- Devel - libvirt List Archives

[libvirt] [Question] vhost-user hotplug: multiqueue hotplug seems doesn't pass mq and vectors
by Gaohaifeng (A) 01 Dec '16

01 Dec '16

Hi ALL: Recently I review the vhost-user hotplug codes, when attach a vhost-user nic with multiqueue, it seems that it doesn't pass mq=on and vectors num to qemu. Is there any special consideration or just a small mistake? Parts of codes: qemuDomainAttachNetDevice(vhostfdSize is 0 when type is vhost-user) qemuBuildNicDevStr: if (usingVirtio && vhostfdSize > 1) { if (net->info.type == VIR_DOMAIN_DEVICE_ADDRESS_TYPE_CCW) { /* ccw provides a one to one relation of fds to queues and * does not support the vectors option */ virBufferAddLit(&buf, ",mq=on"); } else { /* As advised at http://www.linux-kvm.org/page/Multiqueue * we should add vectors=2*N+2 where N is the vhostfdSize */ virBufferAsprintf(&buf, ",mq=on,vectors=%zu", 2 * vhostfdSize + 2); } } Thanks. Haifeng Gao

2 4

[libvirt] [PATCH] qemu: Prevent detaching SCSI controller used by hostdev
by Eric Farman 30 Nov '16

30 Nov '16

Consider a guest started with the following XML snippet: <controller type='scsi' model='virtio-scsi' index='0'/> <hostdev mode='subsystem' type='scsi'> <source> <adapter name='scsi_host0'/> <address bus='0' target='8' unit='1074151456'/> </source> </hostdev> If we attach/create a virtio-scsi hostdev device but forget to include a virtio-scsi controller, one is silently created for us. (See qemuDomainFindOrCreateSCSIDiskController for context.) Detaching the hostdev, followed by the controller, works well and the guest behaves appropriately. If we detach the virtio-scsi controller device first, Libvirt silently detaches any associated hostdevs for us too. But all is not well, as the guest is unable to receive new virtio-scsi devices (the attach commands succeed, but devices never appear within the guest), nor even be shutdown, after this point. It appears that something is tied up in the host virtio layer. While I investigate this, we can see if a controller is being used by any hostdevs, and prevent the detach if it would lead us down this path. $ virsh detach-device guest_one_virtio_scsi scsicontroller.xml error: Failed to detach device from scsicontroller.xml error: operation failed: device cannot be detached: device is busy $ virsh detach-device guest_one_virtio_scsi scsidisk.xml Device detached successfully $ virsh detach-device guest_one_virtio_scsi scsicontroller.xml Device detached successfully Signed-off-by: Eric Farman <farman(a)linux.vnet.ibm.com> Reviewed-by: Bjoern Walk <bwalk(a)linux.vnet.ibm.com> Reviewed-by: Boris Fiuczynski <fiuczy(a)linux.vnet.ibm.com> --- src/qemu/qemu_hotplug.c | 10 ++++++++++ 1 file changed, 10 insertions(+) diff --git a/src/qemu/qemu_hotplug.c b/src/qemu/qemu_hotplug.c index b03e618..5db7abf 100644 --- a/src/qemu/qemu_hotplug.c +++ b/src/qemu/qemu_hotplug.c @@ -4443,6 +4443,7 @@ static bool qemuDomainDiskControllerIsBusy(virDomainObjPtr vm, { size_t i; virDomainDiskDefPtr disk; + virDomainHostdevDefPtr hostdev; for (i = 0; i < vm->def->ndisks; i++) { disk = vm->def->disks[i]; @@ -4465,6 +4466,15 @@ static bool qemuDomainDiskControllerIsBusy(virDomainObjPtr vm, return true; } + for (i = 0; i < vm->def->nhostdevs; i++) { + hostdev = vm->def->hostdevs[i]; + if (!virHostdevIsSCSIDevice(hostdev) || + detach->type != VIR_DOMAIN_CONTROLLER_TYPE_SCSI) + continue; + if (hostdev->info->addr.drive.controller == detach->idx) + return true; + } + return false; } -- 2.8.4

