April 2014 - Devel - libvirt List Archives

[libvirt] [PATCH v2] qemu: don't check for backing chains for formats w/o snapshot support
by Martin Kletzander 24 Apr '14

24 Apr '14

Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=1019926 Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=868673 Signed-off-by: Martin Kletzander <mkletzan(a)redhat.com> --- Notes: Smaller and cleaner variant of v1 [1] after discussion with Eric [1] https://www.redhat.com/archives/libvir-list/2014-April/msg00716.html src/qemu/qemu_domain.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/src/qemu/qemu_domain.c b/src/qemu/qemu_domain.c index 45ed872..5687075 100644 --- a/src/qemu/qemu_domain.c +++ b/src/qemu/qemu_domain.c @@ -2268,11 +2268,14 @@ qemuDomainCheckDiskPresence(virQEMUDriverPtr driver, VIR_DEBUG("Checking for disk presence"); for (i = vm->def->ndisks; i > 0; i--) { disk = vm->def->disks[i - 1]; + enum virStorageFileFormat format = virDomainDiskGetFormat(disk); if (!virDomainDiskGetSource(disk)) continue; - if (qemuDomainDetermineDiskChain(driver, vm, disk, false) >= 0 && + if ((format < VIR_STORAGE_FILE_NONE || + format >= VIR_STORAGE_FILE_BACKING) && + qemuDomainDetermineDiskChain(driver, vm, disk, false) >= 0 && qemuDiskChainCheckBroken(disk) >= 0) continue; -- 1.9.2

2 2

[libvirt] [PATCH 00/18] Get rid of virStorageFileMetadata
by Peter Krempa 24 Apr '14

24 Apr '14

Peter Krempa (18): qemu: unexport qemuDiskChainCheckBroken util: storagefile: Always store raw backing name in the metadata util: storage: Don't force path canonicalization when loading metadata util: storage: Remove obsolete argument virStorageFileGetMetadataInternal util: storage: Move checking of the actual backing image to the worker storage: util: Clean up arguments of virStorageFileGetMetadataInternal util: storage: Rename "path" to "relPath" in virStorageFileMetadata util: storagefile: Rename "canonPath" to "path" in virStorageFileMetadata util: virstoragefile: Don't use "backingStore" directly virstoragefile: Kill "backingStore" field from virStorageFileMetadata util: storagefile: Add function to free a virStorageSourcePrt util: storagefile: Add fields from virStorageMetadata to virStorageSource maint: Switch over from struct virStorageFileMetadata to virStorageSource util: virstorage: Kill struct virStorageFileMetadata util: virstoragefile: Rename backingMeta to backingStore storage: Move disk->backingChain to the recursive disk->src.backingStore util: virstoragefile: Don't mangle data stored about directories util: storage: Invert the way recursive metadata retrieval works src/conf/domain_conf.c | 17 +- src/conf/domain_conf.h | 1 - src/libvirt_private.syms | 2 +- src/qemu/qemu_domain.c | 58 ++--- src/qemu/qemu_domain.h | 2 - src/qemu/qemu_driver.c | 44 ++-- src/security/security_selinux.c | 2 +- src/security/virt-aa-helper.c | 9 +- src/storage/storage_backend_fs.c | 4 +- src/storage/storage_backend_gluster.c | 4 +- src/util/virstoragefile.c | 476 ++++++++++++++++------------------ src/util/virstoragefile.h | 88 +++---- tests/virstoragetest.c | 178 ++++++++----- 13 files changed, 435 insertions(+), 450 deletions(-) -- 1.9.1

2 46

[libvirt] KVM connection issue
by Vikas Kokare 24 Apr '14

24 Apr '14

We are using Libvirt Java API version 0.5.0 to connect RHEL 5 KVM, and fetch KVM environment attributes. While doing it, we are seeing the following exception: *this function is not supported by the connection driver: qemuGetSchedulerTypeStack trace: org.libvirt.ErrorHandler.processError(Unknown Source) org.libvirt.Connect.processError(Unknown Source) org.libvirt.Domain.processError(Unknown Source) org.libvirt.Domain.getSchedulerType(Unknown Source)* The machine that is running the Java code to connect to the KVM is also RHEL 5. Is this related to certain libvirt version mismatch or something not supported? -Vikas

1 0

[libvirt] [PATCHv2 1/1] Set pci-ohci as the USB default controller for PPC64.
by Li Zhang 24 Apr '14

