- Devel - libvirt List Archives

[libvirt] [PATCH] qemu: Use proper agent entering function when freezing filesystems
by Peter Krempa 02 Oct '12

02 Oct '12

When doing snapshots, the filesystem freeze function used the agent entering function that expects the qemud_driver unlocked. This might cause a deadlock of the qemu driver if the agent does not respond. The only call path of this function has the qemud_driver locked, so this patch changes the entering functions to those expecting the driver locked. --- src/qemu/qemu_driver.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c index b0a0bb5..50f44ce 100644 --- a/src/qemu/qemu_driver.c +++ b/src/qemu/qemu_driver.c @@ -10375,6 +10375,7 @@ qemuDomainSnapshotIsAllowed(virDomainObjPtr vm) return true; } +/* this function expects the driver lock to be held by the caller */ static int qemuDomainSnapshotFSFreeze(struct qemud_driver *driver, virDomainObjPtr vm) { @@ -10393,9 +10394,9 @@ qemuDomainSnapshotFSFreeze(struct qemud_driver *driver, return -1; } - qemuDomainObjEnterAgent(driver, vm); + qemuDomainObjEnterAgentWithDriver(driver, vm); freezed = qemuAgentFSFreeze(priv->agent); - qemuDomainObjExitAgent(driver, vm); + qemuDomainObjExitAgentWithDriver(driver, vm); return freezed; } -- 1.7.12

2 2

[libvirt] only 256 active networks
by Dan Kenigsberg 02 Oct '12

02 Oct '12

Hi List, Laine, Maybe you can help here. We define 256 bridged networks such as <network> <name>vdsm-test244</name> <uuid>c3acd4c3-37be-3f9d-ee6b-317513015f9e</uuid> <forward mode='bridge'/> <bridge name='test244' /> </network> and end up killing `virsh net-list` with error: Failed to list active networks error: too many remote undefineds: 257 > 256 error: Reconnected to the hypervisor What is this limitation? Why is it there? How can this be extended/avoided? Dan.

2 2

[libvirt] [PATCH] lxc: Correctly report active cgroups
by Michal Privoznik 02 Oct '12

02 Oct '12

There was an inverted return value in lxcCgroupControllerActive(). The function assumes cgroups are active and do couple of checks to prove that. If any of them fails, false is returned. Therefore, at the end, after all checks are done we must return true, not false. --- src/lxc/lxc_driver.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/src/lxc/lxc_driver.c b/src/lxc/lxc_driver.c index ae5163e..87305db 100644 --- a/src/lxc/lxc_driver.c +++ b/src/lxc/lxc_driver.c @@ -1645,7 +1645,7 @@ static bool lxcCgroupControllerActive(virLXCDriverPtr driver, if (driver->cgroupControllers & (1 << controller)) return true; #endif - return false; + return true; } -- 1.7.8.6

2 2

[libvirt] [PATCH 1/2] build: avoid journald on rhel 5
by Eric Blake 02 Oct '12

02 Oct '12

Commit f6430390 broke builds on RHEL 5, where glibc (2.5) is too old to support mkostemp (2.7) or htole64 (2.9). While gnulib has mkostemp, it still lacks htole64; and it's not worth dragging in replacements on systems where journald is unlikely to exist in the first place, so we just use an extra configure-time check as our witness of whether to attempt compiling the code. * src/util/logging.c (virLogParseOutputs): Don't attempt to compile journald on older glibc. * configure.ac (AC_CHECK_FUNCS_ONCE): Check for htole64. --- Pushing under the build-breaker rule. configure.ac | 2 +- src/util/logging.c | 16 ++++++++++------ 2 files changed, 11 insertions(+), 7 deletions(-) diff --git a/configure.ac b/configure.ac index 13967e9..c791ed0 100644 --- a/configure.ac +++ b/configure.ac @@ -172,7 +172,7 @@ AC_CHECK_SIZEOF([long]) dnl Availability of various common functions (non-fatal if missing), dnl and various less common threadsafe functions AC_CHECK_FUNCS_ONCE([cfmakeraw geteuid getgid getgrnam_r getmntent_r \ - getpwuid_r getuid initgroups kill mmap newlocale posix_fallocate \ + getpwuid_r getuid htole64 initgroups kill mmap newlocale posix_fallocate \ posix_memalign regexec sched_getaffinity]) dnl Availability of pthread functions (if missing, win32 threading is diff --git a/src/util/logging.c b/src/util/logging.c index 0ce18f1..7a73553 100644 --- a/src/util/logging.c +++ b/src/util/logging.c @@ -50,6 +50,12 @@ #include "virtime.h" #include "intprops.h" +/* Journald output is only supported on Linux new enough to expose + * htole64. */ +#if HAVE_SYSLOG_H && defined(__linux__) && HAVE_HTOLE64 +# define USE_JOURNALD 1 +#endif + #define VIR_FROM_THIS VIR_FROM_NONE VIR_ENUM_DECL(virLogSource) @@ -1029,7 +1035,7 @@ virLogAddOutputToSyslog(virLogPriority priority, } -# ifdef __linux__ +# if USE_JOURNALD # define IOVEC_SET_STRING(iov, str) \ do { \ struct iovec *_i = &(iov); \ @@ -1197,7 +1203,7 @@ static int virLogAddOutputToJournald(int priority) } return 0; } -# endif /* __linux__ */ +# endif /* USE_JOURNALD */ #endif /* HAVE_SYSLOG_H */ #define IS_SPACE(cur) \ @@ -1294,12 +1300,10 @@ virLogParseOutputs(const char *outputs) VIR_FREE(abspath); } else if (STREQLEN(cur, "journald", 8)) { cur += 8; -#if HAVE_SYSLOG_H -# ifdef __linux__ +#if USE_JOURNALD if (virLogAddOutputToJournald(prio) == 0) count++; -# endif /* __linux__ */ -#endif /* HAVE_SYSLOG_H */ +#endif /* USE_JOURNALD */ } else { goto cleanup; } -- 1.7.11.4

2 4

[libvirt] [PATCH] build: avoid infinite autogen loop
by Eric Blake 02 Oct '12

02 Oct '12

Several people have reported that if the .gnulib submodule is dirty, then 'make' will go into an infinite loop attempting to rerun bootstrap, because that never cleans up the dirty submodule. By default, we should halt and make the user investigate, but if the user doesn't know why or care that the submodule is dirty, I also added the ability to 'make CLEAN_SUBMODULE=1' to get things going again. Also, while testing this, I noticed that when a submodule update was needed, 'make' would first run autoreconf, then bootstrap (which reruns autoreconf); adding a strategic dependency allows for less work. * .gnulib: Update to latest, for maint.mk improvements. * cfg.mk (_autogen): Also hook maint.mk, to run before autoreconf. * autogen.sh (bootstrap): Refuse to run if gnulib is dirty, unless user requests discarding gnulib changes. --- * .gnulib 440a1db...b493832 (39): > sockets, sys_stat: restore AC_C_INLINE > localeconv tests: Avoid test failure on OpenIndiana. > havelib: Follow libtool developments. > ChangeLog: mention who reported the problem. > fstatat.c: fix a compile-impeding typo > extern-inline: provide a -Wundef safe config.h > hash-pjw: relax license to LGPLv2+ > autoupdate > maint.mk: fix strict vs. lazy variable issues with RELEASE > maint.mk: formatting changes > maint.mk: provide "make upload" to ease uploading > maint.mk: factor the validation of RELEASE_TYPE > maint.mk: silent rules > localename: port gl_locale_name_thread_unsafe to FreeBSD > binary-io, eealloc, mbfile, mbiter, mbutil, xsize: better 'inline' > pipe-filter-gi, pipe-filter-ii: better use of 'inline' > fdutimensat: omit unnecessary AC_C_INLINE > fchmodat, fchownat, fstatat: use extern-inline > acl, mbchar, priv-set: use extern-inline > sockets, sys_stat: remove AC_C_INLINE in MSVC-only cases > tls-tests: omit unnecessary 'inline' > utimens-tests: avoid unnecessary 'inline' > misc: don't limit commentary to inline functions > non-recursive-gnulib-prefix-hack: new module > ChangeLog: fix indentation > maint.mk: generalize _gl_tight_scope for non-recursive make > maint.mk: exempt trailing blanks found in "binary" files > maint.mk: sc_prohibit_path_max_allocation: don't FP for UNIX_PATH_MAX > maint.mk: teach sc_prohibit_magic_number_exit to accept 77 > maint.mk: relax sc_prohibit_strcmp, to avoid a false positive > localcharset: work around Mac OS X bug with UTF-8 and MB_CUR_MAX > doc: document sticky-EOF issue > poll: fix poll(0, NULL, msec) > poll: fix for systems that can't recv() on a non-socket > poll/select: document portability problems not fixed by Gnulib. > update from texinfo > Fix typo in previous patch: 1 -> 4. > fcntl-h: check for AIX 7.1 bug with O_NOFOLLOW and O_CREAT > net_if: give more details about the bug being fixed .gnulib | 2 +- autogen.sh | 11 +++++++++++ cfg.mk | 16 ++++++++-------- 3 files changed, 20 insertions(+), 9 deletions(-) diff --git a/.gnulib b/.gnulib index 440a1db..b493832 160000 --- a/.gnulib +++ b/.gnulib @@ -1 +1 @@ -Subproject commit 440a1dbe523e37f206252cb034c3a62f26867e42 +Subproject commit b4938324b7113c9d73f8390d21f3ecbd842e74b9 diff --git a/autogen.sh b/autogen.sh index 72f1e7a..d183397 100755 --- a/autogen.sh +++ b/autogen.sh @@ -63,11 +63,22 @@ bootstrap_hash() if test -d .git; then curr_status=.git-module-status t=$(bootstrap_hash; git diff .gnulib) + case $t:${CLEAN_SUBMODULE+set} in + *:set) ;; + *-dirty*) + echo "error: gnulib submodule is dirty, please investigate" 2>&1 + echo "set env-var CLEAN_SUBMODULE to discard gnulib changes" 2>&1 + exit 1 ;; + esac if test "$t" = "$(cat $curr_status 2>/dev/null)" \ && test -f "po/Makevars"; then # good, it's up to date, all we need is autoreconf autoreconf -if else + if test ${CLEAN_SUBMODULE+set}; then + echo cleaning up submodules... + git submodule foreach 'git clean -dfqx && git reset --hard' + fi echo running bootstrap$no_git... ./bootstrap$no_git --bootstrap-sync && bootstrap_hash > $curr_status \ || { echo "Failed to bootstrap, please investigate."; exit 1; } diff --git a/cfg.mk b/cfg.mk index bbfd4a2..e1fbf4f 100644 --- a/cfg.mk +++ b/cfg.mk @@ -700,10 +700,17 @@ ifeq (0,$(MAKELEVEL)) _clean_requested = $(filter %clean,$(MAKECMDGOALS)) ifeq (1,$(_update_required)$(_clean_requested)) $(info INFO: gnulib update required; running ./autogen.sh first) -Makefile: _autogen +maint.mk Makefile: _autogen endif endif +# It is necessary to call autogen any time gnulib changes. Autogen +# reruns configure, then we regenerate all Makefiles at once. +.PHONY: _autogen +_autogen: + $(srcdir)/autogen.sh + ./config.status + # Give credit where due: # Ensure that each commit author email address (possibly mapped via # git log's .mailmap) appears in our AUTHORS file. @@ -718,13 +725,6 @@ sc_check_author_list: && echo '$(ME): committer(s) not listed in AUTHORS' >&2; \ test $$fail = 0 -# It is necessary to call autogen any time gnulib changes. Autogen -# reruns configure, then we regenerate all Makefiles at once. -.PHONY: _autogen -_autogen: - $(srcdir)/autogen.sh - ./config.status - # regenerate HACKING as part of the syntax-check syntax-check: $(top_srcdir)/HACKING -- 1.7.1

