November 2018 - Devel - libvirt List Archives

[libvirt] [PATCH] openvswitch: Add new port VLAN mode "dot1q-tunnel"
by luzhipeng＠uniudc.com 28 Nov '18

28 Nov '18

From: ZhiPeng Lu <luzhipeng(a)uniudc.com> Signed-off-by: ZhiPeng Lu <luzhipeng(a)uniudc.com> --- docs/formatnetwork.html.in | 13 +++++++------ docs/schemas/networkcommon.rng | 1 + src/conf/netdev_vlan_conf.c | 2 +- src/util/virnetdevopenvswitch.c | 5 +++++ src/util/virnetdevvlan.h | 1 + 5 files changed, 15 insertions(+), 7 deletions(-) diff --git a/docs/formatnetwork.html.in b/docs/formatnetwork.html.in index 363a72b..de223e6 100644 --- a/docs/formatnetwork.html.in +++ b/docs/formatnetwork.html.in @@ -688,16 +688,17 @@ For network connections using Open vSwitch it is also possible - to configure 'native-tagged' and 'native-untagged' VLAN modes + to configure 'native-tagged' and 'native-untagged' and + 'dot1q-tunnel' VLAN modes Since 1.1.0. This is done with the optional <code>nativeMode</code> attribute on the <code><tag></code> subelement: <code>nativeMode</code> - may be set to 'tagged' or 'untagged'. The <code>id</code> - attribute of the <code><tag></code> subelement - containing <code>nativeMode</code> sets which VLAN is considered - to be the "native" VLAN for this interface, and + may be set to 'tagged' or 'untagged' or 'dot1q-tunnel'. + The <code>id</code> attribute of the <code><tag></code> + subelement containing <code>nativeMode</code> sets which VLAN is + considered to be the "native" VLAN for this interface, and the <code>nativeMode</code> attribute determines whether or not - traffic for that VLAN will be tagged. + traffic for that VLAN will be tagged or QinQ. <code><vlan></code> elements can also be specified in diff --git a/docs/schemas/networkcommon.rng b/docs/schemas/networkcommon.rng index 2699555..11c48ff 100644 --- a/docs/schemas/networkcommon.rng +++ b/docs/schemas/networkcommon.rng @@ -223,6 +223,7 @@ <choice> <value>tagged</value> <value>untagged</value> + <value>dot1q-tunnel</value> </choice> </attribute> </optional> diff --git a/src/conf/netdev_vlan_conf.c b/src/conf/netdev_vlan_conf.c index dff49c6..79710d9 100644 --- a/src/conf/netdev_vlan_conf.c +++ b/src/conf/netdev_vlan_conf.c @@ -29,7 +29,7 @@ #define VIR_FROM_THIS VIR_FROM_NONE VIR_ENUM_IMPL(virNativeVlanMode, VIR_NATIVE_VLAN_MODE_LAST, - "default", "tagged", "untagged") + "default", "tagged", "untagged", "dot1q-tunnel") int virNetDevVlanParse(xmlNodePtr node, xmlXPathContextPtr ctxt, virNetDevVlanPtr def) diff --git a/src/util/virnetdevopenvswitch.c b/src/util/virnetdevopenvswitch.c index 8fe06fd..ba5e333 100644 --- a/src/util/virnetdevopenvswitch.c +++ b/src/util/virnetdevopenvswitch.c @@ -91,6 +91,11 @@ virNetDevOpenvswitchConstructVlans(virCommandPtr cmd, virNetDevVlanPtr virtVlan) virCommandAddArg(cmd, "vlan_mode=native-untagged"); virCommandAddArgFormat(cmd, "tag=%d", virtVlan->nativeTag); break; + case VIR_NATIVE_VLAN_MODE_DOT1Q_TUNNEL: + virCommandAddArg(cmd, "vlan_mode=dot1q-tunnel"); + virCommandAddArg(cmd, "other_config:qinq-ethtype=802.1q"); + virCommandAddArgFormat(cmd, "tag=%d", virtVlan->nativeTag); + break; case VIR_NATIVE_VLAN_MODE_DEFAULT: default: break; diff --git a/src/util/virnetdevvlan.h b/src/util/virnetdevvlan.h index be85f59..0667f9d 100644 --- a/src/util/virnetdevvlan.h +++ b/src/util/virnetdevvlan.h @@ -29,6 +29,7 @@ typedef enum { VIR_NATIVE_VLAN_MODE_DEFAULT = 0, VIR_NATIVE_VLAN_MODE_TAGGED, VIR_NATIVE_VLAN_MODE_UNTAGGED, + VIR_NATIVE_VLAN_MODE_DOT1Q_TUNNEL, VIR_NATIVE_VLAN_MODE_LAST } virNativeVlanMode; -- 1.8.3.1

