January 2009 - Devel - libvirt List Archives

[libvirt] [PATCH] maint: update list of error function names and adapt
by Jim Meyering 21 Jan '09

21 Jan '09

I updated the list of error function names in Makefile.maint, and exposed 3 more unmarked diagnostics. Going through the old/new lists, I also found two now-unused (#define'd) "functions". I originally did this on top of Dan's patch series, but it also applies to the trunk, modulo some offset differences in xen_unified.c. >From 8124c5ae693fdad477df6b38dc5b75b4cd8f36cf Mon Sep 17 00:00:00 2001 From: Jim Meyering <meyering(a)redhat.com> Date: Mon, 19 Jan 2009 09:25:02 +0100 Subject: [PATCH] maint: update list of error function names and adapt * Makefile.maint (msg_gen_function): Update list. Remove some now-unused names, add some new ones. * src/hash.c (virHashError): Remove unused definition. * src/openvz_conf.h (openvzLog): Likewise. * src/xen_unified.c: Mark newly-exposed diagnostics for translation. --- Makefile.maint | 8 ++++---- src/hash.c | 5 ----- src/openvz_conf.h | 8 -------- src/xen_unified.c | 8 ++++---- 4 files changed, 8 insertions(+), 21 deletions(-) diff --git a/Makefile.maint b/Makefile.maint index 2ac6dd7..35eddd1 100644 --- a/Makefile.maint +++ b/Makefile.maint @@ -390,6 +390,7 @@ msg_gen_function += ERROR0 msg_gen_function += REMOTE_DEBUG msg_gen_function += ReportError msg_gen_function += VIR_FREE +msg_gen_function += VIR_INFO msg_gen_function += VIR_USE_CPU msg_gen_function += errorf msg_gen_function += lxcError @@ -401,9 +402,7 @@ msg_gen_function += qemudDispatchClientFailure msg_gen_function += qemudLog msg_gen_function += qemudReportError msg_gen_function += regerror -msg_gen_function += remoteDispatchError msg_gen_function += remoteDispatchFormatError -msg_gen_function += umlError msg_gen_function += umlLog msg_gen_function += umlReportError msg_gen_function += virConfError @@ -414,18 +413,19 @@ msg_gen_function += virLibDomainError msg_gen_function += virLog msg_gen_function += virNetworkReportError msg_gen_function += virNodeDeviceReportError +msg_gen_function += virProxyError msg_gen_function += virRaiseError msg_gen_function += virReportErrorHelper +msg_gen_function += virReportSystemError msg_gen_function += virSexprError msg_gen_function += virStorageLog msg_gen_function += virStorageReportError -msg_gen_function += virTestError msg_gen_function += virXMLError msg_gen_function += virXenInotifyError msg_gen_function += virXenStoreError msg_gen_function += virXendError -msg_gen_function += virxmlRpcError msg_gen_function += vshCloseLogFile +msg_gen_function += xenUnifiedError msg_gen_function += xenXMError # Uncomment the following and run "make syntax-check" to see diagnostics diff --git a/src/hash.c b/src/hash.c index b960640..bde3a0b 100644 --- a/src/hash.c +++ b/src/hash.c @@ -32,10 +32,6 @@ /* #define DEBUG_GROW */ -#define virHashError(conn, code, fmt...) \ - virReportErrorHelper(conn, VIR_FROM_NONE, code, __FILE__, \ - __FUNCTION__, __LINE__, fmt) - /* * A single entry in the hash table */ @@ -593,4 +589,3 @@ void *virHashSearch(virHashTablePtr table, virHashSearcher iter, const void *dat } return (NULL); } - diff --git a/src/openvz_conf.h b/src/openvz_conf.h index 1030d17..8e02056 100644 --- a/src/openvz_conf.h +++ b/src/openvz_conf.h @@ -34,14 +34,6 @@ enum { OPENVZ_WARN, OPENVZ_ERR }; -#define openvzLog(level, msg...) { if(level == OPENVZ_WARN) \ - fprintf(stderr, "\nWARNING: ");\ - else \ - fprintf(stderr, "\nERROR: ");\ - fprintf(stderr, "\n\t");\ - fprintf(stderr, msg);\ - fprintf(stderr, "\n"); } - #define openvzError(conn, code, fmt...) \ virReportErrorHelper(conn, VIR_FROM_OPENVZ, code, __FILE__, \ __FUNCTION__, __LINE__, fmt) diff --git a/src/xen_unified.c b/src/xen_unified.c index e0a446a..66d5fc2 100644 --- a/src/xen_unified.c +++ b/src/xen_unified.c @@ -1,7 +1,7 @@ /* * xen_unified.c: Unified Xen driver. * - * Copyright (C) 2007, 2008 Red Hat, Inc. + * Copyright (C) 2007, 2008, 2009 Red Hat, Inc. * * See COPYING.LIB for the License of this software * @@ -235,7 +235,7 @@ xenUnifiedOpen (virConnectPtr conn, virConnectAuthPtr auth, int flags) /* Allocate per-connection private data. */ if (VIR_ALLOC(priv) < 0) { - xenUnifiedError (NULL, VIR_ERR_NO_MEMORY, "allocating private data"); + xenUnifiedError (NULL, VIR_ERR_NO_MEMORY, _("allocating private data")); return VIR_DRV_OPEN_ERROR; } if (virMutexInit(&priv->lock) < 0) { @@ -247,7 +247,7 @@ xenUnifiedOpen (virConnectPtr conn, virConnectAuthPtr auth, int flags) /* Allocate callback list */ if (VIR_ALLOC(cbList) < 0) { - xenUnifiedError (NULL, VIR_ERR_NO_MEMORY, "allocating callback list"); + xenUnifiedError (NULL, VIR_ERR_NO_MEMORY, _("allocating callback list")); virMutexDestroy(&priv->lock); VIR_FREE(priv); return VIR_DRV_OPEN_ERROR; @@ -1533,7 +1533,7 @@ xenUnifiedAddDomainInfo(xenUnifiedDomainInfoListPtr list, list->count++; return 0; memory_error: - xenUnifiedError (NULL, VIR_ERR_NO_MEMORY, "allocating domain info"); + xenUnifiedError (NULL, VIR_ERR_NO_MEMORY, _("allocating domain info")); if (info) VIR_FREE(info->name); VIR_FREE(info); -- 1.6.1.258.g7ff14