24 Apr '14

From: Li Zhang <zhlcindy(a)linux.vnet.ibm.com> PPC64 prefers to set pci-ohci controller as default USB controller. Currently, libvirt is using legacy USB controller as default. There are problems with VGA which can't work correctly with USB Keyboard and USB Mouse. While providing -nodefaults, ppc64 should be specifying the usb controller explicitly in place of using the legacy controller(-usb). Qemu spapr initialization code when sees "-usb" adds a USB Keyboard and USB Mouse by default. And libvirt has added a USB keyboard and USB mouse. A recent fix in the in qemu VGA code uncoverd this problem, which resulted in addition of extra keyboard and mouse to the qemu machine. This patch is to set pci-ohci as USB default controller. Signed-off-by: Li Zhang <zhlcindy(a)linux.vnet.ibm.com> --- v2 -> v1: * Rephrase the commit message. (Nikunj A Dadhania) src/qemu/qemu_command.c | 5 +++-- tests/qemuxml2argvdata/qemuxml2argv-pseries-usb-default.args | 2 +- 2 files changed, 4 insertions(+), 3 deletions(-) diff --git a/src/qemu/qemu_command.c b/src/qemu/qemu_command.c index 379c094..46e851a 100644 --- a/src/qemu/qemu_command.c +++ b/src/qemu/qemu_command.c @@ -8466,8 +8466,9 @@ qemuBuildCommandLine(virConnectPtr conn, } else if (cont->type == VIR_DOMAIN_CONTROLLER_TYPE_USB && cont->model == -1 && !qemuDomainMachineIsQ35(def) && - (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_PIIX3_USB_UHCI) || - def->os.arch == VIR_ARCH_PPC64)) { + (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_PIIX3_USB_UHCI) || + (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_PCI_OHCI) && + def->os.arch == VIR_ARCH_PPC64))) { if (usblegacy) { virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s", _("Multiple legacy USB controllers are " diff --git a/tests/qemuxml2argvdata/qemuxml2argv-pseries-usb-default.args b/tests/qemuxml2argvdata/qemuxml2argv-pseries-usb-default.args index 3a21b76..eac7c56 100644 --- a/tests/qemuxml2argvdata/qemuxml2argv-pseries-usb-default.args +++ b/tests/qemuxml2argvdata/qemuxml2argv-pseries-usb-default.args @@ -3,5 +3,5 @@ LC_ALL=C PATH=/bin HOME=/home/test USER=test LOGNAME=test QEMU_AUDIO_DRV=none \ -nographic -nodefconfig -nodefaults \ -chardev socket,id=charmonitor,path=/tmp/test-monitor,server,nowait \ -mon chardev=charmonitor,id=monitor,mode=readline -no-acpi -boot c \ --usb -chardev pty,id=charserial0 \ +-device pci-ohci,id=usb,bus=pci,addr=0x1 -chardev pty,id=charserial0 \ -device spapr-vty,chardev=charserial0,reg=0x30000000 -- 1.8.2.1

2 3

[libvirt] [PATCH] Qemu fails to restore when guest has selinux static context
by Shivaprasad G Bhat 23 Apr '14

23 Apr '14

The file handle passed from libvirt to qemu is not set to the context of the guest. So, set the selinux context of fd before passing it to qemu Signed-off-by: Shivaprasad G Bhat<sbhat(a)linux.vnet.ibm.com> --- src/qemu/qemu_driver.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c index 4bb4819..8cf3666 100644 --- a/src/qemu/qemu_driver.c +++ b/src/qemu/qemu_driver.c @@ -5522,6 +5522,9 @@ qemuDomainRestoreFlags(virConnectPtr conn, goto cleanup; def = NULL; + if (virSecurityManagerSetImageFDLabel(driver->securityManager, vm->def, fd) < 0) + goto cleanup; + if (qemuDomainObjBeginJob(driver, vm, QEMU_JOB_MODIFY) < 0) goto cleanup;

