- Devel - libvirt List Archives

[libvirt] [PATCH] Revert "Error out on missing machine type in machine configs"
by Ján Tomko 25 Feb '16

25 Feb '16

Revert commit 55e6d8cd9eac7eb2aaa4d221585e9402cf7269d5. It unconditionally required a machine type for all machine types even though qemu is the only emulator using them. Reverting this re-introduces the crash when someone fiddles with libvirt's machine configs with /etc/, but fixes persistent domains for other drivers. Breaks https://bugzilla.redhat.com/show_bug.cgi?id=1267256 again. --- A proper fix would be too invasive for the freeze, we would either * have to revert commit f1a89a8 which relaxed the requirement to execute the emulator * or (partially) revert commits a8b628e and 3d92a00 which reintroduced them to post parse callbacks and keep track of which parser requires the machine type in virDomainDefParserConfig. src/conf/domain_conf.c | 6 ------ 1 file changed, 6 deletions(-) diff --git a/src/conf/domain_conf.c b/src/conf/domain_conf.c index 3b15cb4..79758d4 100644 --- a/src/conf/domain_conf.c +++ b/src/conf/domain_conf.c @@ -14851,12 +14851,6 @@ virDomainDefParseXML(xmlDocPtr xml, goto error; } VIR_FREE(capsdata); - } else { - if (!def->os.machine) { - virReportError(VIR_ERR_INTERNAL_ERROR, "%s", - _("Missing machine type")); - goto error; - } } /* Extract domain name */ -- 2.4.10

2 5

[libvirt] [PATCH] libxl: implement virDomainInterfaceStats
by Joao Martins 25 Feb '16