2 2

[libvirt] Tiny additional RHEL patch
by Daniel Veillard 21 Jan '09

21 Jan '09

Small addition to the spec file, basically if on Fedora we should not activate RHEL-5 Xen specific support, but on RHEL and possibly CentOS the flag should be passed to %configure. The test is somehow reversed because it's hard to detect building on a RHEL platform from the spec file, and that should cope well with CentOS too. Now for other rpm based environment, I welcome improvements! Daniel -- Daniel Veillard | libxml Gnome XML XSLT toolkit http://xmlsoft.org/ daniel(a)veillard.com | Rpmfind RPM search engine http://rpmfind.net/ http://veillard.com/ | virtualization library http://libvirt.org/

2 2

[libvirt] [PATCH 3/4] use monitor fd for domain shutdown
by Guido Günther 21 Jan '09

21 Jan '09

This way we don't have to bother about stdin/stdout/stderr on reconnect. Since we dup stdin/stderr on the logfile we need to reparse it for the monitor path. Cheers, -- Guido --- src/domain_conf.h | 1 + src/qemu_driver.c | 155 +++++++++++++++++++++++++++-------------------------- 2 files changed, 80 insertions(+), 76 deletions(-) diff --git a/src/domain_conf.h b/src/domain_conf.h index 45b3e10..c236a66 100644 --- a/src/domain_conf.h +++ b/src/domain_conf.h @@ -465,6 +465,7 @@ struct _virDomainObj { int stderr_fd; int stderr_watch; int monitor; + int monitor_watch; char *monitorpath; int monitorWatch; int logfile; diff --git a/src/qemu_driver.c b/src/qemu_driver.c index 06f444b..734a329 100644 --- a/src/qemu_driver.c +++ b/src/qemu_driver.c @@ -181,6 +181,45 @@ qemudLogFD(virConnectPtr conn, const char* logDir, const char* name) } +static int +qemudLogReadFD(virConnectPtr conn, const char* logDir, const char* name, off_t pos) +{ + char logfile[PATH_MAX]; + mode_t logmode = O_RDONLY; + int ret, fd = -1; + + if ((ret = snprintf(logfile, sizeof(logfile), "%s/%s.log", logDir, name)) + < 0 || ret >= sizeof(logfile)) { + qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR, + _("failed to build logfile name %s/%s.log"), + logDir, name); + return -1; + } + + + if ((fd = open(logfile, logmode)) < 0) { + qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR, + _("failed to create logfile %s: %s"), + logfile, strerror(errno)); + return -1; + } + if (qemudSetCloseExec(fd) < 0) { + qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR, + _("Unable to set VM logfile close-on-exec flag: %s"), + strerror(errno)); + close(fd); + return -1; + } + if (lseek(fd, pos, SEEK_SET) < 0) { + qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR, + _("Unable to seek to %lld in %s: %s"), + pos, logfile, strerror(errno)); + close(fd); + } + return fd; +} + + static void qemudAutostartConfigs(struct qemud_driver *driver) { unsigned int i; @@ -516,11 +555,7 @@ qemudReadMonitorOutput(virConnectPtr conn, int ret; ret = read(fd, buf+got, buflen-got-1); - if (ret == 0) { - qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR, - _("QEMU quit during %s startup\n%s"), what, buf); - return -1; - } + if (ret < 0) { struct pollfd pfd = { .fd = fd, .events = POLLIN }; if (errno == EINTR) @@ -584,6 +619,7 @@ qemudCheckMonitorPrompt(virConnectPtr conn ATTRIBUTE_UNUSED, } static int qemudOpenMonitor(virConnectPtr conn, + struct qemud_driver* driver, virDomainObjPtr vm, const char *monitor) { int monfd; @@ -618,6 +654,12 @@ static int qemudOpenMonitor(virConnectPtr conn, goto error; } + if ((vm->monitor_watch = virEventAddHandle(vm->monitor, 0, + qemudDispatchVMEvent, + driver, NULL)) < 0) + goto error; + + /* Keep monitor open upon success */ if (ret == 0) return ret; @@ -677,6 +719,7 @@ qemudFindCharDevicePTYs(virConnectPtr conn, const char *output, int fd ATTRIBUTE_UNUSED) { + struct qemud_driver* driver = conn->privateData; char *monitor = NULL; size_t offset = 0; int ret, i; @@ -711,7 +754,9 @@ qemudFindCharDevicePTYs(virConnectPtr conn, } /* Got them all, so now open the monitor console */ - ret = qemudOpenMonitor(conn, vm, monitor); + qemuDriverLock(driver); + ret = qemudOpenMonitor(conn, driver, vm, monitor, 0); + qemuDriverUnlock(driver); cleanup: VIR_FREE(monitor); @@ -719,21 +764,23 @@ cleanup: } static int qemudWaitForMonitor(virConnectPtr conn, - virDomainObjPtr vm) { + struct qemud_driver* driver, + virDomainObjPtr vm, off_t pos) +{ char buf[1024]; /* Plenty of space to get startup greeting */ - int ret = qemudReadMonitorOutput(conn, - vm, vm->stderr_fd, - buf, sizeof(buf), - qemudFindCharDevicePTYs, - "console", 