3 2

[libvirt] [PATCH] Fix pci bus naming for PPC
by Daniel P. Berrange 23 Apr '14

23 Apr '14

Recent discussions around naming of 'pci' vs 'pci.0' for PPC made me go back and look at the PPC emulator in every historical version of QEMU since 1.0. The results were worse than I imagined. This patch adds the logic required to make libvirt work with PPC correctly with naming variations across all versions & machine types. Signed-off-by: Daniel P. Berrange <berrange(a)redhat.com> --- src/qemu/qemu_capabilities.c | 80 +++++++++++++++++++++++++++++++++++++++----- src/qemu/qemu_capabilities.h | 3 ++ src/qemu/qemu_command.c | 2 +- 3 files changed, 75 insertions(+), 10 deletions(-) diff --git a/src/qemu/qemu_capabilities.c b/src/qemu/qemu_capabilities.c index a5c6879..b491f58 100644 --- a/src/qemu/qemu_capabilities.c +++ b/src/qemu/qemu_capabilities.c @@ -1928,6 +1928,56 @@ virQEMUCapsGet(virQEMUCapsPtr qemuCaps, } +bool virQEMUCapsHasPCIMultiBus(virQEMUCapsPtr qemuCaps, + virDomainDefPtr def) +{ + bool hasMultiBus = virQEMUCapsGet(qemuCaps, QEMU_CAPS_PCI_MULTIBUS); + + if (hasMultiBus) + return true; + + if (def->os.arch == VIR_ARCH_PPC || + def->os.arch == VIR_ARCH_PPC64) { + /* + * Usage of pci.0 naming: + * + * ref405ep: no pci + * taihu: no pci + * bamboo: 1.1.0 + * mac99: 2.0.0 + * g3beige: 2.0.0 + * prep: 1.4.0 + * pseries: 2.0.0 + * mpc8544ds: forever + * virtex-m507: no pci + * ppce500: 1.6.0 + */ + + if (qemuCaps->version >= 2000000) + return true; + + if (qemuCaps->version >= 1006000 && + STREQ(def->os.machine, "ppce500")) + return true; + + if (qemuCaps->version >= 1004000 && + STREQ(def->os.machine, "prep")) + return true; + + if (qemuCaps->version >= 1001000 && + STREQ(def->os.machine, "bamboo")) + return true; + + if (STREQ(def->os.machine, "mpc8544ds")) + return true; + + return false; + } + + return false; +} + + const char *virQEMUCapsGetBinary(virQEMUCapsPtr qemuCaps) { return qemuCaps->binary; @@ -2809,15 +2859,22 @@ virQEMUCapsInitHelp(virQEMUCapsPtr qemuCaps, uid_t runUid, gid_t runGid) false) < 0) goto cleanup; - /* Currently only x86_64 and i686 support PCI-multibus. */ + /* x86_64 and i686 support PCI-multibus on all machine types + * since forever. For other architectures, it has been changing + * across releases, per machine type, so we can't simply detect + * it here. Thus the rest of the logic is provided in a separate + * helper virQEMUCapsHasPCIMultiBus() which keys off the machine + * stored in virDomainDef and QEMU version number + */ if (qemuCaps->arch == VIR_ARCH_X86_64 || - qemuCaps->arch == VIR_ARCH_I686) { + qemuCaps->arch == VIR_ARCH_I686) virQEMUCapsSet(qemuCaps, QEMU_CAPS_PCI_MULTIBUS); - } else { - /* -no-acpi is not supported on other archs - * even if qemu reports it in -help */ + + /* -no-acpi is not supported on non-x86 + * even if qemu reports it in -help */ + if (qemuCaps->arch != VIR_ARCH_X86_64 && + qemuCaps->arch != VIR_ARCH_I686) virQEMUCapsClear(qemuCaps, QEMU_CAPS_NO_ACPI); - } /* virQEMUCapsExtractDeviceStr will only set additional caps if qemu * understands the 0.13.0+ notion of "-device driver,". */ @@ -2934,9 +2991,14 @@ virQEMUCapsInitArchQMPBasic(virQEMUCapsPtr qemuCaps, goto cleanup; } - /* - * Currently only x86_64 and i686 support PCI-multibus, - * -no-acpi and -no-kvm-pit-reinjection. + /* x86_64 and i686 support PCI-multibus on all machine types + * since forever. For other architectures, it has been changing + * across releases, per machine type, so we can't simply detect + * it here. Thus the rest of the logic is provided in a separate + * helper virQEMUCapsHasPCIMultiBus() which keys off the machine + * stored in virDomainDef and QEMU version number + * + * ACPI/HPET/KVM PIT are also x86 specific */ if (qemuCaps->arch == VIR_ARCH_X86_64 || qemuCaps->arch == VIR_ARCH_I686) { diff --git a/src/qemu/qemu_capabilities.h b/src/qemu/qemu_capabilities.h index df8c978..5c63b81 100644 --- a/src/qemu/qemu_capabilities.h +++ b/src/qemu/qemu_capabilities.h @@ -239,6 +239,9 @@ void virQEMUCapsClear(virQEMUCapsPtr qemuCaps, bool virQEMUCapsGet(virQEMUCapsPtr qemuCaps, enum virQEMUCapsFlags flag); +bool virQEMUCapsHasPCIMultiBus(virQEMUCapsPtr qemuCaps, + virDomainDefPtr def); + char *virQEMUCapsFlagsString(virQEMUCapsPtr qemuCaps); const char *virQEMUCapsGetBinary(virQEMUCapsPtr qemuCaps); diff --git a/src/qemu/qemu_command.c b/src/qemu/qemu_command.c index 91367ab..403c9a9 100644 --- a/src/qemu/qemu_command.c +++ b/src/qemu/qemu_command.c @@ -3076,7 +3076,7 @@ qemuBuildDeviceAddressStr(virBufferPtr buf, goto cleanup; } } else { - if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_PCI_MULTIBUS)) { + if (virQEMUCapsHasPCIMultiBus(qemuCaps, domainDef)) { virBufferAsprintf(buf, ",bus=%s", contAlias); } else { virBufferAddLit(buf, ",bus=pci"); -- 1.9.0

