- Devel - libvirt List Archives

[libvirt] [PATCH v2] vz: implementation of domainCreateXML callback
by Mikhail Feoktistov 16 Dec '15

16 Dec '15

--- diff from v1. Remove call of vzDomainSuspend() in case of VIR_DOMAIN_START_PAUSED flag is set. Now we don't support to create instance in stopped state. src/vz/vz_driver.c | 24 ++++++++++++++++++++++++ 1 file changed, 24 insertions(+) diff --git a/src/vz/vz_driver.c b/src/vz/vz_driver.c index ea1090a..4498e01 100644 --- a/src/vz/vz_driver.c +++ b/src/vz/vz_driver.c @@ -979,6 +979,29 @@ vzDomainUndefine(virDomainPtr domain) return vzDomainUndefineFlags(domain, 0); } +static virDomainPtr +vzDomainCreateXML(virConnectPtr conn, + const char *xml, + unsigned int flags) +{ + virDomainPtr domain; + int ret; + + virCheckFlags(0, NULL); + + domain = vzDomainDefineXMLFlags(conn, xml, 0); + if (domain == NULL) + return domain; + + ret = vzDomainCreate(domain); + if (ret != 0) { + vzDomainUndefine(domain); + return NULL; + } + + return domain; +} + static int vzDomainHasManagedSaveImage(virDomainPtr domain, unsigned int flags) { @@ -1493,6 +1516,7 @@ static virHypervisorDriver vzDriver = { .domainShutdown = vzDomainShutdown, /* 0.10.0 */ .domainCreate = vzDomainCreate, /* 0.10.0 */ .domainCreateWithFlags = vzDomainCreateWithFlags, /* 1.2.10 */ + .domainCreateXML = vzDomainCreateXML, /* 1.2.22 */ .domainReboot = vzDomainReboot, /* 1.3.0 */ .domainDefineXML = vzDomainDefineXML, /* 0.10.0 */ .domainDefineXMLFlags = vzDomainDefineXMLFlags, /* 1.2.12 */ -- 1.8.3.1

2 2

[libvirt] [PATCH v3] pci: Use virPCIDeviceAddress in virPCIDevice
by Andrea Bolognani 16 Dec '15

16 Dec '15

Instead of replicating the information (domain, bus, slot, function) inside the virPCIDevice structure, use the already-existing virPCIDeviceAddress structure. For users of the module, this means that the object returned by virPCIDeviceGetAddress() can no longer be NULL and must no longer be freed by the caller. --- Changes in v3: * don't check the return value of virPCIDeviceGetAddress(), it can no longer be NULL * don't use a variable to store the device address if it's only going to be used a single time Changes in v2: * use virPCIDeviceAddress instead of virPCIDeviceAddressPtr to avoid extra allocations src/util/virhostdev.c | 20 ++--------- src/util/virpci.c | 97 +++++++++++++++++++++++---------------------------- 2 files changed, 47 insertions(+), 70 deletions(-) diff --git a/src/util/virhostdev.c b/src/util/virhostdev.c index de029a0..4065535 100644 --- a/src/util/virhostdev.c +++ b/src/util/virhostdev.c @@ -585,15 +585,12 @@ virHostdevPreparePCIDevices(virHostdevManagerPtr hostdev_mgr, goto cleanup; } - VIR_FREE(devAddr); - if (!(devAddr = virPCIDeviceGetAddress(dev))) - goto cleanup; - /* The device is in use by other active domain if * the dev is in list activePCIHostdevs. VFIO devices * belonging to same iommu group can't be shared * across guests. */ + devAddr = virPCIDeviceGetAddress(dev); if (usesVfio) { if (virPCIDeviceAddressIOMMUGroupIterate(devAddr, virHostdevIsPCINodeDeviceUsed, @@ -728,7 +725,6 @@ virHostdevPreparePCIDevices(virHostdevManagerPtr hostdev_mgr, virObjectUnlock(hostdev_mgr->activePCIHostdevs); virObjectUnlock(hostdev_mgr->inactivePCIHostdevs); virObjectUnref(pcidevs); - VIR_FREE(devAddr); return ret; } @@ -1558,7 +1554,6 @@ int virHostdevPCINodeDeviceDetach(virHostdevManagerPtr hostdev_mgr, virPCIDevicePtr pci) { - virPCIDeviceAddressPtr devAddr = NULL; struct virHostdevIsPCINodeDeviceUsedData data = { hostdev_mgr, NULL, false }; int ret = -1; @@ -1566,10 +1561,7 @@ virHostdevPCINodeDeviceDetach(virHostdevManagerPtr hostdev_mgr, virObjectLock(hostdev_mgr->activePCIHostdevs); virObjectLock(hostdev_mgr->inactivePCIHostdevs); - if (!(devAddr = virPCIDeviceGetAddress(pci))) - goto out; - - if (virHostdevIsPCINodeDeviceUsed(devAddr, &data)) + if (virHostdevIsPCINodeDeviceUsed(virPCIDeviceGetAddress(pci), &data)) goto out; if (virPCIDeviceDetach(pci, hostdev_mgr->activePCIHostdevs, @@ -1581,7 +1573,6 @@ virHostdevPCINodeDeviceDetach(virHostdevManagerPtr hostdev_mgr, out: virObjectUnlock(hostdev_mgr->inactivePCIHostdevs); virObjectUnlock(hostdev_mgr->activePCIHostdevs); - VIR_FREE(devAddr); return ret; } @@ -1589,7 +1580,6 @@ int virHostdevPCINodeDeviceReAttach(virHostdevManagerPtr hostdev_mgr, virPCIDevicePtr pci) { - virPCIDeviceAddressPtr devAddr = NULL; struct virHostdevIsPCINodeDeviceUsedData data = {hostdev_mgr, NULL, false}; int ret = -1; @@ -1597,10 +1587,7 @@ virHostdevPCINodeDeviceReAttach(virHostdevManagerPtr hostdev_mgr, virObjectLock(hostdev_mgr->activePCIHostdevs); virObjectLock(hostdev_mgr->inactivePCIHostdevs); - if (!(devAddr = virPCIDeviceGetAddress(pci))) - goto out; - - if (virHostdevIsPCINodeDeviceUsed(devAddr, &data)) + if (virHostdevIsPCINodeDeviceUsed(virPCIDeviceGetAddress(pci), &data)) goto out; virPCIDeviceReattachInit(pci); @@ -1613,7 +1600,6 @@ virHostdevPCINodeDeviceReAttach(virHostdevManagerPtr hostdev_mgr, out: virObjectUnlock(hostdev_mgr->inactivePCIHostdevs); virObjectUnlock(hostdev_mgr->activePCIHostdevs); - VIR_FREE(devAddr); return ret; } diff --git a/src/util/virpci.c b/src/util/virpci.c index 7ca3fef..6f0cb8c 100644 --- a/src/util/virpci.c +++ b/src/util/virpci.c @@ -56,10 +56,7 @@ VIR_ENUM_IMPL(virPCIELinkSpeed, VIR_PCIE_LINK_SPEED_LAST, "", "2.5", "5", "8") struct _virPCIDevice { - unsigned int domain; - unsigned int bus; - unsigned int slot; - unsigned int function; + virPCIDeviceAddress address; char name[PCI_ADDR_LEN]; /* domain:bus:slot.function */ char id[PCI_ID_LEN]; /* product vendor */ @@ -655,10 +652,10 @@ virPCIDeviceSharesBusWithActive(virPCIDevicePtr dev, virPCIDevicePtr check, void virPCIDeviceList *inactiveDevs = data; /* Different domain, different bus, or simply identical device */ - if (dev->domain != check->domain || - dev->bus != check->bus || - (dev->slot == check->slot && - dev->function == check->function)) + if (dev->address.domain != check->address.domain || + dev->address.bus != check->address.bus || + (dev->address.slot == check->address.slot && + dev->address.function == check->address.function)) return 0; /* same bus, but inactive, i.e. about to be assigned to guest */ @@ -689,7 +686,7 @@ virPCIDeviceIsParent(virPCIDevicePtr dev, virPCIDevicePtr check, void *data) int ret = 0; int fd; - if (dev->domain != check->domain) + if (dev->address.domain != check->address.domain) return 0; if ((fd = virPCIDeviceConfigOpen(check, false)) < 0) @@ -713,7 +710,7 @@ virPCIDeviceIsParent(virPCIDevicePtr dev, virPCIDevicePtr check, void *data) /* if the secondary bus exactly equals the device's bus, then we found * the direct parent. No further work is necessary */ - if (dev->bus == secondary) { + if (dev->address.bus == secondary) { ret = 1; goto cleanup; } @@ -722,10 +719,12 @@ virPCIDeviceIsParent(virPCIDevicePtr dev, virPCIDevicePtr check, void *data) * In this case, what we need to do is look for the "best" match; i.e. * the most restrictive match that still satisfies all of the conditions. */ - if (dev->bus > secondary && dev->bus <= subordinate) { + if (dev->address.bus > secondary && dev->address.bus <= subordinate) { if (*best == NULL) { - *best = virPCIDeviceNew(check->domain, check->bus, check->slot, - check->function); + *best = virPCIDeviceNew(check->address.domain, + check->address.bus, + check->address.slot, + check->address.function); if (*best == NULL) { ret = -1; goto cleanup; @@ -745,8 +744,10 @@ virPCIDeviceIsParent(virPCIDevicePtr dev, virPCIDevicePtr check, void *data) if (secondary > best_secondary) { virPCIDeviceFree(*best); - *best = virPCIDeviceNew(check->domain, check->bus, check->slot, - check->function); + *best = virPCIDeviceNew(check->address.domain, + check->address.bus, + check->address.slot, + check->address.function); if (*best == NULL) { ret = -1; goto cleanup; @@ -970,7 +971,7 @@ virPCIDeviceReset(virPCIDevicePtr dev, ret = virPCIDeviceTryPowerManagementReset(dev, fd); /* Bus reset is not an option with the root bus */ - if (ret < 0 && dev->bus != 0) + if (ret < 0 && dev->address.bus != 0) ret = virPCIDeviceTrySecondaryBusReset(dev, fd, inactiveDevs); if (ret < 0) { @@ -1483,8 +1484,8 @@ virPCIDeviceWaitForCleanup(virPCIDevicePtr dev, const char *matcher) virStrToLong_ui(tmp + 1, &tmp, 16, &function) < 0 || *tmp != '\n') continue; - if (domain != dev->domain || bus != dev->bus || slot != dev->slot || - function != dev->function) + if (domain != dev->address.domain || bus != dev->address.bus || + slot != dev->address.slot || function != dev->address.function) continue; in_matching_device = true; match_depth = strspn(line, " "); @@ -1560,17 +1561,16 @@ virPCIDeviceNew(unsigned int domain, if (VIR_ALLOC(dev) < 0) return NULL; - dev->domain = domain; - dev->bus = bus; - dev->slot = slot; - dev->function = function; + dev->address.domain = domain; + dev->address.bus = bus; + dev->address.slot = slot; + dev->address.function = function; if (snprintf(dev->name, sizeof(dev->name), "%.4x:%.2x:%.2x.%.1x", - dev->domain, dev->bus, dev->slot, - dev->function) >= sizeof(dev->name)) { + domain, bus, slot, function) >= sizeof(dev->name)) { virReportError(VIR_ERR_INTERNAL_ERROR, _("dev->name buffer overflow: %.4x:%.2x:%.2x.%.1x"), - dev->domain, dev->bus, dev->slot, dev->function); + domain, bus, slot, function); goto error; } if (virAsprintf(&dev->path, PCI_SYSFS "devices/%s/config", @@ -1661,25 +1661,14 @@ virPCIDeviceFree(virPCIDevicePtr dev) * @dev: device to get address from * * Take a PCI device on input and return its PCI address. The - * caller must free the returned value when no longer needed. + * returned object is owned by the device and must not be freed. * - * Returns NULL on failure, the device address on success. + * Returns: a pointer to the address, which can never be NULL. */ virPCIDeviceAddressPtr virPCIDeviceGetAddress(virPCIDevicePtr dev) { - - virPCIDeviceAddressPtr pciAddrPtr; - - if (!dev || (VIR_ALLOC(pciAddrPtr) < 0)) - return NULL; - - pciAddrPtr->domain = dev->domain; - pciAddrPtr->bus = dev->bus; - pciAddrPtr->slot = dev->slot; - pciAddrPtr->function = dev->function; - - return pciAddrPtr; + return &(dev->address); } const char * @@ -1890,12 +1879,14 @@ virPCIDeviceListFindIndex(virPCIDeviceListPtr list, virPCIDevicePtr dev) { size_t i; - for (i = 0; i < list->count; i++) - if (list->devs[i]->domain == dev->domain && - list->devs[i]->bus == dev->bus && - list->devs[i]->slot == dev->slot && - list->devs[i]->function == dev->function) + for (i = 0; i < list->count; i++) { + virPCIDevicePtr other = list->devs[i]; + if (other->address.domain == dev->address.domain && + other->address.bus == dev->address.bus && + other->address.slot == dev->address.slot && + other->address.function == dev->address.function) return i; + } return -1; } @@ -1910,10 +1901,11 @@ virPCIDeviceListFindByIDs(virPCIDeviceListPtr list, size_t i; for (i = 0; i < list->count; i++) { - if (list->devs[i]->domain == domain && - list->devs[i]->bus == bus && - list->devs[i]->slot == slot && - list->devs[i]->function == function) + virPCIDevicePtr other = list->devs[i]; + if (other->address.domain == domain && + other->address.bus == bus && + other->address.slot == slot && + other->address.function == function) return list->devs[i]; } return NULL; @@ -1944,7 +1936,8 @@ int virPCIDeviceFileIterate(virPCIDevicePtr dev, int direrr; if (virAsprintf(&pcidir, "/sys/bus/pci/devices/%04x:%02x:%02x.%x", - dev->domain, dev->bus, dev->slot, dev->function) < 0) + dev->address.domain, dev->address.bus, + dev->address.slot, dev->address.function) < 0) goto cleanup; if (!(dir = opendir(pcidir))) { @@ -2074,13 +2067,11 @@ virPCIDeviceListPtr virPCIDeviceGetIOMMUGroupList(virPCIDevicePtr dev) { virPCIDeviceListPtr groupList = virPCIDeviceListNew(); - virPCIDeviceAddress devAddr = { dev->domain, dev->bus, - dev->slot, dev->function }; if (!groupList) goto error; - if (virPCIDeviceAddressIOMMUGroupIterate(&devAddr, + if (virPCIDeviceAddressIOMMUGroupIterate(&(dev->address), virPCIDeviceGetIOMMUGroupAddOne, groupList) < 0) goto error; @@ -2294,7 +2285,7 @@ virPCIDeviceIsBehindSwitchLackingACS(virPCIDevicePtr dev) * into play since devices on the root bus can't P2P without going * through the root IOMMU. */ - if (dev->bus == 0) { + if (dev->address.bus == 0) { return 0; } else { virReportError(VIR_ERR_INTERNAL_ERROR, -- 2.5.0