2 3

[libvirt] [PATCH v2] Add systemd journal support
by Daniel P. Berrange 02 Oct '12

02 Oct '12

From: "Daniel P. Berrange" <berrange(a)redhat.com> Add support for logging to the systemd journal, using its simple client library. The benefit over syslog is that it accepts structured log data, so the journald can store individual items like code file/line/func separately from the string message. Tools which require structured log data can then query the journal to extract exactly what they desire without resorting to string parsing While systemd provides a simple client library for logging, it is more convenient for libvirt to directly write its own client code. This lets us build up the iovec's on the stack, avoiding the need to alloc memory when writing log messages. Changed in v2: - Add virFormatIntDecimal instead of using snprintf - Add comment about mkostemp - Fix declaration of linestr var to use size of linenr - Wrap in #ifdef __linux__ Signed-off-by: Daniel P. Berrange <berrange(a)redhat.com> --- src/libvirt_private.syms | 1 + src/util/logging.c | 184 +++++++++++++++++++++++++++++++++++++++++++++++ src/util/logging.h | 1 + src/util/util.c | 30 ++++++++ src/util/util.h | 5 +- 5 files changed, 220 insertions(+), 1 deletion(-) diff --git a/src/libvirt_private.syms b/src/libvirt_private.syms index dab607a..eebc52a 100644 --- a/src/libvirt_private.syms +++ b/src/libvirt_private.syms @@ -1233,6 +1233,7 @@ virFileUnlock; virFileWaitForDevices; virFileWriteStr; virFindFileInPath; +virFormatIntDecimal; virGetGroupID; virGetGroupName; virGetHostname; diff --git a/src/util/logging.c b/src/util/logging.c index cdd94fd..9e44880 100644 --- a/src/util/logging.c +++ b/src/util/logging.c @@ -35,6 +35,10 @@ #if HAVE_SYSLOG_H # include <syslog.h> #endif +#include <sys/socket.h> +#if HAVE_SYS_UN_H +# include <sys/un.h> +#endif #include "virterror_internal.h" #include "logging.h" @@ -44,6 +48,7 @@ #include "threads.h" #include "virfile.h" #include "virtime.h" +#include "intprops.h" #define VIR_FROM_THIS VIR_FROM_NONE @@ -146,6 +151,8 @@ virLogOutputString(virLogDestination ldest) return "syslog"; case VIR_LOG_TO_FILE: return "file"; + case VIR_LOG_TO_JOURNALD: + return "journald"; } return "unknown"; } @@ -1020,6 +1027,177 @@ virLogAddOutputToSyslog(virLogPriority priority, } return 0; } + + +# ifdef __linux__ +# define IOVEC_SET_STRING(iov, str) \ + do { \ + struct iovec *_i = &(iov); \ + _i->iov_base = (char*)str; \ + _i->iov_len = strlen(str); \ + } while (0) + +# define IOVEC_SET_INT(iov, buf, val) \ + do { \ + struct iovec *_i = &(iov); \ + _i->iov_base = virFormatIntDecimal(buf, sizeof(buf), val); \ + _i->iov_len = strlen(buf); \ + } while (0) + +static int journalfd = -1; + +static void +virLogOutputToJournald(virLogSource source, + virLogPriority priority, + const char *filename, + int linenr, + const char *funcname, + const char *timestamp ATTRIBUTE_UNUSED, + unsigned int flags, + const char *rawstr, + const char *str ATTRIBUTE_UNUSED, + void *data ATTRIBUTE_UNUSED) +{ + virCheckFlags(VIR_LOG_STACK_TRACE,); + int buffd = -1; + size_t niov = 0; + struct msghdr mh; + struct sockaddr_un sa; + union { + struct cmsghdr cmsghdr; + uint8_t buf[CMSG_SPACE(sizeof(int))]; + } control; + struct cmsghdr *cmsg; + /* We use /dev/shm instead of /tmp here, since we want this to + * be a tmpfs, and one that is available from early boot on + * and where unprivileged users can create files. */ + char path[] = "/dev/shm/journal.XXXXXX"; + char priostr[INT_BUFSIZE_BOUND(priority)]; + char linestr[INT_BUFSIZE_BOUND(linenr)]; + + /* First message takes upto 4 iovecs, and each + * other field needs 3, assuming they don't have + * newlines in them + */ +# define IOV_SIZE (4 + (5 * 3)) + struct iovec iov[IOV_SIZE]; + + if (strchr(rawstr, '\n')) { + uint64_t nstr; + /* If 'str' containes a newline, then we must + * encode the string length, since we can't + * rely on the newline for the field separator + */ + IOVEC_SET_STRING(iov[niov++], "MESSAGE\n"); + nstr = htole64(strlen(rawstr)); + iov[niov].iov_base = (char*)&nstr; + iov[niov].iov_len = sizeof(nstr); + niov++; + } else { + IOVEC_SET_STRING(iov[niov++], "MESSAGE="); + } + IOVEC_SET_STRING(iov[niov++], rawstr); + IOVEC_SET_STRING(iov[niov++], "\n"); + + IOVEC_SET_STRING(iov[niov++], "PRIORITY="); + IOVEC_SET_INT(iov[niov++], priostr, priority); + IOVEC_SET_STRING(iov[niov++], "\n"); + + IOVEC_SET_STRING(iov[niov++], "LIBVIRT_SOURCE="); + IOVEC_SET_STRING(iov[niov++], virLogSourceTypeToString(source)); + IOVEC_SET_STRING(iov[niov++], "\n"); + + IOVEC_SET_STRING(iov[niov++], "CODE_FILE="); + IOVEC_SET_STRING(iov[niov++], filename); + IOVEC_SET_STRING(iov[niov++], "\n"); + + IOVEC_SET_STRING(iov[niov++], "CODE_LINE="); + IOVEC_SET_INT(iov[niov++], linestr, linenr); + IOVEC_SET_STRING(iov[niov++], "\n"); + + IOVEC_SET_STRING(iov[niov++], "CODE_FUNC="); + IOVEC_SET_STRING(iov[niov++], funcname); + IOVEC_SET_STRING(iov[niov++], "\n"); + + memset(&sa, 0, sizeof(sa)); + sa.sun_family = AF_UNIX; + if (!virStrcpy(sa.sun_path, "/run/systemd/journal/socket", sizeof(sa.sun_path))) + return; + + memset(&mh, 0, sizeof(mh)); + mh.msg_name = &sa; + mh.msg_namelen = offsetof(struct sockaddr_un, sun_path) + strlen(sa.sun_path); + mh.msg_iov = iov; + mh.msg_iovlen = niov; + + if (sendmsg(journalfd, &mh, MSG_NOSIGNAL) >= 0) + return; + + if (errno != EMSGSIZE && errno != ENOBUFS) + return; + + /* Message was too large, so dump to temporary file + * and pass an FD to the journal + */ + + /* NB: mkostemp is not declared async signal safe by + * POSIX, but this is Linux only code and the GLibc + * impl is safe enough, only using open() and inline + * asm to read a timestamp (falling back to gettimeofday + * on some arches + */ + if ((buffd = mkostemp(path, O_CLOEXEC|O_RDWR)) < 0) + return; + + if (unlink(path) < 0) + goto cleanup; + + if (writev(buffd, iov, niov) < 0) + goto cleanup; + + mh.msg_iov = NULL; + mh.msg_iovlen = 0; + + memset(&control, 0, sizeof(control)); + mh.msg_control = &control; + mh.msg_controllen = sizeof(control); + + cmsg = CMSG_FIRSTHDR(&mh); + cmsg->cmsg_level = SOL_SOCKET; + cmsg->cmsg_type = SCM_RIGHTS; + cmsg->cmsg_len = CMSG_LEN(sizeof(int)); + memcpy(CMSG_DATA(cmsg), &buffd, sizeof(int)); + + mh.msg_controllen = cmsg->cmsg_len; + + sendmsg(journalfd, &mh, MSG_NOSIGNAL); + +cleanup: + VIR_LOG_CLOSE(buffd); +} + + +static void virLogCloseJournald(void *data ATTRIBUTE_UNUSED) +{ + VIR_LOG_CLOSE(journalfd); +} + + +static int virLogAddOutputToJournald(int priority) +{ + if ((journalfd = socket(AF_UNIX, SOCK_DGRAM, 0)) < 0) + return -1; + if (virSetInherit(journalfd, false) < 0) { + VIR_LOG_CLOSE(journalfd); + return -1; + } + if (virLogDefineOutput(virLogOutputToJournald, virLogCloseJournald, NULL, + priority, VIR_LOG_TO_JOURNALD, NULL, 0) < 0) { + return -1; + } + return 0; +} +# endif /* __linux__ */ #endif /* HAVE_SYSLOG_H */ #define IS_SPACE(cur) \ @@ -1114,6 +1292,12 @@ virLogParseOutputs(const char *outputs) count++; VIR_FREE(name); VIR_FREE(abspath); + } else if (STREQLEN(cur, "journald", 8)) { + cur += 8; +#if HAVE_SYSLOG_H + if (virLogAddOutputToJournald(prio) == 0) + count++; +#endif /* HAVE_SYSLOG_H */ } else { goto cleanup; } diff --git a/src/util/logging.h b/src/util/logging.h index a3ea821..4fe0c8e 100644 --- a/src/util/logging.h +++ b/src/util/logging.h @@ -80,6 +80,7 @@ typedef enum { VIR_LOG_TO_STDERR = 1, VIR_LOG_TO_SYSLOG, VIR_LOG_TO_FILE, + VIR_LOG_TO_JOURNALD, } virLogDestination; typedef enum { diff --git a/src/util/util.c b/src/util/util.c index 28f9ae3..43fdaf1 100644 --- a/src/util/util.c +++ b/src/util/util.c @@ -2124,6 +2124,36 @@ virDoubleToStr(char **strp, double number) return ret; } + +/** + * Format @val as a base-10 decimal number, in the + * buffer @buf of size @buflen. To allocate a suitable + * sized buffer, the INT_BUFLEN(int) macro should be + * used + * + * Returns pointer to start of the number in @buf + */ +char * +virFormatIntDecimal(char *buf, size_t buflen, int val) +{ + char *p = buf + buflen - 1; + *p = '\0'; + if (val >= 0) { + do { + *--p = '0' + (val % 10); + val /= 10; + } while (val != 0); + } else { + do { + *--p = '0' - (val % 10); + val /= 10; + } while (val != 0); + *--p = '-'; + } + return p; +} + + const char *virEnumToString(const char *const*types, unsigned int ntypes, int type) diff --git a/src/util/util.h b/src/util/util.h index 5ab36ed..4316ab1 100644 --- a/src/util/util.h +++ b/src/util/util.h @@ -210,7 +210,10 @@ char *virStrcpy(char *dest, const char *src, size_t destbytes) # define virStrcpyStatic(dest, src) virStrcpy((dest), (src), sizeof(dest)) int virDoubleToStr(char **strp, double number) - ATTRIBUTE_NONNULL(1) ATTRIBUTE_RETURN_CHECK; + ATTRIBUTE_NONNULL(1) ATTRIBUTE_RETURN_CHECK; + +char *virFormatIntDecimal(char *buf, size_t buflen, int val) + ATTRIBUTE_NONNULL(1) ATTRIBUTE_RETURN_CHECK; int virDiskNameToIndex(const char* str); char *virIndexToDiskName(int idx, const char *prefix); -- 1.7.11.4

