May 2015 - Devel - libvirt List Archives

[libvirt] [PATCH] libxl: provide impl for nodeGetSecurityModel
by Jim Fehlig 18 May '15

18 May '15

Currently, the libxl driver does not support any security drivers. When the qemu driver has no security driver configued, nodeGetSecurityModel succeeds but returns an empty virSecurityModel object. Do the same in the libxl driver instead of reporting this function is not supported by the connection driver: virNodeGetSecurityModel Signed-off-by: Jim Fehlig <jfehlig(a)suse.com> --- I was reminded of this today when looking through a libvirtd log. The system was running a test script that among other things called 'virsh dominfo'. Each time dominfo was called, the log was spammed with "this function is not supported by the connection driver: virNodeGetSecurityModel". src/libxl/libxl_driver.c | 18 ++++++++++++++++++ 1 file changed, 18 insertions(+) diff --git a/src/libxl/libxl_driver.c b/src/libxl/libxl_driver.c index 60c139e..d6b20ae 100644 --- a/src/libxl/libxl_driver.c +++ b/src/libxl/libxl_driver.c @@ -5027,6 +5027,23 @@ libxlDomainMigrateConfirm3Params(virDomainPtr domain, return libxlDomainMigrationConfirm(driver, vm, flags, cancelled); } +static int libxlNodeGetSecurityModel(virConnectPtr conn, + virSecurityModelPtr secmodel) +{ + memset(secmodel, 0, sizeof(*secmodel)); + + if (virNodeGetSecurityModelEnsureACL(conn) < 0) + return -1; + + /* + * Currently the libxl driver does not support security model. + * Similar to the qemu driver, treat this as success and simply + * return no data in secmodel. Avoids spamming the libvirt log + * with "this function is not supported by the connection driver: + * virNodeGetSecurityModel" + */ + return 0; +} static virHypervisorDriver libxlHypervisorDriver = { .name = LIBXL_DRIVER_NAME, @@ -5122,6 +5139,7 @@ static virHypervisorDriver libxlHypervisorDriver = { .domainMigratePerform3Params = libxlDomainMigratePerform3Params, /* 1.2.6 */ .domainMigrateFinish3Params = libxlDomainMigrateFinish3Params, /* 1.2.6 */ .domainMigrateConfirm3Params = libxlDomainMigrateConfirm3Params, /* 1.2.6 */ + .nodeGetSecurityModel = libxlNodeGetSecurityModel, /* 1.2.16 */ }; static virConnectDriver libxlConnectDriver = { -- 2.3.7

3 2

[libvirt] [PATCH] util: make it more robust to calculate timeout value
by zhang bo 18 May '15

18 May '15

When we change system clock to years ago, a certain CPU may use up 100% cputime. The reason is that in function virEventPollCalculateTimeout(), we assign the unsigned long long result to an INT variable, *timeout = then - now; // timeout is INT, and then/now are long long if (*timeout < 0) *timeout = 0; there's a chance that variable @then minus variable @now may be a very large number that overflows INT value expression, then *timeout will be negative and be assigned to 0. Next the 'poll' in function virEventPollRunOnce() will get into an 'endless' while loop there. thus, the cpu that virEventPollRunOnce() thread runs on will go up to 100%. Although as we discussed before in https://www.redhat.com/archives/libvir-list/2015-May/msg00400.html it should be prohibited to set-time while other applications are running, but it does seems to have no harm to make the codes more robust. Signed-off-by: Wang Yufei <james.wangyufei(a)huawei.com> Signed-off-by: Zhang Bo <oscar.zhangbo(a)huawei.com> --- src/util/vireventpoll.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/src/util/vireventpoll.c b/src/util/vireventpoll.c index ffda206..5f5a149 100644 --- a/src/util/vireventpoll.c +++ b/src/util/vireventpoll.c @@ -357,9 +357,10 @@ static int virEventPollCalculateTimeout(int *timeout) return -1; EVENT_DEBUG("Schedule timeout then=%llu now=%llu", then, now); - *timeout = then - now; - if (*timeout < 0) + if (then < now) *timeout = 0; + else + *timeout = (then - now) & 0x7FFFFFFF; } else { *timeout = -1; } -- 1.7.12.4

4 5

[libvirt] C# PInvoke Library done, Nuget Package released
by scott lee 17 May '15

17 May '15

Libvirt C# Nuget package released at https://www.nuget.org/packages/Libvirt_Pinvoke/ This will add all necessary dll's and includes a full 1 to 1 mapping of the libvirt API 1.2.13 (http://libvirt.org/html/) All Structures, enums, delegates, and functions are included with the same naming and function.

