- Devel - libvirt List Archives

[libvirt] [PATCH] pciSharesBusWithActive fails to find multiple devices on bus
by Chris Wright 23 Jul '10

23 Jul '10

The first conditional is always true which means the iterator will never find another device on the same bus. if (dev->domain != check->domain || dev->bus != check->bus || ----> (check->slot == check->slot && check->function == check->function)) <----- The goal of that check is to verify that the device is either: in a different pci domain on a different bus is the same identical device This means libvirt may issue a secondary bus reset when there are devices on that bus that actively in use by the host or another guest. Not good. Fix the typo. Cc: Alex Williamson <alex.williamson(a)redhat.com> Cc: Don Dutile <ddutile(a)redhat.com> Cc: Chris Lalancette <clalance(a)redhat.com> Signed-off-by: Chris Wright <chrisw(a)redhat.com> --- src/util/pci.c | 5 +++-- 1 files changed, 3 insertions(+), 2 deletions(-) diff --git a/src/util/pci.c b/src/util/pci.c index b2e1673..6d0ca24 100644 --- a/src/util/pci.c +++ b/src/util/pci.c @@ -446,10 +446,11 @@ pciSharesBusWithActive(pciDevice *dev, pciDevice *check, void *data) { pciDeviceList *activeDevs = data; + /* Different domain, different bus, or simply identical device */ if (dev->domain != check->domain || dev->bus != check->bus || - (check->slot == check->slot && - check->function == check->function)) + (dev->slot == check->slot && + dev->function == check->function)) return 0; if (activeDevs && !pciDeviceListFind(activeDevs, check))

4 3

[libvirt] libvir: Domain error <domName>: invalid argument in virDomainGetXMLDesc
by Priya 23 Jul '10

23 Jul '10

All, I just downloaded and installed *libvirt-bin* and this is my first email to the group, so I apologize in advance if my questions are too naive! So, I am trying to access VMs running on* xen3.2 hypervisor* (host with kernel *2.6.24-24-xen*) using *virt-manager* installed on a machine running*Ubuntu 9.04 * (client machine). I got to the point where I can connect to my virtual machine and my domain-0 through the* virt-manager* client (meaning that I already have *libvirt*,* ssh-askpass* downloaded and the *libvirtd* and * xend* running on my host machine), and on the client machine I can view the window showing the statistics of CPU utilization of my domain-0 and user domain. The attachment shows a shapshot of the *virt-manager* on my client machine. Now when I try to access the GUI of either of domain-0 or the user domain (by selecting the specific domain and clicking open) I get the following error message in a pop-up on my client and in the libvirtd console on the host: "*libvir: Domain error xen_2: invalid argument in virDomainGetXMLDesc*" I tried to Google this error, but I don't find any clear answers. It *seems*to be a bug in the *libvirt* implementation that has been detected and resolved already. And I am confused as to why is it that such a recent version is still buggy? Should I have downloded another version instead? I desperately wish to resolve this problem and would appreciate any guidance. Many thanks! -- Priya Bhat

