- Devel - libvirt List Archives

[libvirt] [PATCH python v2] fix libvirt headers list
by Dmitry Guryanov 28 Oct '14

28 Oct '14

Since libvirt.h has been split out, generator.py should be fixed accordingly. So add full list of header files. Signed-off-by: Dmitry Guryanov <dguryanov(a)parallels.com> --- Changes in v2: * fix list format generator.py | 26 ++++++++++++++++++-------- 1 file changed, 18 insertions(+), 8 deletions(-) diff --git a/generator.py b/generator.py index 01ab441..c66c6d4 100755 --- a/generator.py +++ b/generator.py @@ -29,6 +29,21 @@ import xml.sax debug = 0 onlyOverrides = False +libvirt_headers = [ + "libvirt", + "libvirt-domain", + "libvirt-domain-snapshot", + "libvirt-event", + "libvirt-host", + "libvirt-interface", + "libvirt-network", + "libvirt-nodedev", + "libvirt-nwfilter", + "libvirt-secret", + "libvirt-storage", + "libvirt-stream", +] + def getparser(): # Attach parser to an unmarshalling object. return both objects. target = docParser() @@ -112,9 +127,7 @@ class docParser(xml.sax.handler.ContentHandler): self.function_return_field = attrs['field'] elif tag == 'enum': # enums come from header files, hence virterror.h - if (attrs['file'] == "libvirt" or - attrs['file'] == "virterror" or - attrs['file'] == "virerror"): + if attrs['file'] in libvirt_headers + ["virerror", "virterror"]: enum(attrs['type'],attrs['name'],attrs['value']) elif attrs['file'] == "libvirt-lxc": lxc_enum(attrs['type'],attrs['name'],attrs['value']) @@ -127,11 +140,8 @@ class docParser(xml.sax.handler.ContentHandler): if tag == 'function': # fuctions come from source files, hence 'virerror.c' if self.function is not None: - if (self.function_module == "libvirt" or - self.function_module == "event" or - self.function_module == "virevent" or - self.function_module == "virerror" or - self.function_module == "virterror"): + if self.function_module in libvirt_headers + \ + ["event", "virevent", "virerror", "virterror"]: function(self.function, self.function_descr, self.function_return, self.function_args, self.function_file, self.function_module, -- 1.9.3

2 1

[libvirt] [PATCH python] fix libvirt headers list
by Dmitry Guryanov 28 Oct '14

28 Oct '14

Since libvirt.h has been splitted out, generator.py should be fixed accordingly. So add full list of header files. Signed-off-by: Dmitry Guryanov <dguryanov(a)parallels.com> --- generator.py | 24 ++++++++++++++++-------- 1 file changed, 16 insertions(+), 8 deletions(-) diff --git a/generator.py b/generator.py index 01ab441..5ecbed5 100755 --- a/generator.py +++ b/generator.py @@ -29,6 +29,19 @@ import xml.sax debug = 0 onlyOverrides = False +libvirt_headers = ["libvirt", + "libvirt-domain", + "libvirt-domain-snapshot", + "libvirt-event", + "libvirt-host", + "libvirt-interface", + "libvirt-network", + "libvirt-nodedev", + "libvirt-nwfilter", + "libvirt-secret", + "libvirt-storage", + "libvirt-stream",] + def getparser(): # Attach parser to an unmarshalling object. return both objects. target = docParser() @@ -112,9 +125,7 @@ class docParser(xml.sax.handler.ContentHandler): self.function_return_field = attrs['field'] elif tag == 'enum': # enums come from header files, hence virterror.h - if (attrs['file'] == "libvirt" or - attrs['file'] == "virterror" or - attrs['file'] == "virerror"): + if attrs['file'] in libvirt_headers + ["virerror", "virterror"]: enum(attrs['type'],attrs['name'],attrs['value']) elif attrs['file'] == "libvirt-lxc": lxc_enum(attrs['type'],attrs['name'],attrs['value']) @@ -127,11 +138,8 @@ class docParser(xml.sax.handler.ContentHandler): if tag == 'function': # fuctions come from source files, hence 'virerror.c' if self.function is not None: - if (self.function_module == "libvirt" or - self.function_module == "event" or - self.function_module == "virevent" or - self.function_module == "virerror" or - self.function_module == "virterror"): + if self.function_module in libvirt_headers + \ + ["event", "virevent", "virerror", "virterror"]: function(self.function, self.function_descr, self.function_return, self.function_args, self.function_file, self.function_module, -- 1.9.3

