- Devel - libvirt List Archives

[libvirt] [PATCH 0/4] qemu: Fix qemuProcessInitCpuAffinity()
by Andrea Bolognani 04 Jun '19

04 Jun '19

See detailed explanation in the commit message for patch 1/4 and the corresponding bug report. Andrea Bolognani (4): util: Introduce virBitmapUnion() fixup? util: Optimize virBitmapUnion() util: Introduce virNumaNodesetToCPUset() qemu: Fix qemuProcessInitCpuAffinity() src/libvirt_private.syms | 2 ++ src/qemu/qemu_process.c | 7 ++++- src/util/virbitmap.c | 32 +++++++++++++++++++++++ src/util/virbitmap.h | 4 +++ src/util/virnuma.c | 55 ++++++++++++++++++++++++++++++++++++++++ src/util/virnuma.h | 2 ++ tests/virbitmaptest.c | 37 +++++++++++++++++++++++++++ 7 files changed, 138 insertions(+), 1 deletion(-) -- 2.21.0

3 16

[libvirt] [PATCH] util: Propagate numad failures correctly
by Andrea Bolognani 04 Jun '19

04 Jun '19

Right now, if numad fails, we raise an error but return an empty string to the caller instead of a NULL pointer, which means processing will continue and the user will see # virsh start guest error: Failed to start domain guest error: invalid argument: Failed to parse bitmap '' instead of a more reasonable # virsh start guest error: Failed to start domain guest error: operation failed: Failed to query numad for the advisory nodeset Make sure the user gets a better error message. https://bugzilla.redhat.com/show_bug.cgi?id=1716387 Signed-off-by: Andrea Bolognani <abologna(a)redhat.com> --- src/util/virnuma.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) diff --git a/src/util/virnuma.c b/src/util/virnuma.c index dd3fb7519e..fee0dfcd5a 100644 --- a/src/util/virnuma.c +++ b/src/util/virnuma.c @@ -66,10 +66,12 @@ virNumaGetAutoPlacementAdvice(unsigned short vcpus, virCommandSetOutputBuffer(cmd, &output); - if (virCommandRun(cmd, NULL) < 0) - virReportError(VIR_ERR_INTERNAL_ERROR, "%s", + if (virCommandRun(cmd, NULL) < 0) { + virReportError(VIR_ERR_OPERATION_FAILED, "%s", _("Failed to query numad for the " "advisory nodeset")); + VIR_FREE(output); + } return output; } -- 2.21.0

3 2

[libvirt] [PATCH] news: Drop empty section
by Andrea Bolognani 03 Jun '19

03 Jun '19

Signed-off-by: Andrea Bolognani <abologna(a)redhat.com> --- Pushed as trivial. docs/news.xml | 2 -- 1 file changed, 2 deletions(-) diff --git a/docs/news.xml b/docs/news.xml index 86f5ffb3f6..f95dfdf3b4 100644 --- a/docs/news.xml +++ b/docs/news.xml @@ -56,8 +56,6 @@ </section> </release> <release version="v5.4.0" date="2019-06-03"> - <section title="New features"> - </section> <section title="Security"> <change> <summary> -- 2.21.0

1 0

[libvirt] [PATCH v2 0/7] Check TSC frequency before starting QEMU
by Jiri Denemark 03 Jun '19

03 Jun '19

Version 2: - new patches (1 and 2) fixing qemuargv2xmltest, which was probing host CPU When migrating a domain with invtsc CPU feature enabled, the TSC frequency of the destination host must match the frequency used when the domain was started on the source host or the destination host has to support TSC scaling. If the frequencies do not match and the destination host does not support TSC scaling, QEMU will fail to set the right TSC frequency when starting vCPUs on the destination and thus migration will fail. However, this is quite late since both host might have spent significant time transferring memory and perhaps even storage data. By adding the check to libvirt we can let migration fail before any data starts to be sent over. If for some reason libvirt is unable to detect the host's TSC frequency or scaling support, we'll just let QEMU try and the migration will either succeed or fail later. Luckily, we mandate TSC frequency to be explicitly set in the domain XML to even allow migration of domains with invtsc. We can just check whether the requested frequency is compatible with the current host before starting QEMU. And to let libvirt client decide whether it should even start the migration to a specific host, this series adds host's TSC frequency and scaling support to the host CPU capabilities XML. https://bugzilla.redhat.com/show_bug.cgi?id=1641702 Jiri Denemark (7): qemu: Make virQEMUCapsProbeHostCPUForEmulator more generic qemuargv2xmltest: Use mocked virQEMUCapsProbeHostCPU util: Add virHostCPUGetTscInfo conf: Report TSC frequency in host CPU capabilities cpu_x86: Fix placement of *CheckFeature functions cpu_x86: Probe TSC frequency and scaling support qemu: Check TSC frequency before starting QEMU src/conf/cpu_conf.c | 48 +++++++++++++++++++++ src/conf/cpu_conf.h | 2 + src/cpu/cpu_x86.c | 81 ++++++++++++++++++++---------------- src/qemu/qemu_capabilities.c | 14 +++---- src/qemu/qemu_capspriv.h | 5 +-- src/qemu/qemu_process.c | 53 +++++++++++++++++++++++ src/util/virhostcpu.c | 71 +++++++++++++++++++++++++++++++ src/util/virhostcpu.h | 11 +++++ tests/Makefile.am | 3 +- tests/qemuargv2xmltest.c | 3 +- tests/qemucpumock.c | 5 +-- 11 files changed, 243 insertions(+), 53 deletions(-) -- 2.21.0