2 2

[libvirt] [PATCH v3 0/8] libxl: domain statistics support
by Joao Martins 15 Dec '15

15 Dec '15

Hey, [ Apologies for resending this series before receiving comments, but I found two issues (namely on patch 3 and 4) and thus I am reposting it as v3 having those fixed. ] This series bring support for various statistics about domains regarding CPU, Memory, Network Interfaces, Block and Jobs. Not all of the statistics are implemented: qdisk support is missing in this series and some of the memory statistics aren't available. With this series we further implement 7 more functions of libvirt APIs. It is organized as follows: * Patch 1, 2: implements cpu/memory statistics. * Patch 3: implements network statistics * Patch 4: adds helper method virDiskNameParse as an extension to virDiskNameToIndex (New) * Patch 5: VBD block statistics. QDisk will follow up in a separate series regarding QEMU monitor integration. * Patch 6: implement fetching all domain statistics * Patch 7, 8: implements Job information statistics. Overall it looks big but 70% of the patch is due to #5 and #6 but doesn't add necessarily more complexity to the driver I believe. Patch #7 and #8 are of special importance because GetJobInfo and GetJobStats are now used in Openstack Kilo to monitor live-migration progress. These two patches together with the p2p migration I sent earlier let us sucessfully live-migrate with Openstack Kilo. Furthermore with this series we get to support nova diagnostics. Tested this series on 4.4.3 and 4.5 setups plus Openstack Kilo. Individual patches contain the changelog and comments addressed since v1. Thanks! Joao Martins (8): libxl: implement virDomainGetCPUStats libxl: implement virDomainMemorystats libxl: implement virDomainInterfaceStats util: add virDiskNameParse to handle disk and partition idx libxl: implement virDomainBlockStats libxl: implement virConnectGetAllDomainStats libxl: implement virDomainGetJobInfo libxl: implement virDomainGetJobStats configure.ac | 2 +- src/libvirt_private.syms | 1 + src/libxl/libxl_domain.c | 54 +++ src/libxl/libxl_domain.h | 6 + src/libxl/libxl_driver.c | 960 +++++++++++++++++++++++++++++++++++++++++++++++ src/util/virutil.c | 41 +- src/util/virutil.h | 1 + tests/utiltest.c | 56 +++ 8 files changed, 1116 insertions(+), 5 deletions(-) -- 2.1.4

3 29

[libvirt] [PATCH v2] pci: Use virPCIDeviceAddress in virPCIDevice
by Andrea Bolognani 15 Dec '15

15 Dec '15

Instead of replicating the information (domain, bus, slot, function) inside the virPCIDevice structure, use the already-existing virPCIDeviceAddress structure. Outside of the module, the only visible change is that the return value of virPCIDeviceGetAddress() must no longer be freed by the caller. --- Changes in v2: * use virPCIDeviceAddress instead of virPCIDeviceAddressPtr to avoid extra allocations src/util/virhostdev.c | 4 --- src/util/virpci.c | 95 +++++++++++++++++++++++---------------------------- 2 files changed, 43 insertions(+), 56 deletions(-) diff --git a/src/util/virhostdev.c b/src/util/virhostdev.c index de029a0..173ac58 100644 --- a/src/util/virhostdev.c +++ b/src/util/virhostdev.c @@ -585,7 +585,6 @@ virHostdevPreparePCIDevices(virHostdevManagerPtr hostdev_mgr, goto cleanup; } - VIR_FREE(devAddr); if (!(devAddr = virPCIDeviceGetAddress(dev))) goto cleanup; @@ -728,7 +727,6 @@ virHostdevPreparePCIDevices(virHostdevManagerPtr hostdev_mgr, virObjectUnlock(hostdev_mgr->activePCIHostdevs); virObjectUnlock(hostdev_mgr->inactivePCIHostdevs); virObjectUnref(pcidevs); - VIR_FREE(devAddr); return ret; } @@ -1581,7 +1579,6 @@ virHostdevPCINodeDeviceDetach(virHostdevManagerPtr hostdev_mgr, out: virObjectUnlock(hostdev_mgr->inactivePCIHostdevs); virObjectUnlock(hostdev_mgr->activePCIHostdevs); - VIR_FREE(devAddr); return ret; } @@ -1613,7 +1610,6 @@ virHostdevPCINodeDeviceReAttach(virHostdevManagerPtr hostdev_mgr, out: virObjectUnlock(hostdev_mgr->inactivePCIHostdevs); virObjectUnlock(hostdev_mgr->activePCIHostdevs); - VIR_FREE(devAddr); return ret; } diff --git a/src/util/virpci.c b/src/util/virpci.c index bececb5..bb874ac 100644 --- a/src/util/virpci.c +++ b/src/util/virpci.c @@ -56,10 +56,7 @@ VIR_ENUM_IMPL(virPCIELinkSpeed, VIR_PCIE_LINK_SPEED_LAST, "", "2.5", "5", "8") struct _virPCIDevice { - unsigned int domain; - unsigned int bus; - unsigned int slot; - unsigned int function; + virPCIDeviceAddress address; char name[PCI_ADDR_LEN]; /* domain:bus:slot.function */ char id[PCI_ID_LEN]; /* product vendor */ @@ -655,10 +652,10 @@ virPCIDeviceSharesBusWithActive(virPCIDevicePtr dev, virPCIDevicePtr check, void virPCIDeviceList *inactiveDevs = data; /* Different domain, different bus, or simply identical device */ - if (dev->domain != check->domain || - dev->bus != check->bus || - (dev->slot == check->slot && - dev->function == check->function)) + if (dev->address.domain != check->address.domain || + dev->address.bus != check->address.bus || + (dev->address.slot == check->address.slot && + dev->address.function == check->address.function)) return 0; /* same bus, but inactive, i.e. about to be assigned to guest */ @@ -689,7 +686,7 @@ virPCIDeviceIsParent(virPCIDevicePtr dev, virPCIDevicePtr check, void *data) int ret = 0; int fd; - if (dev->domain != check->domain) + if (dev->address.domain != check->address.domain) return 0; if ((fd = virPCIDeviceConfigOpen(check, false)) < 0) @@ -713,7 +710,7 @@ virPCIDeviceIsParent(virPCIDevicePtr dev, virPCIDevicePtr check, void *data) /* if the secondary bus exactly equals the device's bus, then we found * the direct parent. No further work is necessary */ - if (dev->bus == secondary) { + if (dev->address.bus == secondary) { ret = 1; goto cleanup; } @@ -722,10 +719,12 @@ virPCIDeviceIsParent(virPCIDevicePtr dev, virPCIDevicePtr check, void *data) * In this case, what we need to do is look for the "best" match; i.e. * the most restrictive match that still satisfies all of the conditions. */ - if (dev->bus > secondary && dev->bus <= subordinate) { + if (dev->address.bus > secondary && dev->address.bus <= subordinate) { if (*best == NULL) { - *best = virPCIDeviceNew(check->domain, check->bus, check->slot, - check->function); + *best = virPCIDeviceNew(check->address.domain, + check->address.bus, + check->address.slot, + check->address.function); if (*best == NULL) { ret = -1; goto cleanup; @@ -745,8 +744,10 @@ virPCIDeviceIsParent(virPCIDevicePtr dev, virPCIDevicePtr check, void *data) if (secondary > best_secondary) { virPCIDeviceFree(*best); - *best = virPCIDeviceNew(check->domain, check->bus, check->slot, - check->function); + *best = virPCIDeviceNew(check->address.domain, + check->address.bus, + check->address.slot, + check->address.function); if (*best == NULL) { ret = -1; goto cleanup; @@ -970,7 +971,7 @@ virPCIDeviceReset(virPCIDevicePtr dev, ret = virPCIDeviceTryPowerManagementReset(dev, fd); /* Bus reset is not an option with the root bus */ - if (ret < 0 && dev->bus != 0) + if (ret < 0 && dev->address.bus != 0) ret = virPCIDeviceTrySecondaryBusReset(dev, fd, inactiveDevs); if (ret < 0) { @@ -1483,8 +1484,8 @@ virPCIDeviceWaitForCleanup(virPCIDevicePtr dev, const char *matcher) virStrToLong_ui(tmp + 1, &tmp, 16, &function) < 0 || *tmp != '\n') continue; - if (domain != dev->domain || bus != dev->bus || slot != dev->slot || - function != dev->function) + if (domain != dev->address.domain || bus != dev->address.bus || + slot != dev->address.slot || function != dev->address.function) continue; in_matching_device = true; match_depth = strspn(line, " "); @@ -1560,17 +1561,16 @@ virPCIDeviceNew(unsigned int domain, if (VIR_ALLOC(dev) < 0) return NULL; - dev->domain = domain; - dev->bus = bus; - dev->slot = slot; - dev->function = function; + dev->address.domain = domain; + dev->address.bus = bus; + dev->address.slot = slot; + dev->address.function = function; if (snprintf(dev->name, sizeof(dev->name), "%.4x:%.2x:%.2x.%.1x", - dev->domain, dev->bus, dev->slot, - dev->function) >= sizeof(dev->name)) { + domain, bus, slot, function) >= sizeof(dev->name)) { virReportError(VIR_ERR_INTERNAL_ERROR, _("dev->name buffer overflow: %.4x:%.2x:%.2x.%.1x"), - dev->domain, dev->bus, dev->slot, dev->function); + domain, bus, slot, function); goto error; } if (virAsprintf(&dev->path, PCI_SYSFS "devices/%s/config", @@ -1661,25 +1661,14 @@ virPCIDeviceFree(virPCIDevicePtr dev) * @dev: device to get address from * * Take a PCI device on input and return its PCI address. The - * caller must free the returned value when no longer needed. + * returned object is owned by the device and must not be freed. * * Returns NULL on failure, the device address on success. */ virPCIDeviceAddressPtr virPCIDeviceGetAddress(virPCIDevicePtr dev) { - - virPCIDeviceAddressPtr pciAddrPtr; - - if (!dev || (VIR_ALLOC(pciAddrPtr) < 0)) - return NULL; - - pciAddrPtr->domain = dev->domain; - pciAddrPtr->bus = dev->bus; - pciAddrPtr->slot = dev->slot; - pciAddrPtr->function = dev->function; - - return pciAddrPtr; + return &(dev->address); } const char * @@ -1890,12 +1879,14 @@ virPCIDeviceListFindIndex(virPCIDeviceListPtr list, virPCIDevicePtr dev) { size_t i; - for (i = 0; i < list->count; i++) - if (list->devs[i]->domain == dev->domain && - list->devs[i]->bus == dev->bus && - list->devs[i]->slot == dev->slot && - list->devs[i]->function == dev->function) + for (i = 0; i < list->count; i++) { + virPCIDevicePtr other = list->devs[i]; + if (other->address.domain == dev->address.domain && + other->address.bus == dev->address.bus && + other->address.slot == dev->address.slot && + other->address.function == dev->address.function) return i; + } return -1; } @@ -1910,10 +1901,11 @@ virPCIDeviceListFindByIDs(virPCIDeviceListPtr list, size_t i; for (i = 0; i < list->count; i++) { - if (list->devs[i]->domain == domain && - list->devs[i]->bus == bus && - list->devs[i]->slot == slot && - list->devs[i]->function == function) + virPCIDevicePtr other = list->devs[i]; + if (other->address.domain == domain && + other->address.bus == bus && + other->address.slot == slot && + other->address.function == function) return list->devs[i]; } return NULL; @@ -1944,7 +1936,8 @@ int virPCIDeviceFileIterate(virPCIDevicePtr dev, int direrr; if (virAsprintf(&pcidir, "/sys/bus/pci/devices/%04x:%02x:%02x.%x", - dev->domain, dev->bus, dev->slot, dev->function) < 0) + dev->address.domain, dev->address.bus, + dev->address.slot, dev->address.function) < 0) goto cleanup; if (!(dir = opendir(pcidir))) { @@ -2074,13 +2067,11 @@ virPCIDeviceListPtr virPCIDeviceGetIOMMUGroupList(virPCIDevicePtr dev) { virPCIDeviceListPtr groupList = virPCIDeviceListNew(); - virPCIDeviceAddress devAddr = { dev->domain, dev->bus, - dev->slot, dev->function }; if (!groupList) goto error; - if (virPCIDeviceAddressIOMMUGroupIterate(&devAddr, + if (virPCIDeviceAddressIOMMUGroupIterate(&(dev->address), virPCIDeviceGetIOMMUGroupAddOne, groupList) < 0) goto error; @@ -2294,7 +2285,7 @@ virPCIDeviceIsBehindSwitchLackingACS(virPCIDevicePtr dev) * into play since devices on the root bus can't P2P without going * through the root IOMMU. */ - if (dev->bus == 0) { + if (dev->address.bus == 0) { return 0; } else { virReportError(VIR_ERR_INTERNAL_ERROR, -- 2.5.0

2 3

[libvirt] [PATCH 0/2] notify about reverting to snapshot
by Dmitry Andreev 15 Dec '15

15 Dec '15

Reverting to snapshot may change domain configuration but there will be no events about that. Lack of the event become a problem for virt-manager https://bugzilla.redhat.com/show_bug.cgi?id=1081148 This patch-set introduces new event and emits it in qemuDomainRevertToSnapshot. Dmitry Andreev (2): Introduce new VIR_DOMAIN_EVENT_DEFINED_FROM_SNAPSHOT event qemu: emit 'defined' event after reverted to snapshot examples/object-events/event-test.c | 2 ++ include/libvirt/libvirt-domain.h | 1 + src/qemu/qemu_driver.c | 5 +++++ tools/virsh-domain.c | 3 ++- 4 files changed, 10 insertions(+), 1 deletion(-) -- 1.8.3.1

2 5

[libvirt] libvirt qcow2 internal snapshot without vm state
by Vasiliy Tolstov 15 Dec '15

15 Dec '15

Hi! I'm try to grep sources, but can't understand, how libvirt create internal snapshot in the same qcow2 file without memory state? As i understand libvirt via qmp/hmp runs savevm command via qemu. But i don't understand how libvirt skips memory state in this case? -- Vasiliy Tolstov, e-mail: v.tolstov(a)selfip.ru

2 2

Re: [libvirt] [Qemu-devel] [Patch V2 1/2] x86, mce: Basic support to add LMCE support to QEMU
by Eduardo Habkost 15 Dec '15