2 3

[libvirt] [PATCH 00/11] RFC: refactor internal driver architecture
by Daniel P. Berrange 28 Oct '14

28 Oct '14

Currently we have a bit of a craz setup for activating drivers for virConnectPtr. We iterate over the primary hypervisor drivers until one matches and activates. This bit is fine. Then we do the same for the network, storage, interface, etc secondary drivers. This is where the fun starts. In the case of the stateless drivers that run outside of libvirtd, they always want the secondary drivers to match the primary driver. Unfortunately if they don't register any dummy secondary driver they'll get given the reomte driver instead. The probing of secondary drivers is really unhelpful here. In the case of stateful drivers that run inside of libvirtd, there is (almost) always only one driver instance that they want to work with. So at best the probing of secondary drivers is a waste of time. With the increasing feature set though we're getting tight dependancies from the hypervisor driver to the secondary driver. eg QEMU depends on the specific network driver we have because it calls various internal APIs it exposes. We also have circular dependancy problems during libvirtd startup: https://www.redhat.com/archives/libvir-list/2014-January/msg01143.html There the storage driver needs a virConnectPtr instance in order to talk to the secret driver. At the same time the storage driver must run before the hypervisor driver because the hypervisor driver needs to use it for autostart of VMs. This is a horrible mess. The solution I think is to remove the concept of probing for secondary drivers entirely. Instead I suggest creating a struct virDriver { ... } which contains a pointer to virHypervisorDriver, virNetworkDriver, virInterfaceDriver, etc. now virConnectPtr will only reference the single virDriverPtr and when we open a hypervisor driver, we immediately get a full set of APIs for everything. This way all the hypervisor drivers will always know exactly what secondary driver they are working with. We'll be able to remove all the xxxPrivateData elements from the virConnectPtr. The stateless drivers can just access the main privateData. The stateful drivers can avoid any use of privateData at all - just access their global static variable with the driverState. This won't quite solve the circular dependancy problem in that mail thread above. By removing the directly reliance on the virConnectPtr driver table though, we'll be able to update the code more easily so it can fetch secrets directly and thus the storage driver wont need to do virConnectOpen("qemu:///system"). The patches that follow illustrate the first part of cleaning up use of the xxxPrivateData fields. This isn't complete but I wanted to show the direction it would go before I spend the time todo the full job. Incidentally our code is buggy as hell when it comes to using the xxxPrivateData fields. The PHyp driver actually abuses the networkPrivateData field for all its APIs. Alot of code wil directly access the global state driver state already ignoring the privateData fields already. So they are really more pain than they are worth and leading to buggy code. Daniel P. Berrange (11): Clean up remote driver connection open code Update remote driver to always use privateData Update ESX driver to always use privateData Update Hyper-V driver to always use privateData Move phyp internal info out of the header file Remove abuse of networkPrivateData in phyp driver Update Parallels driver to always use privateData Update Test driver to always use privateData Remove use of storagePrivateData from storage driver Remove use of networkPrivateData from network driver Remove use of networkPrivateData from netcf driver src/esx/esx_device_monitor.c | 6 +- src/esx/esx_interface_driver.c | 16 +- src/esx/esx_network_driver.c | 20 +- src/esx/esx_nwfilter_driver.c | 6 +- src/esx/esx_secret_driver.c | 6 +- src/esx/esx_storage_backend_iscsi.c | 26 +- src/esx/esx_storage_backend_vmfs.c | 36 +-- src/esx/esx_storage_driver.c | 44 ++-- src/hyperv/hyperv_device_monitor.c | 6 +- src/hyperv/hyperv_interface_driver.c | 6 +- src/hyperv/hyperv_network_driver.c | 6 +- src/hyperv/hyperv_nwfilter_driver.c | 6 +- src/hyperv/hyperv_secret_driver.c | 6 +- src/hyperv/hyperv_storage_driver.c | 6 +- src/interface/interface_backend_netcf.c | 71 ++---- src/network/bridge_driver.c | 435 ++++++++++++++------------------ src/parallels/parallels_network.c | 2 - src/parallels/parallels_storage.c | 12 +- src/parallels/parallels_utils.h | 1 + src/phyp/phyp_driver.c | 225 ++++++++--------- src/phyp/phyp_driver.h | 53 ---- src/remote/remote_driver.c | 153 +++-------- src/rpc/gendispatch.pl | 28 +- src/storage/storage_driver.c | 194 +++++++------- src/test/test_driver.c | 24 +- 25 files changed, 537 insertions(+), 857 deletions(-) -- 2.1.0

