June 2016 - Devel - libvirt List Archives

[libvirt] [PATCH] storage: Adjust qemu-img switches check
by John Ferlan 03 Jun '16

03 Jun '16

Since we support QEMU 0.12 and later, checking for support of specific flags added prior to that isn't necessary. Thus start with the base of having the "-o options" available for the qemu-img create option and then determine whether we have the compat option for qcow2 files (which would be necessary up through qemu 2.0 where the default changes to compat 0.11). NOTE: Keeping old tests around since it's still possible to create in the old format. Signed-off-by: John Ferlan <jferlan(a)redhat.com> --- src/storage/storage_backend.c | 66 ++++++++++++++++--------------------------- 1 file changed, 24 insertions(+), 42 deletions(-) diff --git a/src/storage/storage_backend.c b/src/storage/storage_backend.c index 3a23cd7..eaa6f65 100644 --- a/src/storage/storage_backend.c +++ b/src/storage/storage_backend.c @@ -869,9 +869,19 @@ virStoragePloopResize(virStorageVolDefPtr vol, return ret; } +/* Flag values shared w/ storagevolxml2argvtest.c. + * Since it's still possible to provide the old format args, just + * keep them; however, prefix with an "X_" (similar to qemu_capabilities.c) + * to indicate the are older. + * + * QEMU_IMG_BACKING_FORMAT_OPTIONS (added in qemu 0.11) + * QEMU_IMG_BACKING_FORMAT_OPTIONS_COMPAT + * was made necessary due to 2.0 change to change the default + * qcow2 file format from 0.10 to 1.1. + */ enum { - QEMU_IMG_BACKING_FORMAT_NONE = 0, - QEMU_IMG_BACKING_FORMAT_FLAG, + X_QEMU_IMG_BACKING_FORMAT_NONE = 0, + X_QEMU_IMG_BACKING_FORMAT_FLAG, QEMU_IMG_BACKING_FORMAT_OPTIONS, QEMU_IMG_BACKING_FORMAT_OPTIONS_COMPAT, }; @@ -904,46 +914,18 @@ virStorageBackendQemuImgSupportsCompat(const char *qemuimg) static int virStorageBackendQEMUImgBackingFormat(const char *qemuimg) { - char *help = NULL; - char *start; - char *end; - char *tmp; - int ret = -1; - int exitstatus; - virCommandPtr cmd = virCommandNewArgList(qemuimg, "-h", NULL); - - virCommandAddEnvString(cmd, "LC_ALL=C"); - virCommandSetOutputBuffer(cmd, &help); - virCommandClearCaps(cmd); - - /* qemuimg doesn't return zero exit status on -h, - * therefore we need to provide pointer for storing - * exit status, although we don't parse it any later */ - if (virCommandRun(cmd, &exitstatus) < 0) - goto cleanup; + /* As of QEMU 0.11 the [-o options] support was added via qemu + * commit id '9ea2ea71', so we start with that base and figure + * out what else we have */ + int ret = QEMU_IMG_BACKING_FORMAT_OPTIONS; + + /* QEMU 2.0 changed to using a format that only QEMU 1.1 and newer + * understands. Since we still support QEMU 0.12 and newer, we need + * to be able to handle the previous format as can be set via a + * compat=0.10 option. */ + if (virStorageBackendQemuImgSupportsCompat(qemuimg)) + ret = QEMU_IMG_BACKING_FORMAT_OPTIONS_COMPAT; - if ((start = strstr(help, " create ")) == NULL || - (end = strstr(start, "\n")) == NULL) { - virReportError(VIR_ERR_INTERNAL_ERROR, - _("unable to parse qemu-img output '%s'"), - help); - goto cleanup; - } - if (((tmp = strstr(start, "-F fmt")) && tmp < end) || - ((tmp = strstr(start, "-F backing_fmt")) && tmp < end)) { - ret = QEMU_IMG_BACKING_FORMAT_FLAG; - } else if ((tmp = strstr(start, "[-o options]")) && tmp < end) { - if (virStorageBackendQemuImgSupportsCompat(qemuimg)) - ret = QEMU_IMG_BACKING_FORMAT_OPTIONS_COMPAT; - else - ret = QEMU_IMG_BACKING_FORMAT_OPTIONS; - } else { - ret = QEMU_IMG_BACKING_FORMAT_NONE; - } - - cleanup: - virCommandFree(cmd); - VIR_FREE(help); return ret; } @@ -1196,7 +1178,7 @@ virStorageBackendCreateQemuImgCmdFromVol(virConnectPtr conn, VIR_FREE(opts); } else { if (info.backingPath) { - if (imgformat == QEMU_IMG_BACKING_FORMAT_FLAG) + if (imgformat == X_QEMU_IMG_BACKING_FORMAT_FLAG) virCommandAddArgList(cmd, "-F", backingType, NULL); else VIR_DEBUG("Unable to set backing store format for %s with %s", -- 2.5.5