2 17

[libvirt] Release of libvirt-5.4.0
by Daniel Veillard 03 Jun '19

03 Jun '19

It's out ! The release is tagged in git, and I provided signed tarball and source rpms to the usual place: https://libvirt.org/sources/ I also cut off a 5.4.0 release of the python bindings but code is same a 5.3.0 one, you can find signed tarball and source rpms at: https://libvirt.org/sources/python/ Main theme of this release is security, there is a set of advisory covered so users are invited to update, along with some improvements and bug fixes. Security: - cpu: Introduce support for the md-clear CPUID bit This bit is set when microcode provides the mechanism to invoke a flush of various exploitable CPU buffers by invoking the x86 VERW instruction. CVE-2018-12126, CVE-2018-12127, CVE-2018-12130, CVE-2019-11091. - Restrict user access to virt-admin, virtlogd and virtlockd The intended users for these facilities are the root user and the libvirtd service respectively, but these restrictions were not enforced correctly. CVE-2019-10132. Improvements: - test driver: Expand API coverage Several APIs that were missing from the test driver have now been implemented. - Avoid unnecessary static linking Most binaries shipped as part of libvirt, for example virtlogd and libvirt_iohelper, were embedding parts of the library even though they also linked against the libvirt.so dynamic library. This is no longer the case, which results in both the disk and memory footprint being reduced. - qemu: Report stat-htlb-pgalloc and stat-htlb-pgfail balloon stats These stats have been introduced in QEMU 3.0. Bug fixes: - qemu: Fix emulator scheduler support Setting the scheduler for QEMU's main thread before QEMU had a chance to start up other threads was misleading as it would affect other threads (vCPU and I/O) as well. In some particular situations this could also lead to an error when the thread for vCPU #0 was being moved to its cpu,cpuacct cgroup. This was fixed so that the scheduler for the main thread is set after QEMU starts. - apparmor: Allow hotplug of vhost-scsi devices Thanks everybody who contributed to this release, be it with bug reports, patches, reviews, docs ... Enjoy ! Daniel -- Daniel Veillard | Red Hat Developers Tools http://developer.redhat.com/ veillard(a)redhat.com | libxml Gnome XML XSLT toolkit http://xmlsoft.org/ http://veillard.com/ | virtualization library http://libvirt.org/

1 0

[libvirt] [PATCH] maint: Post-release version bump to 5.5.0
by Andrea Bolognani 03 Jun '19

03 Jun '19

Signed-off-by: Andrea Bolognani <abologna(a)redhat.com> --- Pushed as trivial. configure.ac | 2 +- docs/news.xml | 8 ++++++++ 2 files changed, 9 insertions(+), 1 deletion(-) diff --git a/configure.ac b/configure.ac index 7e01b77fc0..1f05055d6f 100644 --- a/configure.ac +++ b/configure.ac @@ -16,7 +16,7 @@ dnl You should have received a copy of the GNU Lesser General Public dnl License along with this library. If not, see dnl <http://www.gnu.org/licenses/>. -AC_INIT([libvirt], [5.4.0], [libvir-list(a)redhat.com] [], [https://libvirt.org] +AC_INIT([libvirt], [5.5.0], [libvir-list(a)redhat.com] [], [https://libvirt.org] AC_CONFIG_SRCDIR([src/libvirt.c]) AC_CONFIG_AUX_DIR([build-aux]) AC_CONFIG_HEADERS([config.h]) diff --git a/docs/news.xml b/docs/news.xml index 5d5dfb93aa..699fde7b52 100644 --- a/docs/news.xml +++ b/docs/news.xml @@ -39,6 +39,14 @@ --> <libvirt> + <release version="v5.5.0" date="unreleased"> + <section title="New features"> + </section> + <section title="Improvements"> + </section> + <section title="Bug fixes"> + </section> + </release> <release version="v5.4.0" date="2019-06-03"> <section title="New features"> </section> -- 2.21.0

1 0

[libvirt] [PATCH 0/5] Check TSC frequency before starting QEMU
by Jiri Denemark 03 Jun '19

03 Jun '19

When migrating a domain with invtsc CPU feature enabled, the TSC frequency of the destination host must match the frequency used when the domain was started on the source host or the destination host has to support TSC scaling. If the frequencies do not match and the destination host does not support TSC scaling, QEMU will fail to set the right TSC frequency when starting vCPUs on the destination and thus migration will fail. However, this is quite late since both host might have spent significant time transferring memory and perhaps even storage data. By adding the check to libvirt we can let migration fail before any data starts to be sent over. If for some reason libvirt is unable to detect the host's TSC frequency or scaling support, we'll just let QEMU try and the migration will either succeed or fail later. Luckily, we mandate TSC frequency to be explicitly set in the domain XML to even allow migration of domains with invtsc. We can just check whether the requested frequency is compatible with the current host before starting QEMU. And to let libvirt client decide whether it should even start the migration to a specific host, this series adds host's TSC frequency and scaling support to the host CPU capabilities XML. https://bugzilla.redhat.com/show_bug.cgi?id=1641702 Jiri Denemark (5): util: Add virHostCPUGetTscInfo conf: Report TSC frequency in host CPU capabilities cpu_x86: Fix placement of *CheckFeature functions cpu_x86: Probe TSC frequency and scaling support qemu: Check TSC frequency before starting QEMU src/conf/cpu_conf.c | 48 ++++++++++++++++++++++++ src/conf/cpu_conf.h | 2 + src/cpu/cpu_x86.c | 81 ++++++++++++++++++++++------------------- src/qemu/qemu_process.c | 53 +++++++++++++++++++++++++++ src/util/virhostcpu.c | 71 ++++++++++++++++++++++++++++++++++++ src/util/virhostcpu.h | 11 ++++++ 6 files changed, 229 insertions(+), 37 deletions(-) -- 2.21.0

