- Devel - libvirt List Archives

[libvirt] [PATCH] Avoid passing NULL to printf %s specifier
by john.levon＠sun.com 19 Jan '09

19 Jan '09

# HG changeset patch # User john.levon(a)sun.com # Date 1232032750 28800 # Node ID 0335828806b5b3855c8eeaee31446a7bd0c02974 # Parent e542223a4c52d978d4507de709f50885920a5b44 Avoid passing NULL to printf %s specifier This is non-portable. Signed-off-by: John Levon <john.levon(a)sun.com> diff --git a/src/internal.h b/src/internal.h --- a/src/internal.h +++ b/src/internal.h @@ -7,6 +7,7 @@ #include <errno.h> #include <limits.h> +#include <verify.h> #ifdef HAVE_SYS_SYSLIMITS_H #include <sys/syslimits.h> @@ -115,6 +116,13 @@ #define ATTRIBUTE_RETURN_CHECK #endif /* __GNUC__ */ +/* + * Use this when passing possibly-NULL strings to printf-a-likes. + */ +#define NULLSTR(s) \ + ((void)verify_true(sizeof *(s) == sizeof (char)), \ + (s) ? (s) : "(null)") + /** * TODO: * diff --git a/src/libvirt.c b/src/libvirt.c --- a/src/libvirt.c +++ b/src/libvirt.c @@ -864,10 +864,10 @@ do_open (const char *name, " port %d\n" " path %s\n", name, - ret->uri->scheme, ret->uri->opaque, - ret->uri->authority, ret->uri->server, - ret->uri->user, ret->uri->port, - ret->uri->path); + NULLSTR(ret->uri->scheme), NULLSTR(ret->uri->opaque), + NULLSTR(ret->uri->authority), NULLSTR(ret->uri->server), + NULLSTR(ret->uri->user), ret->uri->port, + NULLSTR(ret->uri->path)); } else { DEBUG0("no name, allowing driver auto-select"); } @@ -1056,7 +1056,7 @@ virConnectOpenAuth(const char *name, if (virInitialize() < 0) return NULL; - DEBUG("name=%s, auth=%p, flags=%d", name, auth, flags); + DEBUG("name=%s, auth=%p, flags=%d", NULLSTR(name), auth, flags); return do_open (name, auth, flags); }

4 3

[libvirt] Using vmchannel?
by Richard W.M. Jones 19 Jan '09

19 Jan '09

I was just looking at the new vmchannel feature in QEMU and was wondering if we could make this available in libvirt. Firstly, since there isn't much documentation, I should explain how vmchannel works: [1] You pass an extra parameter on the qemu command line: qemu [...] -vmchannel <port>:<dev> where <port> is the TCP port number (see below) and <dev> is a standard qemu device description. As an example: qemu [...] -vmchannel 600:unix:/some/path [2] A new character device appears in the host, eg. Unix domain socket called "/some/path". [3] In the host, a userspace program should open a socket, bind(2) it to /some/path and listen(2) and accept(2) on it. [4] In the guest, any process may connect(2) to TCP 10.0.2.4:600 (or whatever port was selected). Each connection in the guest causes the listener in the host to accept(2). [5] This is only designed for low-bandwidth, low-performance applications. [6] Multiple vmchannels are supported. [7] Host cannot initiate a connection. My plan to implement this would be to add a new network interface type to the domain XML: <interface type='vmchannel'> <source port='600'/> <target path='/some/path'/> </interface> Only Unix domain socket paths would be allowed on the host side, and the path would be expected to exist already with suitable permissions. Note that I think this would also allow guests to communicate with the libvirtd on the host (not by default, of course, but if users wanted to configure it that way), for example: <interface type='vmchannel'> <source port='600'/> <target path='/var/run/libvirt/libvirt-sock'/> </interface> One problem is that it is qemu/kvm-only. It is built on top of virtio, and virtio is meant to become a standard driver subsystem for all full virtualization systems. Thoughts? Rich. -- Richard Jones, Emerging Technologies, Red Hat http://et.redhat.com/~rjones virt-p2v converts physical machines to virtual machines. Boot with a live CD or over the network (PXE) and turn machines into Xen guests. http://et.redhat.com/~rjones/virt-p2v

4 3

[libvirt] [PATCH 0/4] allow kvm/qemu to survive daemon restarts
by Guido Günther 18 Jan '09

18 Jan '09

Hi, attached is the current patch series for the above issue. The biggest change is in the 3rd patch that switches the qemu driver to use the monitor fd to detect daemon shutdown, therefore we can dup the vms stdout/err on the logfile right at startup. Cheers, -- Guido

1 0

[libvirt] kvm: save / restore
by Matthias Pfafferodt 17 Jan '09

17 Jan '09

Hallo, I use kvm-81 and libvirt 0.5.1. I can save a kvm donain but if I want to restore it I get the following error in the log file: unknown migration protocol: stdio I tried it using only kvm and got the same error. How can I save / restore a VM to / from a file? Kind regards Matthias Pfafferodt -- Matthias Pfafferodt - http://www.mapfa.de Matthias.Pfafferodt <at> mapfa.de

3 3

[libvirt] [PATCH] Fix vm define error with back compat console device
by Cole Robinson 16 Jan '09

16 Jan '09

If you define a domain with serial devs > 0 && parallel devs >= serial devs, libvirtd segfaults when trying to set up the back compat console device. We were using a previous loop counter where we shouldn't. The attached patch fixes this. Thanks, Cole

4 15

[libvirt] PATCH: Allow virtual networks to survive daemon restarts
by Daniel P. Berrange 16 Jan '09