1 1

[libvirt] [PATCH 0/3] Clean up some virstorageencryption code
by John Ferlan 03 Jun '16

03 Jun '16

While working through adding luks support for libvirt, I have a few patches that aren't really germane to adding support and rather than drop a large patch series when I'm done - I figured I'd post a few adjustments. In the long run the encryption code probably doesn't work, but I'm trying to move it to the side at least. John Ferlan (3): util: Clean up code formatting in virstorageencryption storage: Split out setting default secret for encryption storage: Split out a helper for encryption checks src/storage/storage_backend.c | 80 ++++++++++++++++++++++++++-------------- src/storage/storage_backend_fs.c | 79 ++++++++++++++++++++++++--------------- src/util/virstorageencryption.c | 58 ++++++++++++++--------------- 3 files changed, 128 insertions(+), 89 deletions(-) -- 2.5.5

1 4

[libvirt] [PATCH 0/2] Fix a virperf related crasher
by Michal Privoznik 03 Jun '16

03 Jun '16

I've ran into a crasher. Worse crasher over RO connection. Michal Privoznik (2): Drop virPerfGetEventFd virPerfEventIsEnabled: Don't crash on shut off domains src/libvirt_private.syms | 1 - src/util/virperf.c | 22 +++------------------- src/util/virperf.h | 3 --- 3 files changed, 3 insertions(+), 23 deletions(-) -- 2.8.3

2 4

[libvirt] Plan for the next release
by Daniel Veillard 03 Jun '16

03 Jun '16

I will be travelling end of this week and most of next week, but I think we can still try to get a release for June 1st, I suggest: - enter freeze this Thrusday 26 - Push an rc2 over the week-end - try to push the 1.3.5 release on Wed June 1st I hope this works for everybody ! Thanks, Daniel -- Daniel Veillard | Open Source and Standards, Red Hat veillard(a)redhat.com | libxml Gnome XML XSLT toolkit http://xmlsoft.org/ http://veillard.com/ | virtualization library http://libvirt.org/

2 1

[libvirt] [PATCH v3] virDomainChrGetDomainPtrsInternal: Return an integer
by Michal Privoznik 02 Jun '16