2 1

[libvirt] [PATCH 0/2] vsh-table: Remove trailing spaces
by Simon Kobyda 28 Nov '18

28 Nov '18

Right now, each line in last column of our table leaves trailing spaces. Let's get rid of it. Simon Kobyda (2): vsh-table: Get rid of trailing spaces tests: Remove trailing spaces in test cases tests/virshtest.c | 10 ++++---- tests/vshtabletest.c | 58 ++++++++++++++++++++++---------------------- tools/vsh-table.c | 6 +++-- 3 files changed, 38 insertions(+), 36 deletions(-) -- 2.17.1

2 3

[libvirt] [PATCH v2] qemu: Add check for whether KVM nesting is enabled
by John Ferlan 28 Nov '18

28 Nov '18

Support for nested KVM is handled via a kernel module configuration adjustment which if done after libvirtd is started and/or the last QEMU capabilities adjustment can result in the inability to start a guest and use nested KVM until the capabilities cache is invalidated. This is because without knowing, the CPU settings for a guest may not add the vmx=on to/for the guest config. Thus, let's fetch and save the setting during initialization and then when the capabilities are checked for various host related adjustments that could affect whether the capabilities cache is updated add a check whether the nested value was set for Intel, AMD, or s390 to force a refetch of the capabilities. Signed-off-by: John Ferlan <jferlan(a)redhat.com> --- v1 was part of an RFC: https://www.redhat.com/archives/libvir-list/2018-November/msg00494.html This patch alters that code slightly to add the check Marc Hartmayer requested for S390 and to use "kvm" in the API names and variables to make it clearer that it's not CapsIsNested but CapsKVMIsNested. If it's felt the new check should slide down further in virQEMUCapsIsValid then that's fine - just let me know what it should follow. src/qemu/qemu_capabilities.c | 45 ++++++++++++++++++++++++++++++++++++ src/qemu/qemu_capspriv.h | 2 ++ tests/qemucapabilitiestest.c | 3 +++ 3 files changed, 50 insertions(+) diff --git a/src/qemu/qemu_capabilities.c b/src/qemu/qemu_capabilities.c index fde27010e4..c377733fe6 100644 --- a/src/qemu/qemu_capabilities.c +++ b/src/qemu/qemu_capabilities.c @@ -40,6 +40,7 @@ #include "virnodesuspend.h" #include "virnuma.h" #include "virhostcpu.h" +#include "virkmod.h" #include "qemu_monitor.h" #include "virstring.h" #include "qemu_hostdev.h" @@ -557,6 +558,7 @@ struct _virQEMUCaps { virObject parent; bool usedQMP; + bool kvmIsNested; char *binary; time_t ctime; @@ -1528,6 +1530,7 @@ virQEMUCapsPtr virQEMUCapsNewCopy(virQEMUCapsPtr qemuCaps) return NULL; ret->usedQMP = qemuCaps->usedQMP; + ret->kvmIsNested = qemuCaps->kvmIsNested; if (VIR_STRDUP(ret->binary, qemuCaps->binary) < 0) goto error; @@ -3587,6 +3590,9 @@ virQEMUCapsLoadCache(virArch hostArch, virQEMUCapsInitHostCPUModel(qemuCaps, hostArch, VIR_DOMAIN_VIRT_KVM); virQEMUCapsInitHostCPUModel(qemuCaps, hostArch, VIR_DOMAIN_VIRT_QEMU); + qemuCaps->kvmIsNested = virXPathBoolean("count(./kvmIsNested) > 0", + ctxt) > 0; + ret = 0; cleanup: VIR_FREE(str); @@ -3806,6 +3812,9 @@ virQEMUCapsFormatCache(virQEMUCapsPtr qemuCaps) if (qemuCaps->sevCapabilities) virQEMUCapsFormatSEVInfo(qemuCaps, &buf); + if (qemuCaps->kvmIsNested) + virBufferAddLit(&buf, "<kvmIsNested/>\n"); + virBufferAdjustIndent(&buf, -2); virBufferAddLit(&buf, "</qemuCaps>\n"); @@ -3846,6 +3855,30 @@ virQEMUCapsSaveFile(void *data, } +static bool +virQEMUCapsKVMIsNested(void) +{ + VIR_AUTOFREE(char *) kConfig = NULL; + + /* Intel, AMD, and s390 related checks */ + if ((kConfig = virKModConfig()) && + (strstr(kConfig, "kvm_intel nested=1") || + strstr(kConfig, "kvm_amd nested=1") || + strstr(kConfig, "kvm nested=1"))) + return true; + return false; +} + + +void +virQEMUCapsClearKVMIsNested(virQEMUCapsPtr qemuCaps) +{ + /* For qemucapabilitiestest to avoid printing the </kvmIsNested> on + * hosts with nested set in the kernel */ + qemuCaps->kvmIsNested = false; +} + + static bool virQEMUCapsIsValid(void *data, void *privData) @@ -3854,6 +3887,7 @@ virQEMUCapsIsValid(void *data, virQEMUCapsCachePrivPtr priv = privData; bool kvmUsable; struct stat sb; + bool kvmIsNested; if (!qemuCaps->binary) return true; @@ -3886,6 +3920,15 @@ virQEMUCapsIsValid(void *data, return false; } + /* Check if someone changed the nested={0|1} value for the kernel from + * the previous time we checked. If so, then refresh the capabilities. */ + kvmIsNested = virQEMUCapsKVMIsNested(); + if (kvmIsNested != qemuCaps->kvmIsNested) { + VIR_WARN("changed kernel nested kvm value was %d", qemuCaps->kvmIsNested); + qemuCaps->kvmIsNested = kvmIsNested; + return false; + } + if (!virQEMUCapsGuestIsNative(priv->hostArch, qemuCaps->arch)) { VIR_DEBUG("Guest arch (%s) is not native to host arch (%s), " "skipping KVM-related checks", @@ -4472,6 +4515,8 @@ virQEMUCapsInitQMP(virQEMUCapsPtr qemuCaps, if (virQEMUCapsInitQMPMonitor(qemuCaps, cmd->mon) < 0) goto cleanup; + qemuCaps->kvmIsNested = virQEMUCapsKVMIsNested(); + if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_KVM)) { virQEMUCapsInitQMPCommandAbort(cmd); if ((rc = virQEMUCapsInitQMPCommandRun(cmd, true)) != 0) { diff --git a/src/qemu/qemu_capspriv.h b/src/qemu/qemu_capspriv.h index 8d1a40fe74..edfe2cd6f6 100644 --- a/src/qemu/qemu_capspriv.h +++ b/src/qemu/qemu_capspriv.h @@ -48,6 +48,8 @@ int virQEMUCapsInitQMPMonitor(virQEMUCapsPtr qemuCaps, qemuMonitorPtr mon); +void virQEMUCapsClearKVMIsNested(virQEMUCapsPtr qemuCaps); + int virQEMUCapsInitQMPMonitorTCG(virQEMUCapsPtr qemuCaps, qemuMonitorPtr mon); diff --git a/tests/qemucapabilitiestest.c b/tests/qemucapabilitiestest.c index 8fe5a55e1d..90942c6fce 100644 --- a/tests/qemucapabilitiestest.c +++ b/tests/qemucapabilitiestest.c @@ -63,6 +63,9 @@ testQemuCaps(const void *opaque) qemuMonitorTestGetMonitor(mon)) < 0) goto cleanup; + /* Don't apply what the host has... force clear for testing purposes */ + virQEMUCapsClearKVMIsNested(capsActual); + if (virQEMUCapsGet(capsActual, QEMU_CAPS_KVM)) { qemuMonitorResetCommandID(qemuMonitorTestGetMonitor(mon)); if (virQEMUCapsInitQMPMonitorTCG(capsActual, -- 2.17.2