16 Jan '09

Currently when we shutdown the virtual networks are all shutdown too. This is less than useful if we're letting guest VMs hang around post shutdown of libvirtd, because it means we're tearing their network connection out from under them. This patch fixes that allowing networks to survive restarts, and be re-detected next time around. When starting a virtual network we write the live config into /var/lib/libvirt/network/$NAME.xml This is because the bridge device name is potentially auto-generated and we need to keep track of that We change dnsmasq args to include an explicit pidfile location /var/run/libvirt/network/$NAME.pid and also tell it to put itself into the background - ie daemonize. This is because we want dnsmasq to survive the daemon. Now, when libvirtd starts up it - Looks for the live config, and if found loads it. - Calls a new method brHasBridge() to see if its desired bridge actually exists (and thus whether the network is running). If it exists,the network is marked active - If DHCP is configured, then reads the dnsmasq PIDfile, and sends kill($PID, 0) to check if the process is actually alive In addition I cleanup the network code to remove the configFile and autostartLink fields in virNetworkObjPtr, so it matches virDomaiObjPtr usage. With all this applied you can now restart the daemon, and virbr0 is left happily running. THis patch depends on the 25 threading patches I sent earlier, so you probably won't be able to apply it in isolation Daniel diff --git a/src/bridge.c b/src/bridge.c --- a/src/bridge.c +++ b/src/bridge.c @@ -163,6 +163,43 @@ int brAddBridge (brControl *ctl ATTRIBUT } #endif +#ifdef SIOCBRDELBR +int +brHasBridge(brControl *ctl, + const char *name) +{ + struct ifreq ifr; + int len; + + if (!ctl || !name) { + errno = EINVAL; + return -1; + } + + if ((len = strlen(name)) >= BR_IFNAME_MAXLEN) { + errno = EINVAL; + return -1; + } + + memset(&ifr, 0, sizeof(struct ifreq)); + + strncpy(ifr.ifr_name, name, len); + ifr.ifr_name[len] = '\0'; + + if (ioctl(ctl->fd, SIOCGIFMTU, &ifr)) + return -1; + + return 0; +} +#else +int +brHasBridge(brControl *ctl, + const char *name) +{ + return EINVAL; +} +#endif + /** * brDeleteBridge: * @ctl: bridge control pointer diff --git a/src/bridge.h b/src/bridge.h --- a/src/bridge.h +++ b/src/bridge.h @@ -50,6 +50,8 @@ int brAddBridge (brContr char **name); int brDeleteBridge (brControl *ctl, const char *name); +int brHasBridge (brControl *ctl, + const char *name); int brAddInterface (brControl *ctl, const char *bridge, @@ -58,6 +60,7 @@ int brDeleteInterface (brContr const char *bridge, const char *iface); + int brAddTap (brControl *ctl, const char *bridge, char **ifname, diff --git a/src/libvirt_bridge.syms b/src/libvirt_bridge.syms --- a/src/libvirt_bridge.syms +++ b/src/libvirt_bridge.syms @@ -9,6 +9,7 @@ brAddBridge; brAddInterface; brAddTap; brDeleteBridge; +brHasBridge; brInit; brSetEnableSTP; brSetForwardDelay; diff --git a/src/libvirt_private.syms b/src/libvirt_private.syms --- a/src/libvirt_private.syms +++ b/src/libvirt_private.syms @@ -190,6 +190,7 @@ virFree; # network_conf.h virNetworkAssignDef; +virNetworkConfigFile; virNetworkDefFormat; virNetworkDefFree; virNetworkDefParseFile; @@ -202,6 +203,7 @@ virNetworkLoadAllConfigs; virNetworkObjListFree; virNetworkDefParseNode; virNetworkRemoveInactive; +virNetworkSaveConfigXML; virNetworkSaveConfig; virNetworkObjLock; virNetworkObjUnlock; diff --git a/src/network_conf.c b/src/network_conf.c --- a/src/network_conf.c +++ b/src/network_conf.c @@ -125,9 +125,6 @@ void virNetworkObjFree(virNetworkObjPtr virNetworkDefFree(net->def); virNetworkDefFree(net->newDef); - VIR_FREE(net->configFile); - VIR_FREE(net->autostartLink); - virMutexDestroy(&net->lock); VIR_FREE(net); @@ -641,31 +638,17 @@ char *virNetworkDefFormat(virConnectPtr return NULL; } -int virNetworkSaveConfig(virConnectPtr conn, - const char *configDir, - const char *autostartDir, - virNetworkObjPtr net) +int virNetworkSaveXML(virConnectPtr conn, + const char *configDir, + virNetworkDefPtr def, + const char *xml) { - char *xml; + char *configFile = NULL; int fd = -1, ret = -1; size_t towrite; int err; - if (!net->configFile && - virAsprintf(&net->configFile, "%s/%s.xml", - configDir, net->def->name) < 0) { - virNetworkReportError(conn, VIR_ERR_NO_MEMORY, NULL); - goto cleanup; - } - if (!net->autostartLink && - virAsprintf(&net->autostartLink, "%s/%s.xml", - autostartDir, net->def->name) < 0) { - virNetworkReportError(conn, VIR_ERR_NO_MEMORY, NULL); - goto cleanup; - } - - if (!(xml = virNetworkDefFormat(conn, - net->newDef ? net->newDef : net->def))) + if ((configFile = virNetworkConfigFile(conn, configDir, def->name)) == NULL) goto cleanup; if ((err = virFileMakePath(configDir))) { @@ -675,19 +658,12 @@ int virNetworkSaveConfig(virConnectPtr c goto cleanup; } - if ((err = virFileMakePath(autostartDir))) { - virReportSystemError(conn, err, - _("cannot create autostart directory '%s'"), - autostartDir); - goto cleanup; - } - - if ((fd = open(net->configFile, + if ((fd = open(configFile, O_WRONLY | O_CREAT | O_TRUNC, S_IRUSR | S_IWUSR )) < 0) { virReportSystemError(conn, errno, _("cannot create config file '%s'"), - net->configFile); + configFile); goto cleanup; } @@ -695,48 +671,64 @@ int virNetworkSaveConfig(virConnectPtr c if (safewrite(fd, xml, towrite) < 0) { virReportSystemError(conn, errno, _("cannot write config file '%s'"), - net->configFile); + configFile); goto cleanup; } if (close(fd) < 0) { virReportSystemError(conn, errno, _("cannot save config file '%s'"), - net->configFile); + configFile); goto cleanup; } ret = 0; cleanup: - VIR_FREE(xml); if (fd != -1) close(fd); + VIR_FREE(configFile); + return ret; } +int virNetworkSaveConfig(virConnectPtr conn, + const char *configDir, + virNetworkDefPtr def) +{ + int ret = -1; + char *xml; + + if (!(xml = virNetworkDefFormat(conn, + def))) + goto cleanup; + + if (virNetworkSaveXML(conn, configDir, def, xml)) + goto cleanup; + + ret = 0; +cleanup: + VIR_FREE(xml); + return ret; +} + + virNetworkObjPtr virNetworkLoadConfig(virConnectPtr conn, virNetworkObjListPtr nets, const char *configDir, const char *autostartDir, - const char *file) + const char *name) { char *configFile = NULL, *autostartLink = NULL; virNetworkDefPtr def = NULL; virNetworkObjPtr net; int autostart; - if (virAsprintf(&configFile, "%s/%s", - configDir, file) < 0) { - virNetworkReportError(conn, VIR_ERR_NO_MEMORY, NULL); + if ((configFile = virNetworkConfigFile(conn, configDir, name)) == NULL) goto error; - } - if (virAsprintf(&autostartLink, "%s/%s", - autostartDir, file) < 0) { - virNetworkReportError(conn, VIR_ERR_NO_MEMORY, NULL); + if ((autostartLink = virNetworkConfigFile(conn, autostartDir, name)) == NULL) goto error; - } if ((autostart = virFileLinkPointsTo(autostartLink, configFile)) < 0) goto error; @@ -744,7 +736,7 @@ virNetworkObjPtr virNetworkLoadConfig(vi if (!(def = virNetworkDefParseFile(conn, configFile))) goto error; - if (!virFileMatchesNameSuffix(file, def->name, ".xml")) { + if (!STREQ(name, def->name)) { virNetworkReportError(conn, VIR_ERR_INTERNAL_ERROR, _("Network config filename '%s'" " does not match network name '%s'"), @@ -755,10 +747,11 @@ virNetworkObjPtr virNetworkLoadConfig(vi if (!(net = virNetworkAssignDef(conn, nets, def))) goto error; - net->configFile = configFile; - net->autostartLink = autostartLink; net->autostart = autostart; + VIR_FREE(configFile); + VIR_FREE(autostartLink); + return net; error: @@ -791,7 +784,7 @@ int virNetworkLoadAllConfigs(virConnectP if (entry->d_name[0] == '.') continue; - if (!virFileHasSuffix(entry->d_name, ".xml")) + if (!virFileStripSuffix(entry->d_name, ".xml")) continue; /* NB: ignoring errors, so one malformed config doesn't @@ -811,27 +804,51 @@ int virNetworkLoadAllConfigs(virConnectP } int virNetworkDeleteConfig(virConnectPtr conn, + const char *configDir, + const char *autostartDir, virNetworkObjPtr net) { - if (!net->configFile || !net->autostartLink) { - virNetworkReportError(conn, VIR_ERR_INTERNAL_ERROR, - _("no config file for %s"), net->def->name); - return -1; - } + char *configFile = NULL; + char *autostartLink = NULL; + + if ((configFile = virNetworkConfigFile(conn, configDir, net->def->name)) == NULL) + goto error; + if ((autostartLink = virNetworkConfigFile(conn, autostartDir, net->def->name)) == NULL) + goto error; /* Not fatal if this doesn't work */ - unlink(net->autostartLink); + unlink(autostartLink); - if (unlink(net->configFile) < 0) { + if (unlink(configFile) < 0) { virReportSystemError(conn, errno, _("cannot remove config file '%s'"), - net->configFile); - return -1; + configFile); + goto error; } return 0; + +error: + VIR_FREE(configFile); + VIR_FREE(autostartLink); + return -1; } +char *virNetworkConfigFile(virConnectPtr conn, + const char *dir, + const char *name) +{ + char *ret = NULL; + + if (virAsprintf(&ret, "%s/%s.xml", dir, name) < 0) { + virNetworkReportError(conn, VIR_ERR_NO_MEMORY, NULL); + return NULL; + } + + return ret; +} + + void virNetworkObjLock(virNetworkObjPtr obj) { virMutexLock(&obj->lock); diff --git a/src/network_conf.h b/src/network_conf.h --- a/src/network_conf.h +++ b/src/network_conf.h @@ -90,9 +90,6 @@ struct _virNetworkObj { unsigned int autostart : 1; unsigned int persistent : 1; - char *configFile; /* Persistent config file path */ - char *autostartLink; /* Symlink path for autostart */ - virNetworkDefPtr def; /* The current definition */ virNetworkDefPtr newDef; /* New definition to activate at shutdown */ }; @@ -139,10 +136,14 @@ char *virNetworkDefFormat(virConnectPtr const virNetworkDefPtr def); +int virNetworkSaveXML(virConnectPtr conn, + const char *configDir, + virNetworkDefPtr def, + const char *xml); + int virNetworkSaveConfig(virConnectPtr conn, const char *configDir, - const char *autostartDir, - virNetworkObjPtr net); + virNetworkDefPtr def); virNetworkObjPtr virNetworkLoadConfig(virConnectPtr conn, virNetworkObjListPtr nets, @@ -156,8 +157,15 @@ int virNetworkLoadAllConfigs(virConnectP const char *autostartDir); int virNetworkDeleteConfig(virConnectPtr conn, + const char *configDir, + const char *autostartDir, virNetworkObjPtr net); +char *virNetworkConfigFile(virConnectPtr conn, + const char *dir, + const char *name); + + void virNetworkObjLock(virNetworkObjPtr obj); void virNetworkObjUnlock(virNetworkObjPtr obj); diff --git a/src/network_driver.c b/src/network_driver.c --- a/src/network_driver.c +++ b/src/network_driver.c @@ -57,6 +57,8 @@ #include "iptables.h" #include "bridge.h" +#define NETWORK_PID_DIR LOCAL_STATE_DIR "/run/libvirt/network" +#define NETWORK_LIB_DIR LOCAL_STATE_DIR "/lib/libvirt/network" #define VIR_FROM_THIS VIR_FROM_NETWORK @@ -106,6 +108,64 @@ static struct network_driver *driverStat static void +networkFindActiveConfigs(struct network_driver *driver) { + unsigned int i; + + for (i = 0 ; i < driver->networks.count ; i++) { + virNetworkObjPtr obj = driver->networks.objs[i]; + virNetworkDefPtr tmp; + char *config; + + virNetworkObjLock(obj); + + if ((config = virNetworkConfigFile(NULL, + NETWORK_LIB_DIR, + obj->def->name)) == NULL) { + virNetworkObjUnlock(obj); + continue; + } + + if (access(config, R_OK) < 0) { + VIR_FREE(config); + virNetworkObjUnlock(obj); + continue; + } + + /* Try and load the live config */ + tmp = virNetworkDefParseFile(NULL,config); + VIR_FREE(config); + if (tmp) { + obj->newDef = obj->def; + obj->def = tmp; + } + + /* If bridge exists, then mark it active */ + if (obj->def->bridge && + brHasBridge(driver->brctl, obj->def->bridge) == 0) { + obj->active = 1; + + /* Finally try and read dnsmasq pid if any DHCP ranges are set */ + if (obj->def->nranges && + virFileReadPid(NETWORK_PID_DIR, obj->def->name, + &obj->dnsmasqPid) == 0) { + + /* Check its still alive */ + if (kill(obj->dnsmasqPid, 0) != 0) + obj->dnsmasqPid = -1; + + /* XXX ideally we'd check this was actually + * the dnsmasq process, not a stale pid file + * with someone else's process. But how ? + */ + } + } + + virNetworkObjUnlock(obj); + } +} + + +static void networkAutostartConfigs(struct network_driver *driver) { unsigned int i; @@ -132,6 +192,7 @@ static int networkStartup(void) { uid_t uid = geteuid(); char *base = NULL; + int err; if (VIR_ALLOC(driverState) < 0) goto error; @@ -181,12 +242,26 @@ networkStartup(void) { VIR_FREE(base); + if ((err = brInit(&driverState->brctl))) { + virReportSystemError(NULL, err, "%s", + _("cannot initialize bridge support")); + goto error; + } + + if (!