02 Jun '16

There's this problem on the recent gcc-6.1: In file included from conf/domain_conf.c:37:0: conf/domain_conf.c: In function 'virDomainChrPreAlloc': conf/domain_conf.c:14109:35: error: potential null pointer dereference [-Werror=null-dereference] return VIR_REALLOC_N(*arrPtr, *cntPtr + 1); ^~ ./util/viralloc.h:158:73: note: in definition of macro 'VIR_REALLOC_N' # define VIR_REALLOC_N(ptr, count) virReallocN(&(ptr), sizeof(*(ptr)), (count), \ ^~~~~ conf/domain_conf.c: In function 'virDomainChrRemove': conf/domain_conf.c:14133:21: error: potential null pointer dereference [-Werror=null-dereference] for (i = 0; i < *cntPtr; i++) { ^~~~~~~ GCC basically fails to see, that the virDomainChrGetDomainPtrsInternal will never actually return NULL because it's never called over a domain char device with _LAST type. But to make it shut up, lets turn this function into returning an integer and check in the callers if a zero value value was returned. Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com> --- diff to v2: - Yet another improvement as suggested in review of v2 src/conf/domain_conf.c | 39 +++++++++++++++++++++++---------------- 1 file changed, 23 insertions(+), 16 deletions(-) diff --git a/src/conf/domain_conf.c b/src/conf/domain_conf.c index 7d46d0b..9f9fdf2 100644 --- a/src/conf/domain_conf.c +++ b/src/conf/domain_conf.c @@ -14038,7 +14038,7 @@ virDomainChrFind(virDomainDefPtr def, /* Return the address within vmdef to be modified when working with a * chrdefptr of the given type. */ -static void +static int ATTRIBUTE_RETURN_CHECK virDomainChrGetDomainPtrsInternal(virDomainDefPtr vmdef, virDomainChrDeviceType type, virDomainChrDefPtr ***arrPtr, @@ -14048,28 +14048,30 @@ virDomainChrGetDomainPtrsInternal(virDomainDefPtr vmdef, case VIR_DOMAIN_CHR_DEVICE_TYPE_PARALLEL: *arrPtr = &vmdef->parallels; *cntPtr = &vmdef->nparallels; - break; + return 0; case VIR_DOMAIN_CHR_DEVICE_TYPE_SERIAL: *arrPtr = &vmdef->serials; *cntPtr = &vmdef->nserials; - break; + return 0; case VIR_DOMAIN_CHR_DEVICE_TYPE_CONSOLE: *arrPtr = &vmdef->consoles; *cntPtr = &vmdef->nconsoles; - break; + return 0; case VIR_DOMAIN_CHR_DEVICE_TYPE_CHANNEL: *arrPtr = &vmdef->channels; *cntPtr = &vmdef->nchannels; - break; + return 0; case VIR_DOMAIN_CHR_DEVICE_TYPE_LAST: - *arrPtr = NULL; - *cntPtr = NULL; break; } + + virReportError(VIR_ERR_INTERNAL_ERROR, + _("Unknown char device type: %d"), type); + return -1; } @@ -14085,14 +14087,13 @@ virDomainChrGetDomainPtrs(const virDomainDef *vmdef, size_t *cntVar = NULL; /* Cast away const; we add it back in the final assignment. */ - virDomainChrGetDomainPtrsInternal((virDomainDefPtr) vmdef, type, - &arrVar, &cntVar); - if (arrVar) { + if (virDomainChrGetDomainPtrsInternal((virDomainDefPtr) vmdef, type, + &arrVar, &cntVar) < 0) { + *arrPtr = NULL; + *cntPtr = 0; + } else { *arrPtr = (const virDomainChrDef **) *arrVar; *cntPtr = *cntVar; - } else { - *arrPtr = NULL; - *cntPtr = 0; } } @@ -14104,7 +14105,9 @@ virDomainChrPreAlloc(virDomainDefPtr vmdef, virDomainChrDefPtr **arrPtr = NULL; size_t *cntPtr = NULL; - virDomainChrGetDomainPtrsInternal(vmdef, chr->deviceType, &arrPtr, &cntPtr); + if (virDomainChrGetDomainPtrsInternal(vmdef, chr->deviceType, + &arrPtr, &cntPtr) < 0) + return -1; return VIR_REALLOC_N(*arrPtr, *cntPtr + 1); } @@ -14116,7 +14119,9 @@ virDomainChrInsertPreAlloced(virDomainDefPtr vmdef, virDomainChrDefPtr **arrPtr = NULL; size_t *cntPtr = NULL; - virDomainChrGetDomainPtrsInternal(vmdef, chr->deviceType, &arrPtr, &cntPtr); + if (virDomainChrGetDomainPtrsInternal(vmdef, chr->deviceType, + &arrPtr, &cntPtr) < 0) + return; VIR_APPEND_ELEMENT_INPLACE(*arrPtr, *cntPtr, chr); } @@ -14128,7 +14133,9 @@ virDomainChrRemove(virDomainDefPtr vmdef, virDomainChrDefPtr ret = NULL, **arrPtr = NULL; size_t i, *cntPtr = NULL; - virDomainChrGetDomainPtrsInternal(vmdef, chr->deviceType, &arrPtr, &cntPtr); + if (virDomainChrGetDomainPtrsInternal(vmdef, chr->deviceType, + &arrPtr, &cntPtr) < 0) + return NULL; for (i = 0; i < *cntPtr; i++) { ret = (*arrPtr)[i]; -- 2.8.3