1 0

[libvirt] Compiling C++ program with libvirt
by adrian wyssen 23 Jul '10

23 Jul '10

Hi everyone I am trying to compile a C++ program which uses libvirt to connect to an ESXi host. I got the code running if I wrote it in C, directly in the hellolibvirt.c example. However, I am now trying to compile that code in a C++ file which is not inside the example directory. I tried to compile with the following command: g++ -g -D debug=1 -c -llibvirt ESXWrapper.cpp g++ -g -D debug=1 -o ESXWrapper ESXWrapper.o The first line passes and creates ESXWrapper.o file. But when I run the second line, it gives me the following error: --- g++ -g -D debug=1 -c -llibvirt ESXWrapper.cpp g++ -g -D debug=1 -o ESXWrapper ESXWrapper.o ESXWrapper.o: In function `VPopCWrapper': /home/wyssen/school/Dropbox/virtualPOPC-1/NEW/vpopcwrapper/ESXWrapper.cpp:124: undefined reference to `virConnectOpenAuth' /home/wyssen/school/Dropbox/virtualPOPC-1/NEW/vpopcwrapper/ESXWrapper.cpp:133: undefined reference to `virConnectGetURI' /home/wyssen/school/Dropbox/virtualPOPC-1/NEW/vpopcwrapper/ESXWrapper.cpp:144: undefined reference to `virConnectClose' /home/wyssen/school/Dropbox/virtualPOPC-1/NEW/vpopcwrapper/ESXWrapper.cpp:124: undefined reference to `virConnectOpenAuth' /home/wyssen/school/Dropbox/virtualPOPC-1/NEW/vpopcwrapper/ESXWrapper.cpp:133: undefined reference to `virConnectGetURI' /home/wyssen/school/Dropbox/virtualPOPC-1/NEW/vpopcwrapper/ESXWrapper.cpp:144: undefined reference to `virConnectClose' ESXWrapper.o: In function `showError': /home/wyssen/school/Dropbox/virtualPOPC-1/NEW/vpopcwrapper/ESXWrapper.cpp:186: undefined reference to `virConnCopyLastError' /home/wyssen/school/Dropbox/virtualPOPC-1/NEW/vpopcwrapper/ESXWrapper.cpp:202: undefined reference to `virResetError' collect2: ld returned 1 exit status make: *** [all] Error 1 --- My makefile is the follwing: --- #vars CC=g++ OUTPUTFILE=ESXWrapper O_FILES=ESXWrapper.o FLAGS=-g -D debug=1 .PHONY: all all: esxwrapper $(CC) $(FLAGS) -o $(OUTPUTFILE) $(O_FILES) esxwrapper: ESXWrapper.cpp VPopCWrapper.h $(CC) $(FLAGS) -c -llibvirt ESXWrapper.cpp .PHONY: run run: all ./bignum .PHONY: clean clean: rm -f $(O_FILES) .PHONY: clean_all clean_all: clean rm -f $(OUTPUTFILE) --- Can anybody give me a hint, how to link correctly to libvirt. And also tell me, if there are special functions to use with C++, other than the examples provided in the source code, which are written in C. Thank you Adrian

2 1

[libvirt] kvm savevm/loadvm
by Serge Hallyn 23 Jul '10

23 Jul '10

Hi, virsh has save/restore to do a one-time save+shutdown. But it does not expose kvm's savevm/restorevm, which let's me do incremental snapshots. Is there a specific reason why that is not exposed in libvirt, or is it just that noone has implemented it? thanks, -serge

4 7

[libvirt] [PATCH] lxc: Ensure to kill init process by sending SIGKILL if needed
by Ryota Ozaki 23 Jul '10

23 Jul '10

Init process may remain after sending SIGTERM for some reason. For example, if original init program is used, it is definitely not killed by SIGTERM. --- src/lxc/lxc_controller.c | 4 ++++ 1 files changed, 4 insertions(+), 0 deletions(-) diff --git a/src/lxc/lxc_controller.c b/src/lxc/lxc_controller.c index 44bcc82..d8b7bc7 100644 --- a/src/lxc/lxc_controller.c +++ b/src/lxc/lxc_controller.c @@ -663,7 +663,11 @@ cleanup: close(containerPty); if (container > 1) { + int status; kill(container, SIGTERM); + if (!(waitpid(container, &status, WNOHANG) == 0 && + WIFEXITED(status))) + kill(container, SIGKILL); waitpid(container, NULL, 0); } return rc; -- 1.6.6.1

2 2

[libvirt] [PATCH] Fix incorrect use of private data in remote driver
by Daniel P. Berrange 22 Jul '10

22 Jul '10

For upstream bug https://bugzilla.redhat.com/show_bug.cgi?id=576796 The remote driver is using the wrong privateData field in a couple of functions. THis is harmless for stateful drivers like QEMU/UML/LXC, but will crash with Xen * src/remote/remote_driver.c: Fix use of privateData field --- src/remote/remote_driver.c | 10 +++++----- 1 files changed, 5 insertions(+), 5 deletions(-) diff --git a/src/remote/remote_driver.c b/src/remote/remote_driver.c index 7052bf1..c9129b1 100644 --- a/src/remote/remote_driver.c +++ b/src/remote/remote_driver.c @@ -4067,7 +4067,7 @@ remoteNetworkIsActive(virNetworkPtr network) int rv = -1; remote_network_is_active_args args; remote_network_is_active_ret ret; - struct private_data *priv = network->conn->privateData; + struct private_data *priv = network->conn->networkPrivateData; remoteDriverLock(priv); @@ -4091,7 +4091,7 @@ remoteNetworkIsPersistent(virNetworkPtr network) int rv = -1; remote_network_is_persistent_args args; remote_network_is_persistent_ret ret; - struct private_data *priv = network->conn->privateData; + struct private_data *priv = network->conn->networkPrivateData; remoteDriverLock(priv); @@ -4622,7 +4622,7 @@ remoteInterfaceIsActive(virInterfacePtr iface) int rv = -1; remote_interface_is_active_args args; remote_interface_is_active_ret ret; - struct private_data *priv = iface->conn->privateData; + struct private_data *priv = iface->conn->interfacePrivateData; remoteDriverLock(priv); @@ -5127,7 +5127,7 @@ remoteStoragePoolIsActive(virStoragePoolPtr pool) int rv = -1; remote_storage_pool_is_active_args args; remote_storage_pool_is_active_ret ret; - struct private_data *priv = pool->conn->privateData; + struct private_data *priv = pool->conn->storagePrivateData; remoteDriverLock(priv); @@ -5151,7 +5151,7 @@ remoteStoragePoolIsPersistent(virStoragePoolPtr pool) int rv = -1; remote_storage_pool_is_persistent_args args; remote_storage_pool_is_persistent_ret ret; - struct private_data *priv = pool->conn->privateData; + struct private_data *priv = pool->conn->storagePrivateData; remoteDriverLock(priv); -- 1.5.5.6 -- |: Red Hat, Engineering, London -o- http://people.redhat.com/berrange/ :| |: http://libvirt.org -o- http://virt-manager.org -o- http://deltacloud.org :| |: http://autobuild.org -o- http://search.cpan.org/~danberr/ :| |: GnuPG: 7D3B9505 -o- F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 :|