25 Feb '16

Introduce support for domainInterfaceStats API call for querying network interface statistics. Consequently it also enables the use of `virsh domifstat <dom> <interface name>` command plus seeing the interfaces names instead of "-" when doing `virsh domiflist <dom>`. After successful guest creation we fill the network interfaces names based on domain, device id and append suffix if it's emulated in the following form: vif<domid>.<devid>[-emu]. We extract the network interfaces info from the libxl_domain_config object in libxlDomainCreateIfaceNames() to generate ifname. On domain cleanup we also clear ifname, in case it was set by libvirt (i.e. being prefixed with "vif"). We also skip these two steps in case the name of the interface was manually inserted by the adminstrator. Since the introdution of netprefix (commit a040ba9), ifnames with a registered prefix will be freed on virDomain{Obj,Def}Format*, thus eliminating the migration issues observed with the reverted commit d2e5538 whereas source and destination would have the same ifname. For getting the interface statistics we resort to virNetInterfaceStats and let libvirt handle the platform specific nits. Note that the latter is not yet supported in FreeBSD. Signed-off-by: Joao Martins <joao.m.martins(a)oracle.com> --- Changes since d2e5538: - Check net->ifname beforehand, preventing a crash on the migration failure path. - Use LIBXL_GENERATED_PREFIX_XEN as opposed to hardcoding "vif" - Bump from 1.3.0 to 1.3.2 - Change commit message to mention the migration issue and netprefix. --- src/libxl/libxl_domain.c | 39 ++++++++++++++++++++++++++++++++++++ src/libxl/libxl_driver.c | 52 ++++++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 91 insertions(+) diff --git a/src/libxl/libxl_domain.c b/src/libxl/libxl_domain.c index 50f7eed..9d0da68 100644 --- a/src/libxl/libxl_domain.c +++ b/src/libxl/libxl_domain.c @@ -762,6 +762,18 @@ libxlDomainCleanup(libxlDriverPrivatePtr driver, } } + if ((vm->def->nnets)) { + size_t i; + + for (i = 0; i < vm->def->nnets; i++) { + virDomainNetDefPtr net = vm->def->nets[i]; + + if (net->ifname && + STRPREFIX(net->ifname, LIBXL_GENERATED_PREFIX_XEN)) + VIR_FREE(net->ifname); + } + } + if (virAsprintf(&file, "%s/%s.xml", cfg->stateDir, vm->def->name) > 0) { if (unlink(file) < 0 && errno != ENOENT && errno != ENOTDIR) VIR_DEBUG("Failed to remove domain XML for %s", vm->def->name); @@ -929,6 +941,32 @@ libxlConsoleCallback(libxl_ctx *ctx, libxl_event *ev, void *for_callback) libxl_event_free(ctx, ev); } +/* + * Create interface names for the network devices in parameter def. + * Names are created with the pattern 'vif<domid>.<devid><suffix>'. + * devid is extracted from the network devices in the d_config + * parameter. User-provided interface names are skipped. + */ +static void +libxlDomainCreateIfaceNames(virDomainDefPtr def, libxl_domain_config *d_config) +{ + size_t i; + + for (i = 0; i < def->nnets && i < d_config->num_nics; i++) { + virDomainNetDefPtr net = def->nets[i]; + libxl_device_nic *x_nic = &d_config->nics[i]; + const char *suffix = + x_nic->nictype != LIBXL_NIC_TYPE_VIF ? "-emu" : ""; + + if (net->ifname) + continue; + + ignore_value(virAsprintf(&net->ifname, + LIBXL_GENERATED_PREFIX_XEN "%d.%d%s", + def->id, x_nic->devid, suffix)); + } +} + /* * Start a domain through libxenlight. @@ -1073,6 +1111,7 @@ libxlDomainStart(libxlDriverPrivatePtr driver, virDomainObjPtr vm, if (libxl_evenable_domain_death(cfg->ctx, vm->def->id, 0, &priv->deathW)) goto cleanup_dom; + libxlDomainCreateIfaceNames(vm->def, &d_config); if ((dom_xml = virDomainDefFormat(vm->def, cfg->caps, 0)) == NULL) goto cleanup_dom; diff --git a/src/libxl/libxl_driver.c b/src/libxl/libxl_driver.c index 404016e..5479441 100644 --- a/src/libxl/libxl_driver.c +++ b/src/libxl/libxl_driver.c @@ -58,6 +58,7 @@ #include "virhostdev.h" #include "network/bridge_driver.h" #include "locking/domain_lock.h" +#include "virstats.h" #define VIR_FROM_THIS VIR_FROM_LIBXL @@ -4662,6 +4663,56 @@ libxlDomainIsUpdated(virDomainPtr dom) } static int +libxlDomainInterfaceStats(virDomainPtr dom, + const char *path, + virDomainInterfaceStatsPtr stats) +{ + libxlDriverPrivatePtr driver = dom->conn->privateData; + virDomainObjPtr vm; + ssize_t i; + int ret = -1; + + if (!(vm = libxlDomObjFromDomain(dom))) + goto cleanup; + + if (virDomainInterfaceStatsEnsureACL(dom->conn, vm->def) < 0) + goto cleanup; + + if (libxlDomainObjBeginJob(driver, vm, LIBXL_JOB_QUERY) < 0) + goto cleanup; + + if (!virDomainObjIsActive(vm)) { + virReportError(VIR_ERR_OPERATION_INVALID, + "%s", _("domain is not running")); + goto endjob; + } + + /* Check the path is one of the domain's network interfaces. */ + for (i = 0; i < vm->def->nnets; i++) { + if (vm->def->nets[i]->ifname && + STREQ(vm->def->nets[i]->ifname, path)) { + ret = 0; + break; + } + } + + if (ret == 0) + ret = virNetInterfaceStats(path, stats); + else + virReportError(VIR_ERR_INVALID_ARG, + _("'%s' is not a known interface"), path); + + endjob: + if (!libxlDomainObjEndJob(driver, vm)) + vm = NULL; + + cleanup: + if (vm) + virObjectUnlock(vm); + return ret; +} + +static int libxlDomainGetTotalCPUStats(libxlDriverPrivatePtr driver, virDomainObjPtr vm, virTypedParameterPtr params, @@ -5539,6 +5590,7 @@ static virHypervisorDriver libxlHypervisorDriver = { .domainGetJobStats = libxlDomainGetJobStats, /* 1.3.1 */ .domainMemoryStats = libxlDomainMemoryStats, /* 1.3.0 */ .domainGetCPUStats = libxlDomainGetCPUStats, /* 1.3.0 */ + .domainInterfaceStats = libxlDomainInterfaceStats, /* 1.3.2 */ .connectDomainEventRegister = libxlConnectDomainEventRegister, /* 0.9.0 */ .connectDomainEventDeregister = libxlConnectDomainEventDeregister, /* 0.9.0 */ .domainManagedSave = libxlDomainManagedSave, /* 0.9.2 */ -- 2.1.4