2 1

[libvirt] [PATCH] nwfilter: Validate rule after parsing
by Stefan Berger 23 Apr '14

23 Apr '14

From: Stefan Berger <stefanb(a)linux.vnet.ibm.com> An IP or IPv6 rule with port specification but without protocol specification cannot be instantiated by ebtables. The documentation points to 'protocol' being required but implementation does not enforce it to be given. Implement a rule validation function that checks whether the rule is valid when it is defined. This for example prevents the definition of rules like: <ip dstportstart='53'> where a protocol attribute would be required for it to be valid and for ebtables to be able to instantiate it. A valid rule then is: <ip protocol='udp' dstportstart='53'> Signed-off-by: Stefan Berger <stefanb(a)linux.vnet.ibm.com> --- src/conf/nwfilter_conf.c | 59 ++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 59 insertions(+) diff --git a/src/conf/nwfilter_conf.c b/src/conf/nwfilter_conf.c index f5a75e4..ce7bc78 100644 --- a/src/conf/nwfilter_conf.c +++ b/src/conf/nwfilter_conf.c @@ -2093,6 +2093,63 @@ virNWFilterRuleDefFixupIPSet(ipHdrDataDefPtr ipHdr) } } + +/* + * virNWFilterRuleValidate + * + * Perform some basic rule validation to prevent rules from being + * defined that cannot be instantiated. + */ +static int +virNWFilterRuleValidate(virNWFilterRuleDefPtr rule) +{ + int ret = 0; + portDataDefPtr portData = NULL; + const char *protocol; + + switch (rule->prtclType) { + case VIR_NWFILTER_RULE_PROTOCOL_IP: + portData = &rule->p.ipHdrFilter.portData; + protocol = "IP"; + /* fall through */ + case VIR_NWFILTER_RULE_PROTOCOL_IPV6: + if (portData == NULL) { + portData = &rule->p.ipv6HdrFilter.portData; + protocol = "IPv6"; + } + if (HAS_ENTRY_ITEM(&portData->dataSrcPortStart) || + HAS_ENTRY_ITEM(&portData->dataDstPortStart) || + HAS_ENTRY_ITEM(&portData->dataSrcPortEnd) || + HAS_ENTRY_ITEM(&portData->dataDstPortEnd)) { + if (HAS_ENTRY_ITEM(&rule->p.ipHdrFilter.ipHdr.dataProtocolID)) { + switch (rule->p.ipHdrFilter.ipHdr.dataProtocolID.u.u8) { + case 6: /* tcp */ + case 17: /* udp */ + case 33: /* dccp */ + case 132: /* sctp */ + break; + default: + ret = -1; + } + } else { + ret = -1; + } + if (ret < 0) { + virReportError(VIR_ERR_INTERNAL_ERROR, + _("%s rule with port specification requires " + "protocol specification with protocol to be " + "either one of tcp(6), udp(17), dccp(33), or " + "sctp(132)"), protocol); + } + } + break; + default: + break; + } + + return ret; +} + static void virNWFilterRuleDefFixup(virNWFilterRuleDefPtr rule) { @@ -2389,6 +2446,8 @@ virNWFilterRuleParse(xmlNodePtr node) virAttr[i].att) < 0) { goto err_exit; } + if (virNWFilterRuleValidate(ret) < 0) + goto err_exit; break; } if (!found) { -- 1.8.1.4