4 7

[libvirt] [PATCH] qemu: Fix post-copy migration on the source
by Jiri Denemark 28 Nov '18

28 Nov '18

Post-copy migration has been broken on the source since commit v3.8.0-245-g32c29f10db which implemented support for pause-before-switchover QEMU migration capability. Even though the migration itself went well, the source did not really know when it switched to the post-copy mode despite the messages logged by MIGRATION event handler. As a result of this, the events emitted by source libvirtd were not accurate and statistics of the completed migration would cover only the pre-copy part of migration. Moreover, if migration failed during the post-copy phase for some reason, the source libvirtd would just happily resume the domain, which could lead to disk corruption. With the pause-before-switchover capability enabled, the order of events emitted by QEMU changed: pause-before-switchover disabled enabled MIGRATION, potcopy-active STOP STOP MIGRATION, pre-switchover MIGRATION, postcopy-active The STOP even handler checks the migration status (postcopy-active) and sets the domain state accordingly. Which is sufficient when pause-before-switchover is disabled, but once we enable it, the migration status is still active when we get STOP from QEMU. Thus the domain state set in the STOP handler has to be corrected once we are notified that migration changed to postcopy-active. This results in two SUSPENDED events to be emitted by the source libvirtd during post-copy migration. The first one with VIR_DOMAIN_EVENT_SUSPENDED_MIGRATED detail, while the second one reports the corrected VIR_DOMAIN_EVENT_SUSPENDED_POSTCOPY detail. This is inevitable because we don't know whether migration will eventually switch to post-copy at the time we emit the first event. https://bugzilla.redhat.com/show_bug.cgi?id=1647365 Signed-off-by: Jiri Denemark <jdenemar(a)redhat.com> --- src/qemu/qemu_process.c | 26 +++++++++++++++++++++++++- 1 file changed, 25 insertions(+), 1 deletion(-) diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c index 622341b8a4..2c89978996 100644 --- a/src/qemu/qemu_process.c +++ b/src/qemu/qemu_process.c @@ -1543,9 +1543,13 @@ static int qemuProcessHandleMigrationStatus(qemuMonitorPtr mon ATTRIBUTE_UNUSED, virDomainObjPtr vm, int status, - void *opaque ATTRIBUTE_UNUSED) + void *opaque) { qemuDomainObjPrivatePtr priv; + virQEMUDriverPtr driver = opaque; + virObjectEventPtr event = NULL; + virQEMUDriverConfigPtr cfg = virQEMUDriverGetConfig(driver); + int reason; virObjectLock(vm); @@ -1562,8 +1566,28 @@ qemuProcessHandleMigrationStatus(qemuMonitorPtr mon ATTRIBUTE_UNUSED, priv->job.current->stats.mig.status = status; virDomainObjBroadcast(vm); + if (status == QEMU_MONITOR_MIGRATION_STATUS_POSTCOPY && + virDomainObjGetState(vm, &reason) == VIR_DOMAIN_PAUSED && + reason == VIR_DOMAIN_PAUSED_MIGRATION) { + VIR_DEBUG("Correcting paused state reason for domain %s to %s", + vm->def->name, + virDomainPausedReasonTypeToString(VIR_DOMAIN_PAUSED_POSTCOPY)); + + virDomainObjSetState(vm, VIR_DOMAIN_PAUSED, VIR_DOMAIN_PAUSED_POSTCOPY); + event = virDomainEventLifecycleNewFromObj(vm, + VIR_DOMAIN_EVENT_SUSPENDED, + VIR_DOMAIN_EVENT_SUSPENDED_POSTCOPY); + + if (virDomainSaveStatus(driver->xmlopt, cfg->stateDir, vm, driver->caps) < 0) { + VIR_WARN("Unable to save status on vm %s after state change", + vm->def->name); + } + } + cleanup: virObjectUnlock(vm); + virObjectEventStateQueue(driver->domainEventState, event); + virObjectUnref(cfg); return 0; } -- 2.19.1