3000); + int logfd; + int ret; - buf[sizeof(buf)-1] = '\0'; + if ((logfd = qemudLogReadFD(conn, driver->logDir, vm->def->name, pos)) + < 0) + return logfd; - if (safewrite(vm->logfile, buf, strlen(buf)) < 0) { - /* Log, but ignore failures to write logfile for VM */ - qemudLog(QEMUD_WARN, _("Unable to log VM console data: %s\n"), + ret = qemudReadMonitorOutput(conn, vm, logfd, buf, sizeof(buf), + qemudFindCharDevicePTYs, + "console", 3000); + if (close(logfd) < 0) + qemudLog(QEMUD_WARN, _("Unable to close logfile: %s\n"), strerror(errno)); - } return ret; } @@ -942,6 +989,7 @@ static int qemudStartVMDaemon(virConnectPtr conn, fd_set keepfd; const char *emulator; pid_t child; + int pos = -1; FD_ZERO(&keepfd); @@ -1034,12 +1082,14 @@ static int qemudStartVMDaemon(virConnectPtr conn, qemudLog(QEMUD_WARN, _("Unable to write argv to logfile %d: %s\n"), errno, strerror(errno)); - vm->stdout_fd = -1; - vm->stderr_fd = -1; + if ((pos = lseek(vm->logfile, 0, SEEK_END)) < 0) + qemudLog(QEMUD_WARN, _("Unable to seek to end of logfile %d: %s\n"), + errno, strerror(errno)); for (i = 0 ; i < ntapfds ; i++) FD_SET(tapfds[i], &keepfd); + vm->stderr_fd = vm->stdout_fd = vm->logfile; ret = virExec(conn, argv, progenv, &keepfd, &child, vm->stdin_fd, &vm->stdout_fd, &vm->stderr_fd, VIR_EXEC_NONBLOCK | VIR_EXEC_DAEMON); @@ -1078,19 +1128,7 @@ static int qemudStartVMDaemon(virConnectPtr conn, } if (ret == 0) { - if (((vm->stdout_watch = virEventAddHandle(vm->stdout_fd, - VIR_EVENT_HANDLE_READABLE | - VIR_EVENT_HANDLE_ERROR | - VIR_EVENT_HANDLE_HANGUP, - qemudDispatchVMEvent, - driver, NULL)) < 0) || - ((vm->stderr_watch = virEventAddHandle(vm->stderr_fd, - VIR_EVENT_HANDLE_READABLE | - VIR_EVENT_HANDLE_ERROR | - VIR_EVENT_HANDLE_HANGUP, - qemudDispatchVMEvent, - driver, NULL)) < 0) || - (qemudWaitForMonitor(conn, vm) < 0) || + if ((qemudWaitForMonitor(conn, driver, vm, pos) < 0) || (qemudDetectVcpuPIDs(conn, vm) < 0) || (qemudInitCpus(conn, vm, migrateFrom) < 0)) { qemudShutdownVMDaemon(conn, driver, vm); @@ -1102,32 +1140,6 @@ static int qemudStartVMDaemon(virConnectPtr conn, return ret; } -static int qemudVMData(struct qemud_driver *driver ATTRIBUTE_UNUSED, - virDomainObjPtr vm, int fd) { - char buf[4096]; - if (vm->pid < 0) - return 0; - - for (;;) { - int ret = read(fd, buf, sizeof(buf)-1); - if (ret < 0) { - if (errno == EAGAIN) - return 0; - return -1; - } - if (ret == 0) { - return 0; - } - buf[ret] = '\0'; - - if (safewrite(vm->logfile, buf, ret) < 0) { - /* Log, but ignore failures to write logfile for VM */ - qemudLog(QEMUD_WARN, _("Unable to log VM console data: %s\n"), - strerror(errno)); - } - } -} - static void qemudShutdownVMDaemon(virConnectPtr conn ATTRIBUTE_UNUSED, struct qemud_driver *driver, virDomainObjPtr vm) { @@ -1141,17 +1153,11 @@ static void qemudShutdownVMDaemon(virConnectPtr conn ATTRIBUTE_UNUSED, qemudLog(QEMUD_ERROR, _("Failed to send SIGTERM to %s (%d): %s\n"), vm->def->name, vm->pid, strerror(errno)); - qemudVMData(driver, vm, vm->stdout_fd); - qemudVMData(driver, vm, vm->stderr_fd); - - virEventRemoveHandle(vm->stdout_watch); - virEventRemoveHandle(vm->stderr_watch); + virEventRemoveHandle(vm->monitor_watch); if (close(vm->logfile) < 0) qemudLog(QEMUD_WARN, _("Unable to close logfile %d: %s\n"), errno, strerror(errno)); - close(vm->stdout_fd); - close(vm->stderr_fd); if (vm->monitor != -1) close(vm->monitor); vm->logfile = -1; @@ -1191,8 +1197,7 @@ qemudDispatchVMEvent(int watch, int fd, int events, void *opaque) { virDomainObjPtr tmpvm = driver->domains.objs[i]; virDomainObjLock(tmpvm); if (virDomainIsActive(tmpvm) && - (tmpvm->stdout_watch == watch || - tmpvm->stderr_watch == watch)) { + tmpvm->monitor_watch == watch) { vm = tmpvm; break; } @@ -1202,16 +1207,14 @@ qemudDispatchVMEvent(int watch, int fd, int events, void *opaque) { if (!vm) goto cleanup; - if (vm->stdout_fd != fd && - vm->stderr_fd != fd) { + if (vm->monitor != fd) { failed = 1; } else { - if (events & VIR_EVENT_HANDLE_READABLE) { - if (qemudVMData(driver, vm, fd) < 0) - failed = 1; - } else { + if (events & (VIR_EVENT_HANDLE_HANGUP | VIR_EVENT_HANDLE_ERROR)) quit = 1; - } + else + qemudLog(QEMUD_ERROR, _("unhandled fd event %d for %s"), + events, vm->def->name); } if (failed || quit) { -- 1.6.0.6

3 5

[libvirt] [PATCH]: remove stdout_fd, stderr_fd from virDomainObj
by Guido Günther 21 Jan '09

