- Devel - libvirt List Archives

[libvirt] ANNOUNCE: virt-bootstrap 1.1.0 released
by Cedric Bosdonnat 31 May '18

31 May '18

I'm happy to announce the release of virt-bootstrap 1.1.0! virt-bootstrap is a tool providing an easy way to setup the root file system for libvirt-based containers. The release can be downloaded from: http://virt-manager.org/download/ This release includes: - safe_untar: check for permissions to set attribs (Radostin Stoyanov) - docker source, support blobs without .tar extension (Radostin Stoyanov) - docker-source, preserve extended file attributes (Radostin Stoyanov) - docker-source, get list of layers without `--raw` (Radostin Stoyanov) - docker-source, void skopeo copy in cache (Radostin Stoyanov) - Show error when guestfs-python or skopeo is not installed (Radostin Stoyanov) - pylint cleanups (Radostin Stoyanov) Thanks to everyone who has contributed to this release through testing, bug reporting, submitting patches, and otherwise sending in feedback! Thanks, -- Cédric

1 0

[libvirt] [PATCH] testUpdateQEMUCaps: Don't leak host cpuData
by Michal Privoznik 31 May '18

31 May '18

When preparing qemuCaps for test cases the following is happening: qemuTestParseCapabilitiesArch() is called, which calls virQEMUCapsLoadCache() which in turn calls virQEMUCapsInitHostCPUModel() which sets qemuCaps->kvmCPU and qemuCaps->tcgCPU. But then the code tries to update the capabilities: testCompareXMLToArgv() calls testUpdateQEMUCaps() which calls virQEMUCapsInitHostCPUModel() again overwriting previously allocated memory. The solution is to free host cpuData in testUpdateQEMUCaps(). Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com> --- Technically this is v2 of [1] but since it implements completely different approach I'm sending it as v1. 1: https://www.redhat.com/archives/libvir-list/2018-May/msg02260.html src/qemu/qemu_capabilities.c | 21 +++++++++++++++++++-- src/qemu/qemu_capspriv.h | 4 ++++ tests/qemuxml2argvtest.c | 3 +++ 3 files changed, 26 insertions(+), 2 deletions(-) diff --git a/src/qemu/qemu_capabilities.c b/src/qemu/qemu_capabilities.c index e2e76e4dd8..ea1ff520d8 100644 --- a/src/qemu/qemu_capabilities.c +++ b/src/qemu/qemu_capabilities.c @@ -1516,12 +1516,19 @@ virQEMUCapsHostCPUDataCopy(virQEMUCapsHostCPUDataPtr dst, static void -virQEMUCapsHostCPUDataClear(virQEMUCapsHostCPUDataPtr cpuData) +virQEMUCapsHostCPUDataClearHostCPU(virQEMUCapsHostCPUDataPtr cpuData) { - qemuMonitorCPUModelInfoFree(cpuData->info); virCPUDefFree(cpuData->reported); virCPUDefFree(cpuData->migratable); virCPUDefFree(cpuData->full); +} + + +static void +virQEMUCapsHostCPUDataClear(virQEMUCapsHostCPUDataPtr cpuData) +{ + qemuMonitorCPUModelInfoFree(cpuData->info); + virQEMUCapsHostCPUDataClearHostCPU(cpuData); memset(cpuData, 0, sizeof(*cpuData)); } @@ -2834,6 +2841,16 @@ virQEMUCapsNewHostCPUModel(void) } +void +virQEMUCapsFreeHostCPUModel(virQEMUCapsPtr qemuCaps, + virDomainVirtType type) +{ + virQEMUCapsHostCPUDataPtr cpuData = virQEMUCapsGetHostCPUData(qemuCaps, type); + + virQEMUCapsHostCPUDataClearHostCPU(cpuData); +} + + void virQEMUCapsInitHostCPUModel(virQEMUCapsPtr qemuCaps, virArch hostArch, diff --git a/src/qemu/qemu_capspriv.h b/src/qemu/qemu_capspriv.h index 0199501c93..fea039ef3a 100644 --- a/src/qemu/qemu_capspriv.h +++ b/src/qemu/qemu_capspriv.h @@ -56,6 +56,10 @@ void virQEMUCapsSetArch(virQEMUCapsPtr qemuCaps, virArch arch); +void +virQEMUCapsFreeHostCPUModel(virQEMUCapsPtr qemuCaps, + virDomainVirtType type); + void virQEMUCapsInitHostCPUModel(virQEMUCapsPtr qemuCaps, virArch hostArch, diff --git a/tests/qemuxml2argvtest.c b/tests/qemuxml2argvtest.c index a6a40e273e..61c7ae59aa 100644 --- a/tests/qemuxml2argvtest.c +++ b/tests/qemuxml2argvtest.c @@ -388,6 +388,9 @@ testUpdateQEMUCaps(const struct testInfo *info, if (testAddCPUModels(info->qemuCaps, info->skipLegacyCPUs) < 0) goto cleanup; + virQEMUCapsFreeHostCPUModel(info->qemuCaps, VIR_DOMAIN_VIRT_KVM); + virQEMUCapsFreeHostCPUModel(info->qemuCaps, VIR_DOMAIN_VIRT_QEMU); + virQEMUCapsInitHostCPUModel(info->qemuCaps, caps->host.arch, VIR_DOMAIN_VIRT_KVM); virQEMUCapsInitHostCPUModel(info->qemuCaps, caps->host.arch, -- 2.16.4

