November 2015 - Devel - libvirt List Archives

[libvirt] [PATCH] Use correct pci addresses during device-detach
by Nitesh_Konkar 20 Nov '15

20 Nov '15

From: Nitesh_Konkar <niteshkonkar(a)in.ibm.com> The attach-device on live and persistent copies can be done independently. Thus devices can end up having different pci addresses in live and persistent copies. The detach device should try to detach the device from their respective addresses instead of using the same from live/persistent. Signed-off-by:nitkon12@linux.vnet.ibm.com --- src/driver-nodedev.h | 1 + src/qemu/qemu_driver.c | 25 ++++++++++--------------- 2 files changed, 11 insertions(+), 15 deletions(-) diff --git a/src/driver-nodedev.h b/src/driver-nodedev.h index e846612..dea79bd 100644 --- a/src/driver-nodedev.h +++ b/src/driver-nodedev.h @@ -59,6 +59,7 @@ typedef char * typedef char * (*virDrvNodeDeviceGetParent)(virNodeDevicePtr dev); + typedef int (*virDrvNodeDeviceNumOfCaps)(virNodeDevicePtr dev); diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c index f133b45..6fd58c2 100644 --- a/src/qemu/qemu_driver.c +++ b/src/qemu/qemu_driver.c @@ -8708,23 +8708,12 @@ static int qemuDomainDetachDeviceFlags(virDomainPtr dom, const char *xml, !(flags & VIR_DOMAIN_AFFECT_LIVE)) parse_flags |= VIR_DOMAIN_DEF_PARSE_INACTIVE; - dev = dev_copy = virDomainDeviceDefParse(xml, vm->def, + dev_copy = virDomainDeviceDefParse(xml, vm->def, caps, driver->xmlopt, parse_flags); - if (dev == NULL) + if (dev_copy == NULL) goto endjob; - if (flags & VIR_DOMAIN_AFFECT_CONFIG && - flags & VIR_DOMAIN_AFFECT_LIVE) { - /* If we are affecting both CONFIG and LIVE - * create a deep copy of device as adding - * to CONFIG takes one instance. - */ - dev_copy = virDomainDeviceDefCopy(dev, vm->def, caps, driver->xmlopt); - if (!dev_copy) - goto endjob; - } - if (priv->qemuCaps) qemuCaps = virObjectRef(priv->qemuCaps); else if (!(qemuCaps = virQEMUCapsCacheLookup(driver->qemuCapsCache, vm->def->emulator))) @@ -8736,6 +8725,13 @@ static int qemuDomainDetachDeviceFlags(virDomainPtr dom, const char *xml, if (!vmdef) goto endjob; + dev = virDomainDeviceDefParse(xml, vmdef, + caps, driver->xmlopt, + parse_flags); + if (!dev) + goto endjob; + + if (virDomainDefCompatibleDevice(vmdef, dev, VIR_DOMAIN_DEVICE_ACTION_DETACH) < 0) goto endjob; @@ -8777,8 +8773,7 @@ static int qemuDomainDetachDeviceFlags(virDomainPtr dom, const char *xml, cleanup: virObjectUnref(qemuCaps); virDomainDefFree(vmdef); - if (dev != dev_copy) - virDomainDeviceDefFree(dev_copy); + virDomainDeviceDefFree(dev_copy); virDomainDeviceDefFree(dev); virDomainObjEndAPI(&vm); virObjectUnref(caps); -- 2.4.0

2 1

[libvirt] [PATCH 0/3] security: misc cleanups
by Ján Tomko 20 Nov '15

20 Nov '15

Ján Tomko (3): security_dac: check if virSecurityDACGetIds returns negative security: remove extra security from function names security_selinux: fix indentation src/security/security_dac.c | 242 ++++++++++++++++++++-------------------- src/security/security_selinux.c | 205 +++++++++++++++++----------------- 2 files changed, 223 insertions(+), 224 deletions(-) -- 2.4.6

2 6

[libvirt] [PATCH 0/2] Fix coverity build and resolve new issue
by John Ferlan 20 Nov '15