2 2

[libvirt] [PATCH] Error out on missing machine type in machine configs
by Ján Tomko 25 Feb '16

25 Feb '16

Commit f1a89a8 allowed parsing configs from /etc/libvirt without validating the emulator capabilities. Check for the presence of os->type.machine even if the VIR_DOMAIN_DEF_PARSE_SKIP_OSTYPE_CHECKS flag is set, otherwise the daemon can crash on carelessly crafted input in the config directory. https://bugzilla.redhat.com/show_bug.cgi?id=1267256 --- src/conf/domain_conf.c | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/src/conf/domain_conf.c b/src/conf/domain_conf.c index 67415fa..5d3fed0 100644 --- a/src/conf/domain_conf.c +++ b/src/conf/domain_conf.c @@ -14841,6 +14841,12 @@ virDomainDefParseXML(xmlDocPtr xml, goto error; } VIR_FREE(capsdata); + } else { + if (!def->os.machine) { + virReportError(VIR_ERR_INTERNAL_ERROR, "%s", + _("Missing machine type")); + goto error; + } } /* Extract domain name */ -- 2.4.10

2 3

[libvirt] [PATCH] util: Avoid calling closedir(NULL)
by Jiri Denemark 25 Feb '16

25 Feb '16

Signed-off-by: Jiri Denemark <jdenemar(a)redhat.com> --- src/util/virnetdev.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/util/virnetdev.c b/src/util/virnetdev.c index ea95552..5c79c42 100644 --- a/src/util/virnetdev.c +++ b/src/util/virnetdev.c @@ -3104,63 +3104,63 @@ static int virNetDevRDMAFeature(const char *ifname, virBitmapPtr *out) { char *eth_devpath = NULL; char *ib_devpath = NULL; char *eth_res_buf = NULL; char *ib_res_buf = NULL; DIR *dirp = NULL; struct dirent *dp; int ret = -1; if (!virFileExists(SYSFS_INFINIBAND_DIR)) return 0; if (!(dirp = opendir(SYSFS_INFINIBAND_DIR))) { virReportSystemError(errno, _("Failed to opendir path '%s'"), SYSFS_INFINIBAND_DIR); - goto cleanup; + return -1; } if (virAsprintf(&eth_devpath, SYSFS_NET_DIR "%s/device/resource", ifname) < 0) goto cleanup; if (!virFileExists(eth_devpath)) goto cleanup; if (virFileReadAll(eth_devpath, RESOURCE_FILE_LEN, &eth_res_buf) < 0) goto cleanup; while (virDirRead(dirp, &dp, SYSFS_INFINIBAND_DIR) > 0) { if (dp->d_name[0] == '.') continue; if (virAsprintf(&ib_devpath, SYSFS_INFINIBAND_DIR "%s/device/resource", dp->d_name) < 0) continue; if (virFileReadAll(ib_devpath, RESOURCE_FILE_LEN, &ib_res_buf) > 0 && STREQ(eth_res_buf, ib_res_buf)) { ignore_value(virBitmapSetBit(*out, VIR_NET_DEV_FEAT_RDMA)); break; } VIR_FREE(ib_devpath); VIR_FREE(ib_res_buf); } ret = 0; cleanup: closedir(dirp); VIR_FREE(eth_devpath); VIR_FREE(ib_devpath); VIR_FREE(eth_res_buf); VIR_FREE(ib_res_buf); return ret; } /** * virNetDevSendEthtoolIoctl * This function sends ethtool ioctl request * * @ifname: name of the interface * @cmd: reference to an ethtool command structure * * Returns 0 on success, -1 on failure. */ -- 2.7.2