3 24

[libvirt] [PATCHv5 0/6] Add non-FreeBSD guest support to Bhyve driver.
by Conrad Meyer 28 Oct '14

28 Oct '14

Drvbhyve hardcodes bhyveload(8) as the host bootloader for guests. bhyveload(8) loader only supports FreeBSD guests. This patch series adds <bootloader> and <bootloader_args> handling to bhyve_command, so libvirt can boot non-FreeBSD guests in Bhyve. Additionally, support for grub-bhyve(1)'s --cons-dev argument is added so that interactive GRUB menus can be manipulated with the domain-configured serial device. See patch logs for further details. Thanks, Conrad Changelog for v5: - Added loader arguments tests to the bhyvexml2argvtest harness - Dropped po/pot file changes per Daniel - Added bootloader, bootloader_args to domain XML schema - Added xml2argv tests for new bhyve bootloader handling - Added xml2argv test for grub with console device Conrad Meyer (6): bhyve: Support /domain/bootloader configuration for non-FreeBSD guests. bhyvexml2argv: Add loader argv tests. domaincommon.rng: Add bootloader, bootloader_arg to os=hvm schema for Bhyve bhyvexml2argv: Add tests for domain-configured bootloader, args bhyve: Add console support for grub-bhyve bootloader bhyvexml2argv: Add test for grub console support docs/drvbhyve.html.in | 94 ++++++++- docs/formatdomain.html.in | 4 +- docs/schemas/domaincommon.rng | 41 ++-- src/bhyve/bhyve_command.c | 215 +++++++++++++++++++-- src/bhyve/bhyve_command.h | 5 +- src/bhyve/bhyve_domain.c | 5 + src/bhyve/bhyve_domain.h | 1 + src/bhyve/bhyve_driver.c | 2 +- src/bhyve/bhyve_process.c | 13 +- .../bhyvexml2argv-acpiapic.ldargs | 1 + tests/bhyvexml2argvdata/bhyvexml2argv-base.ldargs | 1 + .../bhyvexml2argv-bhyveload-explicitargs.args | 3 + .../bhyvexml2argv-bhyveload-explicitargs.ldargs | 1 + .../bhyvexml2argv-bhyveload-explicitargs.xml | 23 +++ .../bhyvexml2argvdata/bhyvexml2argv-console.ldargs | 1 + .../bhyvexml2argv-custom-loader.args | 3 + .../bhyvexml2argv-custom-loader.ldargs | 1 + .../bhyvexml2argv-custom-loader.xml | 24 +++ .../bhyvexml2argv-disk-cdrom-grub.args | 3 + .../bhyvexml2argv-disk-cdrom-grub.ldargs | 2 + .../bhyvexml2argv-disk-cdrom-grub.xml | 23 +++ .../bhyvexml2argv-disk-cdrom.ldargs | 1 + .../bhyvexml2argv-disk-virtio.ldargs | 1 + .../bhyvexml2argv-grub-defaults.args | 3 + .../bhyvexml2argv-grub-defaults.ldargs | 2 + .../bhyvexml2argv-grub-defaults.xml | 23 +++ .../bhyvexml2argvdata/bhyvexml2argv-macaddr.ldargs | 1 + .../bhyvexml2argv-serial-grub.args | 4 + .../bhyvexml2argv-serial-grub.ldargs | 2 + .../bhyvexml2argv-serial-grub.xml | 26 +++ .../bhyvexml2argvdata/bhyvexml2argv-serial.ldargs | 1 + tests/bhyvexml2argvtest.c | 41 +++- 32 files changed, 527 insertions(+), 44 deletions(-) create mode 100644 tests/bhyvexml2argvdata/bhyvexml2argv-acpiapic.ldargs create mode 100644 tests/bhyvexml2argvdata/bhyvexml2argv-base.ldargs create mode 100644 tests/bhyvexml2argvdata/bhyvexml2argv-bhyveload-explicitargs.args create mode 100644 tests/bhyvexml2argvdata/bhyvexml2argv-bhyveload-explicitargs.ldargs create mode 100644 tests/bhyvexml2argvdata/bhyvexml2argv-bhyveload-explicitargs.xml create mode 100644 tests/bhyvexml2argvdata/bhyvexml2argv-console.ldargs create mode 100644 tests/bhyvexml2argvdata/bhyvexml2argv-custom-loader.args create mode 100644 tests/bhyvexml2argvdata/bhyvexml2argv-custom-loader.ldargs create mode 100644 tests/bhyvexml2argvdata/bhyvexml2argv-custom-loader.xml create mode 100644 tests/bhyvexml2argvdata/bhyvexml2argv-disk-cdrom-grub.args create mode 100644 tests/bhyvexml2argvdata/bhyvexml2argv-disk-cdrom-grub.ldargs create mode 100644 tests/bhyvexml2argvdata/bhyvexml2argv-disk-cdrom-grub.xml create mode 100644 tests/bhyvexml2argvdata/bhyvexml2argv-disk-cdrom.ldargs create mode 100644 tests/bhyvexml2argvdata/bhyvexml2argv-disk-virtio.ldargs create mode 100644 tests/bhyvexml2argvdata/bhyvexml2argv-grub-defaults.args create mode 100644 tests/bhyvexml2argvdata/bhyvexml2argv-grub-defaults.ldargs create mode 100644 tests/bhyvexml2argvdata/bhyvexml2argv-grub-defaults.xml create mode 100644 tests/bhyvexml2argvdata/bhyvexml2argv-macaddr.ldargs create mode 100644 tests/bhyvexml2argvdata/bhyvexml2argv-serial-grub.args create mode 100644 tests/bhyvexml2argvdata/bhyvexml2argv-serial-grub.ldargs create mode 100644 tests/bhyvexml2argvdata/bhyvexml2argv-serial-grub.xml create mode 100644 tests/bhyvexml2argvdata/bhyvexml2argv-serial.ldargs -- 1.9.3