3 2

[libvirt] [PATCH (v2)] Fix PCI address assignment if no IDE controller is present
by Daniel P. Berrange 22 Jul '10

22 Jul '10

The previous patch was flawed because it forgot to take into account pre-reserved addresses The PCI slot 1 must be reserved at all times, since PIIX3 is always present, even if no IDE device is in use for guest disks * src/qemu/qemu_conf.c: Always reserve slot 1 for PIIX3 --- src/qemu/qemu_conf.c | 22 +++++++++++++++++----- 1 files changed, 17 insertions(+), 5 deletions(-) diff --git a/src/qemu/qemu_conf.c b/src/qemu/qemu_conf.c index 23e418b..0f1526d 100644 --- a/src/qemu/qemu_conf.c +++ b/src/qemu/qemu_conf.c @@ -2271,19 +2271,23 @@ int qemuDomainPCIAddressSetNextAddr(qemuDomainPCIAddressSetPtr addrs, * - VirtIO balloon * - Host device passthrough * - Watchdog + * + * Prior to this function being invoked, qemuCollectPCIAddress() will have + * added all existing PCI addresses from the 'def' to 'addrs'. Thus this + * function must only try to reserve addresses if info.type == NONE and + * skip over info.type == PCI */ int qemuAssignDevicePCISlots(virDomainDefPtr def, qemuDomainPCIAddressSetPtr addrs) { int i; + bool reservedIDE = false; /* Host bridge */ if (qemuDomainPCIAddressReserveSlot(addrs, 0) < 0) goto error; - /* PIIX3 (ISA bridge, IDE controller, something else unknown, USB controller) - * hardcoded slot=1, multifunction device - */ + /* Verify that first IDE controller (if any) is on the PIIX3, fn 1 */ for (i = 0; i < def->ncontrollers ; i++) { /* First IDE controller lives on the PIIX3 at slot=1, function=1 */ if (def->controllers[i]->type == VIR_DOMAIN_CONTROLLER_TYPE_IDE && @@ -2297,18 +2301,26 @@ qemuAssignDevicePCISlots(virDomainDefPtr def, qemuDomainPCIAddressSetPtr addrs) _("Primary IDE controller must have PCI address 0:0:1.1")); goto error; } + /* If TYPE==PCI, then then qemuCollectPCIAddress() function + * has already reserved the address, so we must skip */ + reservedIDE = true; } else { def->controllers[i]->info.type = VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI; def->controllers[i]->info.addr.pci.domain = 0; def->controllers[i]->info.addr.pci.bus = 0; def->controllers[i]->info.addr.pci.slot = 1; def->controllers[i]->info.addr.pci.function = 1; - if (qemuDomainPCIAddressReserveSlot(addrs, 1) < 0) - goto error; } } } + /* PIIX3 (ISA bridge, IDE controller, something else unknown, USB controller) + * hardcoded slot=1, multifunction device + */ + if (!reservedIDE && + qemuDomainPCIAddressReserveSlot(addrs, 1) < 0) + goto error; + /* First VGA is hardcoded slot=2 */ if (def->nvideos > 0) { if (def->videos[0]->info.type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI) { -- 1.7.1.1