2 1

[libvirt] [PATCH v2 0/2] Enable detection for zero writes
by Martin Kletzander 25 Feb '16

25 Feb '16

This is a second version with few things fixed. The main difference is that now this series uses zeroes instead of zeros. The latter was chosen at first just because there were more occurences in our codebase, but the former is used now to minimise the confusion -- mainly when none of the versions were used when interfacing with the user or mgmt app. The naming is kept from the previous version despite the discussion as I believe this is pretty general. Also the default value is not changed to 'off' in order for it to be general enough (some hypervisors might choose a different default in the future). 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 + tests/qemucapabilitiesdata/caps_2.6.0-1.caps | 1 + .../qemuxml2argv-disk-drive-detect-zeroes.args | 27 +++++++++++++ .../qemuxml2argv-disk-drive-detect-zeroes.xml | 45 ++++++++++++++++++++++ tests/qemuxml2argvtest.c | 3 ++ .../qemuxml2xmlout-disk-drive-detect-zeroes.xml | 1 + tests/qemuxml2xmltest.c | 1 + 17 files changed, 152 insertions(+), 1 deletion(-) create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-disk-drive-detect-zeroes.args create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-disk-drive-detect-zeroes.xml create mode 120000 tests/qemuxml2xmloutdata/qemuxml2xmlout-disk-drive-detect-zeroes.xml -- 2.7.1

2 3

[libvirt] [PATCH V2] Fix bug of attaching redirdev device
by Osier Yang 25 Feb '16

25 Feb '16

RHBZ: https://bugzilla.redhat.com/show_bug.cgi?id=1298070 The corresponding chardev must be attached first, otherwise the the qemu command line won't be complete (missing the host part), --- src/qemu/qemu_hotplug.c | 15 ++++++++++++++- 1 file changed, 14 insertions(+), 1 deletion(-) diff --git a/src/qemu/qemu_hotplug.c b/src/qemu/qemu_hotplug.c index ee305e7..40cead7 100644 --- a/src/qemu/qemu_hotplug.c +++ b/src/qemu/qemu_hotplug.c @@ -1381,6 +1381,7 @@ int qemuDomainAttachRedirdevDevice(virQEMUDriverPtr driver, int ret; qemuDomainObjPrivatePtr priv = vm->privateData; virDomainDefPtr def = vm->def; + char *charAlias = NULL; char *devstr = NULL; if (!virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_DEVICE)) { @@ -1391,6 +1392,10 @@ int qemuDomainAttachRedirdevDevice(virQEMUDriverPtr driver, if (qemuAssignDeviceRedirdevAlias(vm->def, redirdev, -1) < 0) goto error; + + if (virAsprintf(&charAlias, "char%s", redirdev->info.alias) < 0) + goto error; + if (!(devstr = qemuBuildRedirdevDevStr(def, redirdev, priv->qemuCaps))) goto error; @@ -1398,6 +1403,14 @@ int qemuDomainAttachRedirdevDevice(virQEMUDriverPtr driver, goto error; qemuDomainObjEnterMonitor(driver, vm); + if (qemuMonitorAttachCharDev(priv->mon, + charAlias, + &(redirdev->source.chr)) < 0) { + ignore_value(qemuDomainObjExitMonitor(driver, vm)); + goto error; + } + VIR_FREE(charAlias); + ret = qemuMonitorAddDevice(priv->mon, devstr); if (qemuDomainObjExitMonitor(driver, vm) < 0) @@ -1414,9 +1427,9 @@ int qemuDomainAttachRedirdevDevice(virQEMUDriverPtr driver, return 0; error: + VIR_FREE(charAlias); VIR_FREE(devstr); return -1; - } static int -- 2.1.4

