- Devel - libvirt List Archives

[libvirt] [PATCH v2 00/14] Couple of vhost-user fixes and cleanups
by Michal Privoznik 14 Oct '16

14 Oct '16

v2 of: https://www.redhat.com/archives/libvir-list/2016-August/msg00806.html The first patches are mostly code movement and cleanups. diff to v1: - Hopefully all John's review suggestions are worked in now Michal Privoznik (14): virDomainNetDefParseXML: Realign virDomainNetGetActualType: Return type is virDomainNetType qemuBuildInterfaceCommandLine: Move hostdev handling a bit further qemuBuildInterfaceCommandLine: Move vhostuser handling a bit further qemuBuildInterfaceCommandLine: Move from if-else forest to switch qemuDomainAttachNetDevice: Move hostdev handling a bit further qemuDomainAttachNetDevice: Explicitly list allowed types for hotplug qemuBuildHostNetStr: Explicitly enumerate net types qemuBuildChrChardevStr: Introduce @nowait argument qemuBuildVhostuserCommandLine: Reuse qemuBuildChrChardevStr qemuBuildInterfaceCommandLine: Pass proper args qemuBuildVhostuserCommandLine: Unify -netdev creation qemuBuildHostNetStr: Support VIR_DOMAIN_NET_TYPE_VHOSTUSER qemu_hotplug: Support interface type of vhost-user hotplug src/bhyve/bhyve_command.c | 2 +- src/bhyve/bhyve_process.c | 2 +- src/conf/domain_conf.c | 16 +- src/conf/domain_conf.h | 2 +- src/libxl/libxl_domain.c | 2 +- src/libxl/libxl_driver.c | 2 +- src/lxc/lxc_driver.c | 9 +- src/qemu/qemu_command.c | 182 +++++++++++++-------- src/qemu/qemu_hotplug.c | 130 +++++++++++---- src/qemu/qemu_process.c | 13 +- .../qemuxml2argv-net-hostdev-fail.xml | 39 +++++ .../qemuxml2argv-net-vhostuser-fail.xml | 36 ++++ .../qemuxml2argv-net-vhostuser-multiq.args | 6 +- .../qemuxml2argv-net-vhostuser.args | 4 +- tests/qemuxml2argvtest.c | 7 + 15 files changed, 334 insertions(+), 118 deletions(-) create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-net-hostdev-fail.xml create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-net-vhostuser-fail.xml -- 2.8.4

2 31

[libvirt] [PATCH 0/4] vbox: address thread-safety issues.
by Dawid Zamirski 14 Oct '16

14 Oct '16

This patch series solves (at least in my testing) vbox driver thread-safety issues that were also outlined on libvirt-users ML [1] and I was affected with. Those patches try to follow the suggestions made Matthias' [2] in that thread as close as possible. Here's where my patch differs from that suggesions: * vboxGlobalData - still needs to keep reference to ISession and IVirutalBox session because it's apparently not possible to have multiple instances created/destroyed safely with pfnComInitialize and pfnComUninitalize calls on per-connection basis. * as such vboxPrivate (the new struct introduced here) also has references to ISession and IVirutalBox (which are just referenes to the ones from the global) mainly to immitate ISession instance per-connection. Apparently newer VBOX SDKs introduced pfnClinetInitialize that can allegedly create multiple ISessions and we might want to take advantage of that in the future hopefully without making additional changes allover the driver code that this patch did. The gist of the change is in patch 3 and it also contains more in-depth explanation on how the issue is being resolved. Also, please note that patch 4 should be squashed into 3 as it was kept separate only for code-review purposes and 3rd patch won't compile without 4 applied on top. [1] https://www.redhat.com/archives/libvirt-users/2012-April/msg00122.html [2] https://www.redhat.com/archives/libvirt-users/2012-April/msg00125.htmlgq Dawid Zamirski (4): vbox: add vboxPrivate struct. vbox: replace vboxGlobalData with vboxPrivate. vbox: change API (un)initialization logic. vbox: update rest of the code to for prior changes. src/vbox/vbox_common.c | 275 ++++++++++++--------------- src/vbox/vbox_common.h | 32 ++-- src/vbox/vbox_network.c | 51 +++-- src/vbox/vbox_storage.c | 20 +- src/vbox/vbox_tmpl.c | 433 +++++++++++++++++++++--------------------- src/vbox/vbox_uniformed_api.h | 128 +++++++------ 6 files changed, 460 insertions(+), 479 deletions(-) -- 2.7.4

3 19

[libvirt] [PATCH] virLogDefineOutputs: Fix build without syslog.h
by Michal Privoznik 14 Oct '16