20 Nov '15

The first patch is a Coverity build breaker... The second one popped up recently and while some may consider it noise or a false positive, it is the only call that didn't check status. Not sure why it "popped up" after the recent changes, but that does happen. John Ferlan (2): qemu: Fix build error in Coverity environment vbox: Resolve Coverity CHECKED_RETURN src/qemu/qemu_command.h | 2 +- src/vbox/vbox_common.c | 5 +++-- 2 files changed, 4 insertions(+), 3 deletions(-) -- 2.5.0

3 8

[libvirt] [PATCH 0/6] Fixes for memory locking limit
by Andrea Bolognani 20 Nov '15

20 Nov '15

The code dealing with RLIMIT_MEMLOCK contained a few assumptions that hold for x86 but don't necessarily work as well for other platforms, eg. that qemu will need to lock memory only when VFIO passthrough is involved. This series removes such assumptions by removing ad-hoc code and making sure that the function containing the correct logic is called instead; it also implements the platform-specific calculations needed to calculate the correct RLIMIT_MEMLOCK value for ppc64 guests. Patches 1-4 are architecture-agnostic, patches 5-6 are specific to ppc64. Cheers. Andrea Bolognani (6): process: Log when limiting the amount of locked memory qemu: Use qemuDomainRequiresMlock() in qemuBuildCommandLine() qemu: Use qemuDomainRequiresMlock() when attaching PCI hostdev qemu: Reduce memlock limit after detaching hostdev qemu: Always set locked memory limit for ppc64 domains qemu: Add ppc64-specific math to qemuDomainGetMlockLimitBytes() src/qemu/qemu_command.c | 9 ++--- src/qemu/qemu_domain.c | 89 +++++++++++++++++++++++++++++++++++++++++++++++-- src/qemu/qemu_hotplug.c | 29 ++++++++++++---- src/util/virprocess.c | 4 +++ 4 files changed, 116 insertions(+), 15 deletions(-) -- 2.5.0

3 18

[libvirt] [PATCHv4 0/6] Hyper-v crash feature support
by Dmitry Andreev 19 Nov '15

19 Nov '15

A new Hyper-V cpu feature 'hv_crash' was added to QEMU. The feature will become available in v2.5.0. What is changed in v4: * panic model attribute is added even if it wasn't specified by the user [5/6], tests updated. * configuration with multiple panic defices of same model is rejected [6/6] 1-4 commits have no changes. The previous version: https://www.redhat.com/archives/libvir-list/2015-November/msg00457.html Dmitry Andreev (6): conf: add 'model' attribute for panic device with values isa, pseries, hyperv qemu: add support for hv_crash feature as a panic device schema: allow multiple panic devices tests: add tests for panic models and multiple panic devices qemu: set panic device model if not specified by the user conf: reject multiple panic devices of same model docs/formatdomain.html.in | 19 ++- docs/schemas/domaincommon.rng | 13 ++- src/conf/domain_conf.c | 128 +++++++++++++-------- src/conf/domain_conf.h | 15 ++- src/qemu/qemu_command.c | 74 ++++++++++-- src/qemu/qemu_domain.c | 30 ++++- tests/qemuargv2xmltest.c | 1 + .../qemuxml2argv-hyperv-panic.args | 21 ++++ .../qemuxml2argvdata/qemuxml2argv-hyperv-panic.xml | 25 ++++ .../qemuxml2argv-panic-double.args | 21 ++++ .../qemuxml2argvdata/qemuxml2argv-panic-double.xml | 28 +++++ tests/qemuxml2argvdata/qemuxml2argv-panic-isa.xml | 31 +++++ .../qemuxml2argv-panic-no-address.xml | 2 +- .../qemuxml2argv-panic-pseries.xml | 30 +++++ .../qemuxml2argvdata/qemuxml2argv-pseries-disk.xml | 2 +- .../qemuxml2argv-pseries-nvram.xml | 2 +- tests/qemuxml2argvtest.c | 3 + tests/qemuxml2xmloutdata/qemuxml2xmlout-panic.xml | 31 +++++ .../qemuxml2xmlout-pseries-panic-missing.xml | 2 +- .../qemuxml2xmlout-pseries-panic-no-address.xml | 30 +++++ tests/qemuxml2xmltest.c | 8 +- 21 files changed, 446 insertions(+), 70 deletions(-) create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-hyperv-panic.args create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-hyperv-panic.xml create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-panic-double.args create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-panic-double.xml create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-panic-isa.xml create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-panic-pseries.xml create mode 100644 tests/qemuxml2xmloutdata/qemuxml2xmlout-panic.xml create mode 100644 tests/qemuxml2xmloutdata/qemuxml2xmlout-pseries-panic-no-address.xml -- 1.8.3.1

