 --- a/src/bridge.c +++ b/src/bridge.c @@ -49,8 +49,6 @@ #include "util.h" #include "logging.h" -#define MAX_TAP_ID 256 - #define JIFFIES_TO_MS(j) (((j)*1000)/HZ) #define MS_TO_JIFFIES(ms) (((ms)*HZ)/1000) @@ -466,76 +464,52 @@ brAddTap(brControl *ctl, int vnet_hdr, int *tapfd) { - int id, subst, fd; + int fd, len; + struct ifreq ifr = {0}; if (!ctl || !ctl->fd || !bridge || !ifname) return EINVAL; - subst = id = 0; - - if (strstr(*ifname, "%d")) - subst = 1; - if ((fd = open("/dev/net/tun", O_RDWR)) < 0) return errno; - if (vnet_hdr) - vnet_hdr = brProbeVnetHdr(fd); + ifr.ifr_flags = IFF_TAP|IFF_NO_PI; - do { - struct ifreq try; - int len; +#ifdef IFF_VNET_HDR + if (vnet_hdr && brProbeVnetHdr(fd)) + ifr.ifr_flags |= IFF_VNET_HDR; +#endif - memset(&try, 0, sizeof(struct ifreq)); + strncpy(ifr.ifr_name, *ifname, IFNAMSIZ-1); - try.ifr_flags = IFF_TAP|IFF_NO_PI; + if (ioctl(fd, TUNSETIFF, &ifr) < 0) + goto error; -#ifdef IFF_VNET_HDR - if (vnet_hdr) - try.ifr_flags |= IFF_VNET_HDR; -#endif + len = strlen(ifr.ifr_name); + if (len >= BR_IFNAME_MAXLEN - 1) { + errno = EINVAL; + goto error; + } - if (subst) { - len = snprintf(try.ifr_name, BR_IFNAME_MAXLEN, *ifname, id); - if (len >= BR_IFNAME_MAXLEN) { - errno = EADDRINUSE; - goto error; - } - } else { - len = strlen(*ifname); - if (len >= BR_IFNAME_MAXLEN - 1) { - errno = EINVAL; - goto error; - } - - strncpy(try.ifr_name, *ifname, len); - try.ifr_name[len] = '\0'; - } - - if (ioctl(fd, TUNSETIFF, &try) == 0) { - /* We need to set the interface MTU before adding it - * to the bridge, because the bridge will have its - * MTU adjusted automatically when we add the new interface. - */ - if ((errno = brSetInterfaceMtu(ctl, bridge, try.ifr_name))) - goto error; - if ((errno = brAddInterface(ctl, bridge, try.ifr_name))) - goto error; - if ((errno = brSetInterfaceUp(ctl, try.ifr_name, 1))) - goto error; - if (!tapfd && - (errno = ioctl(fd, TUNSETPERSIST, 1))) - goto error; - VIR_FREE(*ifname); - if (!(*ifname = strdup(try.ifr_name))) - goto error; - if (tapfd) - *tapfd = fd; - return 0; - } - - id++; - } while (subst && id <= MAX_TAP_ID); + /* We need to set the interface MTU before adding it + * to the bridge, because the bridge will have its + * MTU adjusted automatically when we add the new interface. + */ + if ((errno = brSetInterfaceMtu(ctl, bridge, ifr.ifr_name))) + goto error; + if ((errno = brAddInterface(ctl, bridge, ifr.ifr_name))) + goto error; + if ((errno = brSetInterfaceUp(ctl, ifr.ifr_name, 1))) + goto error; + if (!tapfd && + (errno = ioctl(fd, TUNSETPERSIST, 1))) + goto error; + VIR_FREE(*ifname); + if (!(*ifname = strdup(ifr.ifr_name))) + goto error; + if (tapfd) + *tapfd = fd; + return 0; error: close(fd);

1 0

[libvirt] ANNOUNCE: New release virtinst 0.500.0
by Cole Robinson 29 Jul '09

29 Jul '09

I'm happy to announce a new virtinst release, version 0.500.0. The release can be downloaded from: http://virt-manager.org/download.html The direct download link is: http://virt-manager.org/download/sources/virtinst/virtinst-0.500.0.tar.gz This release includes: - New virt-install device options --serial, --parallel, and --video - Allow various auth types for libvirt connections (PolicyKit, SASL, ...) - New virt-clone option --auto-clone: generates all needed input. - Option to specify network device model via virt-install --network (Guido Gunther) - New virt-install option --virt-type for specifying hypervisor type (kvm, qemu). --accelerate is now the default behavior: To provision a plain qemu VM on a KVM enabled host, use '--virt-type qemu' - OVF input support for virt-convert - Many bug fixes and improvements Thanks to everyone who has contributed to this release through testing, bug reporting, submitting patches, and otherwise sending in feedback! Thanks, Cole

1 0

[libvirt] [PATCH] qemu: fix missing ifname in qemu argument
by Ryota Ozaki 28 Jul '09

28 Jul '09

Hi, This patch adds missing ifname, which is a tap name for network/bridge configuration, in qemu argument of -net. Without this patch, a specified tap name, e.g., <target dev='tap01'/>, in a domain XML will be ignored. Thanks, ozaki-r >From feeca4f2cecf5c6836f7aec6e004585f3b855bed Mon Sep 17 00:00:00 2001 From: Ryota Ozaki <ozaki.ryota(a)gmail.com> Date: Tue, 28 Jul 2009 23:46:12 +0900 Subject: [PATCH] qemu: fix missing ifname in qemu argument --- src/qemu_conf.c | 6 ++++-- 1 files changed, 4 insertions(+), 2 deletions(-) diff --git a/src/qemu_conf.c b/src/qemu_conf.c index 4fe4e39..48bf928 100644 --- a/src/qemu_conf.c +++ b/src/qemu_conf.c @@ -1187,11 +1187,13 @@ qemuBuildHostNetStr(virConnectPtr conn, switch (net->type) { case VIR_DOMAIN_NET_TYPE_NETWORK: case VIR_DOMAIN_NET_TYPE_BRIDGE: - if (virAsprintf(str, "%stap%cfd=%s,vlan=%d%s%s", + if (virAsprintf(str, "%stap%cfd=%s,vlan=%d%s%s%s%s", prefix ? prefix : "", type_sep, tapfd, vlan, (net->hostnet_name ? ",name=" : ""), - (net->hostnet_name ? net->hostnet_name : "")) < 0) { + (net->hostnet_name ? net->hostnet_name : ""), + (net->ifname ? ",ifname=" : ""), + (net->ifname ? net->ifname : "")) < 0) { virReportOOMError(conn); return -1; } -- 1.6.0.6

2 2