21 Jan '09

Hi, After applying the previous two patches the qemu driver doesn't need these anymore and they can then be removed from lxc/uml as well. -- Guido

3 2

[libvirt] [PATCH 4/4] read saved vm status on libvirtd startup
by Guido Günther 21 Jan '09

21 Jan '09

connect back to running vms. changes: * don't use proc * don't bother about stdin * don't use global qemu_driver directly Cheer, -- Guido --- src/qemu_driver.c | 112 +++++++++++++++++++++++++++++++++++++++++++++-------- 1 files changed, 96 insertions(+), 16 deletions(-) diff --git a/src/qemu_driver.c b/src/qemu_driver.c index 2dedca2..3b01fd4 100644 --- a/src/qemu_driver.c +++ b/src/qemu_driver.c @@ -297,6 +297,86 @@ cleanup: } +static int qemudOpenMonitor(virConnectPtr conn, + struct qemud_driver* driver, + virDomainObjPtr vm, + const char *monitor, + int reconnect); + +/** + * qemudReconnectVMs + * + * Reconnect running vms to the daemon process + */ +static int +qemudReconnectVMs(struct qemud_driver *driver) +{ + int i; + + for (i = 0 ; i < driver->domains.count ; i++) { + virDomainObjPtr vm = driver->domains.objs[i]; + qemudDomainStatusPtr status = NULL; + char *config = NULL; + int rc; + + virDomainObjLock(vm); + if ((rc = virFileReadPid(driver->stateDir, vm->def->name, &vm->pid)) == 0) + DEBUG("Found pid %d for '%s'", vm->pid, vm->def->name); + else + goto next; + + if ((config = virDomainConfigFile(NULL, + driver->stateDir, + vm->def->name)) == NULL) { + qemudLog(QEMUD_ERR, _("Failed to read domain status for %s\n"), + vm->def->name); + goto next_error; + } + + status = qemudDomainStatusParseFile(NULL, driver->caps, config, 0); + if (status) { + vm->newDef = vm->def; + vm->def = status->def; + } else { + qemudLog(QEMUD_ERR, _("Failed to parse domain status for %s\n"), + vm->def->name); + goto next_error; + } + + if ((rc = qemudOpenMonitor(NULL, driver, vm, status->monitorpath, 1)) != 0) { + qemudLog(QEMUD_ERR, _("Failed to reconnect monitor for %s: %d\n"), + vm->def->name, rc); + goto next_error; + } else + vm->monitorpath = status->monitorpath; + + if((vm->logfile = qemudLogFD(NULL, driver->logDir, vm->def->name)) < 0) + return -1; + + vm->stdout_fd = vm->stderr_fd = vm->logfile; + + if (vm->def->id >= driver->nextvmid) + driver->nextvmid = vm->def->id + 1; + + vm->state = status->state; + goto next; + +next_error: + /* we failed to reconnect the vm so remove it's traces */ + vm->def->id = -1; + qemudRemoveDomainStatus(NULL, driver, vm); + virDomainDefFree(vm->def); + vm->def = vm->newDef; + vm->newDef = NULL; +next: + virDomainObjUnlock(vm); + VIR_FREE(status); + VIR_FREE(config); + } + return 0; +} + + /** * qemudStartup: * @@ -396,6 +476,7 @@ qemudStartup(void) { qemu_driver->autostartDir, NULL, NULL) < 0) goto error; + qemudReconnectVMs(qemu_driver); qemudAutostartConfigs(qemu_driver); qemuDriverUnlock(qemu_driver); @@ -488,7 +569,6 @@ qemudActive(void) { */ static int qemudShutdown(void) { - unsigned int i; if (!qemu_driver) return -1; @@ -496,15 +576,6 @@ qemudShutdown(void) { qemuDriverLock(qemu_driver); virCapabilitiesFree(qemu_driver->caps); - /* shutdown active VMs */ - for (i = 0 ; i < qemu_driver->domains.count ; i++) { - virDomainObjPtr dom = qemu_driver->domains.objs[i]; - virDomainObjLock(dom); - if (virDomainIsActive(dom)) - qemudShutdownVMDaemon(NULL, qemu_driver, dom); - virDomainObjUnlock(dom); - } - virDomainObjListFree(&qemu_driver->domains); VIR_FREE(qemu_driver->logDir); @@ -621,7 +692,8 @@ qemudCheckMonitorPrompt(virConnectPtr conn ATTRIBUTE_UNUSED, static int qemudOpenMonitor(virConnectPtr conn, struct qemud_driver* driver, virDomainObjPtr vm, - const char *monitor) { + const char *monitor, + int reconnect) { int monfd; char buf[1024]; int ret = -1; @@ -642,11 +714,19 @@ static int qemudOpenMonitor(virConnectPtr conn, goto error; } - ret = qemudReadMonitorOutput(conn, - vm, monfd, - buf, sizeof(buf), - qemudCheckMonitorPrompt, - "monitor", 10000); + if (!reconnect) { + ret = qemudReadMonitorOutput(conn, + vm, monfd, + buf, sizeof(buf), + qemudCheckMonitorPrompt, + "monitor", 10000); + } else { + vm->monitor = monfd; + ret = 0; + } + + if (ret != 0) + goto error; if (!(vm->monitorpath = strdup(monitor))) { qemudReportError(conn, NULL, NULL, VIR_ERR_NO_MEMORY, -- 1.6.0.6

3 4

[libvirt] [PATCH] Solaris least privilege support
by john.levon＠sun.com 21 Jan '09

