October 2012 - Devel - libvirt List Archives

[libvirt] [PATCH] fix kvm_pv_eoi with kvmclock
by Martin Kletzander 08 Oct '12

08 Oct '12

When both kvmclock and kvm_pv_eoi are configured (either disabled or enabled) libvirt will generate invalid CPU specification due to the fact that even though kvmclock causes the CPU to be specified, it doesn't set have_cpu flag to true (and the new kvm_pv_eoi as well). This patch fixes the issue and adds a test exactly for that to show that it is fixed correctly (and also to keep it that way in the future of course). --- src/qemu/qemu_command.c | 2 ++ .../qemuxml2argv-kvmclock+eoi-disabled.args | 4 ++++ .../qemuxml2argv-kvmclock+eoi-disabled.xml | 27 ++++++++++++++++++++++ tests/qemuxml2argvtest.c | 1 + 4 files changed, 34 insertions(+) create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-kvmclock+eoi-disabled.args create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-kvmclock+eoi-disabled.xml diff --git a/src/qemu/qemu_command.c b/src/qemu/qemu_command.c index 20730a9..09f412e 100644 --- a/src/qemu/qemu_command.c +++ b/src/qemu/qemu_command.c @@ -4210,6 +4210,7 @@ qemuBuildCpuArgStr(const struct qemud_driver *driver, virBufferAsprintf(&buf, "%s,%ckvmclock", have_cpu ? "" : default_model, sign); + have_cpu = true; break; } } @@ -4224,6 +4225,7 @@ qemuBuildCpuArgStr(const struct qemud_driver *driver, virBufferAsprintf(&buf, "%s,%ckvm_pv_eoi", have_cpu ? "" : default_model, sign); + have_cpu = true; } if (virBufferError(&buf)) diff --git a/tests/qemuxml2argvdata/qemuxml2argv-kvmclock+eoi-disabled.args b/tests/qemuxml2argvdata/qemuxml2argv-kvmclock+eoi-disabled.args new file mode 100644 index 0000000..f7ade0a --- /dev/null +++ b/tests/qemuxml2argvdata/qemuxml2argv-kvmclock+eoi-disabled.args @@ -0,0 +1,4 @@ +LC_ALL=C PATH=/bin HOME=/home/test USER=test LOGNAME=test /usr/bin/qemu -S -M pc \ +-cpu qemu32,-kvmclock,-kvm_pv_eoi -m 214 -smp 6 -nographic -monitor \ +unix:/tmp/test-monitor,server,nowait -boot n -net none -serial \ +none -parallel none -usb diff --git a/tests/qemuxml2argvdata/qemuxml2argv-kvmclock+eoi-disabled.xml b/tests/qemuxml2argvdata/qemuxml2argv-kvmclock+eoi-disabled.xml new file mode 100644 index 0000000..a79b09f --- /dev/null +++ b/tests/qemuxml2argvdata/qemuxml2argv-kvmclock+eoi-disabled.xml @@ -0,0 +1,27 @@ +<domain type='qemu'> + <name>QEMUGuest1</name> + <uuid>c7a5fdbd-edaf-9455-926a-d65c16db1809</uuid> + <memory unit='KiB'>219100</memory> + <currentMemory unit='KiB'>219100</currentMemory> + <vcpu placement='static'>6</vcpu> + <os> + <type arch='i686' machine='pc'>hvm</type> + <boot dev='network'/> + </os> + <features> + <acpi/> + <apic eoi='off'/> + <pae/> + </features> + <clock offset='utc'> + <timer name='kvmclock' present='no'/> + </clock> + <on_poweroff>destroy</on_poweroff> + <on_reboot>restart</on_reboot> + <on_crash>destroy</on_crash> + <devices> + <emulator>/usr/bin/qemu</emulator> + <controller type='usb' index='0'/> + <memballoon model='virtio'/> + </devices> +</domain> diff --git a/tests/qemuxml2argvtest.c b/tests/qemuxml2argvtest.c index 0757e37..7732ae1 100644 --- a/tests/qemuxml2argvtest.c +++ b/tests/qemuxml2argvtest.c @@ -407,6 +407,7 @@ mymain(void) DO_TEST("cpu-eoi-enabled", QEMU_CAPS_ENABLE_KVM); DO_TEST("eoi-disabled", NONE); DO_TEST("eoi-enabled", NONE); + DO_TEST("kvmclock+eoi-disabled", QEMU_CAPS_ENABLE_KVM); DO_TEST("hugepages", QEMU_CAPS_MEM_PATH); DO_TEST("disk-cdrom", NONE); -- 1.7.12