2 1

[libvirt] [PATCH 00/18] Implement original label remembering
by Michal Privoznik 28 Nov '18

28 Nov '18

Dear list, there were several attempts in the past to implement this feature, but none of them was successful. The problem is that we change security labels when starting a domain but never record the original labels therefore when restoring the labels back in domain shutdown phase we have to go with root:root or restorecon. This is not user friendly. Now that we have metadata locking implemented we have exclusive access to the files we are touching and therefore can call functions to record the original owner. Since this database needs to be distributed (consider multiple daemons and an network file system) it can't be stored inside a daemon (libvirtd knows nothing about other daemons running on distant hosts). Therefore the next option is to store it with the files themselves - in XATTRs. There is one caveat though. A file can be passed to multiple domains at the same time (for instance an installation ISO), therefore we need a reference counter so that the only the last label restore call actually restores the original owner. A picture is worth more than a thousand words: # chown 5:6 /var/lib/libvirt/images/fd.img # ls -ln /var/lib/libvirt/images/fd.img -rw-r--r-- 1 5 6 2097152 Mar 17 2018 /var/lib/libvirt/images/fd.img # getfattr -d -m - /var/lib/libvirt/images/fd.img (no output) # virsh domblklist fedora Target Source ------------------------------------------------ sda /var/lib/libvirt/images/fedora.qcow2 sdb /var/lib/libvirt/images/fd.img # virsh domblklist gentoo Target Source ---------------------------------------------------------------------- fda /var/lib/libvirt/images/fd.img sda /var/lib/libvirt/images/gentoo.qcow2 # virsh start fedora Domain fedora started # getfattr -d -m - /var/lib/libvirt/images/fd.img trusted.libvirt.security.dac="+5:+6" trusted.libvirt.security.ref_dac="1" # virsh start gentoo Domain gentoo started # getfattr -d -m - /var/lib/libvirt/images/fd.img trusted.libvirt.security.dac="+5:+6" trusted.libvirt.security.ref_dac="2" # virsh shutdown --domain fedora Domain fedora is being shutdown # ls -ln /var/lib/libvirt/images/fd.img -rw-r--r-- 1 0 0 2097152 Mar 17 2018 /var/lib/libvirt/images/fd.img # getfattr -d -m - /var/lib/libvirt/images/fd.img trusted.libvirt.security.dac="+5:+6" trusted.libvirt.security.ref_dac="1" # virsh shutdown --domain gentoo Domain gentoo is being shutdown # getfattr -d -m - /var/lib/libvirt/images/fd.img (no output) # ls -ln /var/lib/libvirt/images/fd.img -rw-r--r-- 1 5 6 2097152 Mar 17 2018 /var/lib/libvirt/images/fd.img Even though I'm showing DAC only in my example, it's the same story with SELinux. Of course, this plays nicely with filesystems that don't support XATTRs, which there are not that much, but unfortunately NFS is one of them :( Michal Prívozník (18): security: Unify header conditionals util: Introduce xattr getter/setter/remover security: Include security_util security_dac: Restore label on failed chown() attempt virSecurityDACTransactionRun: Implement rollback virSecurityDACRestoreAllLabel: Reorder device relabeling virSecurityDACRestoreAllLabel: Restore more labels security_dac: Allow callers to enable/disable label remembering/recall security_dac: Remember old labels virSecurityDACRestoreImageLabelInt: Restore even shared/RO disks security_selinux: Track if transaction is restore security_selinux: Remember old labels security_selinux: Restore label on failed setfilecon() attempt virSecuritySELinuxTransactionRun: Implement rollback virSecuritySELinuxRestoreAllLabel: Reorder device relabeling virSecuritySELinuxRestoreAllLabel: Restore more labels tools: Provide a script to recover fubar'ed XATTRs setup qemu.conf: Allow users to enable/disable label remembering src/libvirt_private.syms | 3 + src/qemu/libvirtd_qemu.aug | 1 + src/qemu/qemu.conf | 6 + src/qemu/qemu_conf.c | 4 + src/qemu/test_libvirtd_qemu.aug.in | 1 + src/security/Makefile.inc.am | 2 + src/security/security_apparmor.h | 6 +- src/security/security_dac.c | 212 +++++++++++++++++------- src/security/security_dac.h | 6 +- src/security/security_driver.h | 6 +- src/security/security_manager.h | 6 +- src/security/security_nop.h | 6 +- src/security/security_selinux.c | 256 +++++++++++++++++++++-------- src/security/security_selinux.h | 6 +- src/security/security_stack.h | 6 +- src/security/security_util.c | 198 ++++++++++++++++++++++ src/security/security_util.h | 32 ++++ src/util/virfile.c | 121 ++++++++++++++ src/util/virfile.h | 11 ++ tools/Makefile.am | 1 + tools/libvirt_recover_xattrs.sh | 89 ++++++++++ 21 files changed, 829 insertions(+), 150 deletions(-) create mode 100644 src/security/security_util.c create mode 100644 src/security/security_util.h create mode 100755 tools/libvirt_recover_xattrs.sh -- 2.18.1