1 0

[libvirt] [PATCH] netdev: assure that SRIOV PF is online before modifying VF params
by Laine Stump 16 May '15

16 May '15

The kernel won't complain if you set the mac address and vlan tag for an SRIOV VF via its PF, and it will even let you assign the PF to a guest using PCI device assignment or macvtap passthrough. But if the PF isn't online, the device won't be usable in the guest. This patch makes sure that it is turned on. Since multiple guests/VFs could use the same PF, there is no point in ever setting the PF *off*line. This resolves: https://bugzilla.redhat.com/show_bug.cgi?id=893738 Originally filed against RHEL6, but present in every version of libvirt until today. --- src/util/virnetdev.c | 11 +++++++++++ 1 file changed, 11 insertions(+) diff --git a/src/util/virnetdev.c b/src/util/virnetdev.c index e14b401..7022dfa 100644 --- a/src/util/virnetdev.c +++ b/src/util/virnetdev.c @@ -2258,6 +2258,17 @@ virNetDevReplaceVfConfig(const char *pflinkdev, int vf, char macstr[VIR_MAC_STRING_BUFLEN]; char *fileData = NULL; int ifindex = -1; + bool pfIsOnline; + + /* Assure that PF is online prior to twiddling with the VF. It + * *should* be, but if the PF isn't online the changes made to the + * VF via the PF won't take effect, yet there will be no error + * reported. + */ + if (virNetDevGetOnline(pflinkdev, &pfIsOnline) < 0) + return ret; + if (!pfIsOnline && virNetDevSetOnline(pflinkdev, true) < 0) + return ret; if (virNetDevGetVfConfig(pflinkdev, vf, &oldmac, &oldvlanid) < 0) goto cleanup; -- 2.1.0

2 4

[libvirt] [PATCH 0/4] Enable support for s390 crypto key mgmt operations
by akrowiak＠linux.vnet.ibm.com 16 May '15

16 May '15

From: Tony Krowiak <akrowiak(a)linux.vnet.ibm.com> The IBM System z Central Processor Assist for Cryptographic Functions (CPACF) hardware provides a set of CPU instructions for use in clear-key encryption, pseudo random number generation, hash functions, and protected-key encryption. The CPACF protected key cryptographic functions operate with a protected key which is encrypted under a unique wrapping key that is stored in the Hardware System Area (HSA) of the machine and can only be accessed by firmware. The wrapping key cannot be accessed by the operating system or application programs. There are two wrapping keys: One for wrapping AES keys and one for wrapping DES/TDES keys. This patch set enables the support for encrypting clear keys under the AES and DES/TDES wrapping keys for guests started on hosts running on s390 hardware that supports key wrapping. Tony Krowiak (4): libvirt: docs: XML to enable/disable protected key mgmt ops libvirt: conf: parse XML for protected key management ops libvirt: qemu: enable/disable protected key management ops libvirt: tests: test protected key mgmt ops support

3 9

[libvirt] [PATCHv2 0/9] qemu: fix device alias usage, ide controllers,
by Laine Stump 16 May '15

16 May '15

V1 was here: http://www.redhat.com/archives/libvir-list/2015-May/msg00124.html This started out with the intent to generate an error/failure on request for an IDE controller on a machinetype that doesn't support IDE (currently anything except 440fx-based machinetypes), or a 2nd IDE controller on a machinetype that only supports one (i.e. 440fx). This led to a few other related fixes, and some "fixes related to the related fixes". I've pushed some of the ACKed patches, and dropped the two patches that changed qemuBuildDeviceAddress and qemuAssignDeviceControllerAlias to use switches instead of "if .. else if .." (03/13 and 04/13) and dropped the two SCSI patches (12/13 and 13/13), but left a couple of ACKed patches in for clarity. Detailed differences from V1 in each patch. Laine Stump (9): conf: utility to return alias of a controller based on type/index qemu: fix exceptions in qemuAssignDeviceControllerAlias qemu: use controller alias when constructing device/controller args qemu: use alias to refer to non-multibus PCI controller qemu: use controller's alias in commandline for scsi-generic qemu: use USB controller's alias instead of qemuUSBId() qemu: remove test for allowing ide controller in s390, rename usb tests qemu: clean up qemuBuildCommandline loop that builds controller args qemu: log error when domain has an unsupported IDE controller src/conf/domain_conf.c | 34 +++ src/conf/domain_conf.h | 3 + src/libvirt_private.syms | 1 + src/qemu/qemu_command.c | 317 +++++++++++---------- src/qemu/qemu_command.h | 6 +- src/qemu/qemu_hotplug.c | 4 +- .../qemuxml2argv-disk-blockio.args | 2 +- .../qemuxml2argvdata/qemuxml2argv-disk-blockio.xml | 1 - .../qemuxml2argv-disk-ide-drive-split.args | 2 +- .../qemuxml2argv-disk-ide-drive-split.xml | 1 - .../qemuxml2argv-disk-sata-device.args | 4 +- .../qemuxml2argv-disk-source-pool-mode.args | 2 +- .../qemuxml2argv-disk-source-pool-mode.xml | 1 - .../qemuxml2argv-disk-source-pool.args | 2 +- .../qemuxml2argv-disk-source-pool.xml | 1 - ...uxml2argv-s390-allow-bogus-usb-controller.args} | 0 ...muxml2argv-s390-allow-bogus-usb-controller.xml} | 3 - ...=> qemuxml2argv-s390-allow-bogus-usb-none.args} | 0 ... => qemuxml2argv-s390-allow-bogus-usb-none.xml} | 0 tests/qemuxml2argvtest.c | 4 +- .../qemuxml2xmlout-disk-source-pool.xml | 1 - 21 files changed, 226 insertions(+), 163 deletions(-) rename tests/qemuxml2argvdata/{qemuxml2argv-s390-piix-controllers.args => qemuxml2argv-s390-allow-bogus-usb-controller.args} (100%) rename tests/qemuxml2argvdata/{qemuxml2argv-s390-piix-controllers.xml => qemuxml2argv-s390-allow-bogus-usb-controller.xml} (86%) rename tests/qemuxml2argvdata/{qemuxml2argv-s390-usb-none.args => qemuxml2argv-s390-allow-bogus-usb-none.args} (100%) rename tests/qemuxml2argvdata/{qemuxml2argv-s390-usb-none.xml => qemuxml2argv-s390-allow-bogus-usb-none.xml} (100%) -- 2.1.0