2 2

[libvirt] [PATCH 1/1] Add note about numeric domain names to manpage
by Dave Allan 02 Oct '12

02 Oct '12

Clarify that domains with numeric names can only be identified by their domain id. --- tools/virsh.pod | 16 ++++++++++------ 1 file changed, 10 insertions(+), 6 deletions(-) diff --git a/tools/virsh.pod b/tools/virsh.pod index 2120429..e30f1b6 100644 --- a/tools/virsh.pod +++ b/tools/virsh.pod @@ -26,12 +26,16 @@ The basic structure of most virsh usage is: virsh [OPTION]... <command> <domain> [ARG]... -Where I<command> is one of the commands listed below, I<domain> is the numeric -domain id, or the domain name, or the domain UUID and I<ARGS> are command -specific options. There are a few exceptions to this rule in the cases where -the command in question acts on all domains, the entire machine, or directly -on the xen hypervisor. Those exceptions will be clear for each of those -commands. +Where I<command> is one of the commands listed below, I<domain> is the +numeric domain id, or the domain name, or the domain UUID and I<ARGS> +are command specific options. There are a few exceptions to this rule +in the cases where the command in question acts on all domains, the +entire machine, or directly on the xen hypervisor. Those exceptions +will be clear for each of those commands. Note: it is permissible to +give numeric names to domains, however, doing so will result in a +domain that can only be identified by domain id. In other words, if a +numeric value is supplied it will be interpreted as a domain id, not +as a name. The B<virsh> program can be used either to run one I<COMMAND> by giving the command and its arguments on the shell command line, or a I<COMMAND_STRING> -- 1.7.11.4

2 1

[libvirt] [PATCH 0/3] Transport network port data during live migration
by Kyle Mestery 01 Oct '12

01 Oct '12

This series of commits has the end goal of allowing per-port data stored in the Open vSwitch DB to be transported during live migration. This is done by first providing a generic infrastructure for transporting network data, adding some utility functions specific to Open vSwitch, and hooking the two together. The framework provided is generic in that other networking data could be transferred as well by simply adding in additional hooks as needed. An example of the XML being transported is shown here: <network> <interface index='1' vporttype='openvswitch' netdata='blah blah blah'/> <interface index='7' vporttype='openvswitch' netdata='blah blah blah'/> </network> ---- - V3 of this patch series fixes issues found by Laine during review around making the XML more generic. - V2 of this patch series fixes some issues found when migrating VMs on standard Linux bridges. Kyle Mestery (3): Add the ability for the Qemu V3 migration protocol to include transporting network configuration. A generic framework is proposed with this patch to allow for the transfer of opaque data. Add utility functions for Open vSwitch to both save per-port data before a live migration, and restore the per-port data after a live migration. Transport Open vSwitch per-port data during live migration by using the utility functions virNetDevOpenvswitchGetMigrateData() and virNetDevOpenvswitchSetMigrateData(). src/libvirt_private.syms | 2 + src/qemu/qemu_migration.c | 293 +++++++++++++++++++++++++++++++++++++++- src/util/virnetdevopenvswitch.c | 70 ++++++++++ src/util/virnetdevopenvswitch.h | 6 + 4 files changed, 369 insertions(+), 2 deletions(-) -- 1.7.11.4

5 15

[libvirt] curious about dnsmasq --conf-file=
by Gene Czarcinski 01 Oct '12

01 Oct '12

I understand why there currently is "--conf-file= " on the dnsmasq command line ... you do not want dnsmasq to use anything in /etc/dnsmasq.conf. OK, but dnsmasq is picking up a lot of command line parameters. Why aren't these parameters put into a libvirt specific conf file such as is done with radvd? Is there some reason that only command line parameters are used? I can understand that it could be a matter of style/personal choice by the individual that originally did it and, it currently works, so why mess with it? But, if a patch was created which changed things from the command line to a conf file, would that be accepted? BTW, I noticed that there are no tests for what is done with radvd but there are tests for the dnsmasq command line options. These tests would no longer apply and thus be deleted. Can anyone cite an example test for parameters put into a conf file? Or, have some of those tests become OBE? I must say that I have found the tests valuable in seeing and debugging additional command line parameters. Gene