2 2

[libvirt] [PATCH] qemu: Fix parsing of x86 CPU models
by Jiri Denemark 08 Oct '12

08 Oct '12

QEMU changed the output of -cpu ? and CPU models can now be followed by its description. Our parser just used both model name and its description as a model name, which made any model with a description unusable with libvirt. --- src/qemu/qemu_capabilities.c | 11 ++++++++--- 1 file changed, 8 insertions(+), 3 deletions(-) diff --git a/src/qemu/qemu_capabilities.c b/src/qemu/qemu_capabilities.c index a5eb995..d16a7bc 100644 --- a/src/qemu/qemu_capabilities.c +++ b/src/qemu/qemu_capabilities.c @@ -379,7 +379,7 @@ typedef int qemuCapsPtr caps); /* Format: - * <arch> <model> + * <arch> <model> <description> * qemu-0.13 encloses some model names in []: * <arch> [<model>] */ @@ -389,6 +389,7 @@ qemuCapsParseX86Models(const char *output, { const char *p = output; const char *next; + const char *end; int ret = -1; do { @@ -416,8 +417,12 @@ qemuCapsParseX86Models(const char *output, goto cleanup; } - if (next) - len = next - p - 1; + end = next ? next - 1 : NULL; + if ((t = strchr(p, ' ')) && (!next || t < next)) + end = t; + + if (end) + len = end - p; else len = strlen(p); -- 1.7.12

2 5

[libvirt] [ANNOUNCE] libvirt-glib 0.1.3 release
by Daniel P. Berrange 08 Oct '12

08 Oct '12

I am pleased to announce that a new release of the libvirt-glib package, version 0.1.3, is now available from ftp://libvirt.org/libvirt/glib/ The packages are GPG signed with Key fingerprint: DAF3 A6FD B26B 6291 2D0E 8E3F BE86 EBB4 1510 4FDF (4096R) New in this release: - Improve documentation for set_memory method in GVirConfigDomain - Handle size units for memory attributes - Add API for current memory - Fix python example to specify virt type - Use 1 GB of RAM in python example instead 1 TB ! - Add API to get hypervisor name and version - Avoid NULL format string for g_set_error - Fix XXX_new_with_xml constructors for config objects - Add misc getters for VNC/SPICE config attributes - Add handling of PMSUSPENDED state - Add handling of WAKEUP events - Require libvirt 0.10.2 minimum - Add API for domain PM wakeup libvirt-glib comprises three distinct libraries: - libvirt-glib - Integrate with the GLib event loop and error handling - libvirt-gconfig - Representation of libvirt XML documents as GObjects - libvirt-gobject - Mapping of libvirt APIs into the GObject type system NB: While libvirt aims to be API/ABI stable forever, with libvirt-glib we are not yet guaranteeing that libvirt-glib libraries are API/ABI permanently stable. As of the 0.0.8 release, we have tentatively frozen the API/ABI with the intent of being longterm stable hereafter, but there is still a small chance we might find flaws requiring an API/ABI change. The likelihood of this is low, however, and we will strive to avoid it. Follow up comments about libvirt-glib should be directed to the regular libvir-list(a)redhat.com development list. Thanks to all the people involved in contributing to this release. Regards, Daniel -- |: http://berrange.com -o- http://www.flickr.com/photos/dberrange/ :| |: http://libvirt.org -o- http://virt-manager.org :| |: http://autobuild.org -o- http://search.cpan.org/~danberr/ :| |: http://entangle-photo.org -o- http://live.gnome.org/gtk-vnc :|

1 0

[libvirt] Core dump caused by misusing openssl in multithread scenario!
by Benjamin Wang (gendwang) 08 Oct '12