2 1

[libvirt] [PATCH v2 0/5] qemu: Forbid old qcow/qcow2 encryption
by Peter Krempa 31 May '18

31 May '18

The old qcow/qcow2 encryption format is so broken that qemu decided to drop it completely. This series forbids the use of such images even with qemus prior to this and removes all the cruft necessary to support it. v2: - fixed check to include the qcow format too - reworded the error message slightly - split second patch into two with proper justification for the user-alias test since checking LUKS there actually makes sense Peter Krempa (5): tests: qemuxml2argv: Drop disk encryption from 'interface-server' test tests: qemuxml2argv: Verify that disk secret alias is correct with user-aliases tests: qemublock: Switch to qcow2+luks in test files qemu: domain: Forbid storage with old QCOW2 encryption qemu: Remove code for setting up disk passphrases src/qemu/qemu_domain.c | 10 ++ src/qemu/qemu_monitor.c | 13 --- src/qemu/qemu_monitor.h | 4 - src/qemu/qemu_monitor_json.c | 28 ------ src/qemu/qemu_monitor_json.h | 4 - src/qemu/qemu_process.c | 103 --------------------- .../file-qcow2-backing-chain-encryption.json | 2 +- .../file-qcow2-backing-chain-encryption.xml | 2 +- ...etwork-qcow2-backing-chain-encryption_auth.json | 2 +- ...network-qcow2-backing-chain-encryption_auth.xml | 2 +- tests/qemumonitorjsontest.c | 2 - tests/qemuxml2argvdata/encrypted-disk-usage.args | 8 +- tests/qemuxml2argvdata/encrypted-disk-usage.xml | 2 +- tests/qemuxml2argvdata/encrypted-disk.args | 8 +- tests/qemuxml2argvdata/encrypted-disk.xml | 2 +- tests/qemuxml2argvdata/interface-server.xml | 3 - tests/qemuxml2argvdata/user-aliases.args | 8 +- tests/qemuxml2argvdata/user-aliases.xml | 2 +- tests/qemuxml2argvtest.c | 7 +- tests/qemuxml2xmloutdata/encrypted-disk.xml | 2 +- tests/qemuxml2xmloutdata/interface-server.xml | 3 - tests/qemuxml2xmltest.c | 6 +- 22 files changed, 46 insertions(+), 177 deletions(-) -- 2.16.2

3 14

[libvirt] [PATCH 00/38] qemu: Refactor secret/TLS setup and add TLS for nbd
by Peter Krempa 31 May '18