2 1

[libvirt] the libnl3-devel problem
by Gene Czarcinski 01 Oct '12

01 Oct '12

Is the libnl3-devel problem where, if it is installed, libvirt will build using libnl3 but, on F17, netcf is still using libnl and the result is ... not pretty. If someone believes this is fixed in 0.10.2, they are in error. Question: Just how hard would be be to get netcf rebuilt and using libnl3? It is just that I keep getting bit in the rear end by this problem ... current NetworkManager requires libnl3-devel to build but, if you then build libvirt without removing libnl3-devel, everything looks good but it is not. You would think that I would remember from the last time I went around with this problem ... but id did not Gene

2 1

[libvirt] [PATCH] Fix (rare) deadlock in QEMU monitor callbacks
by Daniel P. Berrange 30 Sep '12

30 Sep '12

From: "Daniel P. Berrange" <berrange(a)redhat.com> Some users report (very rarely) seeing a deadlock in the QEMU monitor callbacks Thread 10 (Thread 0x7fcd11e20700 (LWP 26753)): at util/threads-pthread.c:85 at conf/domain_conf.c:14256 vm=0x7fcccc00a850) at qemu/qemu_process.c:1026 at qemu/qemu_monitor.c:249 at util/virobject.c:139 at qemu/qemu_monitor.c:860 vm=vm@entry=0x7fcccc00a850, reason=reason@entry=VIR_DOMAIN_SHUTOFF_DESTROYED, flags=flags@entry=0) at qemu/qemu_process.c:4057 flags=<optimized out>) at qemu/qemu_driver.c:1977 domain=domain@entry=0x7fccf00c1830, flags=1) at libvirt.c:2256 At frame #10 we are holding the domain lock, we call into qemuProcessStop() to cleanup QEMU, which triggers the monitor to close, which invokes qemuProcessHandleMonitorDestroy() which tries to obtain the domain lock again. This is a non-recursive lock, hence hang. Since qemuMonitorPtr is a virObject, the unref call in qemuProcessHandleMonitorDestroy no longer needs mutex protection. The assignment of priv->mon = NULL, can be instead done by the caller of qemuMonitorClose(), thus removing all need for locking. Signed-off-by: Daniel P. Berrange <berrange(a)redhat.com> --- src/qemu/qemu_process.c | 15 +++++---------- 1 file changed, 5 insertions(+), 10 deletions(-) diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c index ade64b7..c8c188a 100644 --- a/src/qemu/qemu_process.c +++ b/src/qemu/qemu_process.c @@ -1019,17 +1019,10 @@ no_memory: } -static void qemuProcessHandleMonitorDestroy(qemuMonitorPtr mon, +static void qemuProcessHandleMonitorDestroy(qemuMonitorPtr mon ATTRIBUTE_UNUSED, virDomainObjPtr vm) { - qemuDomainObjPrivatePtr priv; - - virDomainObjLock(vm); - priv = vm->privateData; - if (priv->mon == mon) - priv->mon = NULL; - if (virObjectUnref(vm)) - virDomainObjUnlock(vm); + virObjectUnref(vm); } static int @@ -3995,8 +3988,10 @@ void qemuProcessStop(struct qemud_driver *driver, priv->agentError = false; } - if (priv->mon) + if (priv->mon) { qemuMonitorClose(priv->mon); + priv->mon = NULL; + } if (priv->monConfig) { if (priv->monConfig->type == VIR_DOMAIN_CHR_TYPE_UNIX) -- 1.7.11.2

3 4

[libvirt] [PATCH] hyperv: Fix and improve hypervListAllDomains
by Matthias Bolte 30 Sep '12

30 Sep '12

Use MATCH for all flags checks. hypervMsvmComputerSystemToDomain expects the domain pointer to the initialized to NULL. All items in doms up to the count-th one are valid, no need to double check before freeing them. --- I could no runtime-test this because I've have trouble getting my Hyper-V setup working again. src/hyperv/hyperv_driver.c | 22 +++++++++++++--------- 1 files changed, 13 insertions(+), 9 deletions(-) diff --git a/src/hyperv/hyperv_driver.c b/src/hyperv/hyperv_driver.c index 65934c9..349627d 100644 --- a/src/hyperv/hyperv_driver.c +++ b/src/hyperv/hyperv_driver.c @@ -1302,13 +1302,14 @@ hypervListAllDomains(virConnectPtr conn, virBufferAddLit(&query, MSVM_COMPUTERSYSTEM_WQL_VIRTUAL); /* construct query with filter depending on flags */ - if (!(flags & VIR_CONNECT_LIST_DOMAINS_ACTIVE && - flags & VIR_CONNECT_LIST_DOMAINS_INACTIVE)) { - if (flags & VIR_CONNECT_LIST_DOMAINS_ACTIVE) { + if (!(MATCH(VIR_CONNECT_LIST_DOMAINS_ACTIVE) && + MATCH(VIR_CONNECT_LIST_DOMAINS_INACTIVE))) { + if (MATCH(VIR_CONNECT_LIST_DOMAINS_ACTIVE)) { virBufferAddLit(&query, "and "); virBufferAddLit(&query, MSVM_COMPUTERSYSTEM_WQL_ACTIVE); } - if (flags & VIR_CONNECT_LIST_DOMAINS_INACTIVE) { + + if (MATCH(VIR_CONNECT_LIST_DOMAINS_INACTIVE)) { virBufferAddLit(&query, "and "); virBufferAddLit(&query, MSVM_COMPUTERSYSTEM_WQL_INACTIVE); } @@ -1358,14 +1359,16 @@ hypervListAllDomains(virConnectPtr conn, continue; } - if (VIR_RESIZE_N(doms, ndoms, count, 2) < 0) + if (VIR_RESIZE_N(doms, ndoms, count, 2) < 0) goto no_memory; + domain = NULL; + if (hypervMsvmComputerSystemToDomain(conn, computerSystem, &domain) < 0) goto cleanup; - doms[count++] = domain; + doms[count++] = domain; } if (doms) @@ -1376,13 +1379,14 @@ hypervListAllDomains(virConnectPtr conn, cleanup: if (doms) { for (i = 0; i < count; ++i) { - if (doms[i]) - virDomainFree(doms[i]); + virDomainFree(doms[i]); } + + VIR_FREE(doms); } - VIR_FREE(doms); hypervFreeObject(priv, (hypervObject *)computerSystemList); + return ret; no_memory: -- 1.7.4.1

2 2

[libvirt] [PATCH v2] esx: Fix and improve esxListAllDomains function
by Matthias Bolte 30 Sep '12