3 22

[libvirt] [PATCH] uitl: use "if and only if" instead of "iff" in some comments
by Shanzhi Yu 28 Oct '14

28 Oct '14

>From wiki "In writing, phrases commonly used, with debatable propriety. In logic formulae, logical symbols are used instead of these phrases" In some comments, we use "if and only if", so instead the logical sysbols with phrases in case someone mistaking it as a wrong spelling of "if" Signed-off-by: Shanzhi Yu <shyu(a)redhat.com> --- src/util/virhash.c | 2 +- src/util/virstring.c | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/src/util/virhash.c b/src/util/virhash.c index e3c1880..e3bf1d3 100644 --- a/src/util/virhash.c +++ b/src/util/virhash.c @@ -65,7 +65,7 @@ struct _virHashTable { uint32_t seed; size_t size; size_t nbElems; - /* True iff we are iterating over hash entries. */ + /* True if and only if we are iterating over hash entries. */ bool iterating; /* Pointer to the current entry during iteration. */ virHashEntryPtr current; diff --git a/src/util/virstring.c b/src/util/virstring.c index 3dad9dd..974005d 100644 --- a/src/util/virstring.c +++ b/src/util/virstring.c @@ -519,7 +519,7 @@ virStrncpy(char *dest, const char *src, size_t n, size_t destbytes) return NULL; ret = strncpy(dest, src, n); - /* strncpy NULL terminates iff the last character is \0. Therefore + /* strncpy NULL terminates if and only if the last character is \0. Therefore * force the last byte to be \0 */ dest[n] = '\0'; -- 1.9.3