31 May '18

This series consists of the following changes: 1) refactors to handling of the TLS object and secret alias 2) storage of 'secret' object aliases in the status XML 3) fix of disk-unplug with TLS after libvirtd restart 4) Adding support for TLS for NBD disks (originally used to reproduce problem with detach of TLS object after libvirtd restart) Few parts could be considered separate but since the end of the series builds up on the various pieces it would be very unpleasant to post separately. Peter Krempa (38): qemu: domain: Add helper to check if encrypted secrets can be used with a VM qemu: domain: Reuse code when preparing hostdev auth secrets qemu: domain: Rename qemuDomainSecretDiskCapable qemu: domain: Rename and fix docs for qemuDomainSecretInfoNew qemu: domain: Add new function to set up encrypted secrets only qemu: domain: Setup disk encryption password secret via new helper qemu: domain: Use qemuDomainSecretInfoNewPlain only for unencrypted secrets qemu: domain: Add helpers for partially clearing qemuDomainSecretInfoPtr qemu: domain: Don't delete aliases of secret objects associated with disks qemu: Store and parse disk authentication and encryption secret alias tests: qemustatusxml2xml: Add test data for re-generating LUKS/auth aliases qemu: domain: Regenerate auth/enc secret aliases when restoring status XML qemu: hotplug: Don't try to infer secret object alias/presence qemu: hotplug: Use 'tlsAlias' to see whether to detach the disk qemu: domain: Store and restore TLS object alias of a disk qemu: domain: Regenerate alias for the TLS x509 credential object qemu: domain: Properly setup data relevant for top disk image qemu: domain: don't loop through images in qemuDomainPrepareDiskSourceChain qemu: domain: Split validation and setup of the virStorageSource qemu: domain: aggregate setup of disk drive options for -drive qemu: domain: Separate setup of TLS for VXHS disks from qemuDomainPrepareDiskSourceTLS qemu: domain: Use switch statement in qemuDomainPrepareDiskSourceTLS qemu: domain: Process only one object in qemuDomainPrepareDiskSourceTLS qemu: domain: Forbid TLS setup for disk protocols not supporting it conf: Don't encode matrix of storage protocols supporting TLS in the parser qemu: hotplug: Don't mandate passing of 'secAlias' in qemuDomainGetTLSObjects qemu: hotplug: Allow passing in NULL 'tlsAlias' to qemuDomainGetTLSObjects qemu: domain: Set up disk TLS alias when preparing TLS setup qemu: command: Don't generate alias for TLS private key password secret qemu: command: Pass in alias for TLS object to qemuBuildTLSx509CommandLine qemu: command: Always setup TLS environment if src->haveTLS is on qemu: migration: Don't pass around secAlias qemu: hotplug: Pass around existing secret object alias from qemuDomainAddChardevTLSObjects qemu: hotplug: Remove misleading comment in qemuDomainGetTLSObjects qemu: hotplug: Drop 'secAlias' output parameter from qemuDomainGetTLSObjects qemu: hotplug: Remove TLS alias generation from qemuDomainGetTLSObjects tests: qemu: Rename disk-drive-network-tlsx509-vxhs test qemu: domain: Add support for TLS for NBD with default TLS env docs/schemas/domaincommon.rng | 5 + src/conf/domain_conf.c | 14 +- src/qemu/qemu_command.c | 75 ++- src/qemu/qemu_domain.c | 658 +++++++++++++++------ src/qemu/qemu_domain.h | 20 +- src/qemu/qemu_hotplug.c | 95 ++- src/qemu/qemu_hotplug.h | 5 +- src/qemu/qemu_migration.c | 8 +- src/qemu/qemu_migration_params.c | 23 +- src/qemu/qemu_migration_params.h | 1 - tests/qemublocktest.c | 9 +- .../disk-secinfo-upgrade-in.xml | 517 ++++++++++++++++ .../disk-secinfo-upgrade-out.xml | 538 +++++++++++++++++ tests/qemustatusxml2xmldata/modern-in.xml | 5 + ...9-vxhs.args => disk-drive-network-tlsx509.args} | 9 +- ...509-vxhs.xml => disk-drive-network-tlsx509.xml} | 8 + tests/qemuxml2argvtest.c | 4 +- ...509-vxhs.xml => disk-drive-network-tlsx509.xml} | 8 + tests/qemuxml2xmltest.c | 3 +- 19 files changed, 1663 insertions(+), 342 deletions(-) create mode 100644 tests/qemustatusxml2xmldata/disk-secinfo-upgrade-in.xml create mode 100644 tests/qemustatusxml2xmldata/disk-secinfo-upgrade-out.xml rename tests/qemuxml2argvdata/{disk-drive-network-tlsx509-vxhs.args => disk-drive-network-tlsx509.args} (82%) rename tests/qemuxml2argvdata/{disk-drive-network-tlsx509-vxhs.xml => disk-drive-network-tlsx509.xml} (85%) rename tests/qemuxml2xmloutdata/{disk-drive-network-tlsx509-vxhs.xml => disk-drive-network-tlsx509.xml} (86%) -- 2.16.2