14 Oct '16

Not every system out there has syslog, that's why we check for it in our configure script. However, in 640b58abdf while fixing another issue, some variables and functions are called that are defined only when syslog.h is present. But these function calls/variables were not guarded by #ifdef-s. Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com> --- Pushed under build breaker rule. src/util/virlog.c | 15 +++++++++------ 1 file changed, 9 insertions(+), 6 deletions(-) diff --git a/src/util/virlog.c b/src/util/virlog.c index 52b0eea..8f831fc 100644 --- a/src/util/virlog.c +++ b/src/util/virlog.c @@ -1362,9 +1362,10 @@ virLogFindOutput(virLogOutputPtr *outputs, size_t noutputs, int virLogDefineOutputs(virLogOutputPtr *outputs, size_t noutputs) { - int ret = -1; +#if HAVE_SYSLOG_H int id; char *tmp = NULL; +#endif /* HAVE_SYSLOG_H */ if (virLogInitialize() < 0) return -1; @@ -1372,6 +1373,7 @@ virLogDefineOutputs(virLogOutputPtr *outputs, size_t noutputs) virLogLock(); virLogResetOutputs(); +#if HAVE_SYSLOG_H /* syslog needs to be special-cased, since it keeps the fd in private */ if ((id = virLogFindOutput(outputs, noutputs, VIR_LOG_TO_SYSLOG, current_ident)) != -1) { @@ -1379,20 +1381,21 @@ virLogDefineOutputs(virLogOutputPtr *outputs, size_t noutputs) * holding the lock so it's safe to call openlog and change the message * tag */ - if (VIR_STRDUP_QUIET(tmp, outputs[id]->name) < 0) - goto cleanup; + if (VIR_STRDUP_QUIET(tmp, outputs[id]->name) < 0) { + virLogUnlock(); + return -1; + } VIR_FREE(current_ident); current_ident = tmp; openlog(current_ident, 0, 0); } +#endif /* HAVE_SYSLOG_H */ virLogOutputs = outputs; virLogNbOutputs = noutputs; - ret = 0; - cleanup: virLogUnlock(); - return ret; + return 0; } -- 2.8.4

2 2

[libvirt] [PATCH 0/4] qemu: escape smbios strings passed on commandline
by Peter Krempa 14 Oct '16

14 Oct '16

See patch 3 for the core change. Peter Krempa (4): qemu: command: Fix up coding style of smbios commandine formatters qemu: command: Don't bother reporting errors in smbios formatters qemu: command: escape smbios entry strings schema: smbios: allow any strings docs/schemas/domaincommon.rng | 4 +- src/qemu/qemu_command.c | 129 ++++++++++++++++++++++++------------------ 2 files changed, 76 insertions(+), 57 deletions(-) -- 2.10.0

2 10

[libvirt] [PATCH 0/4] Couple of test driver fixes and improvements
by Michal Privoznik 13 Oct '16

13 Oct '16

BLURB Michal Privoznik (4): testOpenDefault: Rename loop variable testNodeGetCellsFreeMemory: Fix off by one error test driver: Store memory per NUMA node test_driver: Implement virNodeAllocPages src/test/test_driver.c | 198 ++++++++++++++++++++++++++++++++++++++++--------- 1 file changed, 164 insertions(+), 34 deletions(-) -- 2.8.4

2 12

[libvirt] libvirt API / balloon last-update
by Marko Myllynen 13 Oct '16

13 Oct '16

Hi, I see that recently the last-update field has been added to the balloon statistics, this is how it looks like from Python: [(<libvirt.virDomain object at 0x7fb45b7faa50>, {'balloon.rss': 960048L, 'balloon.swap_in': 0L, 'balloon.usable': 1676424L, 'balloon.unused': 1376188L, 'balloon.major_fault': 816L, 'balloon.swap_out': 0L, 'balloon.current': 2097152L, 'balloon.maximum': 2097152L, 'balloon.available': 2048364L, 'balloon.minor_fault': 460160L, 'balloon.last-update': 1476213030L})] As you can see, there's an inconsistency as all the other members use underscore (_), not dash (-). This actually matters with the PCP plugin since PCP metric names can contain _ but not -. Now that last-update is part of the API in 2.3 I guess it's too late to change this but it would nice if in the future the member names could be kept consistent and underscore used for new members, as it has been so far. Thanks, -- Marko Myllynen