(driverState->iptables = iptablesContextNew())) { + networkReportError(NULL, NULL, NULL, VIR_ERR_NO_MEMORY, + "%s", _("failed to allocate space for IP tables support")); + goto error; + } + + if (virNetworkLoadAllConfigs(NULL, &driverState->networks, driverState->networkConfigDir, driverState->networkAutostartDir) < 0) goto error; + networkFindActiveConfigs(driverState); networkAutostartConfigs(driverState); networkDriverUnlock(driverState); @@ -269,23 +344,11 @@ networkActive(void) { */ static int networkShutdown(void) { - unsigned int i; - if (!driverState) return -1; networkDriverLock(driverState); - /* shutdown active networks */ - for (i = 0 ; i < driverState->networks.count ; i++) { - virNetworkObjPtr net = driverState->networks.objs[i]; - virNetworkObjLock(net); - if (virNetworkIsActive(net)) - networkShutdownNetworkDaemon(NULL, driverState, - driverState->networks.objs[i]); - virNetworkObjUnlock(net); - } - /* free inactive networks */ virNetworkObjListFree(&driverState->networks); @@ -309,23 +372,23 @@ networkShutdown(void) { static int networkBuildDnsmasqArgv(virConnectPtr conn, - virNetworkObjPtr network, - const char ***argv) { + virNetworkObjPtr network, + const char *pidfile, + const char ***argv) { int i, len, r; - char buf[PATH_MAX]; + char *pidfileArg; + char buf[1024]; len = 1 + /* dnsmasq */ - 1 + /* --keep-in-foreground */ 1 + /* --strict-order */ 1 + /* --bind-interfaces */ (network->def->domain?2:0) + /* --domain name */ - 2 + /* --pid-file "" */ + 2 + /* --pid-file /var/run/libvirt/network/$NAME.pid */ 2 + /* --conf-file "" */ /*2 + *//* --interface virbr0 */ 2 + /* --except-interface lo */ 2 + /* --listen-address 10.0.0.1 */ - 1 + /* --dhcp-leasefile=path */ (2 * network->def->nranges) + /* --dhcp-range 10.0.0.2,10.0.0.254 */ /* --dhcp-host 01:23:45:67:89:0a,hostname,10.0.0.3 */ (2 * network->def->nhosts) + @@ -339,11 +402,13 @@ networkBuildDnsmasqArgv(virConnectPtr co goto no_memory; \ } while (0) +#define APPEND_ARG_LIT(v, n, s) \ + (v)[(n)] = s + i = 0; APPEND_ARG(*argv, i++, DNSMASQ); - APPEND_ARG(*argv, i++, "--keep-in-foreground"); /* * Needed to ensure dnsmasq uses same algorithm for processing * multiple namedriver entries in /etc/resolv.conf as GLibC. @@ -356,10 +421,11 @@ networkBuildDnsmasqArgv(virConnectPtr co APPEND_ARG(*argv, i++, network->def->domain); } - APPEND_ARG(*argv, i++, "--pid-file"); - APPEND_ARG(*argv, i++, ""); + if (virAsprintf(&pidfileArg, "--pid-file=%s", pidfile) < 0) + goto no_memory; + APPEND_ARG_LIT(*argv, i++, pidfileArg); - APPEND_ARG(*argv, i++, "--conf-file"); + APPEND_ARG(*argv, i++, "--conf-file="); APPEND_ARG(*argv, i++, ""); /* @@ -377,15 +443,6 @@ networkBuildDnsmasqArgv(virConnectPtr co APPEND_ARG(*argv, i++, "--except-interface"); APPEND_ARG(*argv, i++, "lo"); - /* - * NB, dnsmasq command line arg bug means we need to - * use a single arg '--dhcp-leasefile=path' rather than - * two separate args in '--dhcp-leasefile path' style - */ - snprintf(buf, sizeof(buf), "--dhcp-leasefile=%s/lib/libvirt/dhcp-%s.leases", - LOCAL_STATE_DIR, network->def->name); - APPEND_ARG(*argv, i++, buf); - for (r = 0 ; r < network->def->nranges ; r++) { snprintf(buf, sizeof(buf), "%s,%s", network->def->ranges[r].start, @@ -434,7 +491,10 @@ dhcpStartDhcpDaemon(virConnectPtr conn, virNetworkObjPtr network) { const char **argv; - int ret, i; + char *pidfile; + int ret = -1, i, err; + + network->dnsmasqPid = -1; if (network->def->ipAddress == NULL) { networkReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR, @@ -442,13 +502,49 @@ dhcpStartDhcpDaemon(virConnectPtr conn, return -1; } + if ((err = virFileMakePath(NETWORK_PID_DIR)) < 0) { + virReportSystemError(conn, err, + _("cannot create directory %s"), + NETWORK_PID_DIR); + return -1; + } + if ((err = virFileMakePath(NETWORK_LIB_DIR)) < 0) { + virReportSystemError(conn, err, + _("cannot create directory %s"), + NETWORK_LIB_DIR); + return -1; + } + + if (!(pidfile = virFilePid(NETWORK_PID_DIR, network->def->name))) { + virReportOOMError(conn); + return -1; + } + argv = NULL; - if (networkBuildDnsmasqArgv(conn, network, &argv) < 0) + if (networkBuildDnsmasqArgv(conn, network, pidfile, &argv) < 0) { + VIR_FREE(pidfile); return -1; + } - ret = virExec(conn, argv, NULL, NULL, - &network->dnsmasqPid, -1, NULL, NULL, VIR_EXEC_NONBLOCK); + if (virRun(conn, argv, NULL) < 0) + goto cleanup; + /* + * There really is no race here - when dnsmasq daemonizes, + * its leader process stays around until its child has + * actually written its pidfile. So by time virRun exits + * it has waitpid'd and guarenteed the proess has started + * and writtena pid + */ + + if (virFileReadPid(NETWORK_PID_DIR, network->def->name, + &network->dnsmasqPid) < 0) + goto cleanup; + + ret = 0; + +cleanup: + VIR_FREE(pidfile); for (i = 0; argv[i]; i++) VIR_FREE(argv[i]); VIR_FREE(argv); @@ -554,13 +650,6 @@ networkAddIptablesRules(virConnectPtr co virNetworkObjPtr network) { int err; - if (!driver->iptables && !(driver->iptables = iptablesContextNew())) { - networkReportError(conn, NULL, NULL, VIR_ERR_NO_MEMORY, - "%s", _("failed to allocate space for IP tables support")); - return 0; - } - - /* allow DHCP requests through to dnsmasq */ if ((err = iptablesAddTcpInput(driver->iptables, network->def->bridge, 67))) { virReportSystemError(conn, err, @@ -716,12 +805,6 @@ static int networkStartNetworkDaemon(vir return -1; } - if (!driver->brctl && (err = brInit(&driver->brctl))) { - virReportSystemError(conn, err, "%s", - _("cannot initialize bridge support")); - return -1; - } - if ((err = brAddBridge(driver->brctl, &network->def->bridge))) { virReportSystemError(conn, err, _("cannot create bridge '%s'"), @@ -729,7 +812,6 @@ static int networkStartNetworkDaemon(vir return -1; } - if (brSetForwardDelay(driver->brctl, network->def->bridge, network->def->delay) < 0) goto err_delbr; @@ -774,10 +856,22 @@ static int networkStartNetworkDaemon(vir dhcpStartDhcpDaemon(conn, network) < 0) goto err_delbr2; + + /* Persist the live configuration now we have bridge info */ + if (virNetworkSaveConfig(conn, NETWORK_LIB_DIR, network->def) < 0) { + goto err_kill; + } + network->active = 1; return 0; + err_kill: + if (network->dnsmasqPid > 0) { + kill(network->dnsmasqPid, SIGTERM); + network->dnsmasqPid = -1; + } + err_delbr2: networkRemoveIptablesRules(driver, network); @@ -798,16 +892,24 @@ static int networkStartNetworkDaemon(vir } -static int networkShutdownNetworkDaemon(virConnectPtr conn ATTRIBUTE_UNUSED, - struct network_driver *driver, - virNetworkObjPtr network) { +static int networkShutdownNetworkDaemon(virConnectPtr conn, + struct network_driver *driver, + virNetworkObjPtr network) { int err; + char *configFile; networkLog(NETWORK_INFO, _("Shutting down network '%s'\n"), network->def->name); if (!virNetworkIsActive(network)) return 0; + configFile = virNetworkConfigFile(conn, NETWORK_LIB_DIR, network->def->name); + if (!configFile) + return -1; + + unlink(configFile); + VIR_FREE(configFile); + if (network->dnsmasqPid > 0) kill(network->dnsmasqPid, SIGTERM); @@ -824,13 +926,10 @@ static int networkShutdownNetworkDaemon( network->def->bridge, strerror(err)); } + /* See if its still alive and really really kill it */ if (network->dnsmasqPid > 0 && - waitpid(network->dnsmasqPid, NULL, WNOHANG) != network->dnsmasqPid) { + (kill(network->dnsmasqPid, 0) == 0)) kill(network->dnsmasqPid, SIGKILL); - if (waitpid(network->dnsmasqPid, NULL, 0) != network->dnsmasqPid) - networkLog(NETWORK_WARN, - "%s", _("Got unexpected pid for dnsmasq\n")); - } network->dnsmasqPid = -1; network->active = 0; @@ -1048,8 +1147,7 @@ static virNetworkPtr networkDefine(virCo if (virNetworkSaveConfig(conn, driver->networkConfigDir, - driver->networkAutostartDir, - network) < 0) { + network->newDef ? network->newDef : network->def) < 0) { virNetworkRemoveInactive(&driver->networks, network); network = NULL; @@ -1086,7 +1184,10 @@ static int networkUndefine(virNetworkPtr goto cleanup; } - if (virNetworkDeleteConfig(net->conn, network) < 0) + if (virNetworkDeleteConfig(net->conn, + driver->networkConfigDir, + driver->networkAutostartDir, + network) < 0) goto cleanup; virNetworkRemoveInactive(&driver->networks, @@ -1140,7 +1241,7 @@ static int networkDestroy(virNetworkPtr } ret = networkShutdownNetworkDaemon(net->conn, driver, network); - if (!network->configFile) { + if (!network->persistent) { virNetworkRemoveInactive(&driver->networks, network); network = NULL; @@ -1233,9 +1334,10 @@ cleanup: } static int networkSetAutostart(virNetworkPtr net, - int autostart) { + int autostart) { struct network_driver *driver = net->conn->networkPrivateData; virNetworkObjPtr network; + char *configFile = NULL, *autostartLink = NULL; int ret = -1; networkDriverLock(driver); @@ -1251,6 +1353,11 @@ static int networkSetAutostart(virNetwor autostart = (autostart != 0); if (network->autostart != autostart) { + if ((configFile = virNetworkConfigFile(net->conn, driver->networkConfigDir, network->def->name)) == NULL) + goto cleanup; + if ((autostartLink = virNetworkConfigFile(net->conn, driver->networkAutostartDir, network->def->name)) == NULL) + goto cleanup; + if (autostart) { int err; @@ -1261,17 +1368,17 @@ static int networkSetAutostart(virNetwor goto cleanup; } - if (symlink(network->configFile, network->autostartLink) < 0) { + if (symlink(configFile, autostartLink) < 0) { virReportSystemError(net->conn, errno, _("Failed to create symlink '%s' to '%s'"), - network->autostartLink, network->configFile); + autostartLink, configFile); goto cleanup; } } else { - if (unlink(network->autostartLink) < 0 && errno != ENOENT && errno != ENOTDIR) { + if (unlink(autostartLink) < 0 && errno != ENOENT && errno != ENOTDIR) { virReportSystemError(net->conn, errno, _("Failed to delete symlink '%s'"), - network->autostartLink); + autostartLink); goto cleanup; } } @@ -1281,6 +1388,8 @@ static int networkSetAutostart(virNetwor ret = 0; cleanup: + VIR_FREE(configFile); + VIR_FREE(autostartLink); if (network) virNetworkObjUnlock(network); return ret; -- |: Red Hat, Engineering, London -o- http://people.redhat.com/berrange/ :| |: http://libvirt.org -o- http://virt-manager.org -o- http://ovirt.org :| |: http://autobuild.org -o- http://search.cpan.org/~danberr/ :| |: GnuPG: 7D3B9505 -o- F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 :|