21 Jan '09

# HG changeset patch # User john.levon(a)sun.com # Date 1232039546 28800 # Node ID b9d4d60bca87633897cb133461e1415d1223c823 # Parent 25a0c46588d5de1653b16dfed6bc357abf11db77 Solaris least privilege support On Solaris dom0, virtd runs as a privilege barrier: all libvirt connections are routed through it, and it performs the relevant privilege checks for any clients. Signed-off-by: John Levon <john.levon(a)sun.com> diff --git a/qemud/qemud.c b/qemud/qemud.c --- a/qemud/qemud.c +++ b/qemud/qemud.c @@ -84,6 +84,39 @@ #endif +#ifdef __sun +#include <ucred.h> +#include <priv.h> + +#ifndef PRIV_VIRT_MANAGE +#define PRIV_VIRT_MANAGE ((const char *)"virt_manage") +#endif + +#ifndef PRIV_XVM_CONTROL +#define PRIV_XVM_CONTROL ((const char *)"xvm_control") +#endif + +#define PU_RESETGROUPS 0x0001 /* Remove supplemental groups */ +#define PU_CLEARLIMITSET 0x0008 /* L=0 */ + +extern int __init_daemon_priv(int, uid_t, gid_t, ...); + +#define SYSTEM_UID 60 + +static gid_t unix_sock_gid = 60; /* Not used */ +static int unix_sock_rw_mask = 0666; +static int unix_sock_ro_mask = 0666; + +#else + +#define SYSTEM_UID 0 + +static gid_t unix_sock_gid = 0; /* Only root by default */ +static int unix_sock_rw_mask = 0700; /* Allow user only */ +static int unix_sock_ro_mask = 0777; /* Allow world */ + +#endif /* __sun */ + static int godaemon = 0; /* -d: Be a daemon */ static int verbose = 0; /* -v: Verbose mode */ static int timeout = -1; /* -t: Shutdown timeout */ @@ -101,10 +134,6 @@ static char *listen_addr = (char *) LIB static char *listen_addr = (char *) LIBVIRTD_LISTEN_ADDR; static char *tls_port = (char *) LIBVIRTD_TLS_PORT; static char *tcp_port = (char *) LIBVIRTD_TCP_PORT; - -static gid_t unix_sock_gid = 0; /* Only root by default */ -static int unix_sock_rw_mask = 0700; /* Allow user only */ -static int unix_sock_ro_mask = 0777; /* Allow world */ #if HAVE_POLKIT static int auth_unix_rw = REMOTE_AUTH_POLKIT; @@ -638,10 +667,11 @@ static int qemudInitPaths(struct qemud_s static int qemudInitPaths(struct qemud_server *server, char *sockname, char *roSockname, - int maxlen) { + int maxlen) +{ uid_t uid = geteuid(); - if (!uid) { + if (uid == SYSTEM_UID) { if (snprintf (sockname, maxlen, "%s/run/libvirt/libvirt-sock", LOCAL_STATE_DIR) >= maxlen) goto snprintf_error; @@ -1105,6 +1135,29 @@ static int qemudDispatchServer(struct qe return -1; } +#ifdef __sun + { + ucred_t *ucred = NULL; + const priv_set_t *privs; + + if (getpeerucred (fd, &ucred) == -1 || + (privs = ucred_getprivset (ucred, PRIV_EFFECTIVE)) == NULL) { + if (ucred != NULL) + ucred_free (ucred); + close (fd); + return -1; + } + + if (!priv_ismember (privs, PRIV_VIRT_MANAGE)) { + ucred_free (ucred); + close (fd); + return -1; + } + + ucred_free (ucred); + } +#endif /* __sun */ + /* Disable Nagle. Unix sockets will ignore this. */ setsockopt (fd, IPPROTO_TCP, TCP_NODELAY, (void *)&no_slow_start, sizeof no_slow_start); @@ -2239,6 +2292,29 @@ version (const char *argv0) { printf ("%s (%s) %s\n", argv0, PACKAGE_NAME, PACKAGE_VERSION); } + +#ifdef __sun +static void +qemudSetupPrivs (struct qemud_server *server) +{ + chown ("/var/run/libvirt", SYSTEM_UID, SYSTEM_UID); + chown (server->logDir, SYSTEM_UID, SYSTEM_UID); + + if (__init_daemon_priv (PU_RESETGROUPS | PU_CLEARLIMITSET, + SYSTEM_UID, SYSTEM_UID, PRIV_XVM_CONTROL, NULL)) { + fprintf (stderr, "additional privileges are required\n"); + exit (1); + } + + if (priv_set (PRIV_OFF, PRIV_ALLSETS, PRIV_FILE_LINK_ANY, PRIV_PROC_INFO, + PRIV_PROC_SESSION, PRIV_PROC_EXEC, PRIV_PROC_FORK, NULL)) { + fprintf (stderr, "failed to set reduced privileges\n"); + exit (1); + } +} +#else +#define qemudSetupPrivs(a) +#endif /* Print command-line usage. */ static void @@ -2417,6 +2493,20 @@ int main(int argc, char **argv) { sig_action.sa_handler = SIG_IGN; sigaction(SIGPIPE, &sig_action, NULL); + /* Change the group ownership of /var/run/libvirt to unix_sock_gid */ + if (geteuid () == 0) { + const char *rundir = LOCAL_STATE_DIR "/run/libvirt"; + + if (mkdir (rundir, 0755)) { + if (errno != EEXIST) { + VIR_ERROR0 (_("unable to create rundir")); + return (-1); + } + } + } + + qemudSetupPrivs(server); + if (!