2 2

[libvirt] [PATCH v2] qemu: better error message when block job can't succeed
by Eric Blake 28 Oct '14

28 Oct '14

https://bugzilla.redhat.com/show_bug.cgi?id=1140981 reports that the qemu-kvm shipped as part of RHEL 7.0 intentionally cripples block jobs by removing the 'block-stream' QMP command, but that our probe for whether block jobs work was merely whether 'block-job-cancel' exists. As the crippling did not remove job cancellation, we end up trying the qemu command and printing an ugly message: error: Requested operation is not valid: Command 'block-stream' is not found even though our code base was already prepared to print a nice message that jobs are unsupported: error: unsupported configuration: block jobs not supported with this qemu binary Doing one more QMP probe lets us match the reality of crippled qemu. Yes, we really have to cater to two spellings of the command, due to older upstream qemu. * src/qemu/qemu_capabilities.h (QEMU_CAPS_BLOCK_STREAM): New probe. * src/qemu/qemu_capabilities.c (virQEMUCapsProbeQMPCommands): Kill block jobs with nice error message if qemu is crippled. Signed-off-by: Eric Blake <eblake(a)redhat.com> --- src/qemu/qemu_capabilities.c | 11 +++++++++++ src/qemu/qemu_capabilities.h | 1 + 2 files changed, 12 insertions(+) diff --git a/src/qemu/qemu_capabilities.c b/src/qemu/qemu_capabilities.c index 6fcb5c7..53cf05f 100644 --- a/src/qemu/qemu_capabilities.c +++ b/src/qemu/qemu_capabilities.c @@ -271,6 +271,7 @@ VIR_ENUM_IMPL(virQEMUCaps, QEMU_CAPS_LAST, "iothread", "migrate-rdma", "ivshmem", + "block-stream", ); @@ -1424,6 +1425,8 @@ struct virQEMUCapsStringFlags virQEMUCapsCommands[] = { { "transaction", QEMU_CAPS_TRANSACTION }, { "block_job_cancel", QEMU_CAPS_BLOCKJOB_SYNC }, { "block-job-cancel", QEMU_CAPS_BLOCKJOB_ASYNC }, + { "block_stream", QEMU_CAPS_BLOCK_STREAM }, + { "block-stream", QEMU_CAPS_BLOCK_STREAM }, { "dump-guest-memory", QEMU_CAPS_DUMP_GUEST_MEMORY }, { "query-spice", QEMU_CAPS_SPICE }, { "query-kvm", QEMU_CAPS_KVM }, @@ -2195,6 +2198,14 @@ virQEMUCapsProbeQMPCommands(virQEMUCapsPtr qemuCaps, ncommands, commands); virQEMUCapsFreeStringList(ncommands, commands); + /* Some RHEL builds of qemu-kvm have 'block-job-cancel' but no way + * to start a block job; in that case, our error messages are + * nicer if we pretend there is no blockjob support at all. */ + if (!virQEMUCapsGet(qemuCaps, QEMU_CAPS_BLOCK_STREAM)) { + virQEMUCapsClear(qemuCaps, QEMU_CAPS_BLOCKJOB_ASYNC); + virQEMUCapsClear(qemuCaps, QEMU_CAPS_BLOCKJOB_SYNC); + } + /* QMP add-fd was introduced in 1.2, but did not support * management control of set numbering, and did not have a * counterpart -add-fd command line option. We require the diff --git a/src/qemu/qemu_capabilities.h b/src/qemu/qemu_capabilities.h index 0214f30..6486b5b 100644 --- a/src/qemu/qemu_capabilities.h +++ b/src/qemu/qemu_capabilities.h @@ -218,6 +218,7 @@ typedef enum { QEMU_CAPS_OBJECT_IOTHREAD = 176, /* -object iothread */ QEMU_CAPS_MIGRATE_RDMA = 177, /* have rdma migration */ QEMU_CAPS_DEVICE_IVSHMEM = 178, /* -device ivshmem */ + QEMU_CAPS_BLOCK_STREAM = 179, /* block-stream exists */ QEMU_CAPS_LAST, /* this must always be the last item */ } virQEMUCapsFlags; -- 1.9.3