15 Dec '15

On Mon, Dec 14, 2015 at 07:17:27PM -0500, Raj, Ashok wrote: > On Mon, Dec 14, 2015 at 11:37:16PM +0100, Borislav Petkov wrote: > > On Mon, Dec 14, 2015 at 02:11:46PM -0500, Raj, Ashok wrote: > > > This is mostly harmless.. since the MCG_CAP space is shared and has no > > > conflict between vendors. Also just the CAP being set has no effect. > > > > Of course it does - we check SER_P in machine_check_poll() and when > > I emulate an AMD guest and inject errors into it, error handling is > > obviously wrong, see: > > > > https://lkml.kernel.org/r/20151123150355.GE5134@pd.tnic > > > > I can see how this hurts.. since the poller isn't doing cpu model specific > stuff..? > > in the LMCE case, even if you advertise MCG_LMCE_P in MCG_CAP, the guest kernel > wont call intel_init_lmce() only from mce_intel.c.. so the same problem > won't happen. > > but the issue Eduardo mentioned seems like the following. > > New QEMU_LMCE + New KVM_LMCE + New_GUEST_LMCE - No problem > > but if you were to migrage the Guest_LMCE to a non-LMCE supported KVM host > we could run into an issue.. > > is this the compatibility issue that you were looking to fix Eduardo? If I understood you correctly, yes. Also, note that currently kvm_arch_init_vcpu() simply warns about missing capabilities, instead of preventing the VM from running/migrating (as it should). We need to change that, and figure out a good way to report "feature FOO can't be enabled in this host" errors to management software[1]. The main problem is that we don't even have a QMP console available anymore if machine initialization is aborted. CCing libvir-list so they get in the loop. [1] This is similar to what we need for CPUID checks, but the new MCE feature means we need something more generic (that just reports QOM property names, probably?) -- Eduardo

1 0

[libvirt] [PATCHv2 0/3] security: misc cleanups
by Ján Tomko 15 Dec '15

15 Dec '15

v1: https://www.redhat.com/archives/libvir-list/2015-November/msg00781.html new in v2: * split by security driver * more functions renamed Ján Tomko (3): security_dac: remove extra Security from function names security_selinux: remove extra Security from function names security_stack: : remove extra security from function names src/security/security_dac.c | 242 +++++++++++++------------- src/security/security_selinux.c | 375 ++++++++++++++++++++-------------------- src/security/security_stack.c | 68 ++++---- 3 files changed, 338 insertions(+), 347 deletions(-) -- 2.4.6

2 5

[libvirt] volume download/upload but for snapshots
by Vasiliy Tolstov 15 Dec '15

15 Dec '15

Hi! I'm happy with libvirt volume download/upload ability, but also i need to download snapshot, what is the best way to do that? Define volume from snapshot? Or write libvirt function for this? -- Vasiliy Tolstov, e-mail: v.tolstov(a)selfip.ru

2 3

[libvirt] [PATCH v2 0/2] Add support for zero-write detection
by Martin Kletzander 15 Dec '15

15 Dec '15

QEMU supports detect-zeroes option since version 2.1, but we never added support for it in libvirt. If was requested by Vasiliy Tolstov in the list, so I just added it. There are two discussions to be had, optionally. One is to decide whether we should disable detect_zeros='unmap' if discard is not set to 'unmap', but this is getting very hypervisor-specific, so I just documented the behaviour. The other one is the naming. I described why I made the decision for "zeros" instead of "zeroes" the decision in the patch, but I have no problem changing it to what others like better. v2: - format detect_zeroes on the command line instead of detect_zeros v1: - https://www.redhat.com/archives/libvir-list/2015-December/msg00484.html Martin Kletzander (2): conf: Add support of zero-detection for disks qemu: Add support for zero-detection writes docs/formatdomain.html.in | 10 ++++++ docs/schemas/domaincommon.rng | 12 +++++++ src/conf/domain_conf.c | 23 +++++++++++++- src/conf/domain_conf.h | 11 +++++++ src/libvirt_private.syms | 2 ++ src/qemu/qemu_capabilities.c | 2 ++ src/qemu/qemu_capabilities.h | 1 + src/qemu/qemu_command.c | 11 +++++++ tests/qemucapabilitiesdata/caps_2.1.1-1.caps | 1 + tests/qemucapabilitiesdata/caps_2.4.0-1.caps | 1 + tests/qemucapabilitiesdata/caps_2.5.0-1.caps | 1 + .../qemuxml2argv-disk-drive-detect-zeros.args | 27 ++++++++++++++++ .../qemuxml2argv-disk-drive-detect-zeros.xml | 37 ++++++++++++++++++++++ tests/qemuxml2argvtest.c | 4 +++ tests/qemuxml2xmltest.c | 1 + 15 files changed, 143 insertions(+), 1 deletion(-) create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-disk-drive-detect-zeros.args create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-disk-drive-detect-zeros.xml -- 2.6.4

3 7

[libvirt] Exclusive VM lock
by Dmitry Andreev 15 Dec '15

15 Dec '15

Hi, I'm looking for a mechanism to do a sequence of API calls as atomic operation. Is there any way for libvirt's API client to acquire an exclusive VM lock to prevent other client from changing VM state through libvirt?

3 5

[libvirt] [PATCH] qemu: Warn when using vhost-user without shared memory
by Martin Kletzander 15 Dec '15

15 Dec '15

When user configures vhost-user interface and forgets to also configure any shared memory, the search for the root cause of non-operational interface might take unpleasantly long time. Let's enhance user experience by emitting a warning in the logs. Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=1266982 Signed-off-by: Martin Kletzander <mkletzan(a)redhat.com> --- src/qemu/qemu_process.c | 44 ++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 44 insertions(+) diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c index 420196264685..fb471342e790 100644 --- a/src/qemu/qemu_process.c +++ b/src/qemu/qemu_process.c @@ -4542,6 +4542,7 @@ qemuProcessLaunch(virConnectPtr conn, unsigned int hostdev_flags = 0; size_t nnicindexes = 0; int *nicindexes = NULL; + bool check_shmem = false; VIR_DEBUG("vm=%p name=%s id=%d asyncJob=%d " "incoming.launchURI=%s incoming.deferredURI=%s " @@ -4749,6 +4750,49 @@ qemuProcessLaunch(virConnectPtr conn, goto cleanup; } + VIR_DEBUG("Checking for any possible (non-fatal) issues"); + + /* + * For vhost-user to work, the domain has to have some type of + * shared memory configured. We're not the proper once to judge + * whether shared hugepages or shm are enough and will be in the + * future, so we'll just warn in case there is none of that + * configured. Moreover failing would give the false illusion + * that libvirt is really checking that everything works before + * running the domain and not only we are unable to do that, but + * it's also not our aim to do so. + */ + for (i = 0; i < vm->def->nnets; i++) { + if (virDomainNetGetActualType(vm->def->nets[i]) == + VIR_DOMAIN_NET_TYPE_VHOSTUSER) { + check_shmem = true; + break; + } + } + + if (check_shmem) { + bool shmem = vm->def->nshmems; + + /* + * This check is by no means complete. We merely check + * whetere there are *some* hugepages enabled and *some* NUMA + * nodes with shared memory access. + */ + if (!shmem && vm->def->mem.nhugepages) { + for (i = 0; i < virDomainNumaGetNodeCount(vm->def->numa); i++) { + if (virDomainNumaGetNodeMemoryAccessMode(vm->def->numa, i) == + VIR_NUMA_MEM_ACCESS_SHARED) + shmem = true; + break; + } + } + + if (!shmem) { + VIR_WARN("Detected vhost-user interface without any shared memory. " + "The interface might not be operational"); + } + } + VIR_DEBUG("Building emulator command line"); if (!(cmd = qemuBuildCommandLine(conn, driver, vm->def, priv->monConfig, priv->monJSON, priv->qemuCaps, -- 2.6.3

3 7

[libvirt] [PATCH] qemuMonitorJSONEjectMedia: don't stringify the replay at all
by Pavel Hrdina 15 Dec '15

15 Dec '15

Commit 256496e1 introduced a detection if "is locked" in error replay from qemu monitor. Commit c4073657 fixed a memory leak, but it was pointed out by Peter, that this could be done cleaner without stringifing the replay. Signed-off-by: Pavel Hrdina <phrdina(a)redhat.com> --- src/qemu/qemu_monitor_json.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-) diff --git a/src/qemu/qemu_monitor_json.c b/src/qemu/qemu_monitor_json.c index 4ae63cf..d4b6514 100644 --- a/src/qemu/qemu_monitor_json.c +++ b/src/qemu/qemu_monitor_json.c @@ -2223,11 +2223,12 @@ int qemuMonitorJSONEjectMedia(qemuMonitorPtr mon, ret = qemuMonitorJSONCheckError(cmd, reply); if (ret < 0) { - char *replyStr = virJSONValueToString(reply, false); - - if (c_strcasestr(replyStr, "is locked")) - ret = -2; - VIR_FREE(replyStr); + virJSONValuePtr error = virJSONValueObjectGet(reply, "error"); + if (error) { + const char *errorStr = virJSONValueObjectGetString(error, "desc"); + if (errorStr && c_strcasestr(errorStr, "is locked")) + ret = -2; + } } virJSONValueFree(cmd); -- 2.6.4

2 2

[libvirt] qemu logging/tracing option in domain xml
by Dmitry Andreev 15 Dec '15

15 Dec '15

Hi all, QEMU has the '-d' command line option to configure logs/traces. As I can see, <qemu:commandline> is the only way to set it in domain xml. <qemu:commandline> makes a configuration tainted and doesn't allow to update tracing/logging configuration in runtime. The ability to update configuration in runtime will be very useful for us in Virtuozzo for QEMU problem investigation and development. So I want to ask you will you mind if I prepare a patch-set that introduces a new section for domain xml with next scheme: <element name="debug"> <zeroOrMore> <element name="log"> <attribute name="name"> <ref name="filter-param-value"/> </attribute> </element> </zeroOrMore> </element> Is this acceptable?

3 2

[libvirt] [PATCH] pci: Use 'addr' instead of 'dev' for virPCIDeviceAddressPtr
by Andrea Bolognani 15 Dec '15

15 Dec '15

The name 'dev' is more appropriate for virPCIDevicePtr. --- src/util/virpci.c | 7 ++++--- src/util/virpci.h | 4 ++-- 2 files changed, 6 insertions(+), 5 deletions(-) diff --git a/src/util/virpci.c b/src/util/virpci.c index bb874ac..9cc4613 100644 --- a/src/util/virpci.c +++ b/src/util/virpci.c @@ -2655,12 +2655,13 @@ virPCIGetSysfsFile(char *virPCIDeviceName, char **pci_sysfs_device_link) } int -virPCIDeviceAddressGetSysfsFile(virPCIDeviceAddressPtr dev, +virPCIDeviceAddressGetSysfsFile(virPCIDeviceAddressPtr addr, char **pci_sysfs_device_link) { if (virAsprintf(pci_sysfs_device_link, - PCI_SYSFS "devices/%04x:%02x:%02x.%x", dev->domain, - dev->bus, dev->slot, dev->function) < 0) + PCI_SYSFS "devices/%04x:%02x:%02x.%x", + addr->domain, addr->bus, + addr->slot, addr->function) < 0) return -1; return 0; } diff --git a/src/util/virpci.h b/src/util/virpci.h index 6516f05..f3d5676 100644 --- a/src/util/virpci.h +++ b/src/util/virpci.h @@ -159,7 +159,7 @@ virPCIDeviceListPtr virPCIDeviceGetIOMMUGroupList(virPCIDevicePtr dev); int virPCIDeviceAddressGetIOMMUGroupAddresses(virPCIDeviceAddressPtr devAddr, virPCIDeviceAddressPtr **iommuGroupDevices, size_t *nIommuGroupDevices); -int virPCIDeviceAddressGetIOMMUGroupNum(virPCIDeviceAddressPtr dev); +int virPCIDeviceAddressGetIOMMUGroupNum(virPCIDeviceAddressPtr addr); char *virPCIDeviceGetIOMMUGroupDev(virPCIDevicePtr dev); int virPCIDeviceIsAssignable(virPCIDevicePtr dev, @@ -180,7 +180,7 @@ int virPCIGetVirtualFunctionIndex(const char *pf_sysfs_device_link, const char *vf_sysfs_device_link, int *vf_index); -int virPCIDeviceAddressGetSysfsFile(virPCIDeviceAddressPtr dev, +int virPCIDeviceAddressGetSysfsFile(virPCIDeviceAddressPtr addr, char **pci_sysfs_device_link); int virPCIGetNetName(char *device_link_sysfs_path, char **netname); -- 2.5.0

2 2

[libvirt] [PATCH 0/3] vz: allow to create CT from template and boot VM from cdrom
by Mikhail Feoktistov 15 Dec '15

15 Dec '15

Mikhail Feoktistov (3): vz: allow to create container based on template vz: allow to boot VM from cdrom vz: implementation of boot device order src/vz/vz_sdk.c | 174 ++++++++++++++++++++++++++++++++++++++++---------------- 1 file changed, 125 insertions(+), 49 deletions(-) -- 1.8.3.1

1 3

[libvirt] [PATCH] qemuMonitorJSONEjectMedia: Don't leak stringified reply
by Michal Privoznik 15 Dec '15

15 Dec '15

The return value of virJSONValueToString() should be freed when no longer needed. This is not the case after 256496e1. ==26902== 138 bytes in 2 blocks are definitely lost in loss record 1,051 of 1,239 ==26902== at 0x4C29F80: malloc (in /usr/lib64/valgrind/vgpreload_memcheck-amd64-linux.so) ==26902== by 0xAA5F599: strdup (in /lib64/libc-2.21.so) ==26902== by 0x552BAD9: virStrdup (virstring.c:726) ==26902== by 0x54F60A7: virJSONValueToString (virjson.c:1790) ==26902== by 0x1DF6EBB9: qemuMonitorJSONEjectMedia (qemu_monitor_json.c:2225) ==26902== by 0x1DF57A4C: qemuMonitorEjectMedia (qemu_monitor.c:1985) ==26902== by 0x1DF1EF2D: qemuDomainChangeEjectableMedia (qemu_hotplug.c:199) ==26902== by 0x1DF90314: qemuDomainChangeDiskLive (qemu_driver.c:7985) ==26902== by 0x1DF90476: qemuDomainUpdateDeviceLive (qemu_driver.c:8030) ==26902== by 0x1DF91ED7: qemuDomainUpdateDeviceFlags (qemu_driver.c:8677) ==26902== by 0x561785F: virDomainUpdateDeviceFlags (libvirt-domain.c:8559) ==26902== by 0x134210: remoteDispatchDomainUpdateDeviceFlags (remote_dispatch.h:10966) ==26902== 106 bytes in 1 blocks are definitely lost in loss record 1,033 of 1,239 ==26902== at 0x4C29F80: malloc (in /usr/lib64/valgrind/vgpreload_memcheck-amd64-linux.so) ==26902== by 0xAA5F599: strdup (in /lib64/libc-2.21.so) ==26902== by 0x552BAD9: virStrdup (virstring.c:726) ==26902== by 0x54F60A7: virJSONValueToString (virjson.c:1790) ==26902== by 0x1DF6EC0C: qemuMonitorJSONEjectMedia (qemu_monitor_json.c:2227) ==26902== by 0x1DF57A4C: qemuMonitorEjectMedia (qemu_monitor.c:1985) ==26902== by 0x1DF1EF2D: qemuDomainChangeEjectableMedia (qemu_hotplug.c:199) ==26902== by 0x1DF90314: qemuDomainChangeDiskLive (qemu_driver.c:7985) ==26902== by 0x1DF90476: qemuDomainUpdateDeviceLive (qemu_driver.c:8030) ==26902== by 0x1DF91ED7: qemuDomainUpdateDeviceFlags (qemu_driver.c:8677) ==26902== by 0x561785F: virDomainUpdateDeviceFlags (libvirt-domain.c:8559) ==26902== by 0x134210: remoteDispatchDomainUpdateDeviceFlags (remote_dispatch.h:10966) Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com> --- src/qemu/qemu_monitor_json.c | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-) diff --git a/src/qemu/qemu_monitor_json.c b/src/qemu/qemu_monitor_json.c index 84c0be2..4ae63cf 100644 --- a/src/qemu/qemu_monitor_json.c +++ b/src/qemu/qemu_monitor_json.c @@ -2222,10 +2222,13 @@ int qemuMonitorJSONEjectMedia(qemuMonitorPtr mon, if (ret == 0) ret = qemuMonitorJSONCheckError(cmd, reply); - VIR_DEBUG("%s", virJSONValueToString(reply, false)); + if (ret < 0) { + char *replyStr = virJSONValueToString(reply, false); - if (ret < 0 && c_strcasestr(virJSONValueToString(reply, false), "is locked")) - ret = -2; + if (c_strcasestr(replyStr, "is locked")) + ret = -2; + VIR_FREE(replyStr); + } virJSONValueFree(cmd); virJSONValueFree(reply); -- 2.4.10