2 5

[libvirt] [PATCH] loopback is not always just lo
by Pavel Timofeev 30 Nov '16

30 Nov '16

On BSD family OSes (Free/Net/Open/DragonFlyBSD, Mac OS) and Solaris loopback interface is called 'lo0' instead of just 'lo'. --- src/network/bridge_driver.c | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/src/network/bridge_driver.c b/src/network/bridge_driver.c index a3ee3f3..9d94d65 100644 --- a/src/network/bridge_driver.c +++ b/src/network/bridge_driver.c @@ -1002,7 +1002,12 @@ networkDnsmasqConfContents(virNetworkObjPtr network, virBufferAsprintf(&configbuf, "pid-file=%s\n", pidfile); /* dnsmasq will *always* listen on localhost unless told otherwise */ +#ifdef __linux__ virBufferAddLit(&configbuf, "except-interface=lo\n"); +#else + /* BSD family OSes and Solaris call loopback interface as lo0 */ + virBufferAddLit(&configbuf, "except-interface=lo0\n"); +#endif if (dnsmasqCapsGet(caps, DNSMASQ_CAPS_BIND_DYNAMIC)) { /* using --bind-dynamic with only --interface (no -- 2.9.2

5 7

[libvirt] [PATCH v2] cgroup: Use system reported "unlimited" value for comparison
by Viktor Mihajlovski 30 Nov '16

30 Nov '16

With kernel 3.18 (since commit 3e32cb2e0a12b6915056ff04601cf1bb9b44f967) the "unlimited" value for cgroup memory limits has changed once again as its byte value is now computed from a page counter. The new "unlimited" value reported by the cgroup fs is therefore 2**51-1 pages which is (VIR_DOMAIN_MEMORY_PARAM_UNLIMITED - 3072). This results e.g. in virsh memtune displaying 9007199254740988 instead of unlimited for the limits. This patch uses the value of memory.limit_in_bytes from the cgroup memory root which is the system's "real" unlimited value for comparison. See also libvirt commit 231656bbeb9e4d3bedc44362784c35eee21cf0f4 for the history for kernel 3.12 and before. Signed-off-by: Viktor Mihajlovski <mihajlov(a)linux.vnet.ibm.com> --- v2: - removed RFC verbiage from commit message - per Martin's review comment, cache the cgroup memory.limit_in_bytes - used the cgroup detection logic proposed by Martin, much nicer now indeed - other than initially planned, fall back to VIR_DOMAIN_MEMORY_PARAM_UNLIMITED in case of cgroup read failure, since the usual paranoia fits in nicely with the "already initialized" check. With the current code structure this will never be called when the memory controller is not configured or mounted anyway... - while at it, replaced the goto cleanups with direct returns as there's really no cleanup to be done in the GetMemoryxxxLimit functions src/util/vircgroup.c | 67 ++++++++++++++++++++++++++++++++++++++-------------- 1 file changed, 49 insertions(+), 18 deletions(-) diff --git a/src/util/vircgroup.c b/src/util/vircgroup.c index f151193..76d9738 100644 --- a/src/util/vircgroup.c +++ b/src/util/vircgroup.c @@ -2452,6 +2452,46 @@ virCgroupGetBlkioDeviceWeight(virCgroupPtr group, } +/* + * Retrieve the "memory.limit_in_bytes" value from the memory controller + * root dir. This value cannot be modified by userspace and therefore + * is the maximum limit value supported by cgroups on the local system. + * Returns this value scaled to KB or falls back to the original + * VIR_DOMAIN_MEMORY_PARAM_UNLIMITED. Either way, remember the return + * value to avoid unnecessary cgroup filesystem access. + */ +static unsigned long long int virCgroupMemoryUnlimitedKB; + +static unsigned long long int +virCgroupGetMemoryUnlimitedKB(void) +{ + virCgroupPtr group; + unsigned long long int mem_unlimited; + + if (virCgroupMemoryUnlimitedKB) { + return virCgroupMemoryUnlimitedKB; + } else { + mem_unlimited = VIR_DOMAIN_MEMORY_PARAM_UNLIMITED << 10; + + if (virCgroupNew(-1, "/", NULL, -1, &group) < 0) + goto cleanup; + + if (!virCgroupHasController(group, VIR_CGROUP_CONTROLLER_MEMORY)) + goto cleanup; + + ignore_value(virCgroupGetValueU64(group, + VIR_CGROUP_CONTROLLER_MEMORY, + "memory.limit_in_bytes", + &mem_unlimited)); + } + + cleanup: + virCgroupFree(&group); + virCgroupMemoryUnlimitedKB = mem_unlimited >> 10; + return virCgroupMemoryUnlimitedKB; +} + + /** * virCgroupSetMemory: * @@ -2534,20 +2574,17 @@ int virCgroupGetMemoryHardLimit(virCgroupPtr group, unsigned long long *kb) { long long unsigned int limit_in_bytes; - int ret = -1; if (virCgroupGetValueU64(group, VIR_CGROUP_CONTROLLER_MEMORY, "memory.limit_in_bytes", &limit_in_bytes) < 0) - goto cleanup; + return -1; *kb = limit_in_bytes >> 10; - if (*kb > VIR_DOMAIN_MEMORY_PARAM_UNLIMITED) + if (*kb >= virCgroupGetMemoryUnlimitedKB()) *kb = VIR_DOMAIN_MEMORY_PARAM_UNLIMITED; - ret = 0; - cleanup: - return ret; + return 0; } @@ -2596,20 +2633,17 @@ int virCgroupGetMemorySoftLimit(virCgroupPtr group, unsigned long long *kb) { long long unsigned int limit_in_bytes; - int ret = -1; if (virCgroupGetValueU64(group, VIR_CGROUP_CONTROLLER_MEMORY, "memory.soft_limit_in_bytes", &limit_in_bytes) < 0) - goto cleanup; + return -1; *kb = limit_in_bytes >> 10; - if (*kb > VIR_DOMAIN_MEMORY_PARAM_UNLIMITED) + if (*kb >= virCgroupGetMemoryUnlimitedKB()) *kb = VIR_DOMAIN_MEMORY_PARAM_UNLIMITED; - ret = 0; - cleanup: - return ret; + return 0; } @@ -2658,20 +2692,17 @@ int virCgroupGetMemSwapHardLimit(virCgroupPtr group, unsigned long long *kb) { long long unsigned int limit_in_bytes; - int ret = -1; if (virCgroupGetValueU64(group, VIR_CGROUP_CONTROLLER_MEMORY, "memory.memsw.limit_in_bytes", &limit_in_bytes) < 0) - goto cleanup; + return -1; *kb = limit_in_bytes >> 10; - if (*kb > VIR_DOMAIN_MEMORY_PARAM_UNLIMITED) + if (*kb >= virCgroupGetMemoryUnlimitedKB()) *kb = VIR_DOMAIN_MEMORY_PARAM_UNLIMITED; - ret = 0; - cleanup: - return ret; + return 0; } -- 1.9.1