2 3

[libvirt] [PATCH] vircgroupmock: Mock access("/sys/devices/system/cpu/present")
by Michal Privoznik 25 Feb '16

25 Feb '16

There's been a report on the upstream list [1] describing we access /sys/devices/system/cpu/present directly on the host from within our test suite. This may end up in unpredictable results as no all linux systems are required to have that file. Mock access to the file. libvirt.git/tests $ ../run strace vircgrouptest ... access("/sys/devices/system/cpu/present", F_OK) = 0 open("libvirt.git/tests/fakerootdir-DmE8Z9/not/really/sys/fs/cgroup//devices_system_cpu_present", O_RDONLY) = 4 read(4, "8-23,48-159\n", 5121) = 12 read(4, "", 5109) = 0 close(4) = 0 ... Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com> --- tests/vircgroupmock.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/tests/vircgroupmock.c b/tests/vircgroupmock.c index 9ce7d41..756ac51 100644 --- a/tests/vircgroupmock.c +++ b/tests/vircgroupmock.c @@ -541,7 +541,8 @@ int access(const char *path, int mode) ret = realaccess(newpath, mode); free(newpath); } else if (STREQ(path, "/proc/cgroups") || - STREQ(path, "/proc/self/cgroup")) { + STREQ(path, "/proc/self/cgroup") || + STREQ(path, SYSFS_CPU_PRESENT)) { /* These files are readable for all. */ ret = (mode == F_OK || mode == R_OK) ? 0 : -1; } else if (STREQ(path, "/proc/mounts")) { -- 2.4.10

2 1

[libvirt] how to test libvirt?
by Zhangbo (Oscar) 25 Feb '16

25 Feb '16

Hi all: AFAIK, there're several ways to test libvirt: 1) virt-test, which is driven by autotest or avocado-vt, based on qemu-kvm 2) test driver which is parall to other drivers such as qemu_driver and libxl_driver in libvirt. 3) make -c tests, which aims to do low level tests, that is to test function modules of libvirt. 4) libvirt-test-API I've got 2 questions: 1) besides the 4 test suites, are there any other automatic-test tools? 2) is there any testsuite used to test libxl_driver, similar to virt-test? Anyone knows that? Thanks in advance. Oscar.

4 5

[libvirt] [PATCH] build: accomodate selinux 2.5 header API change
by Eric Blake 24 Feb '16