2 7

[libvirt] [PATCH] test_driver: implement virDomainGetHostname
by Ilias Stamatis 03 Jun '19

03 Jun '19

Always return "domain_name" + "host". Signed-off-by: Ilias Stamatis <stamatis.iliass(a)gmail.com> --- src/test/test_driver.c | 22 ++++++++++++++++++++++ 1 file changed, 22 insertions(+) diff --git a/src/test/test_driver.c b/src/test/test_driver.c index 2f58a1da95..aad7bb6036 100644 --- a/src/test/test_driver.c +++ b/src/test/test_driver.c @@ -1910,6 +1910,27 @@ static int testDomainReboot(virDomainPtr domain, return ret; } +static char * +testDomainGetHostname(virDomainPtr domain, + unsigned int flags) +{ + char *ret = NULL; + virDomainObjPtr vm = NULL; + + virCheckFlags(0, NULL); + + if (!(vm = testDomObjFromDomain(domain))) + goto cleanup; + + if (virDomainObjCheckActive(vm) < 0) + goto cleanup; + + ignore_value(virAsprintf(&ret, "%shost", domain->name)); + + cleanup: + return ret; +} + static int testDomainGetInfo(virDomainPtr domain, virDomainInfoPtr info) { @@ -6950,6 +6971,7 @@ static virHypervisorDriver testHypervisorDriver = { .domainGetMaxMemory = testDomainGetMaxMemory, /* 0.1.4 */ .domainSetMaxMemory = testDomainSetMaxMemory, /* 0.1.1 */ .domainSetMemory = testDomainSetMemory, /* 0.1.4 */ + .domainGetHostname = testDomainGetHostname, /* 5.5.0 */ .domainGetInfo = testDomainGetInfo, /* 0.1.1 */ .domainGetState = testDomainGetState, /* 0.9.2 */ .domainGetTime = testDomainGetTime, /* 5.4.0 */ -- 2.21.0

2 2

[libvirt] [PATCH] virDomainSendKey: validate codeset argument
by Ilias Stamatis 03 Jun '19

03 Jun '19

This argument wasn't validated anywhere, neither in the generic implementation nor in the individual drivers. As a result a call to this function with a large enough codeset value prior to this change causes libvirtd to crash. This happens because all drivers call virKeycodeValueTranslate which uses codeset as an index to the virKeymapValues array, causing an out-of-bounds error. Signed-off-by: Ilias Stamatis <stamatis.iliass(a)gmail.com> --- src/libvirt-domain.c | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/src/libvirt-domain.c b/src/libvirt-domain.c index df7e405b3e..c09448927b 100644 --- a/src/libvirt-domain.c +++ b/src/libvirt-domain.c @@ -6837,6 +6837,13 @@ virDomainSendKey(virDomainPtr domain, virCheckNonNullArgGoto(keycodes, error); virCheckPositiveArgGoto(nkeycodes, error); + if (codeset >= VIR_KEYCODE_SET_LAST) { + virReportInvalidArg(codeset, + _("codeset must be less than %d"), + VIR_KEYCODE_SET_LAST); + goto error; + } + if (nkeycodes > VIR_DOMAIN_SEND_KEY_MAX_KEYS) { virReportInvalidArg(nkeycodes, _("nkeycodes must be <= %d"), -- 2.21.0

2 1

[libvirt] [PATCH] qemu: fix pr-helper0 remain
by Jie Wang 01 Jun '19

01 Jun '19

if libvirt receive DISCONNECTED event and set prDaemonRunning to false, and qemuDomainRemoveDiskDevice is performing in the meantime. qemuDomainRemoveDiskDevice will return directly by prDaemonRunning check, so the pr-helper0 object will remain. I think it is no need to check prDaemonRunning in qemuHotplugRemoveManagedPR. Signed-off-by: Jie Wang <wangjie88(a)huawei.com> --- src/qemu/qemu_hotplug.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/src/qemu/qemu_hotplug.c b/src/qemu/qemu_hotplug.c index 34249bd030..5e4a929738 100644 --- a/src/qemu/qemu_hotplug.c +++ b/src/qemu/qemu_hotplug.c @@ -465,8 +465,7 @@ qemuHotplugRemoveManagedPR(virQEMUDriverPtr driver, virErrorPtr orig_err; int ret = -1; - if (!priv->prDaemonRunning || - virDomainDefHasManagedPR(vm->def)) + if (virDomainDefHasManagedPR(vm->def)) return 0; virErrorPreserveLast(&orig_err); -- 2.16.2.windows.1

3 6

[libvirt] [PATCH v2] docs: Drop the external AMD SEV links
by Erik Skultety 31 May '19