3 2

[libvirt] [PATCH] pci: Use virPCIDeviceAddress in virPCIDevice
by Andrea Bolognani 15 Dec '15

15 Dec '15

Instead of replicating the information (domain, bus, slot, function) inside the virPCIDevice structure, use the already-existing virPCIDeviceAddress structure. Outside of the module, the only visible change is that the return value of virPCIDeviceGetAddress() must no longer be freed by the caller. --- Suggested by Laine[1] in the context of a patch series; since the idea is not really tied to the series, sending as a stand-alone cleanup. [1] https://www.redhat.com/archives/libvir-list/2015-December/msg00125.html src/util/virhostdev.c | 4 -- src/util/virpci.c | 182 ++++++++++++++++++++++++++++++++++---------------- src/util/virpci.h | 7 ++ 3 files changed, 133 insertions(+), 60 deletions(-) diff --git a/src/util/virhostdev.c b/src/util/virhostdev.c index de029a0..173ac58 100644 --- a/src/util/virhostdev.c +++ b/src/util/virhostdev.c @@ -585,7 +585,6 @@ virHostdevPreparePCIDevices(virHostdevManagerPtr hostdev_mgr, goto cleanup; } - VIR_FREE(devAddr); if (!(devAddr = virPCIDeviceGetAddress(dev))) goto cleanup; @@ -728,7 +727,6 @@ virHostdevPreparePCIDevices(virHostdevManagerPtr hostdev_mgr, virObjectUnlock(hostdev_mgr->activePCIHostdevs); virObjectUnlock(hostdev_mgr->inactivePCIHostdevs); virObjectUnref(pcidevs); - VIR_FREE(devAddr); return ret; } @@ -1581,7 +1579,6 @@ virHostdevPCINodeDeviceDetach(virHostdevManagerPtr hostdev_mgr, out: virObjectUnlock(hostdev_mgr->inactivePCIHostdevs); virObjectUnlock(hostdev_mgr->activePCIHostdevs); - VIR_FREE(devAddr); return ret; } @@ -1613,7 +1610,6 @@ virHostdevPCINodeDeviceReAttach(virHostdevManagerPtr hostdev_mgr, out: virObjectUnlock(hostdev_mgr->inactivePCIHostdevs); virObjectUnlock(hostdev_mgr->activePCIHostdevs); - VIR_FREE(devAddr); return ret; } diff --git a/src/util/virpci.c b/src/util/virpci.c index bececb5..2818bf7 100644 --- a/src/util/virpci.c +++ b/src/util/virpci.c @@ -56,10 +56,7 @@ VIR_ENUM_IMPL(virPCIELinkSpeed, VIR_PCIE_LINK_SPEED_LAST, "", "2.5", "5", "8") struct _virPCIDevice { - unsigned int domain; - unsigned int bus; - unsigned int slot; - unsigned int function; + virPCIDeviceAddressPtr address; char name[PCI_ADDR_LEN]; /* domain:bus:slot.function */ char id[PCI_ID_LEN]; /* product vendor */ @@ -653,12 +650,14 @@ static int virPCIDeviceSharesBusWithActive(virPCIDevicePtr dev, virPCIDevicePtr check, void *data) { virPCIDeviceList *inactiveDevs = data; + virPCIDeviceAddressPtr devAddr = dev->address; + virPCIDeviceAddressPtr checkAddr = check->address; /* Different domain, different bus, or simply identical device */ - if (dev->domain != check->domain || - dev->bus != check->bus || - (dev->slot == check->slot && - dev->function == check->function)) + if (devAddr->domain != checkAddr->domain || + devAddr->bus != checkAddr->bus || + (devAddr->slot == checkAddr->slot && + devAddr->function == checkAddr->function)) return 0; /* same bus, but inactive, i.e. about to be assigned to guest */ @@ -686,10 +685,12 @@ virPCIDeviceIsParent(virPCIDevicePtr dev, virPCIDevicePtr check, void *data) uint16_t device_class; uint8_t header_type, secondary, subordinate; virPCIDevicePtr *best = data; + virPCIDeviceAddressPtr devAddr = dev->address; + virPCIDeviceAddressPtr checkAddr = check->address; int ret = 0; int fd; - if (dev->domain != check->domain) + if (devAddr->domain != checkAddr->domain) return 0; if ((fd = virPCIDeviceConfigOpen(check, false)) < 0) @@ -713,7 +714,7 @@ virPCIDeviceIsParent(virPCIDevicePtr dev, virPCIDevicePtr check, void *data) /* if the secondary bus exactly equals the device's bus, then we found * the direct parent. No further work is necessary */ - if (dev->bus == secondary) { + if (devAddr->bus == secondary) { ret = 1; goto cleanup; } @@ -722,10 +723,10 @@ virPCIDeviceIsParent(virPCIDevicePtr dev, virPCIDevicePtr check, void *data) * In this case, what we need to do is look for the "best" match; i.e. * the most restrictive match that still satisfies all of the conditions. */ - if (dev->bus > secondary && dev->bus <= subordinate) { + if (devAddr->bus > secondary && devAddr->bus <= subordinate) { if (*best == NULL) { - *best = virPCIDeviceNew(check->domain, check->bus, check->slot, - check->function); + *best = virPCIDeviceNew(checkAddr->domain, checkAddr->bus, + checkAddr->slot, checkAddr->function); if (*best == NULL) { ret = -1; goto cleanup; @@ -745,8 +746,8 @@ virPCIDeviceIsParent(virPCIDevicePtr dev, virPCIDevicePtr check, void *data) if (secondary > best_secondary) { virPCIDeviceFree(*best); - *best = virPCIDeviceNew(check->domain, check->bus, check->slot, - check->function); + *best = virPCIDeviceNew(checkAddr->domain, checkAddr->bus, + checkAddr->slot, checkAddr->function); if (*best == NULL) { ret = -1; goto cleanup; @@ -925,6 +926,7 @@ virPCIDeviceReset(virPCIDevicePtr dev, char *drvName = NULL; int ret = -1; int fd = -1; + virPCIDeviceAddressPtr devAddr = dev->address; if (activeDevs && virPCIDeviceListFind(activeDevs, dev)) { virReportError(VIR_ERR_INTERNAL_ERROR, @@ -970,7 +972,7 @@ virPCIDeviceReset(virPCIDevicePtr dev, ret = virPCIDeviceTryPowerManagementReset(dev, fd); /* Bus reset is not an option with the root bus */ - if (ret < 0 && dev->bus != 0) + if (ret < 0 && devAddr->bus != 0) ret = virPCIDeviceTrySecondaryBusReset(dev, fd, inactiveDevs); if (ret < 0) { @@ -1428,6 +1430,7 @@ virPCIDeviceWaitForCleanup(virPCIDevicePtr dev, const char *matcher) bool in_matching_device; int ret; size_t match_depth; + virPCIDeviceAddressPtr devAddr = dev->address; fp = fopen("/proc/iomem", "r"); if (!fp) { @@ -1483,8 +1486,8 @@ virPCIDeviceWaitForCleanup(virPCIDevicePtr dev, const char *matcher) virStrToLong_ui(tmp + 1, &tmp, 16, &function) < 0 || *tmp != '\n') continue; - if (domain != dev->domain || bus != dev->bus || slot != dev->slot || - function != dev->function) + if (domain != devAddr->domain || bus != devAddr->bus || + slot != devAddr->slot || function != devAddr->function) continue; in_matching_device = true; match_depth = strspn(line, " "); @@ -1547,6 +1550,72 @@ virPCIGetAddrString(unsigned int domain, return ret; } +/** + * virPCIDeviceAddressNew: + * @domain: PCI domain + * @bus: PCI bus + * @slot: PCI slot + * @function: PCI function + * + * Create a new virPCIDeviceAddress object. + * + * Returns: a new address, or NULL on failure. + */ +virPCIDeviceAddressPtr +virPCIDeviceAddressNew(unsigned int domain, + unsigned int bus, + unsigned int slot, + unsigned int function) +{ + virPCIDeviceAddressPtr addr; + + if (VIR_ALLOC(addr) < 0) + return NULL; + + addr->domain = domain; + addr->bus = bus; + addr->slot = slot; + addr->function = function; + + return addr; +} + +/** + * virPCIDeviceAddressCopy: + * @addr: PCI address + * + * Create a copy of a PCI address. + * + * Returns: a copy of @addr, or NULL on failure. + */ +virPCIDeviceAddressPtr +virPCIDeviceAddressCopy(virPCIDeviceAddressPtr addr) +{ + virPCIDeviceAddressPtr copy; + + if (!addr) + return NULL; + + if (VIR_ALLOC(copy) < 0) + return NULL; + + *copy = *addr; + + return copy; +} + +/** + * virPCIDeviceAddressFree: + * @addr: PCI address + * + * Release all resources associated with a PCI address. + */ +void +virPCIDeviceAddressFree(virPCIDeviceAddressPtr addr) +{ + VIR_FREE(addr); +} + virPCIDevicePtr virPCIDeviceNew(unsigned int domain, unsigned int bus, @@ -1560,17 +1629,14 @@ virPCIDeviceNew(unsigned int domain, if (VIR_ALLOC(dev) < 0) return NULL; - dev->domain = domain; - dev->bus = bus; - dev->slot = slot; - dev->function = function; + if (!(dev->address = virPCIDeviceAddressNew(domain, bus, slot, function))) + goto error; if (snprintf(dev->name, sizeof(dev->name), "%.4x:%.2x:%.2x.%.1x", - dev->domain, dev->bus, dev->slot, - dev->function) >= sizeof(dev->name)) { + domain, bus, slot, function) >= sizeof(dev->name)) { virReportError(VIR_ERR_INTERNAL_ERROR, _("dev->name buffer overflow: %.4x:%.2x:%.2x.%.1x"), - dev->domain, dev->bus, dev->slot, dev->function); + domain, bus, slot, function); goto error; } if (virAsprintf(&dev->path, PCI_SYSFS "devices/%s/config", @@ -1627,9 +1693,13 @@ virPCIDeviceCopy(virPCIDevicePtr dev) /* shallow copy to take care of most attributes */ *copy = *dev; + + /* deep copy for the rest */ + copy->address = NULL; copy->path = copy->stubDriver = NULL; copy->used_by_drvname = copy->used_by_domname = NULL; - if (VIR_STRDUP(copy->path, dev->path) < 0 || + if (!(copy->address = virPCIDeviceAddressCopy(dev->address)) || + VIR_STRDUP(copy->path, dev->path) < 0 || VIR_STRDUP(copy->stubDriver, dev->stubDriver) < 0 || VIR_STRDUP(copy->used_by_drvname, dev->used_by_drvname) < 0 || VIR_STRDUP(copy->used_by_domname, dev->used_by_domname) < 0) { @@ -1649,6 +1719,7 @@ virPCIDeviceFree(virPCIDevicePtr dev) if (!dev) return; VIR_DEBUG("%s %s: freeing", dev->id, dev->name); + virPCIDeviceAddressFree(dev->address); VIR_FREE(dev->path); VIR_FREE(dev->stubDriver); VIR_FREE(dev->used_by_drvname); @@ -1661,25 +1732,14 @@ virPCIDeviceFree(virPCIDevicePtr dev) * @dev: device to get address from * * Take a PCI device on input and return its PCI address. The - * caller must free the returned value when no longer needed. + * returned object is owned by the device and must not be freed. * * Returns NULL on failure, the device address on success. */ virPCIDeviceAddressPtr virPCIDeviceGetAddress(virPCIDevicePtr dev) { - - virPCIDeviceAddressPtr pciAddrPtr; - - if (!dev || (VIR_ALLOC(pciAddrPtr) < 0)) - return NULL; - - pciAddrPtr->domain = dev->domain; - pciAddrPtr->bus = dev->bus; - pciAddrPtr->slot = dev->slot; - pciAddrPtr->function = dev->function; - - return pciAddrPtr; + return dev->address; } const char * @@ -1888,14 +1948,18 @@ virPCIDeviceListDel(virPCIDeviceListPtr list, int virPCIDeviceListFindIndex(virPCIDeviceListPtr list, virPCIDevicePtr dev) { + virPCIDeviceAddressPtr devAddr = dev->address; + virPCIDeviceAddressPtr tmpAddr; size_t i; - for (i = 0; i < list->count; i++) - if (list->devs[i]->domain == dev->domain && - list->devs[i]->bus == dev->bus && - list->devs[i]->slot == dev->slot && - list->devs[i]->function == dev->function) + for (i = 0; i < list->count; i++) { + tmpAddr = list->devs[i]->address; + if (tmpAddr->domain == devAddr->domain && + tmpAddr->bus == devAddr->bus && + tmpAddr->slot == devAddr->slot && + tmpAddr->function == devAddr->function) return i; + } return -1; } @@ -1907,13 +1971,16 @@ virPCIDeviceListFindByIDs(virPCIDeviceListPtr list, unsigned int slot, unsigned int function) { + virPCIDeviceAddressPtr tmpAddr; size_t i; for (i = 0; i < list->count; i++) { - if (list->devs[i]->domain == domain && - list->devs[i]->bus == bus && - list->devs[i]->slot == slot && - list->devs[i]->function == function) + tmpAddr = list->devs[i]->address; + if (tmpAddr && + tmpAddr->domain == domain && + tmpAddr->bus == bus && + tmpAddr->slot == slot && + tmpAddr->function == function) return list->devs[i]; } return NULL; @@ -1942,9 +2009,11 @@ int virPCIDeviceFileIterate(virPCIDevicePtr dev, int ret = -1; struct dirent *ent; int direrr; + virPCIDeviceAddressPtr devAddr = dev->address; if (virAsprintf(&pcidir, "/sys/bus/pci/devices/%04x:%02x:%02x.%x", - dev->domain, dev->bus, dev->slot, dev->function) < 0) + devAddr->domain, devAddr->bus, + devAddr->slot, devAddr->function) < 0) goto cleanup; if (!(dir = opendir(pcidir))) { @@ -2074,13 +2143,12 @@ virPCIDeviceListPtr virPCIDeviceGetIOMMUGroupList(virPCIDevicePtr dev) { virPCIDeviceListPtr groupList = virPCIDeviceListNew(); - virPCIDeviceAddress devAddr = { dev->domain, dev->bus, - dev->slot, dev->function }; + virPCIDeviceAddressPtr devAddr = dev->address; if (!groupList) goto error; - if (virPCIDeviceAddressIOMMUGroupIterate(&devAddr, + if (virPCIDeviceAddressIOMMUGroupIterate(devAddr, virPCIDeviceGetIOMMUGroupAddOne, groupList) < 0) goto error; @@ -2286,6 +2354,7 @@ static int virPCIDeviceIsBehindSwitchLackingACS(virPCIDevicePtr dev) { virPCIDevicePtr parent; + virPCIDeviceAddressPtr devAddr = dev->address; if (virPCIDeviceGetParent(dev, &parent) < 0) return -1; @@ -2294,7 +2363,7 @@ virPCIDeviceIsBehindSwitchLackingACS(virPCIDevicePtr dev) * into play since devices on the root bus can't P2P without going * through the root IOMMU. */ - if (dev->bus == 0) { + if (devAddr->bus == 0) { return 0; } else { virReportError(VIR_ERR_INTERNAL_ERROR, @@ -2664,12 +2733,13 @@ virPCIGetSysfsFile(char *virPCIDeviceName, char **pci_sysfs_device_link) } int -virPCIDeviceAddressGetSysfsFile(virPCIDeviceAddressPtr dev, +virPCIDeviceAddressGetSysfsFile(virPCIDeviceAddressPtr addr, char **pci_sysfs_device_link) { if (virAsprintf(pci_sysfs_device_link, - PCI_SYSFS "devices/%04x:%02x:%02x.%x", dev->domain, - dev->bus, dev->slot, dev->function) < 0) + PCI_SYSFS "devices/%04x:%02x:%02x.%x", + addr->domain, addr->bus, + addr->slot, addr->function) < 0) return -1; return 0; } diff --git a/src/util/virpci.h b/src/util/virpci.h index 6516f05..a137a9a 100644 --- a/src/util/virpci.h +++ b/src/util/virpci.h @@ -69,6 +69,13 @@ struct _virPCIEDeviceInfo { virPCIELink *link_sta; /* Actually negotiated capabilities */ }; +virPCIDeviceAddressPtr virPCIDeviceAddressNew(unsigned int domain, + unsigned int bus, + unsigned int slot, + unsigned int function); +virPCIDeviceAddressPtr virPCIDeviceAddressCopy(virPCIDeviceAddressPtr addr); +void virPCIDeviceAddressFree(virPCIDeviceAddressPtr addr); + virPCIDevicePtr virPCIDeviceNew(unsigned int domain, unsigned int bus, unsigned int slot, -- 2.5.0