2 1

[libvirt] [PATCH 00/10] vz: improve locking in getting domain statistics
by Nikolay Shirokovskiy 02 Jun '16

02 Jun '16

Current stats cache uses same lock as domain cache, namely lock of virDomainObjPtr. First this is not necessary, domain cache is used to translate id of object of interest from libvirt to vzsdk one. Besides this domain object is not used. Second this is risky game as lock is dropped on waiting for stat events from sdk deep down on stack so callers throughout the stack should be written very carefully. This patch series makes stats cache self locking and thus making code much more solid. This is done thru introducing extra lock for stats cache. Statistics API functions first use domain cache then drop domain lock, query stats cache and unreference domain object. Patches 1-4 conform API functions to this structure. Patches 5-7 makes subsidiary improvments. Patch 8 finally introduce new lock. Last patches are cleanup[9] and bugfix[10]. Nikolay Shirokovskiy (10): vz: pass string instead of disk definition to block stat function vz: prepare disks names before getting all disks stats vz: move getting stats in vzDomainGetVcpus to the end vz: move getting stats in vzDomainGetInfo to the end vz: use consistent naming for different domain object in vz_driver.c vz: simplify refcount on sdkdom in prlsdkLoadDomain vz: hold stats cache in a pointer in vzDomObj vz: introduce stats cache lock vz: extract on stats cache update into a function vz: fix many stat request issue src/vz/vz_driver.c | 397 +++++++++++++++++++++++++++++++++++------------------ src/vz/vz_sdk.c | 230 +++++++++++++++++++++---------- src/vz/vz_sdk.h | 15 +- src/vz/vz_utils.c | 9 -- src/vz/vz_utils.h | 6 +- 5 files changed, 438 insertions(+), 219 deletions(-) -- 1.8.3.1

1 11

[libvirt] libvirt-tck test failure after commit b3d06987
by Jim Fehlig 02 Jun '16

02 Jun '16

We've noticed libvirt-tck test 100-apply-verify-host.t failing recently on libvirt.git master and I finally got around to bisecting it to commit b3d06987. I haven't looked at the test in detail, but it appears to expect a broadcast address of 10.1.2.255, however finds an address of 0.0.0.0 after commit b3d06987. I'm not terribly familiar with this code, but the following hunk of b3d06987 looks suspect @@ -1039,21 +1039,28 @@ virNetDevCreateNetlinkAddressMessage(int messageType, const char *ifname, virSocketAddr *addr, unsigned int prefix, - virSocketAddr *broadcast) + virSocketAddr *broadcast, + virSocketAddr *peer) { struct nl_msg *nlmsg = NULL; struct ifaddrmsg ifa; unsigned int ifindex; void *addrData = NULL; + void *peerData = NULL; void *broadcastData = NULL; size_t addrDataLen; if (virNetDevGetIPAddressBinary(addr, &addrData, &addrDataLen) < 0) return NULL; - if (broadcast && virNetDevGetIPAddressBinary(broadcast, &broadcastData, - &addrDataLen) < 0) - return NULL; + if (peer && VIR_SOCKET_ADDR_VALID(peer)) { + if (virNetDevGetIPAddressBinary(peer, &peerData, &addrDataLen) < 0) + return NULL; + } else if (broadcast) { + if (virNetDevGetIPAddressBinary(broadcast, &broadcastData, + &addrDataLen) < 0) + return NULL; + } /* Get the interface index */ if ((ifindex = if_nametoindex(ifname)) == 0) Vasiliy, I can look in more detail tomorrow, but in the meantime any suggestions you have would be much appreciated. Regards, Jim