3 4

[libvirt] sharing multiple identical USB devices?
by Dr. Michael J. Chudobiak 16 Jan '09

16 Jan '09

Hi, I have Windows XP running on Fedora 10. I can successfully share a single USB-to-RS232 converter by inserting this in my xml file: <hostdev mode='subsystem' type='usb'> <source> <vendor id='0x0403'/> <product id='0x6001'/> </source> </hostdev> However, I can't seem to get two of these to work at the same time. I want to do: <hostdev mode='subsystem' type='usb'> <source> <vendor id='0x0403'/> <product id='0x6001'/> <address bus='0x03' device='0x02'/> </source> </hostdev> <hostdev mode='subsystem' type='usb'> <source> <vendor id='0x0403'/> <product id='0x6001'/> <address bus='0x03' device='0x03'/> </source> </hostdev> using the "address" stanza to identify the devices uniquely, but it just doesn't work. Is this a bug, or am I doing something wrong? With the two device setup, Windows still only assigns one serial port to one of the devices, and none to the other. - Mike

3 3

[libvirt] Fine grained Access Control in libVirt
by Konrad Eriksson1 16 Jan '09

16 Jan '09

Hi, After some background research it doesn't look like anyone have taken on the task yet to add fine-grained access control to libVirt (only option right now is R/W or RO mode). Desired is an addition to libVirt that enables access control on individual actions and data that can be accessed through the library API. This could take the form of an AC-module that, based on the identity of the caller, checks each call and grants/denies access to carry out the action (could also take parameters in account) and optionally filter the return data. The AC-module could then interface different backend AC solutions (SELinux, RBAC, ...) or alternatively implement an internal scheme. When looking at the libvirt core and driver framework it seems promising to inject these kind of call-out hooks either in libvirt.c or between libvirt.c and the underlying drivers, by doing this AC will be enforced independent of if a local or remote call is done to libVirt. I propose an approach to create an AC-module that conforms to the driver API in libVirt and to inject it in the call-path between libvirt.c and the driver(s) to enable the AC-module to inspect the call before sending it to the real driver. Normal call path: user -> libvirt.c -> driver_x AC-module injected in call path: user -> libvirt.c -> AC-module -> driver_x By doing this it can be loaded/unloaded in run-time and also selectable what driver paths it should enforce (hypervisor(aka. driver), storage, network...). The AC-module can also be made in different flavors for different AC backend (SELinux, RBAC, internal, ...) solutions and the appropriate AC-module could be loaded without needing re-compiling. This approach would also leave a very small footprint in existing libvirt code (only need to inject AC-module driver after normal drivers has been loaded) Feel free to comment and to come with improvement ideas. Freundliche Grüsse / Best regards Konrad Eriksson Trusted Computing / Security & Assurance IBM Zurich Research Laboratory Saeumerstrasse 4 8803 Rueschlikon Switzerland

4 5

[libvirt] NULL deref in xenStoreDomainReleased()
by John Levon 15 Jan '09

15 Jan '09

I got a crash with this stack: ----------------- lwp# 1 / thread# 1 -------------------- 00007fffff312a6c xenStoreDomainReleased () + 24 00007fffff312774 xenStoreWatchEvent () + 6c 00000000004181fb virEventDispatchHandles () + 4cb 0000000000418651 virEventRunOnce () + 139 000000000041bc1e qemudOneLoop () + e 000000000041be7b qemudRunLoop () + 13b 000000000041dd43 main () + 6cb 0000000000416f1c ???????? () It looks to me like activeDomainList became NULL. But we've already removed this watch by this point in xenStoreClose(). I'm not au fait with the event locking - does the watch removal need the event lock, or something? thanks john

1 0

[libvirt] [PATCH] Solaris least privilege support
by john.levon＠sun.com 15 Jan '09