2 1

[libvirt] [PATCHv2 0/7] more of round 2 of virstoragefile refactoring
by Eric Blake 23 Apr '14

23 Apr '14

compared to v1, this is rebased on top of Cole's changes, treats disk partition types slightly differently, and adds a couple patches. 1/7: was 16/n, unreviewed so far 2/7: was 13/n with weak ack 3-4/7: was 14-15/n; commit messages improved to justify it further 5-7: all new patches I'm still working on how best to fold virStorageFileMetadata into virStorageSource; it's a bit tricky since the metadata type is rather awkward (storing type information about the backing file in the parent, rather than in the backing file) Eric Blake (7): conf: manage disk source by struct instead of pieces conf: tweak volume target struct details conf: move volume structs to util/ conf: use common struct in storage volumes conf: track sizes directly in source struct conf: drop redundant parameters during probe conf: modify tracking of encrypted images src/conf/domain_conf.c | 139 ++++++++++++------------------ src/conf/domain_conf.h | 24 ++---- src/conf/snapshot_conf.c | 17 +--- src/conf/storage_conf.c | 40 ++++----- src/conf/storage_conf.h | 51 ++--------- src/esx/esx_storage_backend_iscsi.c | 5 +- src/esx/esx_storage_backend_vmfs.c | 20 ++--- src/parallels/parallels_driver.c | 2 +- src/parallels/parallels_storage.c | 22 ++--- src/phyp/phyp_driver.c | 4 +- src/storage/storage_backend.c | 150 +++++++++++++++------------------ src/storage/storage_backend.h | 11 +-- src/storage/storage_backend_disk.c | 26 +++--- src/storage/storage_backend_fs.c | 32 +++---- src/storage/storage_backend_gluster.c | 12 ++- src/storage/storage_backend_logical.c | 18 ++-- src/storage/storage_backend_mpath.c | 8 +- src/storage/storage_backend_rbd.c | 11 +-- src/storage/storage_backend_scsi.c | 6 +- src/storage/storage_backend_sheepdog.c | 10 +-- src/storage/storage_driver.c | 40 ++++----- src/test/test_driver.c | 16 ++-- src/util/virstoragefile.c | 11 ++- src/util/virstoragefile.h | 30 ++++++- src/vbox/vbox_tmpl.c | 11 +-- tests/storagebackendsheepdogtest.c | 5 +- tests/virstoragetest.c | 4 +- 27 files changed, 322 insertions(+), 403 deletions(-) -- 1.9.0

3 18

[libvirt] [PATCH 0/2] Two more setvcpu fixes
by Ján Tomko 23 Apr '14

23 Apr '14

Fix fix fix. Ján Tomko (2): Save domain status after cpu hotplug Document behavior of setvcpus during guest boot src/libvirt.c | 6 ++++++ src/qemu/qemu_driver.c | 3 +++ tools/virsh.pod | 4 +++- 3 files changed, 12 insertions(+), 1 deletion(-) -- 1.8.3.2

2 5

[libvirt] [PATCH] bhyve: bhyveDomainDefineXML fixes
by Roman Bogorodskiy 23 Apr '14

23 Apr '14

- do not loose new definition for an active domain - do not leak oldDef - do not set dom->id if virDomainSaveConfig() fails - do not call virObjectUnlock(vm) if vm is NULL --- src/bhyve/bhyve_driver.c | 14 ++++++++++---- 1 file changed, 10 insertions(+), 4 deletions(-) diff --git a/src/bhyve/bhyve_driver.c b/src/bhyve/bhyve_driver.c index 0cafe4c..6d681fd 100644 --- a/src/bhyve/bhyve_driver.c +++ b/src/bhyve/bhyve_driver.c @@ -472,17 +472,23 @@ bhyveDomainDefineXML(virConnectPtr conn, const char *xml) def = NULL; vm->persistent = 1; + if (virDomainSaveConfig(BHYVE_CONFIG_DIR, + vm->newDef ? vm->newDef : vm->def) < 0) { + virDomainObjListRemove(privconn->domains, vm); + vm = NULL; + goto cleanup; + } + dom = virGetDomain(conn, vm->def->name, vm->def->uuid); if (dom) dom->id = vm->def->id; - if (virDomainSaveConfig(BHYVE_CONFIG_DIR, vm->def) < 0) - goto cleanup; - cleanup: virObjectUnref(caps); virDomainDefFree(def); - virObjectUnlock(vm); + virDomainDefFree(oldDef); + if (vm) + virObjectUnlock(vm); return dom; } -- 1.9.0

2 2