30 Sep '12

Avoid requesting information such as identity or power state when it is not necessary. Lookup virtual machine list with the required fields (configStatus, name, and config.uuid) to make esxVI_GetVirtualMachineIdentity work. No need to call esxVI_GetNumberOfSnapshotTrees. rootSnapshotTreeList can be tested for emptiness by checking it for NULL. esxVI_LookupRootSnapshotTreeList already does the error reporting, don't overwrite it. Check if autostart is enabled at all before looking up the individual autostart setting of a virtual machine. Reorder VIR_EXPAND_N(doms, ndoms, 1) to avoid leaking the result of the call to virGetDomain if VIR_EXPAND_N fails. Replace VIR_EXPAND_N by VIR_RESIZE_N to avoid quadratic scaling, as in the Hyper-V version of the function. If virGetDomain fails it already reports an error, don't overwrite it with an OOM error. All items in doms up to the count-th one are valid, no need to double check before freeing them. Finally, don't leak autoStartDefaults and powerInfoList. --- v2: Replace VIR_EXPAND_N by VIR_RESIZE_N to avoid quadratic scaling, as in the Hyper-V version of the function. src/esx/esx_driver.c | 116 ++++++++++++++++++++++++++++++++----------------- 1 files changed, 76 insertions(+), 40 deletions(-) diff --git a/src/esx/esx_driver.c b/src/esx/esx_driver.c index 991f03c..cf0551f 100644 --- a/src/esx/esx_driver.c +++ b/src/esx/esx_driver.c @@ -5008,13 +5008,15 @@ esxListAllDomains(virConnectPtr conn, { int ret = -1; esxPrivate *priv = conn->privateData; + bool needIdentity; + bool needPowerState; virDomainPtr dom; virDomainPtr *doms = NULL; size_t ndoms = 0; + esxVI_String *propertyNameList = NULL; esxVI_ObjectContent *virtualMachineList = NULL; esxVI_ObjectContent *virtualMachine = NULL; - esxVI_String *propertyNameList = NULL; - esxVI_AutoStartDefaults *autostart_defaults = NULL; + esxVI_AutoStartDefaults *autoStartDefaults = NULL; esxVI_VirtualMachinePowerState powerState; esxVI_AutoStartPowerInfo *powerInfoList = NULL; esxVI_AutoStartPowerInfo *powerInfo = NULL; @@ -5023,7 +5025,6 @@ esxListAllDomains(virConnectPtr conn, int id; unsigned char uuid[VIR_UUID_BUFLEN]; int count = 0; - int snapshotCount; bool autostart; int state; @@ -5033,7 +5034,7 @@ esxListAllDomains(virConnectPtr conn, * - persistence: all esx machines are persistent * - managed save: esx doesn't support managed save */ - if ((MATCH(VIR_CONNECT_LIST_DOMAINS_TRANSIENT) && + if ((MATCH(VIR_CONNECT_LIST_DOMAINS_TRANSIENT) && !MATCH(VIR_CONNECT_LIST_DOMAINS_PERSISTENT)) || (MATCH(VIR_CONNECT_LIST_DOMAINS_MANAGEDSAVE) && !MATCH(VIR_CONNECT_LIST_DOMAINS_NO_MANAGEDSAVE))) { @@ -5045,23 +5046,49 @@ esxListAllDomains(virConnectPtr conn, goto cleanup; } - if (esxVI_EnsureSession(priv->primary) < 0) + if (esxVI_EnsureSession(priv->primary) < 0) return -1; /* check system default autostart value */ if (MATCH(VIR_CONNECT_LIST_DOMAINS_FILTERS_AUTOSTART)) { if (esxVI_LookupAutoStartDefaults(priv->primary, - &autostart_defaults) < 0) + &autoStartDefaults) < 0) { + goto cleanup; + } + + if (autoStartDefaults->enabled == esxVI_Boolean_True) { + if (esxVI_LookupAutoStartPowerInfoList(priv->primary, + &powerInfoList) < 0) { + goto cleanup; + } + } + } + + needIdentity = MATCH(VIR_CONNECT_LIST_DOMAINS_FILTERS_SNAPSHOT) || + domains != NULL; + + if (needIdentity) { + /* Request required data for esxVI_GetVirtualMachineIdentity */ + if (esxVI_String_AppendValueListToList(&propertyNameList, + "configStatus\0" + "name\0" + "config.uuid\0") < 0) { goto cleanup; + } + } + + needPowerState = MATCH(VIR_CONNECT_LIST_DOMAINS_FILTERS_ACTIVE) || + MATCH(VIR_CONNECT_LIST_DOMAINS_FILTERS_STATE) || + domains != NULL; - if (esxVI_LookupAutoStartPowerInfoList(priv->primary, - &powerInfoList) < 0) + if (needPowerState) { + if (esxVI_String_AppendValueToList(&propertyNameList, + "runtime.powerState") < 0) { goto cleanup; + } } - if (esxVI_String_AppendValueToList(&propertyNameList, - "runtime.powerState") < 0 || - esxVI_LookupVirtualMachineList(priv->primary, propertyNameList, + if (esxVI_LookupVirtualMachineList(priv->primary, propertyNameList, &virtualMachineList) < 0) goto cleanup; @@ -5073,12 +5100,21 @@ esxListAllDomains(virConnectPtr conn, for (virtualMachine = virtualMachineList; virtualMachine != NULL; virtualMachine = virtualMachine->_next) { + if (needIdentity) { + VIR_FREE(name); - VIR_FREE(name); + if (esxVI_GetVirtualMachineIdentity(virtualMachine, &id, + &name, uuid) < 0) { + goto cleanup; + } + } - if (esxVI_GetVirtualMachineIdentity(virtualMachine, &id, &name, uuid) < 0 || - esxVI_GetVirtualMachinePowerState(virtualMachine, &powerState) < 0) - goto cleanup; + if (needPowerState) { + if (esxVI_GetVirtualMachinePowerState(virtualMachine, + &powerState) < 0) { + goto cleanup; + } + } /* filter by active state */ if (MATCH(VIR_CONNECT_LIST_DOMAINS_FILTERS_ACTIVE) && @@ -5090,23 +5126,17 @@ esxListAllDomains(virConnectPtr conn, /* filter by snapshot existence */ if (MATCH(VIR_CONNECT_LIST_DOMAINS_FILTERS_SNAPSHOT)) { + esxVI_VirtualMachineSnapshotTree_Free(&rootSnapshotTreeList); + if (esxVI_LookupRootSnapshotTreeList(priv->primary, uuid, &rootSnapshotTreeList) < 0) { - virReportError(VIR_ERR_INTERNAL_ERROR, - _("Couldn't retrieve snapshot list for " - "domain '%s'"), name); goto cleanup; } - snapshotCount = esxVI_GetNumberOfSnapshotTrees(rootSnapshotTreeList, - true, false); - - esxVI_VirtualMachineSnapshotTree_Free(&rootSnapshotTreeList); - if (!((MATCH(VIR_CONNECT_LIST_DOMAINS_HAS_SNAPSHOT) && - snapshotCount > 0) || + rootSnapshotTreeList != NULL) || (MATCH(VIR_CONNECT_LIST_DOMAINS_NO_SNAPSHOT) && - snapshotCount == 0))) + rootSnapshotTreeList == NULL))) continue; } @@ -5114,19 +5144,18 @@ esxListAllDomains(virConnectPtr conn, if (MATCH(VIR_CONNECT_LIST_DOMAINS_FILTERS_AUTOSTART)) { autostart = false; - for (powerInfo = powerInfoList; powerInfo != NULL; - powerInfo = powerInfo->_next) { - if (STREQ(powerInfo->key->value, virtualMachine->obj->value)) { - if (STRCASEEQ(powerInfo->startAction, "powerOn")) - autostart = true; + if (autoStartDefaults->enabled == esxVI_Boolean_True) { + for (powerInfo = powerInfoList; powerInfo != NULL; + powerInfo = powerInfo->_next) { + if (STREQ(powerInfo->key->value, virtualMachine->obj->value)) { + if (STRCASEEQ(powerInfo->startAction, "powerOn")) + autostart = true; - break; + break; + } } } - autostart = autostart && - autostart_defaults->enabled == esxVI_Boolean_True; - if (!((MATCH(VIR_CONNECT_LIST_DOMAINS_AUTOSTART) && autostart) || (MATCH(VIR_CONNECT_LIST_DOMAINS_NO_AUTOSTART) && @@ -5137,6 +5166,7 @@ esxListAllDomains(virConnectPtr conn, /* filter by domain state */ if (MATCH(VIR_CONNECT_LIST_DOMAINS_FILTERS_STATE)) { state = esxVI_VirtualMachinePowerState_ConvertToLibvirt(powerState); + if (!((MATCH(VIR_CONNECT_LIST_DOMAINS_RUNNING) && state == VIR_DOMAIN_RUNNING) || (MATCH(VIR_CONNECT_LIST_DOMAINS_PAUSED) && @@ -5156,17 +5186,18 @@ esxListAllDomains(virConnectPtr conn, continue; } - if (!(dom = virGetDomain(conn, name, uuid))) + if (VIR_RESIZE_N(doms, ndoms, count, 2) < 0) goto no_memory; + if (!(dom = virGetDomain(conn, name, uuid))) + goto cleanup; + /* Only running/suspended virtual machines have an ID != -1 */ if (powerState != esxVI_VirtualMachinePowerState_PoweredOff) dom->id = id; else dom->id = -1; - if (VIR_EXPAND_N(doms, ndoms, 1) < 0) - goto no_memory; doms[count++] = dom; } @@ -5178,14 +5209,19 @@ esxListAllDomains(virConnectPtr conn, cleanup: if (doms) { for (id = 0; id < count; id++) { - if (doms[id]) - virDomainFree(doms[id]); + virDomainFree(doms[id]); } + + VIR_FREE(doms); } - VIR_FREE(doms); + VIR_FREE(name); + esxVI_AutoStartDefaults_Free(&autoStartDefaults); + esxVI_AutoStartPowerInfo_Free(&powerInfoList); esxVI_String_Free(&propertyNameList); esxVI_ObjectContent_Free(&virtualMachineList); + esxVI_VirtualMachineSnapshotTree_Free(&rootSnapshotTreeList); + return ret; no_memory: -- 1.7.4.1

2 2

[libvirt] TSC scaling interface to management
by Marcelo Tosatti 29 Sep '12

29 Sep '12

HW TSC scaling is a feature of AMD processors that allows a multiplier to be specified to the TSC frequency exposed to the guest. KVM also contains provision to trap TSC ("KVM: Infrastructure for software and hardware based TSC rate scaling" cc578287e3224d0da) or advance TSC frequency. This is useful when migrating to a host with different frequency and the guest is possibly using direct RDTSC instructions for purposes other than measuring cycles (that is, it previously calculated cycles-per-second, and uses that information which is stale after migration). "qemu-x86: Set tsc_khz in kvm when supported" (e7429073ed1a76518) added support for tsc_khz= option in QEMU. I am proposing the following changes so that management applications can work with this: 1) New option for tsc_khz, which is tsc_khz=host (QEMU command line option). Host means that QEMU is responsible for retrieving the TSC frequency of the host processor and use that. Management application does not have to deal with the burden. 2) New subsection with tsc_khz value. Destination host should consult supported features of running kernel and fail if feature is unsupported. It is not necessary to use this tsc_khz setting with modern guests using paravirtual clocks, or when its known that applications make proper use of the time interface provided by operating systems. On the other hand, legacy applications or setups which require no modification and correct operation while virtualized and make use of RDTSC might need this. Therefore it appears that this "tsc_khz=auto" option can be specified only if the user specifies so (it can be a per-guest flag hidden in the management configuration/manual). Sending this email to gather suggestions (or objections) to this interface.