08 Oct '12

Hi, I am running libvirt with ESXi driver in multithread scenario to access ESXi by https. Sometimes a core dump will be generated as following: #0 0x0000003f9b030265 in raise () from /lib64/libc.so.6 #1 0x0000003f9b031d10 in abort () from /lib64/libc.so.6 #2 0x0000003f9b06a84b in __libc_message () from /lib64/libc.so.6 #3 0x0000003f9b072fae in _int_malloc () from /lib64/libc.so.6 #4 0x0000003f9b074cde in malloc () from /lib64/libc.so.6 #5 0x0000003f9b07963b in strerror () from /lib64/libc.so.6 #6 0x0000003fa188032a in ERR_load_ERR_strings () from /lib64/libcrypto.so.6 #7 0x0000003fa187fde9 in ERR_load_crypto_strings () from /lib64/libcrypto.so.6 #8 0x0000003fa48309d9 in SSL_load_error_strings () from /lib64/libssl.so.6 #9 0x00002aaaba8e612e in Curl_ossl_init () from /opt/CSCOppm-unit/hypervisor/libcurl/lib/libcurl.so.4 #10 0x00002aaaba8ee6c1 in curl_global_init () from /opt/CSCOppm-unit/hypervisor/libcurl/lib/libcurl.so.4 #11 0x00002aaaba8ee6f8 in curl_easy_init () from /opt/CSCOppm-unit/hypervisor/libcurl/lib/libcurl.so.4 #12 0x00002aaaba0d932b in esxVI_SessionIsActive (ctx=0x2aaac093ca80, sessionID=0x2aaac06932a0 "`3i\300\252*", userName=0x2aaac0ae6e80 "root", output=0xffffffffffffffff) at esx/esx_vi_methods.generated.c:599 #13 0x00002aaaba0c7a60 in esxStorageVolumeLookupByKey (conn=0x7412, key=0x76c1 <Address 0x76c1 out of bounds>) at esx/esx_storage_driver.c:825 I checked that currently ESXi driver didn't initialize openssl. Because libcurl will not handle openssl for multi-thread. According to openssl API, libvirt should register two methods to support mutli-threads. The detailed description is as following: http://www.openssl.org/docs/crypto/threads.html I have changed code as following: 1. virInitialize() in libvirt.c Old Code: int virInitialize(void) { ... virLogSetFromEnv(); virNetTLSInit(); ... } New Code: int virInitialize(void) { ... virLogSetFromEnv(); virNetTLSInit(); virOpenSSLInit(); ... } 2. In virnetServer.c New Code: pthread_mutex_t *lock_cs; long *lock_count; void virOpenSSLLockCallback(int mode, int type, const char *file ATTRIBUTE_UNUSED, int line ATTRIBUTE_UNUSED) { if (mode & CRYPTO_LOCK) { pthread_mutex_lock(&(lock_cs[type])); lock_count[type]++; } else { pthread_mutex_unlock(&(lock_cs[type])); } } unsigned long virOpenSSLIdCallback(void) { unsigned long ret; ret=(unsigned long)pthread_self(); return(ret); } void virOpenSSLInit(void) { int i; lock_cs=OPENSSL_malloc(CRYPTO_num_locks() * sizeof(pthread_mutex_t)); lock_count=OPENSSL_malloc(CRYPTO_num_locks() * sizeof(long)); for (i=0; i<CRYPTO_num_locks(); i++) { lock_count[i]=0; pthread_mutex_init(&(lock_cs[i]),NULL); } CRYPTO_set_id_callback(virOpenSSLIdCallback); CRYPTO_set_locking_callback(virOpenSSLLockCallback); } To be honest, virOpenSSLInit/ virOpenSSLIdCallback/ virOpenSSLLockCallback should not be defined in this file. But It seems that Makefile generated by autoconfig can't handle the new file recursively. What about this solution? If you have any comments, please feel free to contact me. BTW: If I add a new source/header file, is there a simple way to change Makefile? B.R. Benjamin Wang

3 10

[libvirt] [PATCH v2 0/2] python: cleanup vcpu related binding APIs
by Guannan Ren 08 Oct '12