2 2

[libvirt] [PATCHv3] hypervisor driver for Jailhouse
by Christian Loehle 14 Dec '15

14 Dec '15

>From 818447bdc0c7fb5048367276a5eb64108ad95e2a Mon Sep 17 00:00:00 2001 From: Christian Loehle <cloehle(a)linutronix.de> Date: Mon, 14 Dec 2015 20:13:07 +0100 Subject: [PATCH] Add Jailhouse Driver v3 --- configure.ac | 7 + include/libvirt/virterror.h | 1 + m4/virt-driver-jailhouse.m4 | 35 +++ po/POTFILES.in | 1 + src/Makefile.am | 15 ++ src/conf/domain_conf.c | 3 +- src/conf/domain_conf.h | 1 + src/jailhouse/README | 14 ++ src/jailhouse/jailhouse_driver.c | 511 +++++++++++++++++++++++++++++++++++++++ src/jailhouse/jailhouse_driver.h | 28 +++ src/libvirt.c | 10 + src/util/virerror.c | 1 + 12 files changed, 626 insertions(+), 1 deletion(-) create mode 100644 m4/virt-driver-jailhouse.m4 create mode 100644 src/jailhouse/README create mode 100644 src/jailhouse/jailhouse_driver.c create mode 100644 src/jailhouse/jailhouse_driver.h diff --git a/configure.ac b/configure.ac index 98cf210..445b2ca 100644 --- a/configure.ac +++ b/configure.ac @@ -1087,6 +1087,12 @@ dnl LIBVIRT_DRIVER_CHECK_BHYVE dnl +dnl Checks for Jailhouse driver +dnl + +LIBVIRT_DRIVER_CHECK_JAILHOUSE + +dnl dnl check for shell that understands <> redirection without truncation, dnl needed by src/qemu/qemu_monitor_{text,json}.c. dnl @@ -2830,6 +2836,7 @@ AC_MSG_NOTICE([ ESX: $with_esx]) AC_MSG_NOTICE([ Hyper-V: $with_hyperv]) LIBVIRT_DRIVER_RESULT_VZ LIBVIRT_DRIVER_RESULT_BHYVE +LIBVIRT_DRIVER_RESULT_JAILHOUSE AC_MSG_NOTICE([ Test: $with_test]) AC_MSG_NOTICE([ Remote: $with_remote]) AC_MSG_NOTICE([ Network: $with_network]) diff --git a/include/libvirt/virterror.h b/include/libvirt/virterror.h index 0539e48..bbc7e2e 100644 --- a/include/libvirt/virterror.h +++ b/include/libvirt/virterror.h @@ -128,6 +128,7 @@ typedef enum { VIR_FROM_THREAD = 61, /* Error from thread utils */ VIR_FROM_ADMIN = 62, /* Error from admin backend */ VIR_FROM_LOGGING = 63, /* Error from log manager */ + VIR_FROM_JAILHOUSE = 64, /* Error from Jailhouse driver */ # ifdef VIR_ENUM_SENTINELS VIR_ERR_DOMAIN_LAST diff --git a/m4/virt-driver-jailhouse.m4 b/m4/virt-driver-jailhouse.m4 new file mode 100644 index 0000000..6c72043 --- /dev/null +++ b/m4/virt-driver-jailhouse.m4 @@ -0,0 +1,35 @@ +dnl The Jailhouse driver +dnl +dnl Copyright (C) 2005-2015 Red Hat, Inc. +dnl +dnl This library is free software; you can redistribute it and/or +dnl modify it under the terms of the GNU Lesser General Public +dnl License as published by the Free Software Foundation; either +dnl version 2.1 of the License, or (at your option) any later version. +dnl +dnl This library is distributed in the hope that it will be useful, +dnl but WITHOUT ANY WARRANTY; without even the implied warranty of +dnl MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +dnl Lesser General Public License for more details. +dnl +dnl You should have received a copy of the GNU Lesser General Public +dnl License along with this library. If not, see +dnl <http://www.gnu.org/licenses/>. +dnl + +AC_DEFUN([LIBVIRT_DRIVER_CHECK_JAILHOUSE],[ + AC_ARG_WITH([jailhouse], + [AS_HELP_STRING([--with-jailhouse], + [add Jailhouse support @<:@default=yes@:>@])]) + m4_divert_text([DEFAULTS], [with_jailhouse=yes]) + + if test "$with_jailhouse" = "yes"; then + AC_DEFINE_UNQUOTED([WITH_JAILHOUSE], 1, [whether jailhouse driver is enabled]) + fi + + AM_CONDITIONAL([WITH_JAILHOUSE], [test "$with_jailhouse" = "yes"]) +]) + +AC_DEFUN([LIBVIRT_DRIVER_RESULT_JAILHOUSE],[ + AC_MSG_NOTICE([Jailhouse: $with_jailhouse]) +]) diff --git a/po/POTFILES.in b/po/POTFILES.in index 82e8d3e..29259f4 100644 --- a/po/POTFILES.in +++ b/po/POTFILES.in @@ -61,6 +61,7 @@ src/hyperv/hyperv_wmi.c src/interface/interface_backend_netcf.c src/interface/interface_backend_udev.c src/internal.h +src/jailhouse/jailhouse_driver.c src/libvirt.c src/libvirt-admin.c src/libvirt-domain.c diff --git a/src/Makefile.am b/src/Makefile.am index 7219f7c..af46766 100644 --- a/src/Makefile.am +++ b/src/Makefile.am @@ -622,6 +622,7 @@ DRIVER_SOURCE_FILES = \ $(VMWARE_DRIVER_SOURCES) \ $(XEN_DRIVER_SOURCES) \ $(XENAPI_DRIVER_SOURCES) \ + $(JAILHOUSE_DRIVER_SOURCES) \ $(NULL) STATEFUL_DRIVER_SOURCE_FILES = \ @@ -904,6 +905,11 @@ BHYVE_DRIVER_SOURCES = \ bhyve/bhyve_utils.h \ $(NULL) +JAILHOUSE_DRIVER_SOURCES = \ + jailhouse/jailhouse_driver.c \ + jailhouse/jailhouse_driver.h \ + $(NULL) + NETWORK_DRIVER_SOURCES = \ network/bridge_driver.h network/bridge_driver.c \ network/bridge_driver_platform.h \ @@ -1480,6 +1486,14 @@ libvirt_driver_vz_la_LIBADD = $(PARALLELS_SDK_LIBS) $(LIBNL_LIBS) libvirt_driver_vz_la_SOURCES = $(VZ_DRIVER_SOURCES) endif WITH_VZ +if WITH_JAILHOUSE +noinst_LTLIBRARIES += libvirt_driver_jailhouse.la +libvirt_la_BUILT_LIBADD += libvirt_driver_jailhouse.la +libvirt_driver_jailhouse_la_CFLAGS = \ + -I$(srcdir)/conf $(AM_CFLAGS) +libvirt_driver_jailhouse_la_SOURCES = $(JAILHOUSE_DRIVER_SOURCES) +endif WITH_JAILHOUSE + if WITH_BHYVE noinst_LTLIBRARIES += libvirt_driver_bhyve_impl.la libvirt_driver_bhyve_la_SOURCES = @@ -1845,6 +1859,7 @@ EXTRA_DIST += \ $(HYPERV_DRIVER_EXTRA_DIST) \ $(VZ_DRIVER_SOURCES) \ $(BHYVE_DRIVER_SOURCES) \ + $(JAILHOUSE_DRIVER_SOURCES) \ $(NETWORK_DRIVER_SOURCES) \ $(INTERFACE_DRIVER_SOURCES) \ $(STORAGE_DRIVER_SOURCES) \ diff --git a/src/conf/domain_conf.c b/src/conf/domain_conf.c index 5200c27..926c1a0 100644 --- a/src/conf/domain_conf.c +++ b/src/conf/domain_conf.c @@ -107,7 +107,8 @@ VIR_ENUM_IMPL(virDomainVirt, VIR_DOMAIN_VIRT_LAST, "phyp", "parallels", "bhyve", - "vz") + "vz", + "jailhouse") VIR_ENUM_IMPL(virDomainOS, VIR_DOMAIN_OSTYPE_LAST, "hvm", diff --git a/src/conf/domain_conf.h b/src/conf/domain_conf.h index cec681a..271811f 100644 --- a/src/conf/domain_conf.h +++ b/src/conf/domain_conf.h @@ -225,6 +225,7 @@ typedef enum { VIR_DOMAIN_VIRT_PARALLELS, VIR_DOMAIN_VIRT_BHYVE, VIR_DOMAIN_VIRT_VZ, + VIR_DOMAIN_VIRT_JAILHOUSE, VIR_DOMAIN_VIRT_LAST } virDomainVirtType; diff --git a/src/jailhouse/README b/src/jailhouse/README new file mode 100644 index 0000000..02ba87d --- /dev/null +++ b/src/jailhouse/README @@ -0,0 +1,14 @@ +The jailhouse hypervisor driver for the libvirt project aims to provide +rudimentary support for managing jailhouse with the libvirt library. +The main advantage of this is the possibility to use virt-manager as a GUI to +manage Jailhouse cells. Thus the driver is mainly built around the API calls +that virt-manager uses and needs. +Due to the concept of Jailhouse a lot of libvirt functions can't be realized, +so this driver isn't as full-featured as upstream drivers of the +libvirt project. +Currently the driver relies on the Jailhouse binary, which has to be in $PATH +or passed when connecting a libvirt client to it +(e.g. virt-manager -c jailhouse:///path/to/jailhouse/tools/jailhouse). +Be aware though that the driver doesn't store any information about cells, +so most API calls use "jailhouse cell list" every time they're called to get +the current state. diff --git a/src/jailhouse/jailhouse_driver.c b/src/jailhouse/jailhouse_driver.c new file mode 100644 index 0000000..79e97f0 --- /dev/null +++ b/src/jailhouse/jailhouse_driver.c @@ -0,0 +1,511 @@ +/* + * virJailhouseDriver.c: hypervisor driver for managing Jailhouse cells + * + * Copyright (C) 2015 Linutronix GmbH + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library. If not, see + * <http://www.gnu.org/licenses/>. + * + * Author: Christian Loehle + */ + +#include <config.h> +#include <string.h> +#include "jailhouse_driver.h" +#include "datatypes.h" +#include "virerror.h" +#include "viralloc.h" +#include "virlog.h" +#include "vircommand.h" +#include "virxml.h" +#include "configmake.h" +#include "virfile.h" +#include "virtypedparam.h" +#include "virstring.h" +#include "nodeinfo.h" +#include "capabilities.h" +#include "domain_conf.h" + +#define VIR_FROM_THIS VIR_FROM_JAILHOUSE + +#define IDLENGTH 8 +#define NAMELENGTH 24 +#define STATELENGTH 16 +#define CPULENGTH 24 +#define STATERUNNING 0 +#define STATERUNNINGSTRING "running " +#define STATERUNNINGLOCKED 1 +#define STATERUNNINGLOCKEDSTRING "running/locked " +#define STATESHUTDOWN 2 +#define STATESHUTDOWNSTRING "shut down " +#define STATEFAILED 3 +#define STATEFAILEDSTRING "failed " +#define JAILHOUSEBINARY "jailhouse" + +struct virJailhouseCell { + int id; + char name[NAMELENGTH+1]; + int state; + virBitmapPtr assignedCPUs; + virBitmapPtr failedCPUs; /* currently unused */ + unsigned char uuid[VIR_UUID_BUFLEN]; +}; +typedef struct virJailhouseCell virJailhouseCell; +typedef virJailhouseCell *virJailhouseCellPtr; + +/* + * Because virCommandRunRegex Callback gets called every line + */ +struct virJailhouseCellCallbackData { + size_t ncells; + virJailhouseCellPtr cells; +}; +typedef struct virJailhouseCellCallbackData virJailhouseCellCallbackData; +typedef virJailhouseCellCallbackData *virJailhouseCellCallbackDataPtr; + +/* + * The driver requeries the cells on most calls, it stores the result of the + * last query, so it can copy the UUIDs in the new query if the cell is the + * same(otherwise it just generates a new one). + * not preserving the UUID results in a lot of bugs in libvirts clients. + */ +struct virJailhouseDriver { + size_t lastQueryCellsCount; + virJailhouseCellPtr lastQueryCells; +}; +typedef struct virJailhouseDriver virJailhouseDriver; +typedef virJailhouseDriver *virJailhouseDriverPtr; + +static int virJailhouseParseListOutputCallback(char **const groups, void *data) +{ + virJailhouseCellCallbackDataPtr celldata = (virJailhouseCellCallbackDataPtr) data; + virJailhouseCellPtr cells = celldata->cells; + size_t count = celldata->ncells; + char* endptr = groups[0] + strlen(groups[0]) - 1; + char* state = groups[2]; + if (VIR_EXPAND_N(cells, count, 1)) + return -1; + celldata->ncells++; + + if (virStrToLong_i(groups[0], &endptr, 0, &cells[count-1].id)) + return -1; + if (!virStrcpy(cells[count-1].name, groups[1], NAMELENGTH+1)) + return -1; + if (STREQLEN(state, STATERUNNINGSTRING, STATELENGTH)) + cells[count-1].state = STATERUNNING; + else if (STREQLEN(state, STATESHUTDOWNSTRING, STATELENGTH)) + cells[count-1].state = STATESHUTDOWN; + else if (STREQLEN(state, STATERUNNINGLOCKEDSTRING, STATELENGTH)) + cells[count-1].state = STATERUNNINGLOCKED; + else + cells[count-1].state = STATEFAILED; + virBitmapParse(groups[3], 0, &cells[count-1].assignedCPUs, VIR_DOMAIN_CPUMASK_LEN); + virBitmapParse(groups[4], 0, &cells[count-1].failedCPUs, VIR_DOMAIN_CPUMASK_LEN); + celldata->cells = cells; + return 0; +} + +/* + * calls "jailhouse cell list" and parses the output in an array of virJailhouseCell + * example output: + * ID Name State Assigned CPUs Failed CPUs + * 0 QEMU-VM running 0-3 + */ +static ssize_t +virJailhouseParseListOutput(virJailhouseCellPtr *parsedCells) +{ + int nvars[] = { 5 }; + virJailhouseCellCallbackData callbackData; + const char *regex[] = { "([0-9]{1,8})\\s*([-0-9a-zA-Z]{1,24})\\s*([a-z/ ]{1,16})\\s*([0-9,-]{1,24})?\\s*([0-9,-]{1,24})?