3 7

[libvirt] Patch to default selinuxfs mount point to /sys/fs/selinux
by Daniel J Walsh 29 Sep '12

29 Sep '12

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Currently if you build on a machine that does not support SELinux we end up with the default mount point being /selinux, since this is moved to /sys/fs/selinux, we should start defaulting there. I believe this is causing a bug in libvirt-lxc when /selinux does not exists, even though /sys/fs/selinux exists. -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.12 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://www.enigmail.net/ iEYEARECAAYFAlBl6F8ACgkQrlYvE4MpobPz0ACgxPdENx/KDtQY7YGT7BDXoLP3 AVIAoJwUvTib72U0VJ3dnVoGU0PYjMXZ =XU8L -----END PGP SIGNATURE-----

2 1

[libvirt] virsh --connect not working
by hiren panchasara 29 Sep '12

29 Sep '12

I've installed libvirt-0.9.13 from ports on my freebsd machine. I started libvirtd: $ ps awwux | grep libvirtd root 11470 0.0 0.4 103100 31948 - I 10:41PM 0:00.35 libvirtd -v -d $ sudo virsh --connect qemu:///system error: no connection driver available for No connection for URI qemu:///system error: failed to connect to the hypervisor I am sure I am missing something obvious. Thanks in advance, Hiren

3 8

[libvirt] [PATCH] Fix QEMU test with 1.2.0 help output
by Daniel P. Berrange 28 Sep '12

28 Sep '12

From: "Daniel P. Berrange" <berrange(a)redhat.com> The help output for QEMU 1.2.0 changed 'pci-assign' to 'kvm-pci-assign'. Since the new capabilities code does exact device name matching instead of substring matching, this caused the capabilities to go missing. Pushed as a build break fix Signed-off-by: Daniel P. Berrange <berrange(a)redhat.com> --- src/qemu/qemu_capabilities.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/src/qemu/qemu_capabilities.c b/src/qemu/qemu_capabilities.c index f0442ee..cac4917 100644 --- a/src/qemu/qemu_capabilities.c +++ b/src/qemu/qemu_capabilities.c @@ -1321,6 +1321,8 @@ static struct qemuCapsObjectTypeProps qemuCapsObjectProps[] = { ARRAY_CARDINALITY(qemuCapsObjectPropsVirtioNet) }, { "pci-assign", qemuCapsObjectPropsPciAssign, ARRAY_CARDINALITY(qemuCapsObjectPropsPciAssign) }, + { "kvm-pci-assign", qemuCapsObjectPropsPciAssign, + ARRAY_CARDINALITY(qemuCapsObjectPropsPciAssign) }, { "scsi-disk", qemuCapsObjectPropsScsiDisk, ARRAY_CARDINALITY(qemuCapsObjectPropsScsiDisk) }, { "ide-drive", qemuCapsObjectPropsIDEDrive, -- 1.7.11.4

1 0

[libvirt] [PATCH 0/4] Switch to using QMP for capabilities detection
by Daniel P. Berrange 28 Sep '12

28 Sep '12

This is a fixup of the last patch in my previous series based on the feedback obtained, and some flaws I identified myself

2 12

[libvirt] [PATCH v3 00/20] Add support for detecting QEMU capabilities via QM
by Daniel P. Berrange 28 Sep '12

28 Sep '12

This is an update to: https://www.redhat.com/archives/libvir-list/2012-September/msg01051.html Changes in v3: - Already merged patches dropped - Rebased to latest git

4 45

[libvirt] [PATCH 0/9] Use the systemd journal for logging
by Daniel P. Berrange 28 Sep '12

28 Sep '12

The systemd journal provides a highly efficient way for apps to generate structured log messages. Since systemd will be used by default on future Fedora/RHEL distros (and likely many others), we should take advantage of the journal out of the box for improved logging capabilities. Amongst other things, this allows very easy browsing of the libvirt logs after the fact, filtering on filename / line number / function name and more. This is a big benefit to debugging, compared to current case where we need to decide the filters when starting libvirtd. For example it becomes possible to configure libvirt to include all QEMU logging, and then later ask the journal for only messages related to the QEMU json monitor.

3 27

[libvirt] [PATCH v3] Fix dnsmasq/radvd on bridges not being able to be bound to IPv6 address on "recent" kernels
by Benjamin Cama 28 Sep '12