1 0

[libvirt] [PATCH 0/2] cpu: Add alternative feature spellings to CPU map
by Jiri Denemark 30 Nov '16

30 Nov '16

Jiri Denemark (2): cpu: Remove useless comments from CPU map cpu: Add alternative feature spellings to CPU map src/cpu/cpu_map.xml | 154 ++++++++++++++++++++++++++-------------------------- 1 file changed, 77 insertions(+), 77 deletions(-) -- 2.11.0.rc2

3 5

[libvirt] [PATCH v2 0/4] qemu: assign vfio devices to PCIe addresses when appropriate
by Laine Stump 30 Nov '16

30 Nov '16

Patches 1 and 4 were originally added to the end of the "more PCIe less legacy PCI" patchset in its final incarnation, but all the other patches were ACKed and pushed, while this needed a bit more work, resulting in this "faux V2" - although it's the 2nd time I've posted these patches, their "V1" was really inside the "V11666" of a larger series. In order to implement Alex's suggestion of checking the length of a PCI device's config file to determine if it's an Express device, I first made a utility function to return the length of a file, and then another to return the name of the config file of a virPCIDevice - those are the new patches 2 and 3. I'm open to suggestions about additional checks to put in virFileLength(), as long as they don't involve opening the file. For now I've kept it as simple as possible. (Just keep in mind that I'll be away from my keyboard from Wednesday through Sunday this week). Laine Stump (4): util: new function virFileLength() util: new function virPCIDeviceGetConfigPath() qemu: propagate virQEMUDriver object to qemuDomainDeviceCalculatePCIConnectFlags qemu: assign vfio devices to PCIe addresses when appropriate src/libvirt_private.syms | 2 + src/qemu/qemu_domain.c | 2 +- src/qemu/qemu_domain_address.c | 117 ++++++++++++++++++++++++++++++++++++----- src/qemu/qemu_domain_address.h | 7 ++- src/qemu/qemu_hotplug.c | 19 +++---- src/qemu/qemu_process.c | 13 +++-- src/util/virfile.c | 13 +++++ src/util/virfile.h | 2 + src/util/virpci.c | 6 +++ src/util/virpci.h | 1 + tests/qemuhotplugtest.c | 4 +- 11 files changed, 157 insertions(+), 29 deletions(-) -- 2.9.3

