- Devel - libvirt List Archives

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

15 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 15 May '15

15 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 15 May '15

15 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

[libvirt] [RFC v1 0/6] Live Migration with ephemeral host NIC devices
by Chen Fan 15 May '15

15 May '15

my main goal is to add support migration with host NIC passthrough devices and keep the network connectivity. this series patch base on Shradha's patches on https://www.redhat.com/archives/libvir-list/2012-November/msg01324.html which is add migration support for host passthrough devices. 1) unplug the ephemeral devices before migration 2) do native migration 3) when migration finished, hotplug the ephemeral devices TODO: keep network connectivity on guest level by bonding device. Chen Fan (6): conf: add ephemeral element for hostdev supporting migration qemu: Save ephemeral devices into qemuDomainObjPrivate qemu: add check ephemeral devices only for PCI host devices migration: Migration support for ephemeral hostdevs managedsave: move the domain xml handling forward to stop CPU managedsave: add managedsave support for ephemeral host devices docs/schemas/domaincommon.rng | 10 ++ docs/schemas/network.rng | 5 + src/conf/domain_conf.c | 14 +- src/conf/domain_conf.h | 1 + src/conf/network_conf.c | 13 ++ src/conf/network_conf.h | 1 + src/network/bridge_driver.c | 1 + src/qemu/qemu_command.c | 11 ++ src/qemu/qemu_domain.c | 5 + src/qemu/qemu_domain.h | 3 + src/qemu/qemu_driver.c | 48 +++--- src/qemu/qemu_migration.c | 182 ++++++++++++++++++++- src/qemu/qemu_migration.h | 9 + src/qemu/qemu_process.c | 12 ++ tests/networkxml2xmlin/hostdev-pf.xml | 2 +- tests/networkxml2xmlin/hostdev.xml | 2 +- tests/networkxml2xmlout/hostdev-pf.xml | 2 +- tests/networkxml2xmlout/hostdev.xml | 2 +- .../qemuxml2argv-controller-order.xml | 2 +- .../qemuxml2argv-hostdev-pci-address-device.xml | 2 +- .../qemuxml2argv-hostdev-pci-address.xml | 2 +- .../qemuxml2argv-hostdev-scsi-autogen-address.xml | 22 +-- .../qemuxml2argv-hostdev-scsi-lsi-iscsi-auth.xml | 4 +- .../qemuxml2argv-hostdev-scsi-lsi-iscsi.xml | 4 +- .../qemuxml2argv-hostdev-scsi-lsi.xml | 2 +- .../qemuxml2argv-hostdev-scsi-rawio.xml | 2 +- .../qemuxml2argv-hostdev-scsi-readonly.xml | 2 +- .../qemuxml2argv-hostdev-scsi-sgio.xml | 2 +- .../qemuxml2argv-hostdev-scsi-shareable.xml | 2 +- ...qemuxml2argv-hostdev-scsi-virtio-iscsi-auth.xml | 4 +- .../qemuxml2argv-hostdev-scsi-virtio-iscsi.xml | 4 +- .../qemuxml2argv-hostdev-scsi-virtio-scsi.xml | 2 +- ...emuxml2argv-hostdev-usb-address-device-boot.xml | 2 +- .../qemuxml2argv-hostdev-usb-address-device.xml | 2 +- .../qemuxml2argv-hostdev-usb-address.xml | 2 +- .../qemuxml2argv-hostdev-vfio-multidomain.xml | 2 +- .../qemuxml2argvdata/qemuxml2argv-hostdev-vfio.xml | 2 +- .../qemuxml2argv-net-hostdev-multidomain.xml | 2 +- .../qemuxml2argv-net-hostdev-vfio-multidomain.xml | 2 +- .../qemuxml2argv-net-hostdev-vfio.xml | 2 +- .../qemuxml2argvdata/qemuxml2argv-net-hostdev.xml | 2 +- tests/qemuxml2argvdata/qemuxml2argv-pci-rom.xml | 4 +- ...qemuxml2xmlout-hostdev-scsi-autogen-address.xml | 22 +-- 43 files changed, 340 insertions(+), 83 deletions(-) -- 1.9.3

7 33

[libvirt] [PATCH 00/13] qemu: fix device alias usage, ide controllers, scsi options
by Laine Stump 14 May '15