\\s*" }; + virCommandPtr cmd = virCommandNew(JAILHOUSEBINARY); + virCommandAddArg(cmd, "cell"); + virCommandAddArg(cmd, "list"); + virCommandAddEnvPassCommon(cmd); + callbackData.cells = NULL; + callbackData.ncells = 0; + if (virCommandRunRegex(cmd, 1, regex, nvars, &virJailhouseParseListOutputCallback, &callbackData, NULL) < 0) { + virCommandFree(cmd); + return -1; + } + virCommandFree(cmd); + *parsedCells = callbackData.cells; + return callbackData.ncells; +} + +/* + * Returns the libvirts equivalent of the cell state passed to it + */ +static virDomainState +virJailhouseCellToState(virJailhouseCellPtr cell) +{ + switch (cell->state) { + case STATERUNNING: return VIR_DOMAIN_RUNNING; + case STATERUNNINGLOCKED: return VIR_DOMAIN_RUNNING; + case STATESHUTDOWN: return VIR_DOMAIN_SHUTOFF; + case STATEFAILED: return VIR_DOMAIN_CRASHED; + default: return VIR_DOMAIN_NOSTATE; + } +} + +/* + * Returns a new virDomainPtr filled with the data of the virJailhouseCell + */ +static virDomainPtr +virJailhouseCellToDomainPtr(virConnectPtr conn, virJailhouseCellPtr cell) +{ + virDomainPtr dom = virGetDomain(conn, cell->name, cell->uuid); + dom->id = cell->id; + return dom; +} + +/* + * Check cells for cell and copies UUID if found, otherwise generates a new one, this is to preserve UUID in libvirt + */ +static void virJailhouseSetUUID(virJailhouseCellPtr cells, size_t count, virJailhouseCellPtr cell) +{ + size_t i; + for (i = 0; i < count; i++) { + if (strncmp(cells[i].name, cell->name, NAMELENGTH+1)) + continue; + memcpy(cell->uuid, cells[i].uuid, VIR_UUID_BUFLEN); + return; + } + virUUIDGenerate(cell->uuid); +} + +/* + * Frees the old list of cells, gets the new one and preserves UUID if cells were present in the old + */ +static int +virJailhouseGetCurrentCellList(virConnectPtr conn) +{ + virJailhouseDriverPtr driver = (virJailhouseDriverPtr)conn->privateData; + ssize_t count; + size_t i; + size_t lastCount = driver->lastQueryCellsCount; + virJailhouseCellPtr lastCells = driver->lastQueryCells; + virJailhouseCellPtr cells = NULL; + + count = virJailhouseParseListOutput(&cells); + for (i = 0; i < count; i++) + virJailhouseSetUUID(lastCells, lastCount, cells+i); + for (i = 0; i < lastCount; i++) { + virBitmapFree(lastCells[i].assignedCPUs); + virBitmapFree(lastCells[i].failedCPUs); + } + VIR_FREE(lastCells); + driver->lastQueryCells = cells; + driver->lastQueryCellsCount = count; + return count; +} + +/* + * Converts libvirts virDomainPtr to the internal virJailhouseCell by parsing the "jailhouse cell list" output + * and looking up the name of the virDomainPtr, returns NULL if cell is no longer present + */ +static virJailhouseCellPtr +virDomainPtrToCell(virDomainPtr dom) +{ + virJailhouseDriverPtr driver = dom->conn->privateData; + size_t cellsCount; + size_t i; + if (virJailhouseGetCurrentCellList(dom->conn) == -1) + return NULL; + cellsCount = driver->lastQueryCellsCount; + virJailhouseCellPtr cells = driver->lastQueryCells; + for (i = 0; i < cellsCount; i++) + if (dom->id == cells[i].id) + return cells+i; + return NULL; +} + +static virDrvOpenStatus +jailhouseConnectOpen(virConnectPtr conn, virConnectAuthPtr auth ATTRIBUTE_UNUSED, unsigned int flags) +{ + virCheckFlags(0, VIR_DRV_OPEN_ERROR); + virJailhouseDriverPtr driver; + if (conn->uri->scheme == NULL || + STRNEQ(conn->uri->scheme, "jailhouse")) + return VIR_DRV_OPEN_DECLINED; + if (conn->uri->path != NULL && STRNEQ(conn->uri->path, "/")) { + virReportError(VIR_ERR_INTERNAL_ERROR, + _("Unexpected Jailhouse URI path '%s', try jailhouse:///"), + conn->uri->path); + return VIR_DRV_OPEN_ERROR; + } + if (VIR_ALLOC(driver) < 0) + return VIR_DRV_OPEN_ERROR; + driver->lastQueryCells = NULL; + driver->lastQueryCellsCount = 0; + conn->privateData = driver; + return VIR_DRV_OPEN_SUCCESS; +} + +static int +jailhouseConnectClose(virConnectPtr conn) +{ + + virJailhouseDriverPtr driver = (virJailhouseDriverPtr)conn->privateData; + size_t i; + size_t cellsCount = driver->lastQueryCellsCount; + virJailhouseCellPtr cells = driver->lastQueryCells; + for (i = 0; i < cellsCount; i++) { + virBitmapFree(cells[i].assignedCPUs); + virBitmapFree(cells[i].failedCPUs); + } + VIR_FREE(cells); + VIR_FREE(driver); + conn->privateData = NULL; + return 0; +} + +static int +jailhouseConnectNumOfDomains(virConnectPtr conn) +{ + if (virJailhouseGetCurrentCellList(conn) == -1) + return -1; + return ((virJailhouseDriverPtr)conn->privateData)->lastQueryCellsCount; +} + +static int +jailhouseConnectListDomains(virConnectPtr conn, int * ids, int maxids) +{ + virJailhouseDriverPtr driver = (virJailhouseDriverPtr)conn->privateData; + virJailhouseCellPtr cells; + size_t cellsCount; + size_t i; + if (virJailhouseGetCurrentCellList(conn) == -1) + return -1; + cellsCount = driver->lastQueryCellsCount; + cells = driver->lastQueryCells; + for (i = 0; i < maxids && i < cellsCount; i++) + ids[i] = cells[i].id; + return i; +} + +static int +jailhouseConnectListAllDomains(virConnectPtr conn, virDomainPtr ** domains, unsigned int flags) +{ + virCheckFlags(VIR_CONNECT_LIST_DOMAINS_ACTIVE, 0); + virJailhouseDriverPtr driver = (virJailhouseDriverPtr)conn->privateData; + virJailhouseCellPtr cells; + size_t cellsCount; + size_t i; + if (virJailhouseGetCurrentCellList(conn) == -1) + goto error; + + cellsCount = driver->lastQueryCellsCount; + cells = driver->lastQueryCells; + if (cellsCount == -1) + goto error; + if (VIR_ALLOC_N(*domains, cellsCount+1) < 0) + goto error; + for (i = 0; i < cellsCount; i++) + (*domains)[i] = virJailhouseCellToDomainPtr(conn, cells+i); + (*domains)[cellsCount] = NULL; + return cellsCount; + error: + *domains = NULL; + return -1; +} + +static virDomainPtr +jailhouseDomainLookupByID(virConnectPtr conn, int id) +{ + virJailhouseDriverPtr driver = (virJailhouseDriverPtr)conn->privateData; + size_t cellsCount; + size_t i; + if (virJailhouseGetCurrentCellList(conn) == -1) + return NULL; + + cellsCount = driver->lastQueryCellsCount; + if (cellsCount == -1) + return NULL; + virJailhouseCellPtr cells = driver->lastQueryCells; + for (i = 0; i < cellsCount; i++) + if (cells[i].id == id) + return virJailhouseCellToDomainPtr(conn, cells+i); + virReportError(VIR_ERR_NO_DOMAIN, NULL); + return NULL; +} + +static virDomainPtr +jailhouseDomainLookupByName(virConnectPtr conn, const char *lookupName) +{ + virJailhouseDriverPtr driver = (virJailhouseDriverPtr)conn->privateData; + size_t cellsCount; + size_t i; + if (virJailhouseGetCurrentCellList(conn) == -1) + return NULL; + + cellsCount = driver->lastQueryCellsCount; + if (cellsCount == -1) + return NULL; + virJailhouseCellPtr cells = driver->lastQueryCells; + for (i = 0; i < cellsCount; i++) + if (STREQ(cells[i].name, lookupName)) + return virJailhouseCellToDomainPtr(conn, cells+i); + virReportError(VIR_ERR_NO_DOMAIN, NULL); + return NULL; +} + +static virDomainPtr +jailhouseDomainLookupByUUID(virConnectPtr conn, const unsigned char * uuid) +{ + virJailhouseDriverPtr driver = (virJailhouseDriverPtr)conn->privateData; + size_t cellsCount; + size_t i; + if (virJailhouseGetCurrentCellList(conn) == -1) + return NULL; + cellsCount = driver->lastQueryCellsCount; + if (cellsCount == -1) + return NULL; + virJailhouseCellPtr cells = driver->lastQueryCells; + for (i = 0; i < cellsCount; i++) + if (memcmp(cells[i].uuid, (const char*)uuid, VIR_UUID_BUFLEN) == 0) + return virJailhouseCellToDomainPtr(conn, cells+i); + virReportError(VIR_ERR_NO_DOMAIN, NULL); + return NULL; +} + +/* + * There currently is no straightforward way for the driver to retrieve those, + * so maxMem, memory and cpuTime have dummy values + */ +static int +jailhouseDomainGetInfo(virDomainPtr domain, virDomainInfoPtr info) +{ + virJailhouseCellPtr cell = virDomainPtrToCell(domain); + if (cell == NULL) + return -1; + info->state = virJailhouseCellToState(cell); + info->maxMem = 0; + info->memory = 0; + info->nrVirtCpu = virBitmapCountBits(cell->assignedCPUs); + info->cpuTime = 0; + return 0; +} + +static int +jailhouseDomainGetState(virDomainPtr domain, int *state, + int *reason ATTRIBUTE_UNUSED, unsigned int flags) +{ + virCheckFlags(0, 0); + virJailhouseCellPtr cell = virDomainPtrToCell(domain); + if (cell == NULL) + return -1; + *state = virJailhouseCellToState(cell); + return 0; +} + +static int +jailhouseDomainDestroy(virDomainPtr domain) +{ + int resultcode; + virCommandPtr cmd = virCommandNew(JAILHOUSEBINARY); + virCommandAddArg(cmd, "cell"); + virCommandAddArg(cmd, "shutdown"); + virCommandAddArgFormat(cmd, "%d", domain->id); + virCommandAddEnvPassCommon(cmd); + resultcode = virCommandRun(cmd, NULL); + virCommandFree(cmd); + if (resultcode < 0) + return -1; + return 0; +} + +static int +jailhouseDomainCreate(virDomainPtr domain) +{ + int resultcode; + virCommandPtr cmd = virCommandNew(JAILHOUSEBINARY); + virCommandAddArg(cmd, "cell"); + virCommandAddArg(cmd, "start"); + virCommandAddArgFormat(cmd, "%d", domain->id); + virCommandAddEnvPassCommon(cmd); + resultcode = virCommandRun(cmd, NULL); + virCommandFree(cmd); + if (resultcode < 0) + return -1; + return 0; +} + +/* + * There currently is no reason why it shouldn't be + */ +static int +jailhouseConnectIsAlive(virConnectPtr conn ATTRIBUTE_UNUSED) +{ + return 1; +} + +static int +jailhouseNodeGetInfo(virConnectPtr conn ATTRIBUTE_UNUSED, virNodeInfoPtr info) +{ + return nodeGetInfo(NULL, info); +} + +/* + * Returns a dummy capabilities XML for virt-manager + */ +static char * +jailhouseConnectGetCapabilities(virConnectPtr conn ATTRIBUTE_UNUSED) +{ + virCapsPtr caps = virCapabilitiesNew(VIR_ARCH_NONE, false, false); + char* xml = virCapabilitiesFormatXML(caps); + virObjectUnref(caps); + return xml; +} + +static char * +jailhouseDomainGetXMLDesc(virDomainPtr domain, unsigned int flags) +{ + virCheckFlags(0, NULL); + char* xml; + virDomainDefPtr domainDef = virDomainDefNewFull(domain->name, domain->uuid, domain->id); + xml = virDomainDefFormat(domainDef, 0); + virDomainDefFree(domainDef); + return xml; +} + +static virHypervisorDriver jailhouseHypervisorDriver = { + .name = "jailhouse", + .connectOpen = jailhouseConnectOpen, /* 1.3.1 */ + .connectClose = jailhouseConnectClose, /* 1.3.1 */ + .connectGetCapabilities = jailhouseConnectGetCapabilities, /* 1.3.1 */ + .connectNumOfDomains = jailhouseConnectNumOfDomains, /* 1.3.1 */ + .connectListDomains = jailhouseConnectListDomains, /* 1.3.1 */ + .connectIsAlive = jailhouseConnectIsAlive, /* 1.3.1 */ + .connectListAllDomains = jailhouseConnectListAllDomains, /* 1.3.1 */ + .domainLookupByID = jailhouseDomainLookupByID, /* 1.3.1 */ + .domainLookupByName = jailhouseDomainLookupByName, /* 1.3.1 */ + .domainLookupByUUID = jailhouseDomainLookupByUUID, /* 1.3.1 */ + .domainGetInfo = jailhouseDomainGetInfo, /* 1.3.1 */ + .domainGetState = jailhouseDomainGetState, /* 1.3.1 */ + .domainGetXMLDesc = jailhouseDomainGetXMLDesc, /* 1.3.1 */ + .domainDestroy = jailhouseDomainDestroy, /* 1.3.1 */ + .domainCreate = jailhouseDomainCreate, /* 1.3.1 */ + .nodeGetInfo = jailhouseNodeGetInfo /* 1.3.1 */ +}; + +static virConnectDriver jailhouseConnectDriver = { + .hypervisorDriver = &jailhouseHypervisorDriver, +}; + +int +jailhouseRegister(void) +{ + return virRegisterConnectDriver(&jailhouseConnectDriver, + false); +} diff --git a/src/jailhouse/jailhouse_driver.h b/src/jailhouse/jailhouse_driver.h new file mode 100644 index 0000000..47c17e7 --- /dev/null +++ b/src/jailhouse/jailhouse_driver.h @@ -0,0 +1,28 @@ +/* + * jailhouse_driver.h: hypervisor driver for managing Jailhouse cells + * + * Copyright (C) 2015 Linutronix GmbH + * + * This library is free software; you can redistribute it and/or + * modify it under the terms of the GNU Lesser General Public + * License as published by the Free Software Foundation; either + * version 2.1 of the License, or (at your option) any later version. + * + * This library is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + * Lesser General Public License for more details. + * + * You should have received a copy of the GNU Lesser General Public + * License along with this library. If not, see + * <http://www.gnu.org/licenses/>. + * + * Author: Christian Loehle + */ + +#ifndef JAILHOUSE_DRIVER_H +# define JAILHOUSE_DRIVER_H + +int jailhouseRegister(void); + +#endif diff --git a/src/libvirt.c b/src/libvirt.c index dd58e9c..36d5994 100644 --- a/src/libvirt.c +++ b/src/libvirt.c @@ -98,6 +98,9 @@ #ifdef WITH_BHYVE # include "bhyve/bhyve_driver.h" #endif +#ifdef WITH_JAILHOUSE +# include "jailhouse/jailhouse_driver.h" +#endif #define VIR_FROM_THIS VIR_FROM_NONE @@ -437,6 +440,10 @@ virGlobalInit(void) if (vzRegister() == -1) goto error; # endif +#ifdef WITH_JAILHOUSE + if (jailhouseRegister() == -1) + goto error; +#endif #endif #ifdef WITH_REMOTE if (remoteRegister() == -1) @@ -1038,6 +1045,9 @@ do_open(const char *name, #ifndef WITH_VZ STRCASEEQ(ret->uri->scheme, "parallels") || #endif +#ifndef WITH_JAILHOUSE + STRCASEEQ(ret->uri->scheme, "jailhouse") || +#endif false)) { virReportErrorHelper(VIR_FROM_NONE, VIR_ERR_CONFIG_UNSUPPORTED, __FILE__, __FUNCTION__, __LINE__, diff --git a/src/util/virerror.c b/src/util/virerror.c index 098211a..06741bf 100644 --- a/src/util/virerror.c +++ b/src/util/virerror.c @@ -135,6 +135,7 @@ VIR_ENUM_IMPL(virErrorDomain, VIR_ERR_DOMAIN_LAST, "Thread jobs", "Admin Interface", "Log Manager", + "Jailhouse Driver", ) -- 2.1.4