1 1

[libvirt] [PATCH 0/3] lxc: Implement emulator pin APIs to set/get cpuset
by Wang Rui 28 Oct '14

28 Oct '14

We can specify cpuset for a container defined with the xml like <vcpu placement='static' cpuset='0-3'> to achieve cpu isolation. It works when container is started. But there is no implements we can use to either change or get cpuset. The following patches implement the lxc driver methods for virDomainPinEmulator and virDomainGetEmulatorPinInfo. Also support container startup with emulator affinity info in xml. After these patches, we can set and get libvirt_lxc cpuset. Yue Wenyuan (3): lxc: Implement pin emulator for container startup lxc: Implement emulator pin API in lxc driver lxc: Implement geting emulator pin info API in lxc driver src/lxc/lxc_cgroup.c | 88 ++++++++++++++++++++ src/lxc/lxc_cgroup.h | 7 ++ src/lxc/lxc_controller.c | 4 + src/lxc/lxc_driver.c | 206 +++++++++++++++++++++++++++++++++++++++++++++++ 4 files changed, 305 insertions(+) -- 1.7.12.4

3 9

[libvirt] [PATCH v2 00/12] parallels: rewrite driver with parallels SDK
by Dmitry Guryanov 28 Oct '14

28 Oct '14

This patch series replaces all code, which used prlctl command to interact with parallels cloud server with calls to parallels sdk functions. The model of this driver remain almost the same - in creates a list of virDomainObj objects on connect and then functions, which returns different information get info from this list. Changes in v2: * Rebase to latest libvirt sources * Use only "parallels" prefix for functions in parallelsDriver, so that make check will pass * Update privconn->domains in case we change something from current connection. Alexander Burluka (4): parallels: get domain info with SDK parallels: handle events from parallels server parallels: added function virDomainIsActive() parallels: Add domainCreateWithFlags() function. Dmitry Guryanov (8): parallels: move IS_CT macro to parallels_utils.h parallels: move parallelsDomNotFoundError to parallels_utils.h parallels: reimplement functions, which change domain state parallels: rewrite parallelsApplyConfig with SDK parallels: create VMs and containers with sdk parallels: refactor parallelsDomainDefineXML parallels: add cdroms support parallels: implement domainUndefine and domainUndefineFlags src/parallels/parallels_driver.c | 2456 ++++++++------------------------------ src/parallels/parallels_sdk.c | 2451 +++++++++++++++++++++++++++++++++++++ src/parallels/parallels_sdk.h | 25 + src/parallels/parallels_utils.h | 11 + 4 files changed, 2957 insertions(+), 1986 deletions(-) -- 1.9.3

1 12

[libvirt] [PATCH 00/12] parallels: use SDK in parallels_driver.c
by Dmitry Guryanov 28 Oct '14

28 Oct '14

This patch series replaces all code, which used prlctl command to interact with parallels cloud server with calls to parallels sdk functions. The model of this driver remain almost the same - in creates a list of virDomainObj objects on connect and then functions, which returns different information get info from this list. But functions, which modify domains (change state, define or undefine domain) don't modify this list directly. On connect driver subscribes to event from PCS and modify the list from that events handler. So cached list of domains will be always up-to-date either after modifications from currect connection or after modifications by another tools (like prlctl) or libvirt connections. Alexander Burluka (4): parallels: get domain info with SDK parallels: handle events from parallels server parallels: added function virDomainIsActive() parallels: Add domainCreateWithFlags() function. Dmitry Guryanov (8): parallels: move IS_CT macro to parallels_utils.h parallels: move parallelsDomNotFoundError to parallels_utils.h parallels: reimplement functions, which change domain state parallels: rewrite parallelsApplyConfig with SDK parallels: create VMs and containers with sdk parallels: fix parallelsDomainDefineXML parallels: add cdroms support parallels: implement domainUndefine and domainUndefineFlags src/parallels/parallels_driver.c | 2448 +++++++------------------------------ src/parallels/parallels_sdk.c | 2456 ++++++++++++++++++++++++++++++++++++++ src/parallels/parallels_sdk.h | 19 + src/parallels/parallels_utils.h | 11 + 4 files changed, 2939 insertions(+), 1995 deletions(-) -- 1.9.3