14 May '15

These are intertwined a bit, hence they are all posted in the same series. 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": * 1 and 2 just eliminate redundant code that I found while making the changes. * 3 & 4 change two sets of if else if else if, into switches, so that it's cleaner and simpler to add more cases. * 5 is a utility function that will be used later to make it easier to get the alias of the controller required by a particular device. * 6 adds exceptions for the primary IDE on 440fx and primary SATA on q35 to the function that creates controller alias names (since those are hardcoded in qemu) * 7-9 switch all remaining uses of BufferAsprintf(buf, "blahcontroller%d") with calls to the function that returns the *real* alias of the controller device. * 10 removes a test that verifies we allow an IDE controller in an s390 machine definition, and renames the tests to indicate that although we're checking to make sure we allow usb controllers on s390, that this isn't actually a valid config. (I would add a warning in the appropriate place when this happens, but at the moment I'm sick of looking at the problem :-/) - This is v2 of the following patch that I posted by itself earlier: https://www.redhat.com/archives/libvir-list/2015-April/msg01596.html https://www.redhat.com/archives/libvir-list/2015-May/msg00031.html * 11 makes it an error to define an IDE controller that qemu won't be able to actually create. * 12 and 13 fix two additional prolems I noticed with the creation of device and drive strings for SCSI disks - the first is that we add a "channel=" option even when qemu has told us it doesn't support it, and the 2nd is that we have carried forward a restriction that bus must == 0 for SCSI - this check was added in 2009, and it seems that it is only valid when *not* using -device to define the disk parameters. Laine Stump (13): qemu: use qemuDomainMachineIsI440FX() in appropriate place qemu: eliminate duplicated code in qemuBuildDriveDevStr() qemu: change if to switch in qemuBuildDeviceAddressStr qemu: restructure qemuAssignDeviceControllerAlias conf: utility to return alias of a controller based on type/index qemu: add exceptions for alias names of primary sata/ide controllers qemu: use controller alias when constructing disk/controller args qemu: use alias for all controller ids in qemuBuildControllerDevStr qemu: use controller object alias in commandline for virtio-serial device qemu: remove test for allowing ide controller in s390, rename usb tests qemu: log error when domain has an upsupported IDE controller qemu: only add channel arg to scsi-disk if qemu binary supports it qemu: allow bus != 0 for scsi-disk when -device is used src/conf/domain_conf.c | 34 +++ src/conf/domain_conf.h | 3 + src/libvirt_private.syms | 1 + src/qemu/qemu_command.c | 314 ++++++++++++--------- src/qemu/qemu_command.h | 3 +- src/qemu/qemu_hotplug.c | 4 +- .../qemuxml2argv-disk-blockio.args | 2 +- .../qemuxml2argvdata/qemuxml2argv-disk-blockio.xml | 1 - .../qemuxml2argv-disk-drive-network-iscsi-lun.args | 2 +- .../qemuxml2argv-disk-ide-drive-split.args | 2 +- .../qemuxml2argv-disk-ide-drive-split.xml | 1 - .../qemuxml2argv-disk-sata-device.args | 4 +- .../qemuxml2argv-disk-scsi-disk-split.args | 8 +- .../qemuxml2argv-disk-scsi-disk-vpd.args | 2 +- .../qemuxml2argv-disk-scsi-disk-wwn.args | 4 +- .../qemuxml2argv-disk-scsi-lun-passthrough.args | 12 +- .../qemuxml2argv-disk-scsi-lun-passthrough.xml | 4 +- .../qemuxml2argv-disk-scsi-megasas.args | 2 +- .../qemuxml2argv-disk-scsi-virtio-scsi.args | 2 +- .../qemuxml2argv-disk-scsi-vscsi.args | 2 +- .../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 - .../qemuxml2argv-disk-virtio-scsi-ccw.args | 2 +- .../qemuxml2argv-disk-virtio-scsi-cmd_per_lun.args | 2 +- .../qemuxml2argv-disk-virtio-scsi-max_sectors.args | 2 +- .../qemuxml2argv-disk-virtio-scsi-num_queues.args | 2 +- .../qemuxml2argv-pseries-vio-user-assigned.args | 2 +- .../qemuxml2argvdata/qemuxml2argv-pseries-vio.args | 2 +- ...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 | 29 +- .../qemuxml2xmlout-disk-source-pool.xml | 1 - 36 files changed, 266 insertions(+), 192 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

3 41

[libvirt] Windows Development
by scott lee 14 May '15

14 May '15

I am experimenting with libvirt, and my favorite IDE is visual studio so I looked for an easy method to build the libvirt library on windows and found that the state of cross-platform libraries like Mingw are outdated or difficult to use correctly for developers who want to start coding immediatly with libvirt. I know there is a way, but assembling the right libraries and commands was just too much. So, I found that the libvirt dll’s are actually a part of most distros of linux and can be downloaded and extracted. I just needed to find out what dlls were needed, then search the repos for those files. After the search, I have begun work on a powershell script to get the latest libvirt builds from fedora, download and extract them. After that, you can go to town building libraries against the libvirt api. Check out the script here: https://github.com/smasherprog/Libvirt_Windows_Dll Anyone who links against the libvirt library would be creating a management tool of some sort so a language like C# so after I create this tool, I will write some code to generate the C# code files for the DLLImport calls. The Libvirt header thousands of lines long so I will write some code to extract the calls/structs and generate the PInvoke calls for everything at once and place it into a .cs file so developers can start dewveloping in libvirt fast and easy. Scott Lee

1 0