15 Jan '09

# HG changeset patch # User john.levon(a)sun.com # Date 1231990064 28800 # Node ID 629c101c9ec11f3eb5cb56eb9548c96c33c8daf6 # Parent 0f488fb716b1ab0a1379509b8b3594f32f0ea980 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 +84,25 @@ #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, ...); + +#endif + static int godaemon = 0; /* -d: Be a daemon */ static int verbose = 0; /* -v: Verbose mode */ static int timeout = -1; /* -t: Shutdown timeout */ @@ -638,10 +657,32 @@ 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(); - +#ifdef __sun + char *base = NULL; + + if (virAsprintf (&base, "%s/run/libvirt", LOCAL_STATE_DIR) == -1) { + VIR_ERROR0(_("Out of memory")); + return -1; + } + if (mkdir (base, 0755)) { + if (errno != EEXIST) { + VIR_ERROR0(_("unable to create rundir")); + free (base); + exit(-1); + } + } + + free (base); +#endif + +#ifdef __sun + if (uid == 60) { +#else if (!uid) { +#endif if (snprintf (sockname, maxlen, "%s/run/libvirt/libvirt-sock", LOCAL_STATE_DIR) >= maxlen) goto snprintf_error; @@ -1105,6 +1146,29 @@ 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); @@ -2140,6 +2204,10 @@ remoteReadConfigFile (struct qemud_serve if (auth_unix_rw == REMOTE_AUTH_POLKIT) unix_sock_rw_mask = 0777; #endif +#ifdef __sun + unix_sock_rw_mask = 0666; +#endif + if (remoteConfigGetAuth(conf, "auth_unix_ro", &auth_unix_ro, filename) < 0) goto free_and_fail; if (remoteConfigGetAuth(conf, "auth_tcp", &auth_tcp, filename) < 0) @@ -2239,6 +2307,31 @@ 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", 60, 60); + chown ("/var/run/libvirt/libvirt-sock", 60, 60); + chmod ("/var/run/libvirt/libvirt-sock", 0666); + chown (server->logDir, 60, 60); + + if (__init_daemon_priv (PU_RESETGROUPS | PU_CLEARLIMITSET, + 60, 60, 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 +2510,8 @@ int main(int argc, char **argv) { sig_action.sa_handler = SIG_IGN; sigaction(SIGPIPE, &sig_action, NULL); + qemudSetupPrivs(server); + if (!(server = qemudInitialize(sigpipe[0]))) { ret = 2; goto error2; diff --git a/qemud/remote.c b/qemud/remote.c --- a/qemud/remote.c +++ b/qemud/remote.c @@ -424,6 +424,15 @@ remoteDispatchOpen (struct qemud_server flags = args->flags; if (client->readonly) flags |= VIR_CONNECT_RO; +#ifdef __sun + /* + * On Solaris, all clients are forced to go via virtd. As a result, + * virtd must indicate it really does want to connect to the + * hypervisor. + */ + name = "xen:///"; +#endif + client->conn = flags & VIR_CONNECT_RO ? virConnectOpenReadOnly (name) diff --git a/src/libvirt.c b/src/libvirt.c --- a/src/libvirt.c +++ b/src/libvirt.c @@ -46,6 +46,7 @@ #include "test.h" #endif #ifdef WITH_XEN +#include "xen_internal.h" #include "xen_unified.h" #endif #ifdef WITH_REMOTE @@ -825,6 +826,17 @@ do_open (const char *name, } } +#ifdef __sun + /* + * If we're not libvirtd, force us to go via the daemon, unless we + * want the test hypervisor. + */ + if (name == NULL || !STRCASEEQLEN (name, "test://", 7)) { + if (geteuid() == 0 || !xenHavePrivilege()) + name = "remote+unix:///"; + } +#endif + if (name) { /* Convert xen -> xen:/// for back compat */ if (STRCASEEQ(name, "xen")) diff --git a/src/remote_internal.c b/src/remote_internal.c --- a/src/remote_internal.c +++ b/src/remote_internal.c @@ -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; @@ -5086,8 +5089,7 @@ really_read_buf (virConnectPtr conn, str return -1; } if (err == 0) { - error (in_open ? NULL : conn, - VIR_ERR_RPC, _("socket closed unexpectedly")); + DEBUG("conn %p: socket closed unexpectedly", conn); return -1; } } else { @@ -5101,8 +5103,7 @@ really_read_buf (virConnectPtr conn, str return -1; } if (err == 0) { - error (in_open ? NULL : conn, - VIR_ERR_RPC, _("socket closed unexpectedly")); + DEBUG("conn %p: socket closed unexpectedly", conn); return -1; } } diff --git a/src/virsh.c b/src/virsh.c --- a/src/virsh.c +++ b/src/virsh.c @@ -28,6 +28,7 @@ #include <limits.h> #include <assert.h> #include <errno.h> +#include <signal.h> #include <sys/stat.h> #include <sys/wait.h> #include <inttypes.h> @@ -46,6 +47,7 @@ #include "util.h" static char *progname; +static int sigpipe; #ifndef TRUE #define TRUE 1 @@ -6984,12 +6986,22 @@ vshParseArgv(vshControl *ctl, int argc, return TRUE; } +static void sigpipe_handler(int sig ATTRIBUTE_UNUSED) +{ + sigpipe = 1; + /* + * Force readline() to exit. + */ + close(STDIN_FILENO); +} + int main(int argc, char **argv) { vshControl _ctl, *ctl = &_ctl; char *defaultConn; int ret = TRUE; + struct sigaction sig_action; if (!setlocale(LC_ALL, "")) { perror("setlocale"); @@ -7021,6 +7033,12 @@ main(int argc, char **argv) vshDeinit(ctl); exit(EXIT_FAILURE); } + + sig_action.sa_handler = sigpipe_handler; + sig_action.sa_flags = 0; + sigemptyset(&sig_action.sa_mask); + + sigaction(SIGPIPE, &sig_action, NULL); if (!vshInit(ctl)) { vshDeinit(ctl); @@ -7061,6 +7079,13 @@ main(int argc, char **argv) fputc('\n', stdout); /* line break after alone prompt */ } + /* + * If the connection over a socket failed abruptly, it's probably + * due to not having the right privileges. + */ + if (sigpipe) + vshError(ctl, TRUE, _("failed to connect (insufficient privileges?)")); + vshDeinit(ctl); exit(ret ? EXIT_SUCCESS : EXIT_FAILURE); } 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 @@ -283,8 +283,8 @@ 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 +293,7 @@ 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 +318,12 @@ 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")); }

3 4

[libvirt] FYI: Mercurial repository with read-only CVS mirror
by Daniel P. Berrange 15 Jan '09

15 Jan '09

FYI. After a number of requests I have setup a anonymously accessible mercurial repository containing a read-only mirror of libvirt CVS. It is accessible via HTTP only by running: cd $HOME hg clone http://libvirt.org/hg/libvirt-mirror/ The conversion is done using Tailor, and synced every 15 minutes. It is an incremental sync, so you can pull down new changes as they appear by then doing cd $HOME cd libvirt-mirror hg pull -u Regards, Daniel -- |: Red Hat, Engineering, London -o- http://people.redhat.com/berrange/ :| |: http://libvirt.org -o- http://virt-manager.org -o- http://ovirt.org :| |: http://autobuild.org -o- http://search.cpan.org/~danberr/ :| |: GnuPG: 7D3B9505 -o- F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 :|