1 0

[libvirt] [PATCH 0/2] Add support for zero-write detection
by Martin Kletzander 14 Dec '15

14 Dec '15

QEMU supports detect-zeroes option since version 2.1, but we never added support for it in libvirt. If was requested by Vasiliy Tolstov in the list, so I just added it. There are two discussions to be had, optionally. One is to decide whether we should disable detect_zeros='unmap' if discard is not set to 'unmap', but this is getting very hypervisor-specific, so I just documented the behaviour. The other one is the naming. I described why I made the decision for "zeros" instead of "zeroes" the decision in the patch, but I have no problem changing it to what others like better. Martin Kletzander (2): conf: Add support of zero-detection for disks qemu: Add support for zero-detection writes docs/formatdomain.html.in | 10 ++++++ docs/schemas/domaincommon.rng | 12 +++++++ src/conf/domain_conf.c | 23 +++++++++++++- src/conf/domain_conf.h | 11 +++++++ src/libvirt_private.syms | 2 ++ src/qemu/qemu_capabilities.c | 2 ++ src/qemu/qemu_capabilities.h | 1 + src/qemu/qemu_command.c | 11 +++++++ tests/qemucapabilitiesdata/caps_2.1.1-1.caps | 1 + tests/qemucapabilitiesdata/caps_2.4.0-1.caps | 1 + tests/qemucapabilitiesdata/caps_2.5.0-1.caps | 1 + .../qemuxml2argv-disk-drive-detect-zeros.args | 27 ++++++++++++++++ .../qemuxml2argv-disk-drive-detect-zeros.xml | 37 ++++++++++++++++++++++ tests/qemuxml2argvtest.c | 4 +++ tests/qemuxml2xmltest.c | 1 + 15 files changed, 143 insertions(+), 1 deletion(-) create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-disk-drive-detect-zeros.args create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-disk-drive-detect-zeros.xml -- 2.6.4

3 6

[libvirt] [PATCH] virNetDevMacVLanTapSetup: Drop @multiqueue argument
by Michal Privoznik 14 Dec '15

14 Dec '15

Firstly, there's a bug (or typo) in the only place where we call this function: @multiqueue is set whenever @tapfdSize is greater than zero, while in fact the condition should have been 'greater than one'. Then, secondly, since the condition depends on just one variable, that we are even passing down to the function, we can move the condition into the function and drop useless argument. Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com> --- src/util/virnetdevmacvlan.c | 17 ++++++++--------- 1 file changed, 8 insertions(+), 9 deletions(-) diff --git a/src/util/virnetdevmacvlan.c b/src/util/virnetdevmacvlan.c index 8fc71af..496416e 100644 --- a/src/util/virnetdevmacvlan.c +++ b/src/util/virnetdevmacvlan.c @@ -289,12 +289,11 @@ virNetDevMacVLanTapOpen(const char *ifname, * @tapfd: array of file descriptors of the macvtap tap * @tapfdSize: number of file descriptors in @tapfd * @vnet_hdr: whether to enable or disable IFF_VNET_HDR - * @multiqueue: whether to enable or disable IFF_MULTI_QUEUE * - * Turn on the IFF_VNET_HDR flag if requested and available, but make sure it's - * off otherwise. Similarly, turn on IFF_MULTI_QUEUE if requested, but if it - * can't be set, consider it a fatal error (rather than ignoring as with - * @vnet_hdr). + * Turn on the IFF_VNET_HDR flag if requested and available, but make sure + * it's off otherwise. Similarly, turn on IFF_MULTI_QUEUE if @tapfdSize is + * greater than one, but if it can't be set, consider it a fatal error + * (rather than ignoring as with @vnet_hdr). * * A fatal error is defined as the VNET_HDR flag being set but it cannot * be turned off for some reason. This is reported with -1. Other fatal @@ -304,7 +303,7 @@ virNetDevMacVLanTapOpen(const char *ifname, * Returns 0 on success, -1 in case of fatal error. */ static int -virNetDevMacVLanTapSetup(int *tapfd, size_t tapfdSize, bool vnet_hdr, bool multiqueue) +virNetDevMacVLanTapSetup(int *tapfd, size_t tapfdSize, bool vnet_hdr) { unsigned int features; struct ifreq ifreq; @@ -335,12 +334,12 @@ virNetDevMacVLanTapSetup(int *tapfd, size_t tapfdSize, bool vnet_hdr, bool multi } # ifdef IFF_MULTI_QUEUE - if (multiqueue) + if (tapfdSize > 1) new_flags |= IFF_MULTI_QUEUE; else new_flags &= ~IFF_MULTI_QUEUE; # else - if (multiqueue) { + if (tapfdSize > 1) { virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s", _("Multiqueue devices are not supported on this system")); return -1; @@ -870,7 +869,7 @@ int virNetDevMacVLanCreateWithVPortProfile(const char *tgifname, if (virNetDevMacVLanTapOpen(cr_ifname, tapfd, tapfdSize, 10) < 0) goto disassociate_exit; - if (virNetDevMacVLanTapSetup(tapfd, tapfdSize, vnet_hdr, tapfdSize > 0) < 0) { + if (virNetDevMacVLanTapSetup(tapfd, tapfdSize, vnet_hdr) < 0) { VIR_FORCE_CLOSE(rc); /* sets rc to -1 */ goto disassociate_exit; } -- 2.4.10

2 1

[libvirt] [PATCH v2 0/7] Add multiqueue support for macvtaps
by Michal Privoznik 14 Dec '15

14 Dec '15

Patches 1, 2, 3, 6, 7 have been ACKed in previous round. However, I did slightly change them to reflect Laine's review suggestions. Patch 4 has not been ACKed yet, patch 5 is new. Michal Privoznik (7): virNetDevMacVLanCreateWithVPortProfile: Turn vnet_hdr into flag virNetDevMacVLanTapOpen: Slightly rework virNetDevMacVLanTapOpen: Rework to support multiple FDs virNetDevMacVLanTapSetup: Rework to support multiple FDs virNetDevMacVLanTapSetup: Allow enabling of IFF_MULTI_QUEUE virNetDevMacVLanCreateWithVPortProfile: Rework to support multiple FDs qemu: Enable multiqueue for macvtaps src/lxc/lxc_process.c | 3 +- src/qemu/qemu_command.c | 65 ++++++++++------ src/qemu/qemu_command.h | 2 + src/qemu/qemu_hotplug.c | 16 ++-- src/util/virnetdevmacvlan.c | 185 ++++++++++++++++++++++---------------------- src/util/virnetdevmacvlan.h | 7 +- 6 files changed, 153 insertions(+), 125 deletions(-) -- 2.4.10

3 12

[libvirt] how to pass qemu drive option
by Vasiliy Tolstov 14 Dec '15

14 Dec '15

I want to pass to my drive detect-zeros=on how can i do that in libvirt? I'm use lvm with virtio-scsi for sda disk. -- Vasiliy Tolstov, e-mail: v.tolstov(a)selfip.ru

2 2

[libvirt] [PATCH] qemu_agent: fix deadlock in qemuProcessHandleAgentEOF
by Wang Yufei 14 Dec '15

14 Dec '15

We shutdown a VM A by qemu agent,meanwhile an agent EOF of VM A happened, there's a chance that deadlock occurred: qemuProcessHandleAgentEOF in main thread A) priv->agent = NULL; //A happened before B //deadlock when we get agent lock which's held by worker thread qemuAgentClose(agent); qemuDomainObjExitAgent called by qemuDomainShutdownFlags in worker thread B) hasRefs = virObjectUnref(priv->agent); //priv->agent is NULL, return false if (hasRefs) virObjectUnlock(priv->agent); //agent lock will not be released here So I close agent first, then set priv->agent NULL to fix the deadlock. Signed-off-by: Wang Yufei <james.wangyufei(a)huawei.com> Reviewed-by: Ren Guannan <renguannan(a)huawei.com> --- src/qemu/qemu_process.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c index f2586a1..8c9622e 100644 --- a/src/qemu/qemu_process.c +++ b/src/qemu/qemu_process.c @@ -150,11 +150,10 @@ qemuProcessHandleAgentEOF(qemuAgentPtr agent, goto unlock; } + qemuAgentClose(agent); priv->agent = NULL; virObjectUnlock(vm); - - qemuAgentClose(agent); return; unlock: -- 1.8.3.4