2 11

[libvirt] [PATCH v2 0/5] Cleanup storage migration code a bit
by Jiri Denemark 15 May '15

15 May '15

Version 2: - move QEMU-only data from virDomainDiskDef to a private data object Jiri Denemark (5): Add privateData to virDomainDiskDef Rename virDomainHasBlockjob as qemuDomainHasBlockjob Move QEMU-only fields from virDomainDiskDef into privateData qemu: Keep track of what disks are being migrated qemu: Don't give up on first error in qemuMigrationCancelDriverMirror src/conf/domain_conf.c | 43 ++++------------------- src/conf/domain_conf.h | 19 +++------- src/libvirt_private.syms | 1 - src/parallels/parallels_sdk.c | 4 +-- src/qemu/qemu_blockjob.c | 48 ++++++++++++++----------- src/qemu/qemu_command.c | 4 +-- src/qemu/qemu_domain.c | 81 ++++++++++++++++++++++++++++++++++++++++++- src/qemu/qemu_domain.h | 25 +++++++++++++ src/qemu/qemu_driver.c | 15 ++++---- src/qemu/qemu_migration.c | 42 +++++++++++----------- src/qemu/qemu_process.c | 17 +++++---- src/vbox/vbox_common.c | 4 +-- src/vbox/vbox_tmpl.c | 6 ++-- src/vmx/vmx.c | 2 +- src/xenconfig/xen_sxpr.c | 6 ++-- src/xenconfig/xen_xl.c | 2 +- src/xenconfig/xen_xm.c | 4 +-- 17 files changed, 201 insertions(+), 122 deletions(-) -- 2.4.0

2 9

[libvirt] Windows Development
by scott lee 15 May '15

15 May '15

I just finished the C# PInvoke portion for the libvirt library. https://github.com/smasherprog/Libvirt_Windows_Dll If you download the repo, and execute the powershell script, all dlls/headers are pulled, and a single C# PInvoke file is generated from the libvirt header files which contains all enums, callbacks, functions, etc. It is a complete parse of all headers libvirt-domain-snapshot.h libvirt-domain.h libvirt-event.h libvirt-host.h libvirt-interface.h libvirt-lxc.h libvirt-network.h libvirt-nodedev.h libvirt-nwfilter.h libvirt-qemu.h libvirt-secret.h libvirt-storage.h libvirt-stream.h libvirt.h virterror.h into a single .cs file that can be used for development -- thanks to the clang compiler! Now, I'll create a demo solution for this as well.