2 22

[libvirt] [PATCH 0/2] lxc: Fix a bug related to IPv{4, 6} gateway persistent setting.
by Julio Faracco 27 Nov '18

27 Nov '18

This serie fixes a bug related to IPv{4,6} gateway settings when it is defined and used with multiple network definitions. Basically, this data is being carried on to the next network settings because the pointer is not being cleaned up/initialized properly. This serie add a new test case to cover this scenario too. It will be so important to network index implemented on LXC 3.X. Julio Faracco (2): lxc: Initializing IPv6 and IPv4 gateway to overwrite old settings. tests: Adding test case to include multiple network definitions. src/lxc/lxc_native.c | 3 ++ .../lxcconf2xml-miscnetwork-v3.config | 23 ++++++++++ .../lxcconf2xml-miscnetwork.config | 23 ++++++++++ .../lxcconf2xml-miscnetwork.xml | 45 +++++++++++++++++++ tests/lxcconf2xmltest.c | 2 + 5 files changed, 96 insertions(+) create mode 100644 tests/lxcconf2xmldata/lxcconf2xml-miscnetwork-v3.config create mode 100644 tests/lxcconf2xmldata/lxcconf2xml-miscnetwork.config create mode 100644 tests/lxcconf2xmldata/lxcconf2xml-miscnetwork.xml -- 2.19.1

2 3

[libvirt] [PATCH] qemu: Arm architectures don't have a default machine type
by Andrea Bolognani 27 Nov '18