(server = qemudInitialize(sigpipe[0]))) { ret = 2; goto error2; diff --git a/src/remote_internal.c b/src/remote_internal.c --- a/src/remote_internal.c +++ b/src/remote_internal.c @@ -90,7 +90,7 @@ #define MAGIC 999 /* private_data->magic if OK */ #define DEAD 998 /* private_data->magic if dead/closed */ -static int inside_daemon = 0; +int inside_daemon = 0; struct private_data { int magic; /* Should be MAGIC or DEAD. */ @@ -903,18 +903,21 @@ remoteOpen (virConnectPtr conn, } /* - * If URI is NULL, then do a UNIX connection - * possibly auto-spawning unprivileged server - * and probe remote server for URI + * If URI is NULL, then do a UNIX connection possibly auto-spawning + * unprivileged server and probe remote server for URI. On Solaris, + * this isn't supported, but we may be privileged enough to connect + * to the UNIX socket anyway. */ if (!conn->uri) { DEBUG0("Auto-probe remote URI"); rflags |= VIR_DRV_OPEN_REMOTE_UNIX; +#ifndef __sun if (getuid() > 0) { DEBUG0("Auto-spawn user daemon instance"); rflags |= VIR_DRV_OPEN_REMOTE_USER; rflags |= VIR_DRV_OPEN_REMOTE_AUTOSTART; } +#endif } priv->magic = DEAD; diff --git a/src/xen_internal.c b/src/xen_internal.c --- a/src/xen_internal.c +++ b/src/xen_internal.c @@ -26,6 +26,17 @@ #include <errno.h> #include <sys/utsname.h> +#ifdef __sun +#include <sys/systeminfo.h> + +#include <priv.h> + +#ifndef PRIV_XVM_CONTROL +#define PRIV_XVM_CONTROL ((const char *)"xvm_control") +#endif + +#endif /* __sun */ + /* required for dom0_getdomaininfo_t */ #include <xen/dom0_ops.h> #include <xen/version.h> @@ -35,10 +46,6 @@ #ifdef HAVE_XEN_SYS_PRIVCMD_H #include <xen/sys/privcmd.h> #endif -#endif - -#ifdef __sun -#include <sys/systeminfo.h> #endif /* required for shutdown flags */ @@ -3387,3 +3394,17 @@ xenHypervisorGetVcpuMax(virDomainPtr dom return maxcpu; } +/** + * xenHavePrivilege() + * + * Return true if the current process should be able to connect to Xen. + */ +int +xenHavePrivilege() +{ +#ifdef __sun + return priv_ineffect (PRIV_XVM_CONTROL); +#else + return getuid () == 0; +#endif +} diff --git a/src/xen_internal.h b/src/xen_internal.h --- a/src/xen_internal.h +++ b/src/xen_internal.h @@ -104,4 +104,6 @@ int xenHypervisorNodeGetCellsFreeMem int startCell, int maxCells); +int xenHavePrivilege(void); + #endif /* __VIR_XEN_INTERNAL_H__ */ diff --git a/src/xen_unified.c b/src/xen_unified.c --- a/src/xen_unified.c +++ b/src/xen_unified.c @@ -231,6 +231,16 @@ xenUnifiedOpen (virConnectPtr conn, virC xenUnifiedPrivatePtr priv; virDomainEventCallbackListPtr cbList; +#ifdef __sun + extern int inside_daemon; + /* + * Only the libvirtd instance can open this driver. + * Everything else falls back to the remote driver. + */ + if (!inside_daemon) + return VIR_DRV_OPEN_DECLINED; +#endif + if (conn->uri == NULL) { if (!xenUnifiedProbe()) return VIR_DRV_OPEN_DECLINED; @@ -283,8 +293,8 @@ xenUnifiedOpen (virConnectPtr conn, virC priv->proxy = -1; - /* Hypervisor is only run as root & required to succeed */ - if (getuid() == 0) { + /* Hypervisor is only run with privilege & required to succeed */ + if (xenHavePrivilege()) { DEBUG0("Trying hypervisor sub-driver"); if (drivers[XEN_UNIFIED_HYPERVISOR_OFFSET]->open(conn, auth, flags) == VIR_DRV_OPEN_SUCCESS) { @@ -293,7 +303,7 @@ xenUnifiedOpen (virConnectPtr conn, virC } } - /* XenD is required to suceed if root. + /* XenD is required to succeed if privileged. * If it fails as non-root, then the proxy driver may take over */ DEBUG0("Trying XenD sub-driver"); @@ -318,12 +328,12 @@ xenUnifiedOpen (virConnectPtr conn, virC DEBUG0("Activated XS sub-driver"); priv->opened[XEN_UNIFIED_XS_OFFSET] = 1; } else { - if (getuid() == 0) - goto fail; /* XS is mandatory as root */ + if (xenHavePrivilege()) + goto fail; /* XS is mandatory when privileged */ } } else { - if (getuid() == 0) { - goto fail; /* XenD is mandatory as root */ + if (xenHavePrivilege()) { + goto fail; /* XenD is mandatory when privileged */ } else { #if WITH_PROXY DEBUG0("Trying proxy sub-driver"); diff --git a/src/xend_internal.c b/src/xend_internal.c --- a/src/xend_internal.c +++ b/src/xend_internal.c @@ -42,7 +42,7 @@ #include "buf.h" #include "uuid.h" #include "xen_unified.h" -#include "xen_internal.h" /* for DOM0_INTERFACE_VERSION */ +#include "xen_internal.h" #include "xs_internal.h" /* To extract VNC port & Serial console TTY */ #include "memory.h" @@ -151,9 +151,10 @@ do_connect(virConnectPtr xend) s = -1; /* - * Connecting to XenD as root is mandatory, so log this error + * Connecting to XenD when privileged is mandatory, so log this + * error */ - if (getuid() == 0) { + if (xenHavePrivilege()) { virXendError(xend, VIR_ERR_INTERNAL_ERROR, "%s", _("failed to connect to xend")); } diff --git a/src/xs_internal.c b/src/xs_internal.c --- a/src/xs_internal.c +++ b/src/xs_internal.c @@ -35,7 +35,7 @@ #include "uuid.h" #include "xen_unified.h" #include "xs_internal.h" -#include "xen_internal.h" /* for xenHypervisorCheckID */ +#include "xen_internal.h" #ifdef __linux__ #define XEN_HYPERVISOR_SOCKET "/proc/xen/privcmd" @@ -299,11 +299,11 @@ xenStoreOpen(virConnectPtr conn, if (priv->xshandle == NULL) { /* - * not being able to connect via the socket as a normal user - * is rather normal, this should fallback to the proxy (or + * not being able to connect via the socket as an unprivileged + * user is rather normal, this should fallback to the proxy (or * remote) mechanism. */ - if (getuid() == 0) { + if (xenHavePrivilege()) { virXenStoreError(NULL, VIR_ERR_NO_XEN, "%s", _("failed to connect to Xen Store")); }