1 0

[libvirt] [PATCH] docs: add dimm address document in docs
by Luyao Huang 15 May '15

15 May '15

As we have a new device address type 'dimm', add some document and an example to our docs. Signed-off-by: Luyao Huang <lhuang(a)redhat.com> --- docs/formatdomain.html.in | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/docs/formatdomain.html.in b/docs/formatdomain.html.in index e0b6ba7..0e908a1 100644 --- a/docs/formatdomain.html.in +++ b/docs/formatdomain.html.in @@ -2876,6 +2876,13 @@ attributes: <code>iobase</code> and <code>irq</code>. <span class="since">Since 1.2.1</span> </dd> + <dt><code>type='dimm'</code></dt> + <dd>DIMM addresses, for memory device, have the following additional + attributes: <code>slot</code> (a value between 0 and 4294967295, + inclusive), and <code>base</code> (a hex value between 0 and + 0xffffffffffffffff, inclusive). + <span class="since">Since 1.2.14</span> + </dd> </dl> <h4><a name="elementsControllers">Controllers</a></h4> @@ -6038,6 +6045,7 @@ qemu-kvm -net nic,model=? /dev/null <size unit='KiB'>524287</size> <node>0</node> </target> + <address type='dimm' slot='0' base='0x0'/> </memory> <memory model='dimm'> <source> -- 1.8.3.1

3 2

[libvirt] [PATCHv2] virsh: fix no error output when parse cpulist fail
by Luyao Huang 15 May '15

15 May '15

When we pass a invalid cpulist or the lastcpu in the cpulist exceed the maxcpu, we cannot get any error. like this: # virsh vcpupin test3 1 aaa # virsh vcpupin test3 1 1000 Because virBitmapParse() use virReportError() to set the error message, vshCommandRun would output the error in vshReportError, but in the meantime it is overwriten by the virResetLastError in virDomainFree. If we want use the error which set by virReportError(), we need vshSaveLibvirtError to help us. However the error from virBitmap is not clearly enough, i chose use vshError to output error when parse failed. Signed-off-by: Luyao Huang <lhuang(a)redhat.com> --- v2: Add the check in vshParseCPUList, because this will make get last cpu more easier when the cpulist is a bitmap. tools/virsh-domain.c | 21 +++++++++++++++------ 1 file changed, 15 insertions(+), 6 deletions(-) diff --git a/tools/virsh-domain.c b/tools/virsh-domain.c index 14e1e35..20f8c75 100644 --- a/tools/virsh-domain.c +++ b/tools/virsh-domain.c @@ -6348,7 +6348,7 @@ vshPrintPinInfo(unsigned char *cpumap, size_t cpumaplen) } static unsigned char * -vshParseCPUList(int *cpumaplen, const char *cpulist, int maxcpu) +vshParseCPUList(vshControl *ctl, int *cpumaplen, const char *cpulist, int maxcpu) { unsigned char *cpumap = NULL; virBitmapPtr map = NULL; @@ -6358,8 +6358,17 @@ vshParseCPUList(int *cpumaplen, const char *cpulist, int maxcpu) return NULL; virBitmapSetAll(map); } else { - if (virBitmapParse(cpulist, '\0', &map, maxcpu) < 0) - return NULL; + if ((virBitmapParse(cpulist, '\0', &map, 1024) < 0) || + virBitmapIsAllClear(map)) { + vshError(ctl, _("Invalid cpulist '%s'"), cpulist); + goto cleanup; + } + int lastcpu = virBitmapLastSetBit(map); + if (lastcpu >= maxcpu) { + vshError(ctl, _("CPU %d in cpulist '%s' exceed the maxcpu %d"), + lastcpu, cpulist, maxcpu); + goto cleanup; + } } if (virBitmapToData(map, &cpumap, cpumaplen) < 0) @@ -6458,7 +6467,7 @@ cmdVcpuPin(vshControl *ctl, const vshCmd *cmd) } } else { /* Pin mode: pinning specified vcpu to specified physical cpus*/ - if (!(cpumap = vshParseCPUList(&cpumaplen, cpulist, maxcpu))) + if (!(cpumap = vshParseCPUList(ctl, &cpumaplen, cpulist, maxcpu))) goto cleanup; if (flags == -1) { @@ -6577,7 +6586,7 @@ cmdEmulatorPin(vshControl *ctl, const vshCmd *cmd) } /* Pin mode: pinning emulator threads to specified physical cpus*/ - if (!(cpumap = vshParseCPUList(&cpumaplen, cpulist, maxcpu))) + if (!(cpumap = vshParseCPUList(ctl, &cpumaplen, cpulist, maxcpu))) goto cleanup; if (flags == -1) @@ -6862,7 +6871,7 @@ cmdIOThreadPin(vshControl *ctl, const vshCmd *cmd) if ((maxcpu = vshNodeGetCPUCount(ctl->conn)) < 0) goto cleanup; - if (!(cpumap = vshParseCPUList(&cpumaplen, cpulist, maxcpu))) + if (!(cpumap = vshParseCPUList(ctl, &cpumaplen, cpulist, maxcpu))) goto cleanup; if (virDomainPinIOThread(dom, iothread_id, -- 1.8.3.1

3 2