27 Nov '18

Commit 26cfb1a3cd39 introduced a list of preferred machine types, which are meant to reflect QEMU's per-architecture historical defaults and shield libvirt users from any changes in that area. Arm architectures, however, never had a default machine type associated with them, so if no machine type is requested by the user we should just go through the usual path of picking the first one reported by the 'query-machines' QMP command instead of using integratorcp straight away. Reported-by: Kashyap Chamarthy <kchamart(a)redhat.com> Signed-off-by: Andrea Bolognani <abologna(a)redhat.com> --- src/qemu/qemu_capabilities.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/src/qemu/qemu_capabilities.c b/src/qemu/qemu_capabilities.c index fde27010e4..09980467b1 100644 --- a/src/qemu/qemu_capabilities.c +++ b/src/qemu/qemu_capabilities.c @@ -2200,10 +2200,10 @@ static const char *preferredMachines[] = NULL, /* VIR_ARCH_NONE (not a real arch :) */ "clipper", /* VIR_ARCH_ALPHA */ NULL, /* VIR_ARCH_ARMV6L (no QEMU impl) */ - "integratorcp", /* VIR_ARCH_ARMV7L */ - "integratorcp", /* VIR_ARCH_ARMV7B */ + NULL, /* VIR_ARCH_ARMV7L (no default machine type) */ + NULL, /* VIR_ARCH_ARMV7B (no default machine type) */ - "integratorcp", /* VIR_ARCH_AARCH64 */ + NULL, /* VIR_ARCH_AARCH64 (no default machine type) */ "axis-dev88", /* VIR_ARCH_CRIS */ "pc", /* VIR_ARCH_I686 */ NULL, /* VIR_ARCH_ITANIUM (doesn't exist in QEMU any more) */ -- 2.19.1

3 3

[libvirt] [PATCHv10 0/4] Introduce x86 Cache Monitoring Technology (CMT)
by Wang Huaqiang 27 Nov '18

27 Nov '18

These patches are the remaining part for the CMT enabling series, and most of the series have been merged. This series is addressing John's review comments and suggestions, which are https://www.redhat.com/archives/libvir-list/2018-November/msg00907.html https://www.redhat.com/archives/libvir-list/2018-November/msg00510.html https://www.redhat.com/archives/libvir-list/2018-November/msg00561.html Change lists: Changes in v10: -. Add tag (virResctrlMonitorType) in qemuDomainGetResctrlMonData, thus qemuDomainGetResctrlMonData could be reused for MBM. -. Using VIR_APPEND_ELEMENT to append virQEMUResctrlMonDataPtr list. -. Add qemuDomainFreeResctrlMonData. -. Add virResctrlMonitorFreeStats. -. Return a list of virResctrlMonitorStatsPtr instead of a virResctrlMonitorStats array in virResctrlMonitorGetStats. Changes in V9: -. Addressing code review comments form John. -. Refined the names for new data structure and new functions. -. Merged qemuDomainGetStatsCpuResMonitorPerTag and qemuDomainGetStatsCpuResMonitor, and refined new function name based on the fact that we only support cache monitor now. Wang Huaqiang (4): util: Return a list of pointer in virResctrlMonitorGetStats util: Add function to free monitor statistical data qemu: Report cache occupancy (CMT) with domstats docs: Updated news.xml for CMT docs/news.xml | 12 ++++ src/libvirt-domain.c | 12 ++++ src/libvirt_private.syms | 1 + src/qemu/qemu_driver.c | 183 ++++++++++++++++++++++++++++++++++++++++++++++- src/util/virresctrl.c | 26 +++++-- src/util/virresctrl.h | 8 ++- tools/virsh.pod | 14 ++++ 7 files changed, 248 insertions(+), 8 deletions(-) -- 2.7.4

3 11

[libvirt] [PATCH RFC 00/22] Move process code to qemu_process
by Chris Venteicher 27 Nov '18