2 12

[libvirt] [PATCH 0/2] libxl: a few minor fixes
by Jim Fehlig 19 Nov '15

19 Nov '15

A few minor fixes for patches 1 and 2 of Joao's series adding stats support to the libxl driver https://www.redhat.com/archives/libvir-list/2015-November/msg00489.html I only noticed the minor flaws while reviewing patches later in the series, after already committing patches 1 and 2. Jim Fehlig (2): libxl: unref libxlDriverConfig object libxl: don't unlock virDomainObj if refcnt is 0 src/libxl/libxl_driver.c | 26 +++++++++++++++++--------- 1 file changed, 17 insertions(+), 9 deletions(-) -- 2.1.4

3 7

[libvirt] qemuParseCommandLine and virDomainDefPostParse (and virDomaniDefAddImplicitControllers)
by Laine Stump 19 Nov '15

19 Nov '15

While playing with the idea of forcing explicit USB controller models, I ended up with the qemuargv2xml test failing; it ended up that this was because changes I had made to qemuDomainDefPostParse() were affecting the XML produced by qemuParseCommandline(). The reason - after constructing a virDomainDef object by parsing a qemu commandline, qemuParseCommandline() calls two functions that are supposed to be called after parsing domain XML - virDomainDefPostParse() (which calls qemuDomainDefPostParse()). In my opinion, qemuParseCommandLine() shouldn't be calling virDomainDefPostParse() (or virDomainefAddImplicitControllers(), which it calls in the wrong order relative to virDomainDefPostParse() BTW). The reasons are: 1) this is causing the argv-to-xml conversion to include things in the XML that were not on the original commandline, in particular "default" devices like PCI and USB controllers (added in qemuDomainDefPostParse() based on machinetype) as well as disk, smartcard, virtio-serial, and hostdev-scsi controllers in virDomainDefAddImplicitControllers() (why the duality there, anyway?) 2) If the output of argv-to-xml is used for a virDomainDefine, those post-parse functions will be called then, and the implicit/auto devices will be added at that time anyway, so in practice nothing is gained by adding them during argv-to-xml. Does anyone else have an opinion about this?

3 5

[libvirt] [PATCH v2] vz: implementation of domainReboot callback
by Mikhail Feoktistov 19 Nov '15