1 0

[libvirt] [PATCH] Introduce virt-console
by john.levon＠sun.com 15 Jan '09

15 Jan '09

# HG changeset patch # User John Levon <john.levon(a)sun.com> # Date 1231946128 28800 # Node ID dd17b3062611925baa2698ff5923579d0f2cd34e # Parent a0d98d39955f4f304d318c7c780742ab929eb351 Introduce virt-console Separate console handling out into a separate binary to allow management of privileges. Signed-off-by: John Levon <john.levon(a)sun.com> diff --git a/src/Makefile.am b/src/Makefile.am --- a/src/Makefile.am +++ b/src/Makefile.am @@ -479,12 +479,16 @@ libvirt_test_la_LDFLAGS = $(test_LDFLAGS libvirt_test_la_LDFLAGS = $(test_LDFLAGS) libvirt_test_la_CFLAGS = $(COVERAGE_CFLAGS) +libexec_PROGRAMS = virt-console + +virt_console_SOURCES = console.c +virt_console_CFLAGS = $(COVERAGE_CFLAGS) +virt_console_LDFLAGS = $(WARN_CFLAGS) $(COVERAGE_LDFLAGS) +virt_console_LDADD = libvirt.la ../gnulib/lib/libgnu.la $(VIRSH_LIBS) + bin_PROGRAMS = virsh -virsh_SOURCES = \ - console.c console.h \ - virsh.c - +virsh_SOURCES = virsh.c virsh_LDFLAGS = $(WARN_CFLAGS) $(COVERAGE_LDFLAGS) virsh_LDADD = \ $(STATIC_BINARIES) \ @@ -554,8 +558,6 @@ virsh_win_icon.$(OBJEXT): virsh_win_icon --output-format coff --output $@ endif -libexec_PROGRAMS = - if WITH_STORAGE_DISK if WITH_LIBVIRTD libexec_PROGRAMS += libvirt_parthelper diff --git a/src/console.c b/src/console.c --- a/src/console.c +++ b/src/console.c @@ -1,7 +1,10 @@ /* - * console.c: A dumb serial console client + * virt-console: client for connecting to domain consoles. * * Copyright (C) 2007, 2008 Red Hat, Inc. + * + * Copyright 2008 Sun Microsystems, Inc. All rights reserved. + * Use is subject to license terms. * * This library is free software; you can redistribute it and/or * modify it under the terms of the GNU Lesser General Public @@ -22,34 +25,65 @@ #include <config.h> -#ifndef __MINGW32__ - #include <stdio.h> +#include <errno.h> +#include <stdlib.h> +#include <termios.h> +#include <signal.h> #include <sys/types.h> -#include <sys/stat.h> +#include <unistd.h> #include <fcntl.h> -#include <termios.h> #include <poll.h> -#include <string.h> -#include <errno.h> -#include <unistd.h> -#include <signal.h> - -#include "console.h" +#include <getopt.h> +#include <locale.h> + +#include <libxml/parser.h> +#include <libxml/tree.h> +#include <libxml/xpath.h> + #include "internal.h" +#include "memory.h" #include "util.h" +#include "xml.h" /* ie Ctrl-] as per telnet */ #define CTRL_CLOSE_BRACKET '\35' -static int got_signal = 0; -static void do_signal(int sig ATTRIBUTE_UNUSED) { +static int got_signal; +static int verbose; +static const char *dom_name; +static const char *conn_name; + +#ifdef __sun +#include <stropts.h> +#include <priv.h> + +#define PU_RESETGROUPS 0x0001 /* Remove supplemental groups */ +#define PU_LIMITPRIVS 0x0002 /* L=P */ +#define PU_INHERITPRIVS 0x0004 /* I=P */ +#define PU_CLEARLIMITSET 0x0008 /* L=0 */ + +extern int __init_suid_priv(int, ...); +extern int __priv_bracket(priv_op_t); + +#ifndef PRIV_XVM_CONTROL +#define PRIV_XVM_CONTROL ((const char *)"xvm_control") +#endif +#ifndef PRIV_VIRT_MANAGE +#define PRIV_VIRT_MANAGE ((const char *)"virt_manage") +#endif + +#endif /* __sun */ + +static void +do_signal(int sig ATTRIBUTE_UNUSED) +{ got_signal = 1; } #ifndef HAVE_CFMAKERAW static void -cfmakeraw (struct termios *attr) +cfmakeraw(struct termios *attr) { attr->c_iflag &= ~(IGNBRK | BRKINT | PARMRK | ISTRIP | INLCR | IGNCR | ICRNL | IXON); @@ -57,46 +91,432 @@ cfmakeraw (struct termios *attr) attr->c_lflag &= ~(ECHO | ECHONL | ICANON | ISIG | IEXTEN); attr->c_cflag &= ~(CSIZE | PARENB); attr->c_cflag |= CS8; + +#ifdef __sun + attr->c_cc[VMIN] = 0; + attr->c_cc[VTIME] = 0; +#endif } #endif /* !HAVE_CFMAKERAW */ -int vshRunConsole(const char *tty) { - int ttyfd, ret = -1; - struct termios ttyattr, rawattr; +static void make_tty_raw(int fd, struct termios *attr) +{ + struct termios ttyattr; + struct termios rawattr; + + if (attr == NULL) + attr = &ttyattr; + + if (tcgetattr(fd, attr) < 0) { + fprintf(stderr, _("Unable to get tty attributes: %s\n"), + strerror(errno)); + exit(EXIT_FAILURE); + } + + rawattr = *attr; + cfmakeraw(&rawattr); + + if (tcsetattr(STDIN_FILENO, TCSAFLUSH, &rawattr) < 0) { + fprintf(stderr, _("Unable to set tty attributes: %s\n"), + strerror(errno)); + exit(EXIT_FAILURE); + } +} + +#ifdef __sun + +/* + * Verify that the real user invoking this setuid-root executable has + * the needed privilege, then drop root and all the privileges we don't + * need before continuing. + */ +static void +setup_perms(void) +{ + if (seteuid(getuid()) == -1) { + perror("seteuid failed"); + exit(1); + } + + if (!priv_ineffect(PRIV_VIRT_MANAGE)) { + fprintf(stderr, "virt-console: permission denied\n"); + exit(1); + } + + if (seteuid(0) == -1) { + perror("seteuid failed"); + exit(1); + } + + /* + * We need to be able to talk to libvirt and open pty's. + */ + __init_suid_priv(PU_RESETGROUPS | PU_CLEARLIMITSET, + PRIV_VIRT_MANAGE, PRIV_FILE_DAC_READ, + PRIV_FILE_DAC_WRITE, NULL); +} + +#else +#define setup_perms() +#define __priv_bracket(a) +#endif + +static void +usage(int exitval) +{ + FILE *f = stderr; + + if (exitval == EXIT_SUCCESS) + f = stdout; + + fprintf(f, _("usage: virt-console [options] domain\n\n" + " options:\n" + " -c | --connect <uri> hypervisor connection URI\n" + " -h | --help this help\n" + " -v | --verbose be verbose\n\n")); + + exit(exitval); +} + +static void +parse_args(int argc, char *argv[]) +{ + int idx = 0; + int arg; + + struct option opt[] = { + { "connect", 1, 0, 'c' }, + { "help", 0, 0, 'h' }, + { "verbose", 0, 0, 'v' }, + { 0, 0, 0, 0 } + }; + + while ((arg = getopt_long(argc, argv, "c:hv", opt, &idx)) != -1) { + switch (arg) { + case 'c': + conn_name = optarg; + break; + case 'h': + usage(EXIT_SUCCESS); + break; + case 'v': + verbose = 1; + break; + default: + usage(EXIT_FAILURE); + break; + } + } + + if ((argc - optind) != 1) + usage(EXIT_FAILURE); + + dom_name = argv[optind]; + + if (conn_name == NULL) + conn_name = getenv("VIRSH_DEFAULT_CONNECT_URI"); +} + +static int +get_domain(virConnectPtr *conn, virDomainPtr *dom, + virDomainInfo *info, int lookup_by_id) +{ + int ret = 0; + int id; + + __priv_bracket(PRIV_ON); + + printf("1\n"); + *conn = virConnectOpenAuth(conn_name, virConnectAuthPtrDefault, + VIR_CONNECT_RO); + printf("2\n"); + if (*conn == NULL) { + fprintf(stderr, _("Failed to connect to the hypervisor")); + exit(EXIT_FAILURE); + } + + *dom = virDomainLookupByName(*conn, dom_name); + + if (*dom == NULL) + *dom = virDomainLookupByUUIDString(*conn, dom_name); + if (*dom == NULL && lookup_by_id && + virStrToLong_i(dom_name, NULL, 10, &id) == 0 && id >= 0) + *dom = virDomainLookupByID(*conn, id); + + if (*dom == NULL) + goto out; + + if (info != NULL) { + if (virDomainGetInfo(*dom, info) < 0) + goto out; + } + + ret = 1; + +out: + if (ret == 0) { + if (*dom != NULL) + virDomainFree(*dom); + virConnectClose(*conn); + } + __priv_bracket(PRIV_OFF); + return ret; +} + +static void +put_domain(virConnectPtr conn, virDomainPtr dom) +{ + __priv_bracket(PRIV_ON); + if (dom != NULL) + virDomainFree(dom); + virConnectClose(conn); + __priv_bracket(PRIV_OFF); +} + +static char * +get_domain_tty(void) +{ + xmlXPathContextPtr ctxt = NULL; + xmlDocPtr xml = NULL; + virConnectPtr conn = NULL; + virDomainPtr dom = NULL; + char *doc = NULL; + char *tty = NULL; + + if (!get_domain(&conn, &dom, NULL, 1)) { + fprintf(stderr, _("Couldn't find domain \"%s\".\n"), dom_name); + exit(EXIT_FAILURE); + } + + __priv_bracket(PRIV_ON); + doc = virDomainGetXMLDesc(dom, 0); + __priv_bracket(PRIV_OFF); + + put_domain(conn, dom); + conn = NULL; + dom = NULL; + + if (doc == NULL) + goto cleanup; + + xml = xmlReadDoc((const xmlChar *) doc, "domain.xml", NULL, + XML_PARSE_NOENT | XML_PARSE_NONET | + XML_PARSE_NOWARNING); + if (xml == NULL) + goto cleanup; + + ctxt = xmlXPathNewContext(xml); + if (ctxt == NULL) + goto cleanup; + + tty = virXPathString(conn, "string(/domain/devices/console/@tty)", ctxt); + +cleanup: + VIR_FREE(doc); + xmlXPathFreeContext(ctxt); + if (xml != NULL) + xmlFreeDoc(xml); + return tty; +} + +static int +domain_is_running(void) +{ + virConnectPtr conn = NULL; + virDomainPtr dom = NULL; + virDomainInfo info; + int ret = -1; + + if (!get_domain(&conn, &dom, &info, 1)) + return -1; + + switch (info.state) { + case VIR_DOMAIN_RUNNING: + case VIR_DOMAIN_BLOCKED: + case VIR_DOMAIN_PAUSED: + ret = 1; + break; + + case VIR_DOMAIN_NOSTATE: + case VIR_DOMAIN_CRASHED: + case VIR_DOMAIN_SHUTDOWN: + case VIR_DOMAIN_SHUTOFF: + ret = 0; + break; + + default: + break; + } + + put_domain(conn, dom); + return ret; +} + +static int +check_for_reboot(void) +{ + virConnectPtr conn = NULL; + virDomainPtr dom = NULL; + virDomainInfo info; + int tries = 0; + int ret = 0; + +retry: + if (dom != NULL) + put_domain(conn, dom); + + /* + * Domain ID will vary across reboot, so don't lookup by a given ID. + */ + if (!get_domain(&conn, &dom, &info, 0)) + return 0; + + switch (info.state) { + case VIR_DOMAIN_RUNNING: + case VIR_DOMAIN_BLOCKED: + case VIR_DOMAIN_PAUSED: + ret = 1; + goto out; + break; + + case VIR_DOMAIN_CRASHED: + if (verbose) + fprintf(stderr, _("Domain \"%s\" has crashed."), dom_name); + goto out; + break; + + case VIR_DOMAIN_NOSTATE: + default: + break; + + case VIR_DOMAIN_SHUTDOWN: + if (verbose) + fprintf(stderr, _("Domain \"%s\" is shutting down.\n"), dom_name); + tries = 0; + break; + + case VIR_DOMAIN_SHUTOFF: + if (verbose) + fprintf(stderr, _("Domain \"%s\" has shut down."), dom_name); + goto out; + break; + } + + tries++; + if (tries == 1) { + goto retry; + } else if (tries < 6) { + sleep(1); + goto retry; + } + +out: + put_domain(conn, dom); + return ret; +} + +static int +open_tty(void) +{ + struct termios ttyattr; + char *tty; + int ttyfd; + + if ((tty = get_domain_tty()) == NULL) { + if (domain_is_running() != 1) { + fprintf(stderr, _("Domain \"%s\" is not running.\n"), dom_name); + exit(EXIT_FAILURE); + } + + fprintf(stderr, + _("Couldn't get console for domain \"%s\"\n"), dom_name); + exit(EXIT_FAILURE); + } + + __priv_bracket(PRIV_ON); + if ((ttyfd = open(tty, O_NOCTTY | O_RDWR)) < 0) { + fprintf(stderr, _("Unable to open tty %s: %s\n"), + tty, strerror(errno)); + exit(EXIT_FAILURE); + } + + if (lockf(ttyfd, F_TLOCK, 0) == -1) { + if (errno == EACCES || errno == EAGAIN) { + fprintf(stderr, + _("Console for domain \"%s\" is in use.