3 3

[libvirt] [PATCH v2 0/4] Final round of build fixes
by Michal Privoznik 02 Jun '16

02 Jun '16

diff to v1: - Some patches from the original patch set are pushed now - For the other ones, I've managed to work in review suggestions Michal Privoznik (4): virNetDevBridgeGet: Don't require users to virNetDevSetupControl ppc64Compute: Avoid possible NULL dereference virDomainChrGetDomainPtrsInternal: Return an integer virDomainFormatSchedDef: Avoid false positive NULL dereference src/conf/domain_conf.c | 44 ++++++++++++++++++++++++++++++-------------- src/cpu/cpu_ppc64.c | 12 ++++++------ src/util/virnetdevbridge.c | 35 ++++++++++++----------------------- 3 files changed, 48 insertions(+), 43 deletions(-) -- 2.8.3

2 10

[libvirt] Python listDefinedDomains broken?
by Marc Richter 02 Jun '16

02 Jun '16

Hi everyone, please, don't roast me if I'm asking stupid things or fail to provide info, since I'm a freshman, using libvirt. I have two servers, both running the following versions: Compiled against library: libvirt 1.3.1 Using library: libvirt 1.3.1 Using API: QEMU 1.3.1 Running hypervisor: QEMU 0.14.0 I'm trying to create a Python program, which gets a list of all defined domains for each host and give a list for domains only defined on one of these hosts. This Python script is running on a third host (my workstation), using these versions: Python: 3.5.1 libvirt-python: 1.3.1 When I'm running virsh on my workstation (version 1.3.1), like this: virsh -c "qemu+ssh://root@host1/system?keyfile=/home/user/.ssh/id_rsa_libvirt" --readonly list --all Then a domain named "winxp_ausgeschaltet" is within that list: Id Name State ---------------------------------------------------- ... - winxp_ausgeschaltet shut off When I do the same for the second host, it is also in that list: Id Name State ---------------------------------------------------- ... 7 winxp_ausgeschaltet running When I do the following in Python, this domain is listed for the first host (where it's state is "shut off"), but not for the second one: import libvirt import sys class kvmhost: def __init__(self, host, keyfile): self.conn = self.connect_kvm(host, keyfile) self.doms = sorted(self.get_doms()) def connect_kvm(self, host, keyfile): try: conn = libvirt.openReadOnly('qemu+ssh://root@' + host + '/system?keyfile=' + keyfile) except libvirt.libvirtError as lve: print('Error: ' + str(lve)) sys.exit(1) return conn def get_doms(self): try: alldoms = self.conn.listDefinedDomains() except libvirt.libvirtError as lve: print('Error: ' + str(lve)) self.conn.close() sys.exit(1) return alldoms host1 = kvmhost('host1', '/path/to/ssh_keyfile') host2 = kvmhost('host2', '/path/to/ssh_keyfile') print(host1.doms) print(host2.doms) What am I doing wrong here or: is there a missbehavior in libvirt or the python module? Using listAllDomains instead of listDefinedDomains gives: libvirt: Remote Driver error : unknown procedure: 273 What is the difference between listAllDomains and listDefinedDomains at all? Thanks for your help. Marc

3 3

[libvirt] [PATCH v2 0/2] qemu: expand domain memory statistics
by Derbyshev Dmitriy 01 Jun '16

01 Jun '16

From: Derbyshev Dmitry <dderbyshev(a)virtuozzo.com> QEMU reports timestamp and available along with other memory statistics. This information was not saved into domain statistics. Changes since v1: * Enum numeration fixed * Macro getting "usage" field fixed Derbyshev Dmitry (2): qemu: expand domain memory statistics with 'usable' qemu: expand domain memory statistics with 'last-update' timestamp include/libvirt/libvirt-domain.h | 11 ++++++++++- src/libvirt-domain.c | 5 +++++ src/qemu/qemu_monitor_json.c | 23 +++++++++++++---------- tools/virsh-domain-monitor.c | 4 ++++ 4 files changed, 32 insertions(+), 11 deletions(-) -- 1.9.5.msysgit.0

2 3