19 Nov '15

Diff from v1. 1. Add virCheckFlags() call in vzDomainReboot --- src/vz/vz_driver.c | 8 ++++++++ src/vz/vz_sdk.c | 8 ++++++++ src/vz/vz_sdk.h | 1 + 3 files changed, 17 insertions(+) diff --git a/src/vz/vz_driver.c b/src/vz/vz_driver.c index 0a968b9..39f58a4 100644 --- a/src/vz/vz_driver.c +++ b/src/vz/vz_driver.c @@ -923,6 +923,13 @@ static int vzDomainShutdown(virDomainPtr domain) return prlsdkDomainChangeState(domain, prlsdkStop); } +static int vzDomainReboot(virDomainPtr domain, + unsigned int flags) +{ + virCheckFlags(0, -1); + return prlsdkDomainChangeState(domain, prlsdkRestart); +} + static int vzDomainIsActive(virDomainPtr domain) { virDomainObjPtr dom = NULL; @@ -1486,6 +1493,7 @@ static virHypervisorDriver vzDriver = { .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ + .domainReboot = vzDomainReboot, /* 1.2.22 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ .domainUndefine = vzDomainUndefine, /* 1.2.10 */ diff --git a/src/vz/vz_sdk.c b/src/vz/vz_sdk.c index 89c9e89..bf71e96 100644 --- a/src/vz/vz_sdk.c +++ b/src/vz/vz_sdk.c @@ -1831,6 +1831,14 @@ PRL_RESULT prlsdkSuspend(PRL_HANDLE sdkdom) return waitJob(job); } +PRL_RESULT prlsdkRestart(PRL_HANDLE sdkdom) +{ + PRL_HANDLE job = PRL_INVALID_HANDLE; + + job = PrlVm_Restart(sdkdom); + return waitJob(job); +} + int prlsdkDomainChangeStateLocked(vzConnPtr privconn, virDomainObjPtr dom, diff --git a/src/vz/vz_sdk.h b/src/vz/vz_sdk.h index ebe4591..88ee7d9 100644 --- a/src/vz/vz_sdk.h +++ b/src/vz/vz_sdk.h @@ -41,6 +41,7 @@ PRL_RESULT prlsdkStop(PRL_HANDLE sdkdom); PRL_RESULT prlsdkPause(PRL_HANDLE sdkdom); PRL_RESULT prlsdkResume(PRL_HANDLE sdkdom); PRL_RESULT prlsdkSuspend(PRL_HANDLE sdkdom); +PRL_RESULT prlsdkRestart(PRL_HANDLE sdkdom); typedef PRL_RESULT (*prlsdkChangeStateFunc)(PRL_HANDLE sdkdom); int -- 1.8.3.1

3 2

[libvirt] [PATCH] vz: allow only en-us keymap for VNC
by Mikhail Feoktistov 19 Nov '15

19 Nov '15

In virtuozzo we support only en-us keymap for VMs and containers. If keymap is specified than check that it's en-us, otherwise show error message. --- src/vz/vz_sdk.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) diff --git a/src/vz/vz_sdk.c b/src/vz/vz_sdk.c index 89c9e89..526c0f2 100644 --- a/src/vz/vz_sdk.c +++ b/src/vz/vz_sdk.c @@ -2220,10 +2220,11 @@ static int prlsdkCheckGraphicsUnsupportedParams(virDomainDefPtr def) return -1; } - if (gr->data.vnc.keymap != 0) { + if (gr->data.vnc.keymap != 0 && + STRNEQ(gr->data.vnc.keymap, "en-us")) { virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s", - _("vz driver doesn't support " - "keymap setting for VNC graphics.")); + _("vz driver supports only " + "\"en-us\" keymap for VNC graphics.")); return -1; } -- 1.8.3.1

2 1

[libvirt] [PATCH] set UID and GID according to dynamicOwnership
by Matteo Cafasso 19 Nov '15

19 Nov '15

When called with dynamicOwnership set, the qemuOpenFileAs function ignores it resulting in files not created with proper permissions. The issue is reported in: https://www.redhat.com/archives/libvirt-users/2015-November/msg00039.html and also mentioned in: https://www.redhat.com/archives/libvir-list/2015-November/msg00453.html Signed-off-by: Matteo Cafasso <noxdafox(a)gmail.com> --- src/qemu/qemu_driver.c | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c index 92a9961..b3b59b6 100644 --- a/src/qemu/qemu_driver.c +++ b/src/qemu/qemu_driver.c @@ -2932,6 +2932,11 @@ qemuOpenFileAs(uid_t fallback_uid, gid_t fallback_gid, if (path_shared <= 0 || dynamicOwnership) vfoflags |= VIR_FILE_OPEN_FORCE_OWNER; + if (dynamicOwnership) { + uid = fallback_uid; + gid = fallback_gid; + } + if (stat(path, &sb) == 0) { /* It already exists, we don't want to delete it on error */ need_unlink = false; -- 2.6.2

2 1