7 10

[libvirt] [PATCH 0/7 RFC] RHEL-5 support, and a few preparatory fixes
by Markus Armbruster 20 Jan '09

20 Jan '09

This patch series attempts to make libvirt just work on RHEL-5. Right now it doesn't, mostly because libvirt relies on version number checks in a couple of places, and RHEL-5's version numbers aren't the whole truth due to various backports of later stuff. The first two patches are plain bug fixes: [PATCH 1/7] Fix network device inconsistency between xm and sxpr [PATCH 2/7] Fix graphics configuration inconsistency between xm and sxpr The next three add a few special cases for RHEL-5: [PATCH 3/7] New configure option --with-rhel5-api [PATCH 4/7] Fix HVM network device configuration for RHEL-5 [PATCH 5/7] Fix PVFB device configuration for RHEL-5 [PATCH 6/7] Enable NUMA support for RHEL-5 The last one drops a counter-productive error check: [PATCH 7/7] Don't treat missing topology information as error See the patches for more detailed descriptions. I'm not proposing this for immediate commit, as I'm still testing. But I'd appreciate review: is this the right way to do it? Thanks, Markus

4 15

[libvirt] [PATCH 2/4] daemonize qemu processes
by Guido Günther 20 Jan '09

20 Jan '09

so they doen't get killed when the libvirtd quits unexpectedly changes: * longer timeouts when waiting for procs to show up * check if process exists before killing it * use virKill this one basically got acked before but makes little sense without the follow up patches. Cheers, -- Guido --- src/qemu_driver.c | 39 ++++++++++++++++++++++++++------------- 1 files changed, 26 insertions(+), 13 deletions(-) diff --git a/src/qemu_driver.c b/src/qemu_driver.c index 1619154..06f444b 100644 --- a/src/qemu_driver.c +++ b/src/qemu_driver.c @@ -941,6 +941,7 @@ static int qemudStartVMDaemon(virConnectPtr conn, unsigned int qemuCmdFlags; fd_set keepfd; const char *emulator; + pid_t child; FD_ZERO(&keepfd); @@ -1039,12 +1040,26 @@ static int qemudStartVMDaemon(virConnectPtr conn, for (i = 0 ; i < ntapfds ; i++) FD_SET(tapfds[i], &keepfd); - ret = virExec(conn, argv, progenv, &keepfd, &vm->pid, + ret = virExec(conn, argv, progenv, &keepfd, &child, vm->stdin_fd, &vm->stdout_fd, &vm->stderr_fd, - VIR_EXEC_NONBLOCK); - if (ret == 0) + VIR_EXEC_NONBLOCK | VIR_EXEC_DAEMON); + + /* wait for qemu process to to show up */ + if (ret == 0) { + int retries = 100; + while (retries) { + if ((ret = virFileReadPid(driver->stateDir, vm->def->name, &vm->pid)) == 0) + break; + usleep(100*1000); + retries--; + } + if (ret) + qemudLog(QEMUD_WARN, _("Domain %s didn't show up\n"), vm->def->name); + } + + if (ret == 0) { vm->state = migrateFrom ? VIR_DOMAIN_PAUSED : VIR_DOMAIN_RUNNING; - else + } else vm->def->id = -1; for (i = 0 ; argv[i] ; i++) @@ -1121,7 +1136,10 @@ static void qemudShutdownVMDaemon(virConnectPtr conn ATTRIBUTE_UNUSED, qemudLog(QEMUD_INFO, _("Shutting down VM '%s'\n"), vm->def->name); - kill(vm->pid, SIGTERM); + if (virKillProcess(vm->pid, 0) == 0 && + virKillProcess(vm->pid, SIGTERM) < 0) + qemudLog(QEMUD_ERROR, _("Failed to send SIGTERM to %s (%d): %s\n"), + vm->def->name, vm->pid, strerror(errno)); qemudVMData(driver, vm, vm->stdout_fd); qemudVMData(driver, vm, vm->stderr_fd); @@ -1141,15 +1159,10 @@ static void qemudShutdownVMDaemon(virConnectPtr conn ATTRIBUTE_UNUSED, vm->stderr_fd = -1; vm->monitor = -1; - if (waitpid(vm->pid, NULL, WNOHANG) != vm->pid) { - kill(vm->pid, SIGKILL); - if (waitpid(vm->pid, NULL, 0) != vm->pid) { - qemudLog(QEMUD_WARN, - "%s", _("Got unexpected pid, damn\n")); - } - } - qemudRemoveDomainStatus(conn, driver, vm); + /* shut it off for sure */ + virKillProcess(vm->pid, SIGKILL); + qemudRemoveDomainStatus(conn, driver, vm); vm->pid = -1; vm->def->id = -1; vm->state = VIR_DOMAIN_SHUTOFF; -- 1.6.0.6

2 2

[libvirt] [PATCH 1/4] save domstate as string in status file
by Guido Günther 20 Jan '09