\n"), dom_name); + } else { + fprintf(stderr, _("Unable to lock tty %s: %s\n"), + tty, strerror(errno)); + } + exit(EXIT_FAILURE); + } + __priv_bracket(PRIV_OFF); + + VIR_FREE(tty); + +#ifdef __sun + /* + * The pty may come from either xend (with pygrub) or xenconsoled. + * It may have tty semantics set up, or not. While it isn't + * strictly necessary to have those semantics here, it is good to + * have a consistent state that is the same as under Linux. + * + * If tcgetattr fails, they have not been set up, so go ahead and + * set them up now, by pushing the ptem and ldterm streams modules. + */ + if (tcgetattr(ttyfd, &ttyattr) < 0) { + ioctl(ttyfd, I_PUSH, "ptem"); + ioctl(ttyfd, I_PUSH, "ldterm"); + tcgetattr(ttyfd, &ttyattr); + } + + cfmakeraw(&ttyattr); + tcsetattr(ttyfd, TCSANOW, &ttyattr); +#endif + + return ttyfd; +} + +static void +close_tty(int ttyfd) +{ + while (lockf(ttyfd, F_ULOCK, 0) && errno == EINTR) + ; + close(ttyfd); +} + +int +main(int argc, char *argv[]) +{ + struct termios ttyattr; + struct termios rawattr; void (*old_sigquit)(int); void (*old_sigterm)(int); void (*old_sigint)(int); void (*old_sighup)(int); void (*old_sigpipe)(int); - - - /* We do not want this to become the controlling TTY */ - if ((ttyfd = open(tty, O_NOCTTY | O_RDWR)) < 0) { - fprintf(stderr, _("unable to open tty %s: %s\n"), - tty, strerror(errno)); + int ret = EXIT_FAILURE; + int retrying = 0; + int ttyfd; + + setup_perms(); + + if (!setlocale(LC_ALL, "")) { + perror("setlocale"); return -1; } - /* Put STDIN into raw mode so that stuff typed - does not echo to the screen (the TTY reads will - result in it being echoed back already), and - also ensure Ctrl-C, etc is blocked, and misc - other bits */ - if (tcgetattr(STDIN_FILENO, &ttyattr) < 0) { - fprintf(stderr, _("unable to get tty attributes: %s\n"), - strerror(errno)); - goto closetty; - } - - rawattr = ttyattr; - cfmakeraw(&rawattr); - - if (tcsetattr(STDIN_FILENO, TCSAFLUSH, &rawattr) < 0) { - fprintf(stderr, _("unable to set tty attributes: %s\n"), - strerror(errno)); - goto closetty; - } - + if (!bindtextdomain(GETTEXT_PACKAGE, LOCALEBASEDIR)) { + perror("bindtextdomain"); + return -1; + } + + if (!textdomain(GETTEXT_PACKAGE)) { + perror("textdomain"); + return -1; + } + + parse_args(argc, argv); /* Trap all common signals so that we can safely restore the original terminal settings on STDIN before the @@ -107,8 +527,36 @@ int vshRunConsole(const char *tty) { old_sigint = signal(SIGINT, do_signal); old_sighup = signal(SIGHUP, do_signal); old_sigpipe = signal(SIGPIPE, do_signal); - got_signal = 0; - + +retry: + + ttyfd = open_tty(); + + if (!retrying && verbose) { + printf("Connected to domain %s\n", dom_name); + printf("Escape character is '^]'\n"); + } + + /* + * Put STDIN into raw mode so that stuff typed does not echo to the + * screen (the TTY reads will result in it being echoed back + * already), and also ensure Ctrl-C, etc is blocked, and misc other + * bits + */ + if (tcgetattr(STDIN_FILENO, &ttyattr) < 0) { + fprintf(stderr, _("unable to get tty attributes: %s\n"), + strerror(errno)); + exit(EXIT_FAILURE); + } + + rawattr = ttyattr; + cfmakeraw(&rawattr); + + if (tcsetattr(STDIN_FILENO, TCSAFLUSH, &rawattr) < 0) { + fprintf(stderr, _("unable to set tty attributes: %s\n"), + strerror(errno)); + exit(EXIT_FAILURE); + } /* Now lets process STDIN & tty forever.... */ for (; !got_signal ;) { @@ -127,7 +575,7 @@ int vshRunConsole(const char *tty) { if (errno == EINTR || errno == EAGAIN) continue; - fprintf(stderr, _("failure waiting for I/O: %s\n"), + fprintf(stderr, _("Failure waiting for I/O: %s\n"), strerror(errno)); goto cleanup; } @@ -142,15 +590,21 @@ int vshRunConsole(const char *tty) { int got, sent = 0, destfd; if ((got = read(fds[i].fd, buf, sizeof(buf))) < 0) { - fprintf(stderr, _("failure reading input: %s\n"), + fprintf(stderr, _("Failure reading input: %s\n"), strerror(errno)); goto cleanup; } - /* Quit if end of file, or we got the Ctrl-] key */ - if (!got || - (got == 1 && - buf[0] == CTRL_CLOSE_BRACKET)) + if (!got) { + tcsetattr(STDIN_FILENO, TCSAFLUSH, &ttyattr); + if (!check_for_reboot()) + goto done; + close_tty(ttyfd); + retrying = 1; + goto retry; + } + + if (got == 1 && buf[0] == CTRL_CLOSE_BRACKET) goto done; /* Data from stdin goes to the TTY, @@ -162,23 +616,31 @@ int vshRunConsole(const char *tty) { while (sent < got) { int done; - if ((done = safewrite(destfd, buf + sent, got - sent)) - <= 0) { - fprintf(stderr, _("failure writing output: %s\n"), + if ((done = safewrite(destfd, buf + sent, got - sent)) <= 0) { + fprintf(stderr, _("Failure writing output: %s\n"), strerror(errno)); goto cleanup; } sent += done; } + } else if (fds[i].revents & POLLHUP) { + tcsetattr(STDIN_FILENO, TCSAFLUSH, &ttyattr); + if (!check_for_reboot()) + goto done; + close_tty(ttyfd); + retrying = 1; + goto retry; } else { /* Any other flag from poll is an error condition */ goto cleanup; } } } - done: - ret = 0; - - cleanup: + +done: + ret = EXIT_SUCCESS; +cleanup: + tcsetattr(STDIN_FILENO, TCSANOW, &ttyattr); + close_tty(ttyfd); /* Restore original signal handlers */ signal(SIGQUIT, old_sigpipe); @@ -187,14 +649,18 @@ int vshRunConsole(const char *tty) { signal(SIGQUIT, old_sigterm); signal(SIGQUIT, old_sigquit); - /* Put STDIN back into the (sane?) state we found - it in before starting */ - tcsetattr(STDIN_FILENO, TCSAFLUSH, &ttyattr); - - closetty: - close(ttyfd); - - return ret; -} - -#endif /* !__MINGW32__ */ + if (verbose) + printf("\nConnection to domain %s closed.", dom_name); + printf("\n"); + + exit(ret); +} + +/* + * Local variables: + * indent-tabs-mode: nil + * c-indent-level: 4 + * c-basic-offset: 4 + * tab-width: 4 + * End: + */ diff --git a/src/console.h b/src/console.h deleted file mode 100644 --- a/src/console.h +++ /dev/null @@ -1,32 +0,0 @@ -/* - * console.c: A dumb serial console client - * - * Copyright (C) 2007 Red Hat, Inc. - * - * This library is free software; you can redistribute it and/or - * modify it under the terms of the GNU Lesser General Public - * License as published by the Free Software Foundation; either - * version 2.1 of the License, or (at your option) any later version. - * - * This library is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU - * Lesser General Public License for more details. - * - * You should have received a copy of the GNU Lesser General Public - * License along with this library; if not, write to the Free Software - * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA - * - * Daniel Berrange <berrange(a)redhat.com> - */ - -#ifndef __VIR_CONSOLE_H__ -#define __VIR_CONSOLE_H__ - -#ifndef __MINGW32__ - -int vshRunConsole(const char *tty); - -#endif /* !__MINGW32__ */ - -#endif /* __VIR_CONSOLE_H__ */ diff --git a/src/util.c b/src/util.c --- a/src/util.c +++ b/src/util.c @@ -240,7 +240,7 @@ __virExec(virConnectPtr conn, childout = *outfd; } #ifndef ENABLE_DEBUG - } else { + } else if (!(flags & VIR_EXEC_NO_PIPE)) { childout = null; #endif } @@ -271,7 +271,7 @@ __virExec(virConnectPtr conn, childerr = *errfd; } #ifndef ENABLE_DEBUG - } else { + } else if (!(flags & VIR_EXEC_NO_PIPE)) { childerr = null; #endif } @@ -652,6 +652,15 @@ virExec(virConnectPtr conn, return -1; } +int +virExecNoPipe(virConnectPtr conn, + char **argv ATTRIBUTE_UNUSED, + pid_t *retpid ATTRIBUTE_UNUSED) +{ + ReportError(conn, NULL, NULL, VIR_ERR_INTERNAL_ERROR, __FUNCTION__); + return -1; +} + #endif /* __MINGW32__ */ /* Like gnulib's fread_file, but read no more than the specified maximum diff --git a/src/util.h b/src/util.h --- a/src/util.h +++ b/src/util.h @@ -36,6 +36,7 @@ enum { VIR_EXEC_NONE = 0, VIR_EXEC_NONBLOCK = (1 << 0), VIR_EXEC_DAEMON = (1 << 1), + VIR_EXEC_NO_PIPE = (1 << 2) }; int virExec(virConnectPtr conn, diff --git a/src/virsh.c b/src/virsh.c --- a/src/virsh.c +++ b/src/virsh.c @@ -29,6 +29,7 @@ #include <assert.h> #include <errno.h> #include <sys/stat.h> +#include <sys/wait.h> #include <inttypes.h> #include <libxml/parser.h> @@ -42,7 +43,6 @@ #include "internal.h" #include "buf.h" -#include "console.h" #include "util.h" static char *progname; @@ -469,6 +469,7 @@ static const vshCmdInfo info_console[] = static const vshCmdOptDef opts_console[] = { {"domain", VSH_OT_DATA, VSH_OFLAG_REQ, gettext_noop("domain name, id or uuid")}, + {"verbose", VSH_OT_BOOL, 0, gettext_noop("verbose console")}, {NULL, 0, 0, NULL} }; @@ -477,49 +478,39 @@ static int static int cmdConsole(vshControl *ctl, const vshCmd *cmd) { - xmlDocPtr xml = NULL; - xmlXPathObjectPtr obj = NULL; - xmlXPathContextPtr ctxt = NULL; - virDomainPtr dom; + const char *argv[] = { BINDIR "/virt-console", NULL, NULL, NULL, NULL }; + int verbose = vshCommandOptBool(cmd, "verbose"); int ret = FALSE; - char *doc; - - if (!vshConnectionUsability(ctl, ctl->conn, TRUE)) - return FALSE; - - if (!(dom = vshCommandOptDomain(ctl, cmd, NULL))) - return FALSE; - - doc = virDomainGetXMLDesc(dom, 0); - if (!doc) - goto cleanup; - - xml = xmlReadDoc((const xmlChar *) doc, "domain.xml", NULL, - XML_PARSE_NOENT | XML_PARSE_NONET | - XML_PARSE_NOWARNING); - free(doc); - if (!xml) - goto cleanup; - ctxt = xmlXPathNewContext(xml); - if (!ctxt) - goto cleanup; - - obj = xmlXPathEval(BAD_CAST "string(/domain/devices/console/@tty)", ctxt); - if ((obj != NULL) && ((obj->type == XPATH_STRING) && - (obj->stringval != NULL) && (obj->stringval[0] != 0))) { - if (vshRunConsole((const char *)obj->stringval) == 0) - ret = TRUE; - } else { - vshPrintExtra(ctl, "%s", _("No console available for domain\n")); - } - xmlXPathFreeObject(obj); - - cleanup: - xmlXPathFreeContext(ctxt); - if (xml) - xmlFreeDoc(xml); - virDomainFree(dom); - return ret; + pid_t pid; + int argpos = 1; + + if (!vshConnectionUsability(ctl, ctl->conn, TRUE)) + return FALSE; + + if (verbose) + argv[argpos++] = "--verbose"; + + if (ctl->name != NULL) { + argv[argpos++] = "--connect"; + argv[argpos++] = ctl->name; + } + + if (!(argv[argpos] = vshCommandOptString(cmd, "domain", NULL))) { + vshError(ctl, FALSE, _("Undefined domain name or id.")); + return FALSE; + } + + ret = virExec(ctl->conn, argv, NULL, 0, &pid, STDIN_FILENO, NULL, + NULL, VIR_EXEC_NO_PIPE); + + if (ret == -1) { + vshError(ctl, FALSE, _("Couldn't execute virt-console.")); + return FALSE; + } + + waitpid(pid, NULL, 0); + + return TRUE; } #else /* __MINGW32__ */ @@ -4583,6 +4574,7 @@ cmdTTYConsole(vshControl *ctl, const vsh goto cleanup; } vshPrint(ctl, "%s\n", (const char *)obj->stringval); + ret = TRUE; cleanup: xmlXPathFreeObject(obj);