31 May '19

One of the current SEV document links went dead as AMD moved the resource to another place (document store), so there's probably very little point in maintaining 3rd party links if the resources are being moved. Signed-off-by: Erik Skultety <eskultet(a)redhat.com> --- docs/formatdomaincaps.html.in | 16 +++------------- 1 file changed, 3 insertions(+), 13 deletions(-) diff --git a/docs/formatdomaincaps.html.in b/docs/formatdomaincaps.html.in index 432d7aed5e..bcd2b3354b 100644 --- a/docs/formatdomaincaps.html.in +++ b/docs/formatdomaincaps.html.in @@ -514,19 +514,9 @@ encrypted with a key unique to that VM.</p> <p> - For more details on the SEV feature and how to use it with libvirt have - a look at the following resources: - <dl> - <dd> - <a href="https://support.amd.com/TechDocs/55766_SEV-KM_API_Specification.pdf">SEV API spec</a> - </dd> - <dd> - <a href="http://amd-dev.wpengine.netdna-cdn.com/wordpress/media/2013/12/AMD_Memory_E…">SEV White Paper</a> - </dd> - <dd> - <a href="formatdomain.html#launchSecurity">SEV in domain XML</a> - </dd> - </dl> + For more details on the SEV feature, please follow resources in the + AMD developer's document store. In order to use SEV with libvirt have + a look at <a href="formatdomain.html#launchSecurity">SEV in domain XML</a> </p> <dl> -- 2.20.1

2 1

[libvirt] [PATCH] maint: Add support for .ctags.d
by Jiri Denemark 31 May '19

31 May '19

Depending on the way ctags was compiled, it may look for .ctags.d/*.ctags files rather than .ctags for reading configuration. Signed-off-by: Jiri Denemark <jdenemar(a)redhat.com> --- .ctags.d/libvirt.ctags | 1 + 1 file changed, 1 insertion(+) create mode 120000 .ctags.d/libvirt.ctags diff --git a/.ctags.d/libvirt.ctags b/.ctags.d/libvirt.ctags new file mode 120000 index 0000000000..eefd37f401 --- /dev/null +++ b/.ctags.d/libvirt.ctags @@ -0,0 +1 @@ +../.ctags \ No newline at end of file -- 2.21.0

2 1

[libvirt] [PATCH 0/2] news: Update for 5.4.0 release
by Andrea Bolognani 31 May '19

31 May '19

Hopefully I've done a decent enough job at summarizing, especially with the security stuff: if not, please let me know! Andrea Bolognani (2): news: Reformat overgrown line news: Update for 5.4.0 release docs/news.xml | 61 ++++++++++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 60 insertions(+), 1 deletion(-) -- 2.21.0

2 3

[libvirt] Entering freeze for libvirt 5.4.0
by Daniel Veillard 31 May '19

31 May '19

This is tagged in git, I have put the signed tarball and source RPM at the usual place: https://libvirt.org/sources/ I did not build the binaries rpms, first it broke because I was still on F28 c.f. Jan's patch, and second it's true to provides little value, it's doubtful people would just install my binaries as is. Will likely stick to sources as part of releases in the future. Release looks fine in my limited testing, https://ci.centos.org/view/libvirt/ is of a rare uniform green color, which is promising too ! Please give it a try, I will likely cut RC2 on Thrusday, and then if everything goes fine push the GA over the w.e. or Monday, thanks, Daniel -- Daniel Veillard | Red Hat Developers Tools http://developer.redhat.com/ veillard(a)redhat.com | libxml Gnome XML XSLT toolkit http://xmlsoft.org/ http://veillard.com/ | virtualization library http://libvirt.org/

1 1

[libvirt] Question about save and restore OVS per-port data in live migration
by wangxin (U) 31 May '19

31 May '19

Hi When we try to migrate a guest with 12 NICs, we find libvirt will try to get per-port data in finish phase, it makes the guest paused for several hundred milliseconds. qemuMigrationCookieAddNetwork -> virNetDevOpenvswitchGetMigrateData for per-nic -> run 'ovs-vsctl --if-exists get Interface vnetx external_ids:Port' (about 25ms) In our case, there's no portData available and it seems there is no need to run the cmd. Questions: 1. Is it necessary to get/set per-port data in migration finish phase? It will increase the downtime. 2. Is there any way (i.e. a switch) to not run the OVS cmd if there's no portData available?

2 4

[libvirt] [PATCH] docs: Replace the SEV links with a single one to AMD's document store
by Erik Skultety 31 May '19

31 May '19

The SEV whitepaper link stopped working as the resource has been moved. But all the SEV-related stuff like spec, whitepaper, presentations, etc. are being provided at https://developer.amd.com/sev/ which may look like a more stable resource (in terms of not being removed in the foreseeable future) so rather than providing and maintaining links to individual resources, let's use this one instead. Signed-off-by: Erik Skultety <eskultet(a)redhat.com> --- docs/formatdomaincaps.html.in | 5 +---- 1 file changed, 1 insertion(+), 4 deletions(-) diff --git a/docs/formatdomaincaps.html.in b/docs/formatdomaincaps.html.in index 432d7aed5e..ef42bd57c3 100644 --- a/docs/formatdomaincaps.html.in +++ b/docs/formatdomaincaps.html.in @@ -518,10 +518,7 @@ a look at the following resources: <dl> <dd> - <a href="https://support.amd.com/TechDocs/55766_SEV-KM_API_Specification.pdf">SEV API spec</a> - </dd> - <dd> - <a href="http://amd-dev.wpengine.netdna-cdn.com/wordpress/media/2013/12/AMD_Memory_E…">SEV White Paper</a> + <a href="https://developer.amd.com/sev/">SEV further reading</a> </dd> <dd> <a href="formatdomain.html#launchSecurity">SEV in domain XML</a> -- 2.20.1