3 16

[libvirt] [PATCH] daemon: Install virt-guest-shutdown.target properly
by Michal Privoznik 30 Nov '16

30 Nov '16

When trying to install libvirtd from sources I've noticed the following failure: /usr/bin/install: cannot stat 'virt-guest-shutdown.target': No such file or directory Makefile:2792: recipe for target 'install-init-systemd' failed make[3]: *** [install-init-systemd] Error 1 make[3]: *** Waiting for unfinished jobs.... The problem is that while other files around that location in Makefile are firstly generated into the builddir and only after that installed, virt-guest-shutdown.target file is not generated at all and should be installed from the srcdir. This was introduced in 01079727. Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com> --- Pushed under build-breaker & trivial rules. daemon/Makefile.am | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/daemon/Makefile.am b/daemon/Makefile.am index 463db6e..60c7368 100644 --- a/daemon/Makefile.am +++ b/daemon/Makefile.am @@ -457,7 +457,7 @@ install-init-systemd: install-sysconfig libvirtd.service $(MKDIR_P) $(DESTDIR)$(SYSTEMD_UNIT_DIR) $(INSTALL_DATA) libvirtd.service \ $(DESTDIR)$(SYSTEMD_UNIT_DIR)/libvirtd.service - $(INSTALL_DATA) virt-guest-shutdown.target \ + $(INSTALL_DATA) $(srcdir)/virt-guest-shutdown.target \ $(DESTDIR)$(SYSTEMD_UNIT_DIR)/virt-guest-shutdown.target uninstall-init-systemd: uninstall-sysconfig -- 2.8.4

2 1

[libvirt] [PATCH] virt-gnutls.m4: fixed check for gnutls_rnd and gnutls_cipher_encrypt
by Daniel P. Berrange 30 Nov '16

30 Nov '16

From: Nikos Mavrogiannopoulos <nmav(a)redhat.com> --- m4/virt-gnutls.m4 | 3 +++ 1 file changed, 3 insertions(+) I thought this was onlist already, but there was a typo in To, so neither the original patch or my ACK made it to the list. I pushed as a build fix. diff --git a/m4/virt-gnutls.m4 b/m4/virt-gnutls.m4 index b850828..5bca950 100644 --- a/m4/virt-gnutls.m4 +++ b/m4/virt-gnutls.m4 @@ -54,8 +54,11 @@ AC_DEFUN([LIBVIRT_CHECK_GNUTLS],[ #include <gnutls/gnutls.h> ]]) + OLD_LIBS="$LIBS" + LIBS="$LIBS $GNUTLS_LIBS" AC_CHECK_FUNC([gnutls_rnd]) AC_CHECK_FUNC([gnutls_cipher_encrypt]) + LIBS=$OLD_LIBS fi ]) -- 2.9.3

1 0

[libvirt] [PATCH] docs: improve release numbering rule for minor numbers
by Boris Fiuczynski 30 Nov '16

30 Nov '16

Adding first build of year minor number reset to 0. Signed-off-by: Boris Fiuczynski <fiuczy(a)linux.vnet.ibm.com> --- docs/downloads.html.in | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/docs/downloads.html.in b/docs/downloads.html.in index f34e9a6..1f2ec1d 100644 --- a/docs/downloads.html.in +++ b/docs/downloads.html.in @@ -383,7 +383,8 @@ <dd>incremented by 1 for the first release of the year (the Jan 15th release)</dd> <dt><code>minor</code></dt> - <dd>incremented by 1 for each monthly release from git master</dd> + <dd>set to 0 with every major increment or incremented by 1 + for each monthly release from git master</dd> <dt><code>micro</code></dt> <dd>always 0 for releases from git master, incremented by 1 for each stable maintenance release</dd> -- 2.5.5

2 4

[libvirt] [PATCH] fix parsing security labels from virt-aa-helper
by Christian Ehrhardt 30 Nov '16