08 Oct '12

The first version https://www.redhat.com/archives/libvir-list/2012-April/msg00848.html Sorry for so late. I will go on adding error handling and make it strong for python binding APIs if necessary Guannan Ren python: cleanup vcpu related binding APIs python: return error if PyObject obj is NULL for unwrapper helper functions python/libvirt-override.c | 171 ++++++++++++++++++++++++++++++++++++++-------------- python/typewrappers.c | 43 +++++++++++++- 2 files changed, 164 insertions(+), 50 deletions(-)

2 9

[libvirt] [PATCH 0/7] Add support for lock failure action
by Jiri Denemark 08 Oct '12

08 Oct '12

When sanlock is used as a locking driver and sanlock deamon loses access to its lockspace, it automatically kills off the domains that had their locks stored there. Apparently some management apps would like to change this behavior by configuring what should happen when locks are lost. Everything except the sanlock_helper code from patch 7/7 should be stable. The helper code may need some additional work (perhaps even touching other parts of libvirtd internals) because some lock failure action does not work as expected. For example, when automatic disk locks are enabled, the "pause" action will result in libvirtd calling back to sanlock and since it doesn't have access to the lockspace, the call would just hang. However, since currently the only requestor for this feature is VDSM and it doesn't use automatic disk locks, the sanlock_helper could work for them without any modifications. And since current sanlock_helper implementation is supposed to be changed when libvirtd gains administrative interface (see patch 7/7 for more details) we may even just live with its limitations for the time being. Jiri Denemark (7): conf: Generalize life cycle actions to event actions conf: Add on_lockfailure event configuration qemu: Use macro instead of "qemu" in the context of URI scheme locking: Add const char * parameter to avoid ugly typecasts locking: Pass hypervisor driver name when acquiring locks locking: Add support for lock failure action locking: Implement lock failure action in sanlock driver docs/formatdomain.html.in | 37 +++++-- docs/internals/locking.html.in | 8 ++ docs/schemas/domaincommon.rng | 30 +++++- libvirt.spec.in | 1 + src/Makefile.am | 13 ++- src/conf/domain_conf.c | 85 +++++++++------ src/conf/domain_conf.h | 18 +++- src/libvirt_private.syms | 2 + src/locking/domain_lock.c | 36 ++++--- src/locking/domain_lock.h | 4 + src/locking/lock_driver.h | 8 +- src/locking/lock_driver_nop.c | 1 + src/locking/lock_driver_sanlock.c | 114 +++++++++++++++++--- src/locking/lock_manager.c | 10 +- src/locking/lock_manager.h | 1 + src/locking/sanlock_helper.c | 214 ++++++++++++++++++++++++++++++++++++++ src/qemu/qemu_conf.h | 2 + src/qemu/qemu_driver.c | 21 ++-- src/qemu/qemu_hotplug.c | 15 ++- src/qemu/qemu_process.c | 4 +- 20 files changed, 537 insertions(+), 87 deletions(-) create mode 100644 src/locking/sanlock_helper.c -- 1.7.12

2 14

[libvirt] Support for qemu snapshot=on drives in libvirt
by Richard W.M. Jones 08 Oct '12

08 Oct '12

I notice that the qemu driver doesn't support snapshot drives (-drive file=foo,snapshot=on). This is important for libguestfs. Currently libguestfs hacks this using <qemu:arg>. That works fine for static disks in the libvirt XML, but lack of direct support in libvirt is a blocker for adding hotplugging to libguestfs. In qemu, the snapshot=on feature does several things: (a) It creates a temporary qcow2 disk in $TMPDIR. (b) It sets the backing file of the temporary disk to the real disk. (c) When running, writes to the disk go to the temporary overlay, and not to the real disk. (d) I believe it also lets you commit changes from the temporary disk to the backing file using a monitor command ("commit"? "save"?). However we don't need or use this feature. (e) When qemu exits, the temporary qcow2 disk is deleted. So note that when you use snapshot=on there is an implicit dependency on the $TMPDIR environment variable. I don't know of a way to create the snapshot in another directory (at least, not using snapshot=on .. possibly one can do it by creating an explicit overlay disk in libvirt). A simple implementation therefore would be to add a <snapshot/> element to <disk>. It would just add snapshot=on and ignore concerns about $TMPDIR. Or reuse the <readonly/> flag? Note that these disks are writable. A more complex implementation would create an explicit overlay disk in libvirt and clean it up afterwards. This would let us control where the temporary disk is created. XML might look like this: <snapshot [tmp=...] /> I'm also concerned about the time is takes to run the external 'qemu-img create' command, which is non-trivial in some versions of qemu. In libguestfs, every millisecond counts. $ time qemu-img create -f qcow2 test.qcow2 10M Formatting 'test.qcow2', fmt=qcow2 size=10485760 encryption=off cluster_size=65536 real 0m0.610s user 0m0.022s sys 0m0.009s Or, can this be done using existing libvirt features? Any thoughts on this before I implement something ... Rich. -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones Read my programming blog: http://rwmj.wordpress.com Fedora now supports 80 OCaml packages (the OPEN alternative to F#) http://cocan.org/getting_started_with_ocaml_on_red_hat_and_fedora