24 Feb '16

Yet again, selinux has been adding const-correctness; this change is ABI-compatible, but breaks API, which affects us when we try to override things in our testsuite: ../../tests/securityselinuxhelper.c:307:24: error: conflicting types for 'selabel_open' struct selabel_handle *selabel_open(unsigned int backend, ^~~~~~~~~~~~ In file included from ../../tests/securityselinuxhelper.c:32:0: /usr/include/selinux/label.h:73:24: note: previous declaration of 'selabel_open' was here The problem is a new 'const' prior to the second parameter. Fix it the same way we did in commit 292d3f2d: check for the new const at configure time. Signed-off-by: Eric Blake <eblake(a)redhat.com> --- Pushing under the build-breaker rule. m4/virt-selinux.m4 | 17 ++++++++++++++++- tests/securityselinuxhelper.c | 10 ++++++---- 2 files changed, 22 insertions(+), 5 deletions(-) diff --git a/m4/virt-selinux.m4 b/m4/virt-selinux.m4 index 357b758..a65e978 100644 --- a/m4/virt-selinux.m4 +++ b/m4/virt-selinux.m4 @@ -1,6 +1,6 @@ dnl The libselinux.so library dnl -dnl Copyright (C) 2012-2014 Red Hat, Inc. +dnl Copyright (C) 2012-2014, 2016 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 @@ -41,6 +41,21 @@ int setcon(char *context); AC_DEFINE_UNQUOTED([VIR_SELINUX_CTX_CONST], [$lv_cv_setcon_const], [Define to empty or 'const' depending on how SELinux qualifies its security context parameters]) + # ...and again for 2.5 + AC_CACHE_CHECK([for selinux selabel_open parameter type], + [lv_cv_selabel_open_const], + [AC_COMPILE_IFELSE( + [AC_LANG_PROGRAM( + [[ +#include <selinux/selinux.h> +#include <selinux/label.h> +struct selabel_handle *selabel_open(unsigned, struct selinux_opt *, unsigned); + ]])], + [lv_cv_selabel_open_const=''], + [lv_cv_selabel_open_const='const'])]) + AC_DEFINE_UNQUOTED([VIR_SELINUX_OPEN_CONST], [$lv_cv_selabel_open_const], + [Define to empty or 'const' depending on how SELinux qualifies its + selabel_open parameter]) AC_MSG_CHECKING([SELinux mount point]) if test "$with_selinux_mount" = "check" || test -z "$with_selinux_mount"; then diff --git a/tests/securityselinuxhelper.c b/tests/securityselinuxhelper.c index 1252c15..2cfa43f 100644 --- a/tests/securityselinuxhelper.c +++ b/tests/securityselinuxhelper.c @@ -1,5 +1,5 @@ /* - * Copyright (C) 2011-2013 Red Hat, Inc. + * Copyright (C) 2011-2013, 2016 Red Hat, Inc. * * This library is free software; you can redistribute it and/or * modify it under the terms of the GNU Lesser General Public @@ -60,6 +60,7 @@ static const char *(*realselinux_lxc_contexts_path)(void); #if HAVE_SELINUX_LABEL_H static struct selabel_handle *(*realselabel_open)(unsigned int backend, + VIR_SELINUX_OPEN_CONST struct selinux_opt *opts, unsigned nopts); static void (*realselabel_close)(struct selabel_handle *handle); @@ -304,9 +305,10 @@ const char *selinux_lxc_contexts_path(void) #endif #if HAVE_SELINUX_LABEL_H -struct selabel_handle *selabel_open(unsigned int backend, - struct selinux_opt *opts, - unsigned nopts) +struct selabel_handle * +selabel_open(unsigned int backend, + VIR_SELINUX_OPEN_CONST struct selinux_opt *opts, + unsigned nopts) { char *fake_handle; -- 2.5.0

1 0

[libvirt] Dealing with GCC 6.0
by Martin Kletzander 24 Feb '16

24 Feb '16

So we recently started compiling with gcc 5 and there's a new thing now and we don't build on gcc 6. I know it's just a development build, but it finds new things, like: fdstream.c: In function 'virFDStreamWrite': fdstream.c:390:29: error: logical 'or' of equal expressions [-Werror=logical-op] if (errno == EAGAIN || errno == EWOULDBLOCK) { ^~ fdstream.c: In function 'virFDStreamRead': fdstream.c:440:29: error: logical 'or' of equal expressions [-Werror=logical-op] if (errno == EAGAIN || errno == EWOULDBLOCK) { ^~ and rpc/virnetsshsession.c: In function 'virNetSSHAuthenticateAgent': rpc/virnetsshsession.c:548:56: error: logical 'and' of equal expressions [-Werror=logical-op] if (ret != LIBSSH2_ERROR_AUTHENTICATION_FAILED && ^~ rpc/virnetsshsession.c: In function 'virNetSSHAuthenticatePrivkey': rpc/virnetsshsession.c:676:57: error: logical 'or' of equal expressions [-Werror=logical-op] if (ret == LIBSSH2_ERROR_PUBLICKEY_UNRECOGNIZED || ^~ The first one is prety easy to fix, that's the only place where we use EWOULDBLOCK, we are using only EAGAIN everywhere else. However, the second one is not that easy to change for me as I'm not a libssh expert and moreover, I would like to know other's opinions on how to tackle that. We can disable the logical-op warning, but it might show something that makes sense (really?). Or we can do bunch of very very ugly "#if"s. Anyone care to share an idea? Have a nice day, Martin

3 4