2 1

[libvirt] [PATCH] Properly report failure to create raw storage volume files
by Laine Stump 22 Jul '10

22 Jul '10

We were previously checking for a return < 0 from virFileOperation(), but that function returns a standard errno, which is 0 on success, or some small positive number on failure. The result was that we wouldn't report failures to create storage volume files; instead they would appear to be created, but then would vanish as soon as a pool-refresh was done (or cause some later error as soon as someone tried to access the volume). The other uses of virFileOperation() were already properly checking for != 0. --- src/storage/storage_backend.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/src/storage/storage_backend.c b/src/storage/storage_backend.c index aba8937..9792eed 100644 --- a/src/storage/storage_backend.c +++ b/src/storage/storage_backend.c @@ -362,7 +362,7 @@ virStorageBackendCreateRaw(virConnectPtr conn ATTRIBUTE_UNUSED, createRawFileOpHook, &hdata, VIR_FILE_OP_FORCE_PERMS | (pool->def->type == VIR_STORAGE_POOL_NETFS - ? VIR_FILE_OP_AS_UID : 0))) < 0) { + ? VIR_FILE_OP_AS_UID : 0))) != 0) { virReportSystemError(createstat, _("cannot create path '%s'"), vol->target.path); -- 1.7.1.1

4 17

Re: [libvirt] A problem in live migration
by Alex Jia 21 Jul '10

21 Jul '10

I am not sure if libvirt has resolved hostname and ip mapping problem, it had better that two hosts have same domain name, such as "dhcp-1-2-3.nay.redhat.com" and "dhcp-4-5-6.nay.redhat.com". In addition, I haven't seen your any storage pool information on two hosts, you need to create a sharing storage such as nfs sharing folder, and your guest is in the storage, of course, you may also manually mount your nfs folder to target host instead of creating storage pool. Best Regards, Alex ----- Original Message ----- From: "姚远" <sihuyy(a)gmail.com> To: libvir-list(a)redhat.com Sent: Wednesday, July 21, 2010 3:39:15 PM GMT +08:00 Beijing / Chongqing / Hong Kong / Urumqi Subject: [libvirt] A problem in live migration In virsh, I use command "migrate --live 1 qemu+ssh:// 10.1.10.60/system " to do live migration,but it shows error "internal error hostname on destination resolved to localhost, but migration requires an FQDN". I don't know why, and who do live migration successfully can help me. Thank you! My source and destination system are all fedora-13-x86_64, and the version of libvirt is 0.8.2 -- libvir-list mailing list libvir-list(a)redhat.com https://www.redhat.com/mailman/listinfo/libvir-list

3 2

[libvirt] [PATCH] Fix PCI address assignment if no IDE controller is present
by Daniel P. Berrange 21 Jul '10

21 Jul '10

The PCI slot 1 must be reserved at all times, since PIIX3 is always present, even if no IDE device is in use for guest disks * src/qemu/qemu_conf.c: Always reserve slot 1 for PIIX3 --- src/qemu/qemu_conf.c | 6 ++++-- 1 files changed, 4 insertions(+), 2 deletions(-) diff --git a/src/qemu/qemu_conf.c b/src/qemu/qemu_conf.c index 23e418b..6ce9c9e 100644 --- a/src/qemu/qemu_conf.c +++ b/src/qemu/qemu_conf.c @@ -2284,6 +2284,10 @@ qemuAssignDevicePCISlots(virDomainDefPtr def, qemuDomainPCIAddressSetPtr addrs) /* PIIX3 (ISA bridge, IDE controller, something else unknown, USB controller) * hardcoded slot=1, multifunction device */ + if (qemuDomainPCIAddressReserveSlot(addrs, 1) < 0) + goto error; + + /* Verify that first IDE controller (if any) is on the PIIX3, fn 1 */ for (i = 0; i < def->ncontrollers ; i++) { /* First IDE controller lives on the PIIX3 at slot=1, function=1 */ if (def->controllers[i]->type == VIR_DOMAIN_CONTROLLER_TYPE_IDE && @@ -2303,8 +2307,6 @@ qemuAssignDevicePCISlots(virDomainDefPtr def, qemuDomainPCIAddressSetPtr addrs) def->controllers[i]->info.addr.pci.bus = 0; def->controllers[i]->info.addr.pci.slot = 1; def->controllers[i]->info.addr.pci.function = 1; - if (qemuDomainPCIAddressReserveSlot(addrs, 1) < 0) - goto error; } } } -- 1.7.1.1

2 1