30 Nov '16

When parsing labels virt-aa-helper does no more pass VIR_DOMAIN_DEF_PARSE_INACTIVE due to dfbc9a83 that tried to mitigate the changes of a89f05ba. For those it had to switch from VIR_DOMAIN_DEF_PARSE_INACTIVE to active since we need the domain id (ctl->def->id) as it is part of the socket path now which is needed for the aa profile. But that turned out to break non apparmor seclabels as well as apparmor seclabels in xmls without labels. In those cases due to VIR_DOMAIN_DEF_PARSE_INACTIVE now not set anymore virSecurityLabelDefParseXML insists on finding labels. Cases: - non-apparmor seclabel - virt-aa-helper breaks - apparmor seclabel without labels on a defined domain - virt-aa-helper breaks This was not spotted due to labels getting dynamically created on definition. So "define, start, stop" works. But "define, edit (add label), start" does not. Now turning back on VIR_DOMAIN_DEF_PARSE_INACTIVE would cause the old bug, so we have to differ those more fine grained. This is done by the new flag VIR_DOMAIN_DEF_PARSE_SKIP_ACTIVE_LABEL which is like VIR_DOMAIN_DEF_PARSE_INACTIVE but only for the security labels. So far only set by virt-aa-helper. Testcase with virt-aa-helper on xml file: virt-aa-helper -d -r -p 0 -u libvirt-<uuid> < your-guest.xml virt-aa-helper: error: could not parse XML virt-aa-helper: error: could not get VM definition (That should have printed a valid apparmor profile) Signed-off-by: Christian Ehrhardt <christian.ehrhardt(a)canonical.com> --- src/conf/domain_conf.c | 6 ++++-- src/conf/domain_conf.h | 3 +++ src/security/virt-aa-helper.c | 1 + 3 files changed, 8 insertions(+), 2 deletions(-) diff --git a/src/conf/domain_conf.c b/src/conf/domain_conf.c index 03506cb..9eb7883 100644 --- a/src/conf/domain_conf.c +++ b/src/conf/domain_conf.c @@ -6626,7 +6626,8 @@ virSecurityLabelDefParseXML(xmlXPathContextPtr ctxt, * if the 'live' VM XML is requested */ if (seclabel->type == VIR_DOMAIN_SECLABEL_STATIC || - (!(flags & VIR_DOMAIN_DEF_PARSE_INACTIVE) && + (!(flags & (VIR_DOMAIN_DEF_PARSE_SKIP_ACTIVE_LABEL | + VIR_DOMAIN_DEF_PARSE_INACTIVE)) && seclabel->type != VIR_DOMAIN_SECLABEL_NONE)) { p = virXPathStringLimit("string(./label[1])", VIR_SECURITY_LABEL_BUFLEN-1, ctxt); @@ -6642,7 +6643,8 @@ virSecurityLabelDefParseXML(xmlXPathContextPtr ctxt, /* Only parse imagelabel, if requested live XML with relabeling */ if (seclabel->relabel && - (!(flags & VIR_DOMAIN_DEF_PARSE_INACTIVE) && + (!(flags & (VIR_DOMAIN_DEF_PARSE_SKIP_ACTIVE_LABEL | + VIR_DOMAIN_DEF_PARSE_INACTIVE)) && seclabel->type != VIR_DOMAIN_SECLABEL_NONE)) { p = virXPathStringLimit("string(./imagelabel[1])", VIR_SECURITY_LABEL_BUFLEN-1, ctxt); diff --git a/src/conf/domain_conf.h b/src/conf/domain_conf.h index 24aa79c..90693c6 100644 --- a/src/conf/domain_conf.h +++ b/src/conf/domain_conf.h @@ -2657,6 +2657,9 @@ typedef enum { VIR_DOMAIN_DEF_PARSE_ABI_UPDATE = 1 << 9, /* skip definition validation checks meant to be executed on define time only */ VIR_DOMAIN_DEF_PARSE_SKIP_VALIDATE = 1 << 10, + /* in regard to security labels, skip parts of the XML that would only be + * present in an active libvirt XML. */ + VIR_DOMAIN_DEF_PARSE_SKIP_ACTIVE_LABEL = 1 << 11, } virDomainDefParseFlags; typedef enum { diff --git a/src/security/virt-aa-helper.c b/src/security/virt-aa-helper.c index 77eeaff..0ca4c83 100644 --- a/src/security/virt-aa-helper.c +++ b/src/security/virt-aa-helper.c @@ -705,6 +705,7 @@ get_definition(vahControl * ctl, const char *xmlStr) ctl->def = virDomainDefParseString(xmlStr, ctl->caps, ctl->xmlopt, NULL, + VIR_DOMAIN_DEF_PARSE_SKIP_ACTIVE_LABEL | VIR_DOMAIN_DEF_PARSE_SKIP_VALIDATE); if (ctl->def == NULL) { -- 2.7.4

4 14