28 Sep '12

Hi, I hit this problem recently when trying to create a bridge with an IPv6 address on a 3.2 kernel: dnsmasq (and, further, radvd) would not bind to the given address, waiting 20s and then giving up with -EADDRNOTAVAIL (resp. exiting immediately with "error parsing or activating the config file", without libvirt noticing it, BTW). This can be reproduced with (I think) any kernel >= 2.6.39 and the following XML (to be used with "virsh net-create"): <network> <name>test-bridge</name> <bridge name='testbr0' /> <ip family='ipv6' address='fd00::1' prefix='64'> </ip> </network> (it happens even when you have an IPv4, too) The problem is that since commit [1] (which, ironically, was made to “help IPv6 autoconfiguration”) the linux bridge code makes bridges behave like “real” devices regarding carrier detection. This makes the bridges created by libvirt, which are started without any up devices, stay with the NO-CARRIER flag set, and thus prevents DAD (Duplicate address detection) from happening, thus letting the IPv6 address flagged as “tentative”. Such addresses cannot be bound to (see RFC 2462), so dnsmasq fails binding to it (for radvd, it detects that "interface XXX is not RUNNING", thus that "interface XXX does not exist, ignoring the interface" (sic)). It seems that this behavior was enhanced somehow with commit [2] by avoiding setting NO-CARRIER on empty bridges, but I couldn't reproduce this behavior on my kernel. Anyway, with the “dummy tap to set MAC address” trick, this wouldn't work. To fix this, the idea is to get the bridge's attached device to be up so that DAD can happen (deactivating DAD altogether is not a good idea, I think). Currently, libvirt creates a dummy TAP device to set the MAC address of the bridge, keeping it down. But even if we set this device up, it is not RUNNING as soon as the tap file descriptor attached to it is closed, thus still preventing DAD. So, we must modify the API a bit, so that we can get the fd, keep the tap device persistent, run the daemons, and close it after DAD has taken place. After that, the bridge will be flagged NO-CARRIER again, but the daemons will be running, even if not happy about the device's state (but we don't really care about the bridge's daemons doing anything when no up interface is connected to it). Other solutions that I envisioned were: * Keeping the *-nic interface up: this would waste an fd for each bridge during all its life. May be acceptable, I don't really know. * Stop using the dummy tap trick, and set the MAC address directly on the bridge: it is possible since quite some time it seems, even if then there is the problem of the bridge not being RUNNING when empty, contrary to what [2] says, so this will need fixing (and this fix only happened in 3.1, so it wouldn't work for 2.6.39) * Using the --interface option of dnsmasq, but I saw somewhere that it's not used by libvirt for backward compatibility. I am not sure this would solve this problem, though, as I don't know how dnsmasq binds itself to it with this option. This is why this patch does what's described earlier. This patch also makes radvd start even if the interface is “missing” (i.e. it is not RUNNING), as it daemonizes before binding to it, and thus sometimes does it after the interface has been brought down by us (by closing the tap fd), and then originally stops. This also makes it stop yelling about it in the logs when the interface is down at a later time. [1] http://git.kernel.org/?p=linux/kernel/git/torvalds/linux.git;a=commit;h=1fa… [2] http://git.kernel.org/?p=linux/kernel/git/torvalds/linux.git;a=commit;h=b64… --- diff --git a/src/network/bridge_driver.c b/src/network/bridge_driver.c index 6e260f7..8d6abd3 100644 --- a/src/network/bridge_driver.c +++ b/src/network/bridge_driver.c @@ -65,6 +65,7 @@ #include "virnetdevtap.h" #include "virnetdevvportprofile.h" #include "virdbus.h" +#include "virfile.h" #define NETWORK_PID_DIR LOCALSTATEDIR "/run/libvirt/network" #define NETWORK_STATE_DIR LOCALSTATEDIR "/lib/libvirt/network" @@ -987,12 +988,15 @@ networkRadvdConfContents(virNetworkObjPtr network, char **configstr) *configstr = NULL; - /* create radvd config file appropriate for this network */ + /* create radvd config file appropriate for this network; + * IgnoreIfMissing allows radvd to start even when the bridge is down + */ virBufferAsprintf(&configbuf, "interface %s\n" "{\n" " AdvSendAdvert on;\n" " AdvManagedFlag off;\n" " AdvOtherConfigFlag off;\n" + " IgnoreIfMissing on;\n" "\n", network->def->bridge); @@ -2061,6 +2065,7 @@ networkStartNetworkVirtual(struct network_driver *driver, virErrorPtr save_err = NULL; virNetworkIpDefPtr ipdef; char *macTapIfName = NULL; + int tapfd = -1; /* Check to see if any network IP collides with an existing route */ if (networkCheckRouteCollision(network) < 0) @@ -2082,10 +2087,13 @@ networkStartNetworkVirtual(struct network_driver *driver, virReportOOMError(); goto err0; } + /* Keep tun fd open and interface up to allow for IPv6 DAD to happen */ if (virNetDevTapCreateInBridgePort(network->def->bridge, &macTapIfName, &network->def->mac, - NULL, NULL, NULL, NULL, - VIR_NETDEV_TAP_CREATE_USE_MAC_FOR_BRIDGE) < 0) { + NULL, &tapfd, NULL, NULL, + VIR_NETDEV_TAP_CREATE_USE_MAC_FOR_BRIDGE + |VIR_NETDEV_TAP_CREATE_IFUP + |VIR_NETDEV_TAP_CREATE_PERSIST) < 0) { VIR_FREE(macTapIfName); goto err0; } @@ -2149,6 +2157,15 @@ networkStartNetworkVirtual(struct network_driver *driver, if (v6present && networkStartRadvd(network) < 0) goto err4; + /* DAD has happened (dnsmasq waits for it), dnsmasq is now bound to the + * bridge's IPv6 address, so we can now set the dummy tun down. + */ + if (tapfd >= 0) { + if (virNetDevSetOnline(macTapIfName, false) < 0) + goto err4; + VIR_FORCE_CLOSE(tapfd); + } + if (virNetDevBandwidthSet(network->def->bridge, network->def->bandwidth) < 0) { virReportError(VIR_ERR_INTERNAL_ERROR, _("cannot set bandwidth limits on %s"), @@ -2187,6 +2204,8 @@ networkStartNetworkVirtual(struct network_driver *driver, save_err = virSaveLastError(); if (macTapIfName) { + if (tapfd >= 0) + VIR_FORCE_CLOSE(tapfd); ignore_value(virNetDevTapDelete(macTapIfName)); VIR_FREE(macTapIfName); } diff --git a/src/uml/uml_conf.c b/src/uml/uml_conf.c index a317bcc..765c5c2 100644 --- a/src/uml/uml_conf.c +++ b/src/uml/uml_conf.c @@ -142,7 +142,8 @@ umlConnectTapDevice(virConnectPtr conn, vm->uuid, NULL, virDomainNetGetActualVirtPortProfile(net), virDomainNetGetActualVlan(net), - VIR_NETDEV_TAP_CREATE_IFUP) < 0) { + VIR_NETDEV_TAP_CREATE_IFUP + |VIR_NETDEV_TAP_CREATE_PERSIST) < 0) { if (template_ifname) VIR_FREE(net->ifname); goto error; diff --git a/src/util/virnetdevtap.c b/src/util/virnetdevtap.c index 9f2dca1..0eadd6c 100644 --- a/src/util/virnetdevtap.c +++ b/src/util/virnetdevtap.c @@ -112,18 +112,20 @@ virNetDevProbeVnetHdr(int tapfd) * * VIR_NETDEV_TAP_CREATE_VNET_HDR * - Enable IFF_VNET_HDR on the tap device + * VIR_NETDEV_TAP_CREATE_PERSIST + * - The device will persist after the file descriptor is closed * * Creates a tap interface. - * If the @tapfd parameter is supplied, the open tap device file - * descriptor will be returned, otherwise the TAP device will be made - * persistent and closed. The caller must use virNetDevTapDelete to - * remove a persistent TAP devices when it is no longer needed. + * If the @tapfd parameter is supplied, the open tap device file descriptor + * will be returned, otherwise the TAP device will be closed. The caller must + * use virNetDevTapDelete to remove a persistent TAP device when it is no + * longer needed. * * Returns 0 in case of success or -1 on failure. */ int virNetDevTapCreate(char **ifname, int *tapfd, - unsigned int flags ATTRIBUTE_UNUSED) + unsigned int flags) { int fd; struct ifreq ifr; @@ -160,7 +162,7 @@ int virNetDevTapCreate(char **ifname, goto cleanup; } - if (!tapfd && + if ((flags & VIR_NETDEV_TAP_CREATE_PERSIST) && (errno = ioctl(fd, TUNSETPERSIST, 1))) { virReportSystemError(errno, _("Unable to set tap device %s to persistent"), @@ -261,14 +263,16 @@ int virNetDevTapDelete(const char *ifname ATTRIBUTE_UNUSED) * - Enable IFF_VNET_HDR on the tap device * VIR_NETDEV_TAP_CREATE_USE_MAC_FOR_BRIDGE * - Set this interface's MAC as the bridge's MAC address + * VIR_NETDEV_TAP_CREATE_PERSIST + * - The device will persist after the file descriptor is closed * * This function creates a new tap device on a bridge. @ifname can be either * a fixed name or a name template with '%d' for dynamic name allocation. * in either case the final name for the bridge will be stored in @ifname. - * If the @tapfd parameter is supplied, the open tap device file - * descriptor will be returned, otherwise the TAP device will be made - * persistent and closed. The caller must use virNetDevTapDelete to remove - * a persistent TAP devices when it is no longer needed. + * If the @tapfd parameter is supplied, the open tap device file descriptor + * will be returned, otherwise the TAP device will be closed. The caller must + * use virNetDevTapDelete to remove a persistent TAP device when it is no + * longer needed. * * Returns 0 in case of success or -1 on failure */ diff --git a/src/util/virnetdevtap.h b/src/util/virnetdevtap.h index fa6a647..980db61 100644 --- a/src/util/virnetdevtap.h +++ b/src/util/virnetdevtap.h @@ -43,6 +43,8 @@ typedef enum { VIR_NETDEV_TAP_CREATE_VNET_HDR = 1 << 1, /* Set this interface's MAC as the bridge's MAC address */ VIR_NETDEV_TAP_CREATE_USE_MAC_FOR_BRIDGE = 1 << 2, + /* The device will persist after the file descriptor is closed */ + VIR_NETDEV_TAP_CREATE_PERSIST = 1 << 3, } virNetDevTapCreateFlags; int virNetDevTapCreateInBridgePort(const char *brname, -- Benjamin Cama <benjamin.cama(a)telecom-bretagne.eu>

2 2

[libvirt] [PATCH] Move most of qemuProcessKill into virProcessKillPainfully
by Daniel P. Berrange 28 Sep '12