1 13

[libvirt] [PowerPC Patch v3] Libvirt CPU enhancements for Power KVM
by Prerna Saxena 28 Oct '14

28 Oct '14

This patch series is a collection of enhancements for PowerPC CPUs on PowerKVM. Series Summary: ========== Patch 1/3 : Introduce a new architecture 'ppc64le' for libvirt. Patch 2/3 : Add libvirt support for VMs running in 'compat' mode on Power KVM. Patch 3/3 : Optimize PVR comparison for PowerPC CPUs. Detail: ==== * PowerPC has traditionally been a Big-endian architecture. However, with PowerPC ISA version 2.07, it can run in Little-endian mode as well. IBM Power8 processors, compliant with ISA 2.07 allow launching VMs in little-endian mode. This is signified by 'ppc64le' architecture. Patch 1 adds this support to libvirt, to allow running VMs based on ppc64le architecture. * Patch 2,3 tweak libvirt to correctly model PowerPC CPUs based on recent PowerKVM implementation. PowerKVM permits VMs with vcpus in the following allowed modes : i) Host native mode: where the vcpu seen in the VM belongs to the same processor generation as the host. Example: A POWER7 host, conforming to PowerISA version 2.06, will run VMs with "power7" vcpus. ii) Binary Compatibility ("compat") mode: PowerISA allows processors to run VMs in binary compatibility ("compat") mode supporting an older version of ISA. As an example: In compatibility mode, a POWER7 host can run a "power6" VM, conforming to power ISA v2.05. Similarly, a POWER8 host can run a "power7" VM conforming to PowerISA v2.06. QEMU has recently added support to explicitly denote a VM running in compatibility mode through commits 6d9412ea & 8dfa3a5e85. Henceforth, VMs of type (i) will be invoked with the QEMU invocation "-cpu host", while VMs of type (ii) will be invoked using "-cpu host, compat=power6". Now, an explicit cpu selection using "-cpu POWER6" is not valid. Instead, the recommended practice is to use the matching compat mode, if the requested cpu type differs from the host. Patches 2-3 address various aspects of this change. * Patch 2 : Adds support for generating the correct command line for QEMU. New xml semantics are introduced to signify this type. * Patch 3 : PowerKVM vCPUs differ uniquely across generations ( such as power6, power7, power8). Each generation signifies a new PowerISA version that exhibits features unique to that generation. The higher order 16 bits of PVR denote the processor generation and the lower order 16 bits denote the cpu chip (sub)version. For all practical purposes of launching a VM, we care about the generation the vCPU will belong to, and not specifically the chip version. In fact, PowerKVM does not seek out specification of a unique chip version(such as POWER7_v2.3) for running a vCPU. This patch updates the libvirt PVR check to reflect this relationship. Changelog: ========= v1 : https://www.redhat.com/archives/libvir-list/2014-June/msg01338.html v2 : http://www.redhat.com/archives/libvir-list/2014-October/msg00351.html Changes since v2: i) CPU compat mode is now introduced under CPU mode 'host-model' as against the original proposal of using cpu mode'custom'. (Thanks for the tip, Dan.) ii) Patch 4/4 that renamed models in cpu_map.xml is dropped. Instead, CPU generations are added to cpu_map.xml as new listings. The PVR check in patch 3 of this series takes care of making the right match happen for compat mode. Regards, -- Prerna Saxena Linux Technology Centre, IBM Systems and Technology Lab, Bangalore, India

2 8