2 2

[libvirt] [PATCH 0/2] tests: Add aarch64-gigabyte sysinfo test case
by Andrea Bolognani 30 May '19

30 May '19

This is intended to validate that using dmidecode on aarch64[1] works as expected. [1] https://www.redhat.com/archives/libvir-list/2019-May/msg00244.html Andrea Bolognani (2): tests: Tweak x86 sysinfo test case tests: Add aarch64-gigabyte sysinfo test case .../sysinfodata/aarch64-gigabytecpuinfo.data | 288 ++++++++ .../sysinfodata/aarch64-gigabytedmidecode.sh | 3 + .../sysinfodata/aarch64-gigabytesysinfo.data | 619 ++++++++++++++++++ .../aarch64-gigabytesysinfo.expect | 1 + .../{dmidecode.sh => x86dmidecode.sh} | 0 tests/sysinfotest.c | 4 +- 6 files changed, 914 insertions(+), 1 deletion(-) create mode 100644 tests/sysinfodata/aarch64-gigabytecpuinfo.data create mode 100755 tests/sysinfodata/aarch64-gigabytedmidecode.sh create mode 100644 tests/sysinfodata/aarch64-gigabytesysinfo.data create mode 100644 tests/sysinfodata/aarch64-gigabytesysinfo.expect rename tests/sysinfodata/{dmidecode.sh => x86dmidecode.sh} (100%) -- 2.21.0

2 3

[libvirt] [PATCH 0/9] util/resctrl cleanups and refactors
by Wang Huaqiang 30 May '19

30 May '19

Plan to support Resctrl Control Monitors, which is a feature introduced by kernel 'resctrl' sub-model. Submit some cleanup and refactoring patches for upcoming memory bandwidth resource monitoring (MBM) monitors. Related MBM RFC is https://www.redhat.com/archives/libvir-list/2019-April/msg01409.html. This RFC is not actively discussed since libvirt already implemented similar resctrl cache monitoring (CMT), and lots details have been discussed and implemented during the work of CMT. The cleanups and refactoring includes: 1. Removing some reluctant lines and white spaces that is existing in current code and not meet the libvirt coding style. 2. Replace 'virResctrlAllocIsEmpty' with @n==0 for indicating no resctrl allocation in configuration file. 3. Private API changes, removed 'virResctrlMonitorGetCacheOccupancy' and exported a new API named 'virResctrlMonitorGetStats' with similar functionality, but with capability to be used for retrieving MBM statistical information. 4. Refactoring 'virResctrlMonitorFreeStats' for more reusing in code. 5. Extend data structure 'virResctrlMonitorStats' with the capability to carry multiple statistical information from monitor. Wang Huaqiang (9): conf: code cleanup, remove empty line and one space conf: code cleanup for return error code directly conf: check if vcpus matches with any 'existing' allocaiton conf: Replace 'virResctrlAllocIsEmpty' with @n==0 for indicating no resctrl allocation in configuration file conf: refactor 'virDomainResctrlVcpuMatch' util: Refactor 'virResctrlMonitorFreeStats' util: Refactor 'virResctrlMonitorStats' util: Extend virresctl API to retrieve multiple monitor statistics util: Remove unused virResctrlMonitorGetCacheOccupancy src/conf/domain_conf.c | 89 +++++++++++++++++++++++------------------------- src/libvirt_private.syms | 2 +- src/qemu/qemu_driver.c | 36 +++++++++++++++----- src/util/virresctrl.c | 68 +++++++++++++++--------------------- src/util/virresctrl.h | 19 ++++++++--- 5 files changed, 113 insertions(+), 101 deletions(-) -- 2.7.4

3 25

[libvirt] [PATCH v2 0/2] virSysinfo: Try dmidecode on aarch64
by Michal Privoznik 30 May '19

30 May '19

v2 of: https://www.redhat.com/archives/libvir-list/2019-May/msg00221.html diff to v1: - Pushed 1/3 and 2/3 from v1 as they were ACKed already - New patch 1/2 which renames virSysinfoReadX86 - Patch 2/2 then uses the renamed function Michal Prívozník (2): virsysinfo: Rename virSysinfoReadX86 to virSysinfoReadDMI virSysinfoReadARM: Try reading DMI table src/libvirt_private.syms | 2 +- src/util/virsysinfo.c | 22 ++++++++++++++++++++-- src/util/virsysinfopriv.h | 2 +- tests/sysinfotest.c | 2 +- 4 files changed, 23 insertions(+), 5 deletions(-) -- 2.21.0

3 5

[libvirt] [PATCH v2] qemu: aquire job in qemuDomainDefineXMLFlags
by Nikolay Shirokovskiy 30 May '19