3 89

[libvirt] [PATCH] audit: Enforce enum switch type cast in virDomainAuditHostdev
by Erik Skultety 31 May '18

31 May '18

There was a missing enum for mdev causing a strange 'unknown device type' warning when hot-plugging mdev. Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=1583927 Signed-off-by: Erik Skultety <eskultet(a)redhat.com> --- src/conf/domain_audit.c | 13 +++++++++++-- 1 file changed, 11 insertions(+), 2 deletions(-) diff --git a/src/conf/domain_audit.c b/src/conf/domain_audit.c index 82868bca76..14138d93af 100644 --- a/src/conf/domain_audit.c +++ b/src/conf/domain_audit.c @@ -361,6 +361,7 @@ virDomainAuditHostdev(virDomainObjPtr vm, virDomainHostdevDefPtr hostdev, virDomainHostdevSubsysPCIPtr pcisrc = &hostdev->source.subsys.u.pci; virDomainHostdevSubsysSCSIPtr scsisrc = &hostdev->source.subsys.u.scsi; virDomainHostdevSubsysSCSIVHostPtr hostsrc = &hostdev->source.subsys.u.scsi_host; + virDomainHostdevSubsysMediatedDevPtr mdevsrc = &hostdev->source.subsys.u.mdev; virUUIDFormat(vm->def->uuid, uuidstr); if (!(vmname = virAuditEncode("vm", vm->def->name))) { @@ -373,9 +374,9 @@ virDomainAuditHostdev(virDomainObjPtr vm, virDomainHostdevDefPtr hostdev, virt = "?"; } - switch (hostdev->mode) { + switch ((virDomainHostdevMode) hostdev->mode) { case VIR_DOMAIN_HOSTDEV_MODE_SUBSYS: - switch (hostdev->source.subsys.type) { + switch ((virDomainHostdevSubsysType) hostdev->source.subsys.type) { case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_PCI: if (virAsprintfQuiet(&address, "%.4x:%.2x:%.2x.%.1x", pcisrc->addr.domain, @@ -419,6 +420,13 @@ virDomainAuditHostdev(virDomainObjPtr vm, virDomainHostdevDefPtr hostdev, goto cleanup; } break; + case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_MDEV: + if (VIR_STRDUP_QUIET(address, mdevsrc->uuidstr) < 0) { + VIR_WARN("OOM while enconding audit message"); + goto cleanup; + } + break; + case VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_LAST: default: VIR_WARN("Unexpected hostdev type while encoding audit message: %d", hostdev->source.subsys.type); @@ -470,6 +478,7 @@ virDomainAuditHostdev(virDomainObjPtr vm, virDomainHostdevDefPtr hostdev, } break; + case VIR_DOMAIN_HOSTDEV_MODE_LAST: default: VIR_WARN("Unexpected hostdev mode while encoding audit message: %d", hostdev->mode); -- 2.14.3

3 5

[libvirt] [PATCH 1/1] conf: fixup USB input bus check
by Xiao Feng Ren 31 May '18