4 9

[libvirt] [PATCH] python: keep consistent handling of Python integer conversion
by Guannan Ren 08 Oct '12

08 Oct '12

libvirt_ulonglongUnwrap requires the integer type of python obj. But libvirt_longlongUnwrap still could handle python obj of Pyfloat_type which causes the float value to be rounded up to an integer. For example >>> dom.setSchedulerParameters({'vcpu_quota': 0.88}) 0 libvirt_longlongUnwrap treats 0.88 as a valid value 0 However >>> dom.setSchedulerParameters({'cpu_shares': 1000.22}) libvirt_ulonglongUnwrap will throw out an error "TypeError: an integer is required" The patch make this consistent. --- python/typewrappers.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/python/typewrappers.c b/python/typewrappers.c index 7580689..d633603 100644 --- a/python/typewrappers.c +++ b/python/typewrappers.c @@ -220,7 +220,7 @@ libvirt_ulongUnwrap(PyObject *obj, unsigned long *val) int libvirt_longlongUnwrap(PyObject *obj, long long *val) { - long long llong_val; + long long llong_val = -1; if (!obj) { PyErr_SetString(PyExc_TypeError, "unexpected type"); @@ -230,7 +230,11 @@ libvirt_longlongUnwrap(PyObject *obj, long long *val) /* If obj is of PyInt_Type, PyLong_AsLongLong * will call PyInt_AsLong() to handle it automatically. */ - llong_val = PyLong_AsLongLong(obj); + if (PyInt_Check(obj) || PyLong_Check(obj)) + llong_val = PyLong_AsLongLong(obj); + else + PyErr_SetString(PyExc_TypeError, "an integer is required"); + if ((llong_val == -1) && PyErr_Occurred()) return -1; -- 1.7.11.2

2 2

[libvirt] [RFC] Private data pointer for virStoragePool and virStorageVol structs
by Matthias Bolte 08 Oct '12

08 Oct '12

Ata Bohra proposed a backend concept for the ESX storage driver to simplify the addition of iSCSI support to the ESX storage driver. The problem here is that most of the storage driver functions need to determine the backend to be used from the pool name. This overhead can be avoided by adding a private data pointer to the virStoragePool and virStorageVol structs that the ESX storage driver can use to store a pointer to the correct backend. struct _virStoragePool { virObject object; virConnectPtr conn; char *name; unsigned char uuid[VIR_UUID_BUFLEN]; void *privateData; virFreeCallback privateDataFreeFunc; }; struct _virStorageVol { virObject object; virConnectPtr conn; char *pool; char *name; char *key; void *privateData; virFreeCallback privateDataFreeFunc; }; virGetStoragePool and virGetStorageVol would set privateData and privateDataFreeFunc as given by the caller. virStoragePoolDispose and virStorageVolDispose would call privateDataFreeFunc if set. Comments? -- Matthias Bolte http://photron.blogspot.com

2 1

[libvirt] ANNOUNCE: libvirt 0.9.6.3 maintenance release
by Cole Robinson 07 Oct '12

07 Oct '12

libvirt 0.9.6.3 maintenance release is now available. This is libvirt 0.9.6 with additional bugfixes that have accumulated upstream since the initial release. This release can be downloaded at: http://libvirt.org/sources/stable_updates/libvirt-0.9.6.3.tar.gz Changes in this version: * security: Fix libvirtd crash possibility For info about past maintenance releases, see: http://wiki.libvirt.org/page/Maintenance_Releases Thanks, Cole

1 0