30 May '19

The function updates vm->newDef and frees previous pesistent definition which can be used by in different thread by some API that unlocks domain to communicate to qemu. So we have an access to freed memory in that thread. Let's acquire job condition in case of updating exiting domain. Signed-off-by: Nikolay Shirokovskiy <nshirokovskiy(a)virtuozzo.com> --- diff from v1 [1]: - pay a bit of attention to the problem Reproducer in the Michal's reply to the [1]. [1] https://www.redhat.com/archives/libvir-list/2019-May/msg00804.html src/qemu/qemu_driver.c | 37 +++++++++++++++++++++++++++++++++++-- 1 file changed, 35 insertions(+), 2 deletions(-) diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c index 42b1ce2..2dd4442 100644 --- a/src/qemu/qemu_driver.c +++ b/src/qemu/qemu_driver.c @@ -7620,6 +7620,7 @@ qemuDomainDefineXMLFlags(virConnectPtr conn, virDomainDefPtr def = NULL; virDomainDefPtr oldDef = NULL; virDomainObjPtr vm = NULL; + virDomainObjPtr exvm = NULL; virDomainPtr dom = NULL; virObjectEventPtr event = NULL; virQEMUDriverConfigPtr cfg; @@ -7647,10 +7648,34 @@ qemuDomainDefineXMLFlags(virConnectPtr conn, if (virDomainDefineXMLFlagsEnsureACL(conn, def) < 0) goto cleanup; + if ((exvm = virDomainObjListFindByUUID(driver->domains, def->uuid))) { + if (qemuDomainObjBeginJob(driver, exvm, QEMU_JOB_MODIFY) < 0) { + virDomainObjEndAPI(&exvm); + goto cleanup; + } + virObjectUnlock(exvm); + } + if (!(vm = virDomainObjListAdd(driver->domains, def, driver->xmlopt, - 0, &oldDef))) + 0, &oldDef))) { + if (exvm) + virObjectLock(exvm); goto cleanup; + } + + if (exvm) { + if (vm != exvm) { + virReportError(VIR_ERR_OPERATION_FAILED, + _("domain '%s' is removed"), def->name); + virObjectLock(exvm); + goto cleanup; + } + + /* we don't need extra reference for same domain */ + virObjectUnref(exvm); + } + def = NULL; vm->persistent = 1; @@ -7670,7 +7695,10 @@ qemuDomainDefineXMLFlags(virConnectPtr conn, /* Brand new domain. Remove it */ VIR_INFO("Deleting domain '%s'", vm->def->name); vm->persistent = 0; - qemuDomainRemoveInactiveJob(driver, vm); + if (exvm) + qemuDomainRemoveInactive(driver, vm); + else + qemuDomainRemoveInactiveJob(driver, vm); } goto cleanup; } @@ -7685,8 +7713,13 @@ qemuDomainDefineXMLFlags(virConnectPtr conn, dom = virGetDomain(conn, vm->def->name, vm->def->uuid, vm->def->id); cleanup: + if (exvm) + qemuDomainObjEndJob(driver, exvm); + virDomainDefFree(oldDef); virDomainDefFree(def); + if (exvm && exvm != vm) + virDomainObjEndAPI(&exvm); virDomainObjEndAPI(&vm); virObjectEventStateQueue(driver->domainEventState, event); virObjectUnref(caps); -- 1.8.3.1

2 1

[libvirt] [PATCH v2 0/2] introduction of version attribute for VFIO live migration
by Yan Zhao 29 May '19

29 May '19

This patchset introduces a version attribute under sysfs of VFIO Mediated devices. This version attribute is used to check whether two mdev devices are compatible. user space software can take advantage of this version attribute to determine whether to launch live migration between two mdev devices. Patch 1 defines version attribute in Documentation/vfio-mediated-device.txt Patch 2 uses GVT as an example to show how to expose version attribute and check device compatibility in vendor driver. v2: 1. renamed patched 1 2. made definition of device version string completely private to vendor driver 3. abandoned changes to sample mdev drivers 4. described intent and usage of version attribute more clearly. Yan Zhao (2): vfio/mdev: add version attribute for mdev device drm/i915/gvt: export mdev device version to sysfs for Intel vGPU Documentation/vfio-mediated-device.txt | 135 ++++++++++++++++++++++ drivers/gpu/drm/i915/gvt/Makefile | 2 +- drivers/gpu/drm/i915/gvt/device_version.c | 84 ++++++++++++++ drivers/gpu/drm/i915/gvt/gvt.c | 51 ++++++++ drivers/gpu/drm/i915/gvt/gvt.h | 6 + 5 files changed, 277 insertions(+), 1 deletion(-) create mode 100644 drivers/gpu/drm/i915/gvt/device_version.c -- 2.17.1

7 45

[libvirt] [PATCH v3 0/2] introduction of migration_version attribute for VFIO live migration
by Yan Zhao 29 May '19

29 May '19

This patchset introduces a migration_version attribute under sysfs of VFIO Mediated devices. This migration_version attribute is used to check migration compatibility between two mdev devices of the same mdev type. Patch 1 defines migration_version attribute in Documentation/vfio-mediated-device.txt Patch 2 uses GVT as an example to show how to expose migration_version attribute and check migration compatibility in vendor driver. v3: 1. renamed version to migration_version 2. let errno to be freely defined by vendor driver 3. let checking mdev_type be prerequisite of migration compatibility check 4. reworded most part of patch 1 5. print detailed error log in patch 2 and generate migration_version string at init time v2: 1. renamed patched 1 2. made definition of device version string completely private to vendor driver 3. reverted changes to sample mdev drivers 4. described intent and usage of version attribute more clearly. Yan Zhao (2): vfio/mdev: add migration_version attribute for mdev device drm/i915/gvt: export migration_version to mdev sysfs for Intel vGPU Documentation/vfio-mediated-device.txt | 113 +++++++++++++ drivers/gpu/drm/i915/gvt/Makefile | 2 +- drivers/gpu/drm/i915/gvt/gvt.c | 39 +++++ drivers/gpu/drm/i915/gvt/gvt.h | 5 + drivers/gpu/drm/i915/gvt/migration_version.c | 167 +++++++++++++++++++ drivers/gpu/drm/i915/gvt/vgpu.c | 13 +- 6 files changed, 336 insertions(+), 3 deletions(-) create mode 100644 drivers/gpu/drm/i915/gvt/migration_version.c -- 2.17.1