31 May '18

This patch fixes the USB input bus check, the bug was introduced by commit 317badb Signed-off-by: Xiao Feng Ren <renxiaof(a)linux.vnet.ibm.com> --- src/conf/domain_conf.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/conf/domain_conf.c b/src/conf/domain_conf.c index 27e2bd50eb..e4d39c06e8 100644 --- a/src/conf/domain_conf.c +++ b/src/conf/domain_conf.c @@ -27878,7 +27878,7 @@ virDomainDeviceIsUSB(virDomainDeviceDefPtr dev) if ((t == VIR_DOMAIN_DEVICE_DISK && dev->data.disk->bus == VIR_DOMAIN_DISK_BUS_USB) || (t == VIR_DOMAIN_DEVICE_INPUT && - dev->data.input->type == VIR_DOMAIN_INPUT_BUS_USB) || + dev->data.input->bus == VIR_DOMAIN_INPUT_BUS_USB) || (t == VIR_DOMAIN_DEVICE_HOSTDEV && dev->data.hostdev->mode == VIR_DOMAIN_HOSTDEV_MODE_SUBSYS && dev->data.hostdev->source.subsys.type == -- 2.16.3

2 1

[libvirt] [PATCH 0/3] Fix more memory leaks
by Michal Privoznik 31 May '18

31 May '18

Ideally, these would be merged before the release. But I don't have strong opinion on that. Michal Privoznik (3): virDomainDefParseXML: Free @tmp when parsing genid qemuxml2argvtest: Don't initialize qemuCaps twice virQEMUCapsSetHostModel: Free cpuData before setting it src/conf/domain_conf.c | 1 + src/qemu/qemu_capabilities.c | 4 ++++ tests/qemuxml2argvtest.c | 2 -- 3 files changed, 5 insertions(+), 2 deletions(-) -- 2.16.1

3 13

[libvirt] [PATCH 0/1] bug: fixup USB input bus check
by Xiao Feng Ren 31 May '18

31 May '18

There's one explicit bug(introduced in commit:317badb) when checking the bus of one input device is USB. Let's fix it. Xiao Feng Ren (1): Fixup USB input bus check src/conf/domain_conf.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) -- 2.16.3

2 3

[libvirt] What is the strategy to update the CPU Models in src/cpu/cpu_map.xml based on?
by Dou Liyang 31 May '18

31 May '18

Hi All, I am not sure about the update strategy of CPU models in libvirt. IMO, It's depend on the CPU model in qemu-kvm, if some CPU models were updated in qemu-kvm. Then, we should modify the src/cpu/cpu_map.xml of libvirt to synchronize? eg: commit cad8054ece28("cpu: Add cpu definition for Intel Sandy Bridge cpu type") in libvirt upstream. https://bugzilla.redhat.com/show_bug.cgi?id=761005 If it's right, I found this commit b3a4f0b1a072("target-i386: add VME to all CPUs") updated the VME feature in qemu upstream. But in the src/cpu /cpu_map.xml of libvirt, It didn't be updated. eg: For the SandyBridge CPU, - In QEmu, it has the 'vme' feature defined by CPUID_VME - In libvirt, it doesn't has the 'vme' in cpu_map.xml file. Why? Thanks, dou

3 9

Re: [libvirt] What is the strategy to update the CPU Models in src/cpu/cpu_map.xml based on?
by Peter Krempa 31 May '18

31 May '18

[re-adding libvir-list] On Wed, May 30, 2018 at 19:36:10 +0800, Dou Liyang wrote: > Hi Peter, > > Sorry to trouble you offline. Please keep conversations on the list, so that it does not get lost in private inboxes. > Recently, I forced on the PCID feature in CPU and > found the SandyBridge didn't include the PCID feature > in libvirt. > > commit cad8054ece285d1712880cd108e8a39f833f7e88 > Author: Peter Krempa <pkrempa(a)redhat.com> > Date: Wed Feb 8 12:43:07 2012 +0100 > > cpu: Add cpu definition for Intel Sandy Bridge cpu type > > I knew the SandyBridge CPU models in QEMU also > didn't include the PCID feature. > > I want to know: > > - If we can't make sure PCID was supported in all SandyBridge? > - If we avoided PCID definition intentionally ? > - If we forgot to add it, and need to be fixed? Qemu added it later than libvirt added the sandy-bridge cpu: commit 434acb817b8ae747f31e91ec152f9f47ac514433 Author: Mao, Junjie <junjie.mao(a)intel.com> Date: Fri Jul 20 07:08:21 2012 +0000 Recognize PCID feature This patch makes Qemu recognize the PCID feature specified from configuration or command line options. Jiri Denemark already covered why we can't/will not change existing models in cpu_map.xml > > Thanks, > dou

3 3