28 Sep '12

From: "Daniel P. Berrange" <berrange(a)redhat.com> In the cgroups APIs we have a virCgroupKillPainfully function which does the loop sending SIGTERM, then SIGKILL and waiting for the process to exit. There is similar functionality for simple processes in qemuProcessKill, but it is tangled with the QEMU code. Untangle it to provide a virProcessKillPainfuly function --- src/libvirt_private.syms | 1 + src/qemu/qemu_driver.c | 8 ++--- src/qemu/qemu_process.c | 79 ++++++++---------------------------------------- src/util/virprocess.c | 57 ++++++++++++++++++++++++++++++++++ src/util/virprocess.h | 2 ++ 5 files changed, 76 insertions(+), 71 deletions(-) diff --git a/src/libvirt_private.syms b/src/libvirt_private.syms index 4635a4d..dab607a 100644 --- a/src/libvirt_private.syms +++ b/src/libvirt_private.syms @@ -1708,6 +1708,7 @@ virPidFileDeletePath; # virprocess.h virProcessAbort; virProcessKill; +virProcessKillPainfully; virProcessTranslateStatus; virProcessWait; diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c index 7ac53ac..22fef7a 100644 --- a/src/qemu/qemu_driver.c +++ b/src/qemu/qemu_driver.c @@ -2006,13 +2006,11 @@ qemuDomainDestroyFlags(virDomainPtr dom, * it now means the job will be released */ if (flags & VIR_DOMAIN_DESTROY_GRACEFUL) { - if (qemuProcessKill(driver, vm, 0) < 0) { - virReportError(VIR_ERR_OPERATION_FAILED, "%s", - _("failed to kill qemu process with SIGTERM")); + if (qemuProcessKill(driver, vm, 0) < 0) goto cleanup; - } } else { - ignore_value(qemuProcessKill(driver, vm, VIR_QEMU_PROCESS_KILL_FORCE)); + if (qemuProcessKill(driver, vm, VIR_QEMU_PROCESS_KILL_FORCE) < 0) + goto cleanup; } /* We need to prevent monitor EOF callback from doing our work (and sending diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c index 3cd30af..70b72af 100644 --- a/src/qemu/qemu_process.c +++ b/src/qemu/qemu_process.c @@ -3877,9 +3877,7 @@ int qemuProcessKill(struct qemud_driver *driver, virDomainObjPtr vm, unsigned int flags) { - int i, ret = -1; - const char *signame = "TERM"; - bool driver_unlocked = false; + int ret; VIR_DEBUG("vm=%s pid=%d flags=%x", vm->def->name, vm->pid, flags); @@ -3891,78 +3889,27 @@ qemuProcessKill(struct qemud_driver *driver, } } - /* This loop sends SIGTERM (or SIGKILL if flags has - * VIR_QEMU_PROCESS_KILL_FORCE and VIR_QEMU_PROCESS_KILL_NOWAIT), - * then waits a few iterations (10 seconds) to see if it dies. If - * the qemu process still hasn't exited, and - * VIR_QEMU_PROCESS_KILL_FORCE is requested, a SIGKILL will then - * be sent, and qemuProcessKill will wait up to 5 seconds more for - * the process to exit before returning. Note that the FORCE mode - * could result in lost data in the guest, so it should only be - * used if the guest is hung and can't be destroyed in any other - * manner. - */ - for (i = 0 ; i < 75; i++) { - int signum; - if (i == 0) { - if ((flags & VIR_QEMU_PROCESS_KILL_FORCE) && - (flags & VIR_QEMU_PROCESS_KILL_NOWAIT)) { - signum = SIGKILL; /* kill it immediately */ - signame="KILL"; - } else { - signum = SIGTERM; /* kindly suggest it should exit */ - } - } else if ((i == 50) & (flags & VIR_QEMU_PROCESS_KILL_FORCE)) { - VIR_WARN("Timed out waiting after SIG%s to process %d, " - "sending SIGKILL", signame, vm->pid); - signum = SIGKILL; /* kill it after a grace period */ - signame="KILL"; - } else { - signum = 0; /* Just check for existence */ - } - - if (virProcessKill(vm->pid, signum) < 0) { - if (errno != ESRCH) { - char ebuf[1024]; - VIR_WARN("Failed to terminate process %d with SIG%s: %s", - vm->pid, signame, - virStrerror(errno, ebuf, sizeof(ebuf))); - goto cleanup; - } - ret = 0; - goto cleanup; /* process is dead */ - } + if ((flags & VIR_QEMU_PROCESS_KILL_NOWAIT)) { + virProcessKill(vm->pid, + (flags & VIR_QEMU_PROCESS_KILL_FORCE) ? + SIGKILL : SIGTERM); + return 0; + } - if (i == 0 && (flags & VIR_QEMU_PROCESS_KILL_NOWAIT)) { - ret = 0; - goto cleanup; - } + if (driver) + qemuDriverUnlock(driver); - if (driver && !driver_unlocked) { - /* THREADS.txt says we can't hold the driver lock while sleeping */ - qemuDriverUnlock(driver); - driver_unlocked = true; - } + ret = virProcessKillPainfully(vm->pid, + !!(flags & VIR_QEMU_PROCESS_KILL_FORCE)); - usleep(200 * 1000); - } - VIR_WARN("Timed out waiting after SIG%s to process %d", signame, vm->pid); -cleanup: - if (driver_unlocked) { - /* We had unlocked the driver, so re-lock it. THREADS.txt says - * we can't have the domain locked when locking the driver, so - * we must first unlock the domain. BUT, before we can unlock - * the domain, we need to add a ref to it in case there aren't - * any active jobs (analysis of all callers didn't reveal such - * a case, but there are too many to maintain certainty, so we - * will do this as a precaution). - */ + if (driver) { virObjectRef(vm); virDomainObjUnlock(vm); qemuDriverLock(driver); virDomainObjLock(vm); virObjectUnref(vm); } + return ret; } diff --git a/src/util/virprocess.c b/src/util/virprocess.c index 958f5f7..c70aa58 100644 --- a/src/util/virprocess.c +++ b/src/util/virprocess.c @@ -235,3 +235,60 @@ int virProcessKill(pid_t pid, int sig) return kill(pid, sig); #endif } + + +/* + * Try to kill the process and verify it has exited + * + * Returns 0 if it was killed gracefully, 1 if it + * was killed forcably, -1 if it is still alive, + * or another error occurred. + */ +int +virProcessKillPainfully(pid_t pid, bool force) +{ + int i, ret = -1; + const char *signame = "TERM"; + + VIR_DEBUG("vpid=%d force=%d", pid, force); + + /* This loop sends SIGTERM, then waits a few iterations (10 seconds) + * to see if it dies. If the process still hasn't exited, and + * @force is requested, a SIGKILL will be sent, and this will + * wait upto 5 seconds more for the process to exit before + * returning. + * + * Note that setting @force could result in dataloss for the process. + */ + for (i = 0 ; i < 75; i++) { + int signum; + if (i == 0) { + signum = SIGTERM; /* kindly suggest it should exit */ + } else if ((i == 50) & force) { + VIR_DEBUG("Timed out waiting after SIGTERM to process %d, " + "sending SIGKILL", pid); + signum = SIGKILL; /* kill it after a grace period */ + signame = "KILL"; + } else { + signum = 0; /* Just check for existence */ + } + + if (virProcessKill(pid, signum) < 0) { + if (errno != ESRCH) { + virReportSystemError(errno, + _("Failed to terminate process %d with SIG%s"), + pid, signame); + goto cleanup; + } + ret = signum == SIGTERM ? 0 : 1; + goto cleanup; /* process is dead */ + } + + usleep(200 * 1000); + } + + VIR_DEBUG("Timed out waiting after SIGKILL to process %d", pid); + +cleanup: + return ret; +} diff --git a/src/util/virprocess.h b/src/util/virprocess.h index 048a73c..d537d92 100644 --- a/src/util/virprocess.h +++ b/src/util/virprocess.h @@ -38,5 +38,7 @@ virProcessWait(pid_t pid, int *exitstatus) int virProcessKill(pid_t pid, int sig); +int virProcessKillPainfully(pid_t pid, bool force); + #endif /* __VIR_PROCESS_H__ */ -- 1.7.11.2

3 5

[libvirt] [PATCH 0/3] non-shared migration without target disk images
by liguang 28 Sep '12

28 Sep '12

by now, if you want a non-shared migration, you have to create same files at target as source side, which seems intorlerable! this issue was reported by Reinier Schoof http://www.redhat.com/archives/libvir-list/2011-December/msg00451.html These patches try to let migrate --copy-storage-* work without pre-exist disk image files at target side. we may also hope qemu to aware of this issue hooks/qemu | 55 ++++++++++++++++++++++++++++++++++++++++++++++++++++++ src/Makefile.am | 3 ++ src/util/hooks.c | 2 +- 3 files changed, 59 insertions(+), 1 deletions(-)

1 3

Re: [libvirt] Coverity automatic detection
by Alex Jia 28 Sep '12

28 Sep '12

The Coverity results were too big for the list filters; So compress and reattach them as an attachment. -- Regards, Alex ----- Original Message ----- From: "Alex Jia" <ajia(a)redhat.com> To: "libvir" <libvir-list(a)redhat.com> Cc: "Alex Jia" <ajia(a)redhat.com> Sent: Thursday, September 27, 2012 2:02:38 PM Subject: [libvirt] Coverity automatic detection This email is automatically generated. The test result is based on the following git commit: 371ddc9 Support Xen sysctl version 9 in Xen 4.2 Analysis summary report: ------------------------ Files analyzed : 340 Total LoC input to cov-analyze : 392401 Functions analyzed : 9785 Paths analyzed : 1186824 Time taken by Coverity analysis : 00:24:37 Defect occurrences found : 296 Total 5 ATOMICITY 53 BAD_SIZEOF 14 CHECKED_RETURN 1 CONSTANT_EXPRESSION_RESULT 6 DEADCODE 14 FORWARD_NULL 2 LOCK 7 NEGATIVE_RETURNS 1 NO_EFFECT 5 NULL_RETURNS 2 OVERRUN_STATIC 121 RESOURCE_LEAK 1 RETURN_LOCAL 16 REVERSE_INULL 3 SECURE_TEMP 2 SIZEOF_MISMATCH 1 STRING_NULL 1 STRING_OVERFLOW 4 TAINTED_SCALAR 2 TAINTED_STRING 21 TOCTOU 8 UNINIT 1 UNREACHABLE 5 UNUSED_VALUE Exceeded path limit of 5000 paths in 0.58% of functions (normally up to 5% of functions encounter this limitation) For details, please see attachment. Regards, Alex

1 0

[libvirt] [PATCH] ARMHF: Fixed Parser for /proc/cpuinfo needs to be adapted for your architecture
by Chuck Short 28 Sep '12

28 Sep '12

Minimal CPU "parser" for armhf to avoid compile time warning. Signed-off-by: Chuck Short <chuck.short(a)canonical.com> --- src/nodeinfo.c | 21 +++++++++++++++++++++ 1 file changed, 21 insertions(+) diff --git a/src/nodeinfo.c b/src/nodeinfo.c index 84a5d66..358fa3b 100644 --- a/src/nodeinfo.c +++ b/src/nodeinfo.c @@ -397,6 +397,27 @@ int linuxNodeInfoCPUPopulate(FILE *cpuinfo, * and parsed in next iteration, because it is not in expected * format and thus lead to error. */ } +# elif defined(__arm__) + char *buf = line; + if (STRPREFIX(buf, "BogoMIPS")) { + char *p; + unsigned int ui; + + buf += 8; + while (*buf && c_isspace(*buf)) + buf++; + + if (*buf != ':' || !buf[1]) { + nodeReportError(VIR_ERR_INTERNAL_ERROR, + "%s", _("parsing cpu MHz from cpuinfo")); + goto cleanup; + } + + if (virStrToLong_ui(buf+1, &p, 10, &ui) == 0 + /* Accept trailing fractional part. */ + && (*p == '\0' || *p == '.' || c_isspace(*p))) + nodeinfo->mhz = ui; + } # elif defined(__s390__) || \ defined(__s390x__) /* s390x has no realistic value for CPU speed, -- 1.7.10.4

2 1