3 8

[libvirt] [PATCH 00/11] qemu: Introduce SMMUv3 IOMMU support
by Andrea Bolognani 28 May '19

28 May '19

This is the ARM equivalent of Intel IOMMU. Andrea Bolognani (11): qemu: Fix switch() statements for virDomainIOMMUModel qemu: Drop 'ret' from qemuBuildIOMMUCommandLine() qemu: Use VIR_AUTOCLEAN() in qemuBuildIOMMUCommandLine() qemu: Move virBuffer inside switch() statement qemu: Move capability checks inside switch() statements qemu: Introduce QEMU_CAPS_MACHINE_VIRT_IOMMU conf: Parse and format SMMUv3 IOMMU qemu: Add validation for SMMUv3 IOMMU qemu: Move capability checks for IOMMU features qemu: Format SMMUv3 IOMMU news: Update for SMMUv3 IOMMU support docs/formatdomain.html.in | 7 +- docs/news.xml | 8 + docs/schemas/domaincommon.rng | 5 +- src/conf/domain_conf.c | 1 + src/conf/domain_conf.h | 1 + src/qemu/qemu_capabilities.c | 8 + src/qemu/qemu_capabilities.h | 1 + src/qemu/qemu_command.c | 63 ++++--- src/qemu/qemu_domain.c | 73 +++++--- .../caps_2.12.0.aarch64.replies | 166 +++++++++++++++-- .../caps_3.0.0.riscv32.replies | 131 +++++++++++++- .../caps_3.0.0.riscv64.replies | 131 +++++++++++++- .../caps_4.0.0.aarch64.replies | 171 ++++++++++++++++-- .../caps_4.0.0.aarch64.xml | 1 + .../caps_4.0.0.riscv32.replies | 129 ++++++++++++- .../caps_4.0.0.riscv64.replies | 129 ++++++++++++- .../iommu-smmuv3.aarch64-latest.args | 31 ++++ tests/qemuxml2argvdata/iommu-smmuv3.xml | 15 ++ tests/qemuxml2argvtest.c | 1 + .../iommu-smmuv3.aarch64-latest.xml | 25 +++ tests/qemuxml2xmltest.c | 1 + 21 files changed, 985 insertions(+), 113 deletions(-) create mode 100644 tests/qemuxml2argvdata/iommu-smmuv3.aarch64-latest.args create mode 100644 tests/qemuxml2argvdata/iommu-smmuv3.xml create mode 100644 tests/qemuxml2xmloutdata/iommu-smmuv3.aarch64-latest.xml -- 2.21.0

2 22

[libvirt] [PATCH v3 0/6] numa: deprecate '-numa node, mem' and default memory distribution
by Igor Mammedov 28 May '19

28 May '19

Changes since v2: - taking in account previous review, implement a way for mgmt to intospect if '-numa node,mem' is supported by machine type as suggested by Daniel at https://www.mail-archive.com/qemu-devel@nongnu.org/msg601220.html * ammend "qom-list-properties" to show property values * add "numa-mem-supported" machine property to reflect if '-numa node,mem=SZ' is supported. It culd be used with '-machine none' or at runtime with --preconfig before numa memory mapping are configured * minor fixes to deprecation documentation mentioning "numa-mem-supported" property 1) "I'm considering to deprecating -mem-path/prealloc CLI options and replacing them with a single memdev Machine property to allow interested users to pick used backend for initial RAM (fixes mixed -mem-path+hostmem backends issues) and as a transition step to modeling initial RAM as a Device instead of (ab)using MemoryRegion APIs." (for more details see: https://www.mail-archive.com/qemu-devel@nongnu.org/msg596314.html) However there is a couple of roadblocks on the way (s390x and numa memory handling). I think I finally thought out a way to hack s390x in migration compatible manner, but I don't see any way to do it for -numa node,mem and default RAM assignement to nodes. Considering both numa usecases aren't meaningfully using NUMA (aside guest side testing) and could be replaced with explicitly used memdev parameter, I'd like to propose removing these fake NUMA friends on new machine types, hence this deprecation. And once the last machie type that supported the option is removed we would be able to remove option altogether. As result of removing deprecated options and replacing initial RAM allocation with 'memdev's (1), QEMU will allocate guest RAM in consistent way, fixing mixed use-case and allowing boards to move towards modelling initial RAM as Device(s). Which in its own turn should allow to cleanup NUMA/HMP/memory accounting code more by dropping ad-hoc node_mem tracking and reusing memory device enumeration instead. Reference to previous versions: * [PATCH 0/2] numa: deprecate -numa node, mem and default memory distribution https://www.mail-archive.com/qemu-devel@nongnu.org/msg600706.html * [PATCH] numa: warn if numa 'mem' option or default RAM splitting between nodes is used. https://www.mail-archive.com/qemu-devel@nongnu.org/msg602136.html * [PATCH v2] numa: warn if numa 'mem' option or default RAM splitting between nodes is used. https://www.spinics.net/linux/fedora/libvir/msg180917.html CC: libvir-list(a)redhat.com CC: ehabkost(a)redhat.com CC: pbonzini(a)redhat.com CC: berrange(a)redhat.com CC: armbru(a)redhat.com Igor Mammedov (6): pc: fix possible NULL pointer dereference in pc_machine_get_device_memory_region_size() qmp: make "qom-list-properties" show initial property values qmp: qmp_qom_list_properties(): ignore empty string options numa: introduce "numa-mem-supported" machine property numa: deprecate 'mem' parameter of '-numa node' option numa: deprecate implict memory distribution between nodes include/hw/boards.h | 1 + hw/arm/virt.c | 1 + hw/core/machine.c | 12 ++++++++++++ hw/i386/pc.c | 7 ++++++- hw/ppc/spapr.c | 1 + numa.c | 5 +++++ qapi/misc.json | 5 ++++- qemu-deprecated.texi | 24 ++++++++++++++++++++++++ qmp.c | 15 +++++++++++++++ 9 files changed, 69 insertions(+), 2 deletions(-) -- 2.7.4