27 Nov '18

Make process code usable outside qemu_capabilities by moving code from qemu_capabilities to qemu_process and exposing public functions. The process code is used to activate a non domain QEMU process for QMP message exchanges. This patch set modifies capabilities to use the new public functions. -- The process code is being decoupled from qemu_capabilities now to support hypervisor baseline and comparison using QMP commands. This patch set was originally submitted as part of the baseline patch set: [libvirt] [PATCH v4 00/37] BaselineHypervisorCPU using QEMU QMP exchanges https://www.redhat.com/archives/libvir-list/2018-November/msg00091.html The baseline and comparison requirements are described here: https://bugzilla.redhat.com/show_bug.cgi?id=1511999 https://bugzilla.redhat.com/show_bug.cgi?id=1511996 I am extracting and resubmitting just the process changes as a stand alone series to try to make review easier. The patch set shows capabilities using the public functions. To see baseline using the public functions... Look at the "qemu_driver:" patches at the end of https://www.redhat.com/archives/libvir-list/2018-November/msg00091.html Also, The "qemu_driver: Support feature expansion via QEMU when baselining cpu" patch might be of particular interest because the same QEMU process is used for both baseline and expansion using QMP commands. -- Many patches were used to isolate code moves and name changes from other actual implementation changes. The patches reuse the pattern of public qemuProcess{Start,Stop} functions and internal static qemuProcess{Init,Launch,ConnectMonitor} functions but adds a "Qmp" suffix to make them unique. A number of patches are about re-partitioning the code into static functions for initialization, process launch and connection monitor stuff. This matches the established pattern in qemu_process and seemed to make sense to do. For concurrency... A thread safe library function creates a unique directory under libDir for each QEMU process (for QMP messaging) to decouple processes in terms of sockets and file system footprint. Every patch should compile independently if applied in sequence. Chris Venteicher (22): qemu_process: Move process code from qemu_capabilities to qemu_process qemu_process: Use qemuProcess prefix qemu_process: Limit qemuProcessNew to const input strings qemu_process: Refer to proc not cmd in process code qemu_process: Use consistent name for stop process function qemu_capabilities: Stop QEMU process before freeing qemu_process: Use qemuProcess struct for a single process qemu_process: Persist stderr in qemuProcess struct qemu_capabilities: Detect caps probe failure by checking monitor ptr qemu_process: Introduce qemuProcessStartQmp qemu_process: Collect monitor code in single function qemu_process: Store libDir in qemuProcess struct qemu_process: Setup paths within qemuProcessInitQmp qemu_process: Stop retaining Qemu Monitor config in qemuProcess qemu_process: Don't open monitor if process failed qemu_process: Cleanup qemuProcess alloc function qemu_process: Cleanup qemuProcessStopQmp function qemu_process: Catch process free before process stop qemu_monitor: Make monitor callbacks optional qemu_process: Enter QMP command mode when starting QEMU Process qemu_process: Use unique directories for QMP processes qemu_process: Stop locking QMP process monitor immediately src/qemu/qemu_capabilities.c | 300 +++++------------------------ src/qemu/qemu_monitor.c | 4 +- src/qemu/qemu_process.c | 356 +++++++++++++++++++++++++++++++++++ src/qemu/qemu_process.h | 37 ++++ tests/qemucapabilitiestest.c | 7 + 5 files changed, 444 insertions(+), 260 deletions(-) -- 2.17.1

3 44

[libvirt] [BUG][PATCH][RRFC][libvirt-python] libvirtError(..., conn=, dom=. vol=, pool=, snap=)
by Philipp Hahn 27 Nov '18

27 Nov '18

Hi, while working on the Python type annotations for the Python libvirt binding I noticed the following code in libvirt-override-virDomainSnapshot.py: > def listAllChildren(self, flags=0): > """List all child snapshots and returns a list of snapshot objects""" ... > raise libvirtError("..., conn=self) "self" is an instance of virDomainSnapshot here. I found two similar cases where "conn" was not a "virConnect" instance in listAllSnapshots() and listAllVolumes(). Compare that with the declaration of libvirtError in libvirt-override.py: > class libvirtError(Exception): > def __init__(self, defmsg, conn=None, dom=None, net=None, pool=None, vol=None): Looking further at the implementation of that method only "defmsg" is used; all other references are not accessed and never stored in an instance variable. Should I add a new "snap" argument to libvirtError.__init__() or should we stop passing those references to the constructor altogether? Patch 2 and 3 might be applied to the current branch already, patch 1 currently depends on my other work. Philipp -- Philipp Hahn Open Source Software Engineer Univention GmbH be open. Mary-Somerville-Str. 1 D-28359 Bremen Tel.: +49 421 22232-0 Fax : +49 421 22232-99 hahn(a)univention.de http://www.univention.de/ Geschäftsführer: Peter H. Ganten HRB 20755 Amtsgericht Bremen Steuer-Nr.: 71-597-02876

3 4