2 1

[libvirt] Cpu Modeling
by Jason J. Herne 13 Oct '16

13 Oct '16

Hi Jiri & Eduardo, You might remember a discussion with David Hildenbrand of IBM on the Qemu mailing list regarding a new Qemu<->Libvirt interface for cpu modeling. I am picking up this work from David and I wanted to confirm that we are still on the same page as to the direction of that interface. For your reference: https://www.redhat.com/archives/libvir-list/2016-June/thread.html#01413 https://lists.gnu.org/archive/html/qemu-devel/2016-09/threads.html#00489 The first link is to the discussion you were directly involved in. The second link is to the final patch set posted to qemu-devel. The cover letter gives a good overview of the interface added to Qemu and the proposed use-case for Libvirt to use this new cpu modeling support. I'll paste in the most relevant section for your convenience: --------------------------------Libvirt usecase---------------------------- Testing for runability: - Simply try to start QEMU with KVM, compat machine, CPU model - Could be done using query-cpu-model-comparison in the future. Identifying host model, e.g. "virsh capabilities" - query-cpu-model-expansion on "host" with "-M none --enable-kvm" <cpu mode='host-model'>: - simply copy the identified host model <cpu mode='host-passthrough'>: - "-cpu host" "virsh cpu-baseline": - query-cpu-model-baseline on two models with "-M none" "virsh cpu-compare": - query-cpu-model-comparison on two models with "-M none" There might be some cenarios where libvirt wants to convert another CPU model to a static variant, this can be done using query-cpu-model-expansion --------------------------------------------------------------------------- Now that I've hopefully refreshed your memory :) I just want to make sure that you are still on-board with this plan. I realize that this new approach does things differently than Libvirt does today for other platforms. Especially x86_64. The big differences are as follows: 1. We will invoke qemu to gather the host cpu data used for virsh capabilities. Today this data seems to be collected directly from the host hardware for most (all?) architectures. 2. virsh cpu-models {arch} will also use a Qemu invocation to gather cpu model data. 3. Most architectures seem to use a "map" (xml file containing cpu model data that ships with Libvirt) to satisfy #1 and #2 above. Our new method does not use this map as it gets all of the data directly from Qemu. 4. virsh cpu-baseline and cpu-compare will now invoke qemu directly as well. Note: I'm not sure exactly how much all of this will apply just to s390 with other architectures moving over to the new interface if/when they want to. Or if we will want to change all architectures to this new interface at the same time. Any guidance? Thanks for your time and consideration. -- -- Jason J. Herne (jjherne(a)linux.vnet.ibm.com)

4 13

[libvirt] [libvirt-perl][PATCH] Add VIR_DOMAIN_VCPU_HOTPLUGGABLE constant
by Michal Privoznik 13 Oct '16

13 Oct '16

Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com> --- Changes | 2 +- Virt.xs | 1 + lib/Sys/Virt/Domain.pm | 4 ++++ 3 files changed, 6 insertions(+), 1 deletion(-) diff --git a/Changes b/Changes index 4985100..5831939 100644 --- a/Changes +++ b/Changes @@ -2,7 +2,7 @@ Revision history for perl module Sys::Virt 2.4.0 2016-00-00 - - XXX + - Add VIR_DOMAIN_VCPU_HOTPLUGGABLE constant 2.3.0 2016-10-06 diff --git a/Virt.xs b/Virt.xs index b4ca049..c1d923c 100644 --- a/Virt.xs +++ b/Virt.xs @@ -8309,6 +8309,7 @@ BOOT: REGISTER_CONSTANT(VIR_DOMAIN_VCPU_CONFIG, VCPU_CONFIG); REGISTER_CONSTANT(VIR_DOMAIN_VCPU_MAXIMUM, VCPU_MAXIMUM); REGISTER_CONSTANT(VIR_DOMAIN_VCPU_GUEST, VCPU_GUEST); + REGISTER_CONSTANT(VIR_DOMAIN_VCPU_HOTPLUGGABLE, VCPU_HOTPLUGGABLE); REGISTER_CONSTANT(VIR_DOMAIN_SHUTDOWN_DEFAULT, SHUTDOWN_DEFAULT); diff --git a/lib/Sys/Virt/Domain.pm b/lib/Sys/Virt/Domain.pm index e37b078..f45fc9e 100644 --- a/lib/Sys/Virt/Domain.pm +++ b/lib/Sys/Virt/Domain.pm @@ -2768,6 +2768,10 @@ Flag to request adjustment of the maximum vCPU value Flag to request the guest VCPU mask +=item Sys::Virt::Domain::VCPU_HOTPLUGGABLE + +Flag to make vcpus added hot(un)pluggable + =back =head2 STATE CHANGE EVENTS -- 2.8.4