2 8

[libvirt] Release of libvirt-1.3.0
by Daniel Veillard 14 Dec '15

14 Dec '15

So as planned, I tagged the release in git and pushed signed tarball and rpms to the usual place: ftp://libvirt.org/libvirt I also pushed the associated python release for 1.3.0 though there was no actual code change except numbering: ftp://libvirt.org/libvirt/python The bump in version minor number comes from the addition of the administration API which nearly made it in 1.2.17 in the summer but was postponed since. That is backed up by serious improvement in virtio support and logging. As usual that release carries a sigificant amount of fixes and smaller improvements too: Features: - virt-admin and administration API (Erik Skultety, Martin Kletzander) - various improvements in virtio devices support (Ján Tomko, Marc-André Lureau) - log daemon, logging improvements and protocol (Daniel P. Berrange) Documentation: - libvirt: Update virDomainSetMemory description (Nikolay Shirokovskiy) - virt-admin: Provide a man page for virt-admin (Erik Skultety) - Enhance documentation of virDomainDetachDevice (Jiri Denemark) - qemu: monitor: Explain logic of qemuMonitorGetCPUInfo (Peter Krempa) - document virCommandRunRegex function (Christian Loehle) - libvirt-domain: Fix typo in debug message (Cole Robinson) - qemu: Explain mlock limit size more in detail (Peter Krempa) - virsh.pod: improve attach-interface section (Pavel Hrdina) - virnetdev: Fix function comments for virNetDevGetFeatures (John Ferlan) - virnetdev: Document reasons for ignoring some SIOCETHTOOL errno values (John Ferlan) Portability: - network: selectively disable -Wcast-align in virNetDevParseDadStatus (Ian Campbell) - log_manager: Include configmake.h last (Michal Privoznik) - virtlogd: Fix build without DBus (Martin Kletzander) - virtlogd: use %llu to print 64bit types (Guido Günther) Bug Fixes: - qemu: fix memory leak in opening log file (Daniel P. Berrange) - qemu: Automatic SCSI controller creation in SCSI disk hotplug broken (Boris Fiuczynski) - qemu: domain: Prevent overflows in memory alignment code (Peter Krempa) - conf: Revert some code to resolve issues for hostdev hotplug (Boris Fiuczynski) - virsh: report errors for empty strings (Ján Tomko) - bridge: check for invalid MAC in networkGetDHCPLeases (Ján Tomko) - qemu_agent: fix deadlock in qemuProcessHandleAgentEOF (Wang Yufei) - include: Install libvirt-common.h (Martin Kletzander) - tools: fix output of list with state-shutoff (Wei Jiangang) - virlogd: fix crash if log file exists and it's larger the maxlen (Pavel Hrdina) - systemd: Escape only needed characters for machined (Martin Kletzander) - logging: remove reference to non-existent augeas files (Daniel P. Berrange) - virtlockd: fix misc memory leaks and other bugs (Daniel P. Berrange) - systemd: Escape machine name for machined (Martin Kletzander) - schema: use a better regex for listen addresses (Ján Tomko) - apparmor: add missing qemu binaries (Guido Günther) - storage: Change virStorageBackendVolOpen to use virFileOpenAs (John Ferlan) - storage: Really fix setting mode for backend exec in NFS root-squash env (John Ferlan) - qemu: Add ppc64-specific math to qemuDomainGetMlockLimitBytes() (Andrea Bolognani) - libxl: don't unlock virDomainObj if refcnt is 0 (Jim Fehlig) - libxl: unref libxlDriverConfig object (Jim Fehlig) - qemu: Fix build error in Coverity environment (John Ferlan) - virSetUIDGID: Don't leak supplementary groups (Richard Weinberger) - locking: Add io_timeout to sanlock (Michal Privoznik) - libvirt-guests: Disable shutdown timeout (Guido Günther) - tpm: adapt sysfs cancel path for new TPM driver (Stefan Berger) - bhyve: monitor: do not override domain's privateData (Roman Bogorodskiy) - storage: Don't assume storage pool exists for FC/SCSI refresh thread (John Ferlan) - domain-conf: reorder usb controllers so the master is first (Pavel Hrdina) - qemu: fix parsing of -sdl arg (Daniel P. Berrange) - qemu: handle floppy disk bus when parsing command line argv (Daniel P. Berrange) - qemu: hotplug: Fix mlock limit handling on memory hotplug (Peter Krempa) - Revert "utils: Remove the logging of errors from virNetDevSendEthtoolIoctl" (Daniel P. Berrange) - qemu: migration: Actually error out on unsupported migration flag (Peter Krempa) - qemu: migration: Properly parse memory hotplug migration flag (Peter Krempa) - network: Remove extraneous ATTRIBUTE_NONNULL for virNetDevWaitDadFinish (John Ferlan) - virnetdev: Check correct return value for virNetDevFeatureAvailable (John Ferlan) - storage: On 'buildVol' failure don't delete the volume (John Ferlan) - Revert "storage: Prior to creating a volume, refresh the pool" (John Ferlan) - qemu: Fix memory leak in qemuProcessStart (Jiri Denemark) - qemu: Use correct type when calling qemuPrepareNVRAM (Jiri Denemark) Improvements: - Revert "libxl: implement virDomainInterfaceStats" (Jim Fehlig) - rpm: explicitly enable & start virtlogd on install (Daniel P. Berrange) - libvirtd: enable virtlockd/virtlogd socket activation on install (Daniel P. Berrange) - logging: validate flags passed from client in virtlogd (Daniel P. Berrange) - logging: change log protocol to be more reusable (Daniel P. Berrange) - logging: preserve driver, dom name & uuid against log file (Daniel P. Berrange) - qemu: include hostname in QEMU log files (Daniel P. Berrange) - rotatingfile: mark log files as close-on-exec (Daniel P. Berrange) - libvirtd: require virtlogd to start before libvirtd (Guido Günther) - schema: Allow > UINT_MAX KiB of memory for NUMA nodes (Peter Krempa) - virsh: remove custom error for cpulist from cmdIOThreadPin (Ján Tomko) - libxl: implement virDomainInterfaceStats (Joao Martins) - tests: Run virnetdaemontest iff WITH_YAJL (Michal Privoznik) - admin: Distribute libvirt-admin.conf (Martin Kletzander) - admin: Rename virAdmConnect to virAdmDaemon (Martin Kletzander) - spec: Temporarily disable new admin-related files (Martin Kletzander) - admin: Include admin_remote.c in the dist package (Martin Kletzander) - build: Create needed folders without dependency tracking (Martin Kletzander) - util: Avoid variable named 'truncate' shadowing global declaration (Martin Kletzander) - conf: Split virDomainObjList into a separate file (Michal Privoznik) - qemu: build command line for virtio-input-host device (Ján Tomko) - qemu: add passed-through input devs to cgroup ACL (Ján Tomko) - security: label the evdev for input device passthrough (Ján Tomko) - conf: add XML for input device passthrough (Ján Tomko) - qemu: add capability for virtio-input-host-device (Ján Tomko) - qemu: build command line for virtio input devices (Ján Tomko) - conf: parse and format virtio input bus in domain XML (Ján Tomko) - qemu: add capabilities for virtio input devices (Ján Tomko) - admin: Introduce virAdmConnectGetLibVersion (Erik Skultety) - admin: Add support for connection close callbacks (Erik Skultety) - admin: Add support for URI aliases (Erik Skultety) - livirt: Move URI alias matching to util (Erik Skultety) - admin: Add URI support and introduce virAdmGetDefaultURI (Erik Skultety) - admin: Do not generate remoteAdminConnect{Open,Close} (Erik Skultety) - admin: Move remote admin API version to a separate module (Erik Skultety) - admin: Introduce virAdmConnectIsAlive (Erik Skultety) - virt-admin: Introduce first working skeleton (Erik Skultety) - admin: introduce virAdmGetVersion (Erik Skultety) - libvirt: Move config getters to util (Erik Skultety) - admin: Introduce libvirt-admin.conf (Erik Skultety) - libvirt: introduce libvirt/libvirt-common.h.in (Erik Skultety) - qemu: add virtio-gpu virgl support (Marc-André Lureau) - qemu: add virtio video device (Marc-André Lureau) - domain: replace bool accel{2d, 3d} with a tristate (Marc-André Lureau) - Replace support{2d,3d} with accel{2d,3d} (Marc-André Lureau) - logging: avoid variables called 'daemon' due to function clash (Daniel P. Berrange) - logging: inhibit virtlogd shutdown while log files are open (Daniel P. Berrange) - qemu: add support for sending QEMU stdout/stderr to virtlogd (Daniel P. Berrange) - qemu: convert monitor to use qemuDomainLogContextPtr indirectly (Daniel P. Berrange) - qemu: convert process stop/attach to use qemuDomainLogContextPtr (Daniel P. Berrange) - qemu: convert qemuLogOperation to take a qemuDomainLogContextPtr (Daniel P. Berrange) - qemu: change qemuDomainTaint APIs to accept qemuDomainLogContextPtr (Daniel P. Berrange) - qemu: convert log file creation to use qemuDomainLogContextPtr (Daniel P. Berrange) - qemu: introduce a qemuDomainLogContext object (Daniel P. Berrange) - qemu: unify code for reporting errors from QEMU log files (Daniel P. Berrange) - qemu: remove writing to QEMU log file for rename operation (Daniel P. Berrange) - logging: add client for virtlogd daemon (Daniel P. Berrange) - logging: introduce log handling protocol (Daniel P. Berrange) - Import stripped down virtlockd code as basis of virtlogd (Daniel P. Berrange) - util: add APIs for reading/writing from/to rotating files (Daniel P. Berrange) - virsh: Try to keep printed XML pretty with change-media (Martin Kletzander) - qemu: Use qemuProcessLaunch in migration Prepare phase (Jiri Denemark) - qemu: Skip starting NBD servers for offline migration (Jiri Denemark) - qemu: Kill QEMU process if Prepare phase fails (Jiri Denemark) - qemu: Separate incoming URI generation from qemuMigrationPrepareAny (Jiri Denemark) - qemu: Introduce qemuProcessFinishStartup (Jiri Denemark) - qemu: Introduce qemuProcessLaunch (Jiri Denemark) - qemu: Introduce qemuProcessInit (Jiri Denemark) - conf: reject multiple panic devices of same model (Dmitry Andreev) - Allow multiple panic devices (Dmitry Andreev) - qemu: add support for hv_crash feature as a panic device (Dmitry Andreev) - tests: add tests for the new panic device attribute - 'model' (Dmitry Andreev) - conf: add 'model' attribute for panic device with values isa, pseries, hyperv (Dmitry Andreev) - conf: refactor code for checking ABI stability of panic device (Dmitry Andreev) - nodedev: report maxCount for virtual_functions capability (Laine Stump) - conf: support reporting maxCount attribute for virtual_functions cap (Laine Stump) - Post-release version bump to 1.3.0 (Pavel Hrdina) - conf: Drop useless check when parsing cpu scheduler info (Peter Krempa) - qemu: pass the asyncJob to qemuProcessStartCPUs (Ján Tomko) - xenapi: Refactor extraction of vcpu count (Peter Krempa) - phyp: Refactor extraction of vcpu count (Peter Krempa) - openvz: Refactor extraction of vcpu count (Peter Krempa) - hyperv: Allocate 'def' via virDomainDefNew (Peter Krempa) - qemuSetupChrSourceCgroup: rename dev to source (Ján Tomko) - Simplify qemuSetupChrSourceCgroup and its callers (Ján Tomko) - rename qemuSetupHostdevCGroup to qemuSetupHostdevCgroup (Ján Tomko) - qemu: handle more machines with a single builtin IDE controller (Guido Günther) - qemu: Always set locked memory limit for ppc64 domains (Andrea Bolognani) - qemu: Use qemuDomainRequiresMlock() when attaching PCI hostdev (Andrea Bolognani) - qemu: Use qemuDomainRequiresMlock() in qemuBuildCommandLine() (Andrea Bolognani) - process: Log when limiting the amount of locked memory (Andrea Bolognani) - vz: implementation of domainReboot callback (Mikhail Feoktistov) - vz: allow only en-us keymap for VNC (Mikhail Feoktistov) - qemu: Close logfd when closing monitor (Jiri Denemark) - qemu: Do not infer flags from other qemuProcessStart arguments (Jiri Denemark) - qemu: Introduce qemuProcessMakeDir (Jiri Denemark) - qemu: Separate balloon code from qemuProcessStart (Jiri Denemark) - qemu: Enter monitor within qemuProcessSetLinkStates (Jiri Denemark) - qemu: Separate raw IO code from qemuProcessStart (Jiri Denemark) - qemu: Separate graphics handling code from qemuProcessStart (Jiri Denemark) - qemu: Separate hook handling code from qemuProcessStart (Jiri Denemark) - qemu: Rename stdin_{fd,path} in qemuProcessStart (Jiri Denemark) - qemu: Use -incoming defer for migrations (Jiri Denemark) - qemu: Add APIs for migrate-incoming QMP command (Jiri Denemark) - qemu: Always set async job when starting a domain (Jiri Denemark) - qemu: Introduce qemuProcessIncomingDef (Jiri Denemark) - qemu: Move incoming URI code to qemu_migration (Jiri Denemark) - qemu: Don't generate migration URI in qemuBuildCommandLine (Jiri Denemark) - qemu: Refactor the code to build -incoming command line (Jiri Denemark) - qemu: Refactor waiting for completed migration on destination (Jiri Denemark) - util: add virDiskNameParse to handle disk and partition idx (Joao Martins) - libxl: implement virDomainMemorystats (Joao Martins) - lxc: Bind mount container TTYs (Richard Weinberger) - lxc: Don't make container's TTY a controlling TTY (Richard Weinberger) - qemu: ppc64: Support memory hotplug without NUMA enabled (Peter Krempa) - qemu: command: Prepare memory device def formatter for missing target node (Peter Krempa) - conf: Prepare making memory device target node optional (Peter Krempa) - qemu: command: Move dimm device checks from formatter to checker (Peter Krempa) - qemu: domain: Add common function to perform memory hotplug checks (Peter Krempa) - qemu: command: Always execute memory device formatter (Peter Krempa) - qemu: command: Make qemuBuildMemoryBackendStr usable without NUMA (Peter Krempa) - libxl: implement virDomainGetCPUStats (Joao Martins) - syntax-check: Add prohibit_space_in_label rule (Andrea Bolognani) - util: remove unnecessary needSize (Chen Hanxiao) - storage: Introduce virStoragePoolObjFindPoolByUUID (John Ferlan) - storage: Change cbdata scsi refresh thread field name (John Ferlan) - storage: Make active boolean (John Ferlan) - qemu: domain: Restructurate control flow in qemuDomainGetMlockLimitBytes (Peter Krempa) - qemu: Fix job entry debug message (Jiri Denemark) - tests: Add QEMU 2.4.0 capabilities (Jiri Denemark) - tests: Remove qemuxmlnstest (Jiri Denemark) - qemu: Fix style in qemuProcessStart (Jiri Denemark) - security: Cleanup DAC driver (Jiri Denemark) - domain-conf: cleanup controller insert function (Pavel Hrdina) - virsh-domain: update attach-interface to support type=hostdev (Pavel Hrdina) - vz: support cpu time in driver's domainGetInfo (Nikolay Shirokovskiy) - qemu: assume various QEMU 0.10 features are always available (Daniel P. Berrange) - qemu: assume -vga is always available (Daniel P. Berrange) - qemu: assume -drive format is always available (Daniel P. Berrange) - qemu: assume -drive cache always uses v2 option names (Daniel P. Berrange) - qemu: assume support for all migration protocols except rdma (Daniel P. Berrange) - qemu: assume vnet-hdr feature is always available (Daniel P. Berrange) - qemu: really remove last traces of Xenner support (Daniel P. Berrange) - qemu: assume -uuid is always available (Daniel P. Berrange) - qemu: assume -name is always available (Daniel P. Berrange) - qemu: assume -drive argument is always available (Daniel P. Berrange) - qemu: handle USB bus in qemuAssignDeviceDiskAliasFixed() (Daniel P. Berrange) - qemu: assume -no-reboot is always available (Daniel P. Berrange) - qemu: assume 'info chardev' is always available (Daniel P. Berrange) - qemu: assume -vnc arg always takes a ':' (Daniel P. Berrange) - qemu: remove all support for kQEMU (Daniel P. Berrange) - qemu: mandate QEMU version 0.12.0 or newer (Daniel P. Berrange) - qemu: hotplug: Reject VFIO hotplug if setting RLIMIT_MEMLOCK fails (Peter Krempa) - qemu: Extract logic to determine the mlock limit size for VFIO (Peter Krempa) - conf: Make @def const in virDomainDefGetMemoryInitial (Peter Krempa) - tests: redo test argv file line wrapping (Daniel P. Berrange) - virnetdev: Use virNetDevSetupControl in virNetDevSendEthtoolIoctl (John Ferlan) - virnetdev: Check for root in virNetDevGetFeatures (John Ferlan) - qemu: add /usr/lib to AC_PATH_PROG for qemu-bridge-helper (Michel Normand) - storage: Pull volume removal from pool in storageVolDeleteInternal (John Ferlan) - storage: Cleanup failures in virStorageBackendCreateRaw (John Ferlan) - storage: Cleanup failures virStorageBackendCreateExecCommand (John Ferlan) - storage: Fix setting mode in virStorageBackendCreateExecCommand (John Ferlan) - Remove new lines from log messages (Jiri Denemark) - qemu: Introduce cleanup label in qemuProcessStart (Jiri Denemark) - qemu: Rename ret variable in qemuProcessStart (Jiri Denemark) - qemu: Rename cleanup label in qemuProcessStart (Jiri Denemark) So thanks everybody for helping with this release, be it with patches, ideas, reports, documentation, etc... As a reminder next release supposedly 1.3.1 will come mid-January due to the end of year slowdown, and then 1.3.2 at the end or February. Enjoy ! 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