3 13

[libvirt] [PATCH] qemu: aquire job in qemuDomainDefineXMLFlags
by Nikolay Shirokovskiy 28 May '19

28 May '19

The function updates vm->newDef and frees previous pesistent definition which can be used by in different thread by some API that unlocks domain to communicate to qemu. So we have an access to freed memory in that thread. Let's acquire job condition when changing persistent xml. Signed-off-by: Nikolay Shirokovskiy <nshirokovskiy(a)virtuozzo.com> --- A bug example (some downstream version of libvirt, nevertheless): (gdb) bt #0 0x00007f3bb77a0533 in virDomainDefHasVcpusOffline (def=def@entry=0x7f3b7802a830) at conf/domain_conf.c:1637 #1 0x00007f3bb77cb32f in virDomainCpuDefFormat (def=0x7f3b7802a830, buf=0x7f3ba7308940) at conf/domain_conf.c:27503 #2 virDomainDefFormatInternal (def=def@entry=0x7f3b7802a830, caps=0x7f3b70000ae0, flags=3, flags@entry=1, buf=buf@entry=0x7f3ba7308940, xmlopt=xmlopt@entry=0x0) at conf/domain_conf.c:27850 #3 0x00007f3bb77ced46 in virDomainDefFormat (def=def@entry=0x7f3b7802a830, caps=<optimized out>, flags=flags@entry=1) at conf/domain_conf.c:28564 #4 0x00007f3bb77d23f9 in virDomainSaveConfig (configDir=0x7f3b981227e0 "/etc/libvirt/qemu", caps=<optimized out>, def=0x7f3b7802a830) at conf/domain_conf.c:28776 #5 0x00007f3ba086b617 in qemuDomainSetMemoryStatsPeriod (dom=<optimized out>, period=30, flags=<optimized out>) at qemu/qemu_driver.c:4839 #6 0x00007f3bb7905613 in virDomainSetMemoryStatsPeriod (domain=domain@entry=0x7f3b74006200, period=30, flags=3) at libvirt-domain.c:2087 (gdb) f 5 #5 0x00007f3ba086b617 in qemuDomainSetMemoryStatsPeriod (dom=<optimized out>, period=30, flags=<optimized out>) at qemu/qemu_driver.c:4839 4839 ret = virDomainSaveConfig(cfg->configDir, driver->caps, persistentDef); (gdb) p vm->def $8 = (virDomainDefPtr) 0x7f3b8400c6f0 (gdb) p vm->newDef $9 = (virDomainDefPtr) 0x7f3b7001de20 (gdb) p persistentDef $10 = (virDomainDefPtr) 0x7f3b7802a830 src/qemu/qemu_driver.c | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c index 42b1ce2..683dcaa 100644 --- a/src/qemu/qemu_driver.c +++ b/src/qemu/qemu_driver.c @@ -7626,6 +7626,7 @@ qemuDomainDefineXMLFlags(virConnectPtr conn, virCapsPtr caps = NULL; unsigned int parse_flags = VIR_DOMAIN_DEF_PARSE_INACTIVE | VIR_DOMAIN_DEF_PARSE_ABI_UPDATE; + bool job = false; virCheckFlags(VIR_DOMAIN_DEFINE_VALIDATE, NULL); @@ -7647,6 +7648,10 @@ qemuDomainDefineXMLFlags(virConnectPtr conn, if (virDomainDefineXMLFlagsEnsureACL(conn, def) < 0) goto cleanup; + if (qemuDomainObjBeginJob(driver, vm, QEMU_JOB_MODIFY) < 0) + goto cleanup; + job = true; + if (!(vm = virDomainObjListAdd(driver->domains, def, driver->xmlopt, 0, &oldDef))) @@ -7685,6 +7690,9 @@ qemuDomainDefineXMLFlags(virConnectPtr conn, dom = virGetDomain(conn, vm->def->name, vm->def->uuid, vm->def->id); cleanup: + if (job) + qemuDomainObjEndJob(driver, vm); + virDomainDefFree(oldDef); virDomainDefFree(def); virDomainObjEndAPI(&vm); -- 1.8.3.1

3 3