2 1

[libvirt] [PATCH V3 0/4] qemu: report actual vcpu state in virVcpuInfo
by Viktor Mihajlovski 13 Oct '16

13 Oct '16

Currently, the virVcpuInfo returned by virDomainGetVcpus() will always report a state of VIR_VCPU_RUNNING for each defined domain vcpu even if the vcpu is currently in the halted state. As the monitor interface is in fact reporting the accurate state, it is rather easy to transport this information with the existing API. This is done by - adding a new state of VIR_VCPU_HALTED - extending the monitor to pass back the halted state for the vcpus - adding a new field to the private domain vcpu object reflecting the halted state for the vcpu - modifying the driver code to report the vcpu state based on the halted indicator - extending virsh vcpuinfo to also display the halted state The vcpu state is however not recorded in the internal XML format, since the state can change asynchronously (without notification). V2 is a rebase on top of Peter Krempa's CPU hotplug modernization. V3 tries to address the review comments by John and Peter. I noticed Peter has posted (we had a tendency for mid-air collisions lately) https://www.redhat.com/archives/libvir-list/2016-October/msg00498.html I've applied the patches on top of mine and have seen no issues so far. Viktor Mihajlovski (4): domain: Add new VCPU state "halted" qemu: Add monitor support for CPU halted state qemu: Add domain support for VCPU halted state qemu: Ensure reported VCPU state is current in driver API include/libvirt/libvirt-domain.h | 1 + src/qemu/qemu_domain.c | 66 ++++++++++++++++++++++++++++++++++++++++ src/qemu/qemu_domain.h | 5 +++ src/qemu/qemu_driver.c | 15 +++++++-- src/qemu/qemu_monitor.c | 6 +++- src/qemu/qemu_monitor.h | 3 ++ src/qemu/qemu_monitor_json.c | 3 ++ src/qemu/qemu_monitor_text.c | 8 ++++- tests/qemumonitorjsontest.c | 8 ++--- tools/virsh-domain.c | 3 +- tools/virsh.pod | 57 ++++++++++++++++++++++++++++++++++ 11 files changed, 166 insertions(+), 9 deletions(-) -- 1.9.1

3 8

[libvirt] [PATCH] virlog: Enclose part of virLogDefineOutputs in HAVE_SYSLOG_H block
by Erik Skultety 13 Oct '16

13 Oct '16

Commit 640b58ab broke the mingw build because it referenced 'current_ident' as well as 'openlog' symbols both of which are declared conditionally within HAVE_SYSLOG_H directive. This patch fixes the broken build. Additional changes like moving all variable declarations in virLogDefineOutputs into the conditional block were necessary to avoid 'unused variable' errors from mingw. Signed-off-by: Erik Skultety <eskultet(a)redhat.com> --- I could've pushed it under build breaker rule but I'd rather be safe than sorry that I missed anything else. Erik src/util/virlog.c | 28 ++++++++++++++-------------- 1 file changed, 14 insertions(+), 14 deletions(-) diff --git a/src/util/virlog.c b/src/util/virlog.c index 14ee701..f52d9d8 100644 --- a/src/util/virlog.c +++ b/src/util/virlog.c @@ -1362,16 +1362,15 @@ virLogFindOutput(virLogOutputPtr *outputs, size_t noutputs, int virLogDefineOutputs(virLogOutputPtr *outputs, size_t noutputs) { - int ret = -1; - int id; + if (virLogInitialize() < 0) + return -1; + + virLogLock(); + virLogResetOutputs(); + +#if HAVE_SYSLOG_H + int id = -1; char *tmp = NULL; - - if (virLogInitialize() < 0) - return -1; - - virLogLock(); - virLogResetOutputs(); - /* syslog needs to be special-cased, since it keeps the fd in private */ if ((id = virLogFindOutput(outputs, noutputs, VIR_LOG_TO_SYSLOG, current_ident)) != -1) { @@ -1379,20 +1378,21 @@ virLogDefineOutputs(virLogOutputPtr *outputs, size_t noutputs) * holding the lock so it's safe to call openlog and change the message * tag */ - if (VIR_STRDUP_QUIET(tmp, outputs[id]->name) < 0) - goto cleanup; + if (VIR_STRDUP_QUIET(tmp, outputs[id]->name) < 0) { + virLogUnlock(); + return -1; + } VIR_FREE(current_ident); current_ident = tmp; openlog(current_ident, 0, 0); } +#endif virLogOutputs = outputs; virLogNbOutputs = noutputs; - ret = 0; - cleanup: virLogUnlock(); - return ret; + return 0; } -- 2.5.5

2 1