20 Jan '09

This one is as posted before. Not really necessary, but makes things look nicer. -- Guido --- src/domain_conf.c | 9 +++++++++ src/domain_conf.h | 2 ++ src/libvirt_private.syms | 2 ++ src/qemu_conf.c | 12 ++++++++---- 4 files changed, 21 insertions(+), 4 deletions(-) diff --git a/src/domain_conf.c b/src/domain_conf.c index 94196e8..8deb992 100644 --- a/src/domain_conf.c +++ b/src/domain_conf.c @@ -142,6 +142,15 @@ VIR_ENUM_IMPL(virDomainHostdevSubsys, VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_LAST, "usb", "pci") +VIR_ENUM_IMPL(virDomainState, VIR_DOMAIN_CRASHED+1, + "nostate", + "running", + "blocked", + "paused", + "shutdown", + "shutoff", + "crashed") + #define virDomainReportError(conn, code, fmt...) \ virReportErrorHelper(conn, VIR_FROM_DOMAIN, code, __FILE__, \ __FUNCTION__, __LINE__, fmt) diff --git a/src/domain_conf.h b/src/domain_conf.h index 1e5c103..45b3e10 100644 --- a/src/domain_conf.h +++ b/src/domain_conf.h @@ -625,5 +625,7 @@ VIR_ENUM_DECL(virDomainHostdevSubsys) VIR_ENUM_DECL(virDomainInput) VIR_ENUM_DECL(virDomainInputBus) VIR_ENUM_DECL(virDomainGraphics) +/* from libvirt.h */ +VIR_ENUM_DECL(virDomainState) #endif /* __DOMAIN_CONF_H */ diff --git a/src/libvirt_private.syms b/src/libvirt_private.syms index c76b6c2..3f83a07 100644 --- a/src/libvirt_private.syms +++ b/src/libvirt_private.syms @@ -94,6 +94,8 @@ virDomainVirtTypeToString; virDomainFSDefFree; virDomainObjLock; virDomainObjUnlock; +virDomainStateTypeToString; +virDomainStateTypeFromString; # domain_event.h diff --git a/src/qemu_conf.c b/src/qemu_conf.c index faf1e12..45108f3 100644 --- a/src/qemu_conf.c +++ b/src/qemu_conf.c @@ -1375,12 +1375,14 @@ qemudDomainStatusParseFile(virConnectPtr conn, } ctxt->node = root; - if((virXPathLong(conn, "string(./@state)", ctxt, &val)) < 0) { + if(!(tmp = virXPathString(conn, "string(./@state)", ctxt))) { qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR, "%s", _("invalid domain state")); goto error; - } else - status->state = (int)val; + } else { + status->state = virDomainStateTypeFromString(tmp); + VIR_FREE(tmp); + } if((virXPathLong(conn, "string(./@pid)", ctxt, &val)) < 0) { qemudReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR, @@ -1433,7 +1435,9 @@ qemudDomainStatusFormat(virConnectPtr conn, char *config_xml = NULL, *xml = NULL; virBuffer buf = VIR_BUFFER_INITIALIZER; - virBufferVSprintf(&buf, "<domstatus state='%d' pid='%d'>\n", vm->state, vm->pid); + virBufferVSprintf(&buf, "<domstatus state='%s' pid='%d'>\n", + virDomainStateTypeToString(vm->state), + vm->pid); virBufferEscapeString(&buf, " <monitor path='%s'/>\n", vm->monitorpath); if (!(config_xml = virDomainDefFormat(conn, -- 1.6.0.6

3 3

[libvirt] [PATCH] don't ignore write failure [was PATCH: 10/25: ...
by Jim Meyering 19 Jan '09

19 Jan '09

"Daniel P. Berrange" <berrange(a)redhat.com> wrote: > The strerror() method is not guarenteed to be re-entrant, which is ... > @@ -759,8 +762,8 @@ static int networkStartNetworkDaemon(vir > > if (network->def->forwardType != VIR_NETWORK_FORWARD_NONE && > !networkEnableIpForwarding()) { > - networkReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR, > - _("failed to enable IP forwarding : %s"), strerror(err)); > + virReportSystemError(conn, errno, "%s", > + _("failed to enable IP forwarding")); > goto err_delbr2; > } > Looking at this chunk and the s/err/errno/ change (that's a bug fix), I spotted another small bug: Any write error in networkEnableIpForwarding was ignored, I added a check for failed close, too, just in case. Jim >From b4abd36014b56cdb33a91ab336eff6450a5d3c8a Mon Sep 17 00:00:00 2001 From: Jim Meyering <meyering(a)redhat.com> Date: Mon, 19 Jan 2009 10:11:15 +0100 Subject: [PATCH] don't ignore write failure * src/network_driver.c (networkEnableIpForwarding): This function would always return 1 (success). Now it returns 0 upon write or close failure, being careful to preserve any errno from a write failure. Add comments. --- src/network_driver.c | 20 ++++++++++++++------ 1 files changed, 14 insertions(+), 6 deletions(-) diff --git a/src/network_driver.c b/src/network_driver.c index 695acfb..51cfbbd 100644 --- a/src/network_driver.c +++ b/src/network_driver.c @@ -1,7 +1,7 @@ /* * driver.c: core driver methods for managing qemu guests * - * Copyright (C) 2006, 2007, 2008 Red Hat, Inc. + * Copyright (C) 2006-2009 Red Hat, Inc. * Copyright (C) 2006 Daniel P. Berrange * * This library is free software; you can redistribute it and/or @@ -774,20 +774,29 @@ networkRemoveIptablesRules(struct network_driver *driver, iptablesSaveRules(driver->iptables); } +/* Enable IP Forwarding. + Return 0 for success, nonzero for failure. + Be careful to preserve any errno value upon failure. */ static int networkEnableIpForwarding(void) { #define PROC_IP_FORWARD "/proc/sys/net/ipv4/ip_forward" - int fd, ret; + int fd; if ((fd = open(PROC_IP_FORWARD, O_WRONLY|O_TRUNC)) == -1) return 0; - if (safewrite(fd, "1\n", 2) < 0) - ret = 0; + if (safewrite(fd, "1\n", 2) < 0) { + int saved_errno = errno; + close (fd); + errno = saved_errno; + return 0; + } - close (fd); + /* Use errno from failed close only if there was no write error. */ + if (close (fd) != 0) + return 0; return 1; @@ -1429,4 +1438,3 @@ int networkRegister(void) { virRegisterStateDriver(&networkStateDriver); return 0; } - -- 1.6.1.258.g7ff14

1 0