5 18

[libvirt] PATCH: Don't stop storage pools on daemon shutdown
by Daniel P. Berrange 15 Jan '09

15 Jan '09

Tearing a guest's storage out from under its feet on libvirtd shutdown is just as bad as tearing out its network :-) This patch removes the code which shuts down storage pool when the daemon shuts down. So NFS mounts stay around, LVM VGs remain active, and iSCSI connections remain logged in. When we then start up again, we happily detect that these resources are already running, and mark the pool as such Daniel diff --git a/src/storage_backend_iscsi.c b/src/storage_backend_iscsi.c --- a/src/storage_backend_iscsi.c +++ b/src/storage_backend_iscsi.c @@ -573,6 +573,7 @@ virStorageBackendISCSIStartPool(virConne virStoragePoolObjPtr pool) { char *portal = NULL; + char *session; if (pool->def->source.host.name == NULL) { virStorageReportError(conn, VIR_ERR_INTERNAL_ERROR, @@ -587,13 +588,17 @@ virStorageBackendISCSIStartPool(virConne return -1; } - if ((portal = virStorageBackendISCSIPortal(conn, pool)) == NULL) - return -1; - if (virStorageBackendISCSILogin(conn, pool, portal) < 0) { + if ((session = virStorageBackendISCSISession(conn, pool)) == NULL) { + if ((portal = virStorageBackendISCSIPortal(conn, pool)) == NULL) + return -1; + if (virStorageBackendISCSILogin(conn, pool, portal) < 0) { + VIR_FREE(portal); + return -1; + } VIR_FREE(portal); - return -1; + } else { + VIR_FREE(session); } - VIR_FREE(portal); return 0; } diff --git a/src/storage_driver.c b/src/storage_driver.c --- a/src/storage_driver.c +++ b/src/storage_driver.c @@ -241,32 +241,10 @@ storageDriverActive(void) { */ static int storageDriverShutdown(void) { - unsigned int i; - if (!driverState) return -1; storageDriverLock(driverState); - /* shutdown active pools */ - for (i = 0 ; i < driverState->pools.count ; i++) { - virStoragePoolObjPtr pool = driverState->pools.objs[i]; - - if (virStoragePoolObjIsActive(pool)) { - virStorageBackendPtr backend; - if ((backend = virStorageBackendForType(pool->def->type)) == NULL) { - storageLog("Missing backend"); - continue; - } - - if (backend->stopPool && - backend->stopPool(NULL, pool) < 0) { - virErrorPtr err = virGetLastError(); - storageLog("Failed to stop storage pool '%s': %s", - pool->def->name, err ? err->message : NULL); - } - virStoragePoolObjClearVols(pool); - } - } /* free inactive pools */ virStoragePoolObjListFree(&driverState->pools); -- |: Red Hat, Engineering, London -o- http://people.redhat.com/berrange/ :| |: http://libvirt.org -o- http://virt-manager.org -o- http://ovirt.org :| |: http://autobuild.org -o- http://search.cpan.org/~danberr/ :| |: GnuPG: 7D3B9505 -o- F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 :|

3 2

Re: [libvirt] Compiling on Windows with MinGW/MSYS
by Richard W.M. Jones 15 Jan '09

15 Jan '09

On Thu, Jan 15, 2009 at 01:23:38PM +0100, Brecht Sanders wrote: > I know I'm outside the envelope. But envelopes need to be pushed > sometimes in order to get things ported. I didn't mean to discourage you, just to note that you're likely the first to try this. > I'm not looking to run Xen on Windows at all. > However, I have implementations with Xen on Linux server where all the > other systems in the network run Windows. So I am looking for a Xen > manager that runs on Windows purely to manage Xen running on another > system. > > How would the remote driver work then exactly? Does it need some sort of > proxy on a Linux system? Right, so you want libvirtd running on the managed target system! This is exactly the scenario that we support on Windows. When you use a connection URI such as xen://somehost/, the connection is routed through the remote driver to the libvirtd daemon listening on the target ('somehost'). There is no need to configure the Xen driver in the Windows client libvirt for this, since everything involving Xen is done on the target. There's lots more to say about this, such as security, port numbers, transports etc. but it's better for you just to read the documentation here: http://libvirt.org/remote.html (Not all transports are supported on Windows -- in particular, ssh transport, and SASL probably don't work at the moment, although we will try to make both work at some point in the future). Rich. -- Richard Jones, Emerging Technologies, Red Hat http://et.redhat.com/~rjones virt-top is 'top' for virtual machines. Tiny program with many powerful monitoring features, net stats, disk stats, logging, etc. http://et.redhat.com/~rjones/virt-top

1 0

[libvirt] [PATCH] Avoid passing NULL to printf %s specifier
by john.levon＠sun.com 15 Jan '09

15 Jan '09

# HG changeset patch # User john.levon(a)sun.com # Date 1231985011 28800 # Node ID 2766ee91dd5ea8e99dac27ce730af0dc46a1d107 # Parent 082e0f7d5de236e69bea177c8d4c4204350144c1 Avoid passing NULL to printf %s specifier This is non-portable. Signed-off-by: John Levon <john.levon(a)sun.com> diff --git a/src/internal.h b/src/internal.h --- a/src/internal.h +++ b/src/internal.h @@ -115,6 +115,8 @@ #define ATTRIBUTE_RETURN_CHECK #endif /* __GNUC__ */ +#define NULLSTR(a) ((a) ? (a) : "(null)") + /** * TODO: * diff --git a/src/libvirt.c b/src/libvirt.c --- a/src/libvirt.c +++ b/src/libvirt.c @@ -874,10 +874,10 @@ do_open (const char *name, " port %d\n" " path %s\n", name, - ret->uri->scheme, ret->uri->opaque, - ret->uri->authority, ret->uri->server, - ret->uri->user, ret->uri->port, - ret->uri->path); + NULLSTR(ret->uri->scheme), NULLSTR(ret->uri->opaque), + NULLSTR(ret->uri->authority), NULLSTR(ret->uri->server), + NULLSTR(ret->uri->user), ret->uri->port, + NULLSTR(ret->uri->path)); } else { DEBUG0("no name, allowing driver auto-select"); }

4 3

[libvirt] Compiling on Windows with MinGW/MSYS
by Brecht Sanders 15 Jan '09

15 Jan '09

Hi, It's been a long time since I tried to compile libvirt on Windows using the MinGW/MSYS environment. I tried it again and I have good news and bad news. The good news is thet when I disable support for most things (--without-xen --without-qemu --without-openvz --without-uml --without-test --without-libvirtd --with-remote-pid-file=none --with-init-scripts=none --with-depends --without-sasl --without-polkit) it now builds fine and even the DLL was built (libvirt-0.dll). The bad news is that when I enable Xen support, it now depends on the xenstore library. Unfortunately I wasn't able to build this library on Windows. Is this dependancy necessary? Does anyone know of xenstore on Windows? Regards Brecht Sanders

2 1

[libvirt] [ANN] sVirt v0.30
by James Morris 15 Jan '09

15 Jan '09

This is to announce the release of v0.30 of sVirt, a project to add security labeling support to Linux-based virtualization. Project page: http://www.selinuxproject.org/page/SVirt Download: http://namei.org/svirt/svirt-0.30.tar.bz2 This release is an update in response to feedback received on the v0.20 release, per the discussion thread at: https://www.redhat.com/archives/libvir-list/2008-December/msg00260.html Also see the TODO page: http://selinuxproject.org/page/SVirt/TODO Change log for v0.30: --------------------- * Fix SELinux build configuration * Simplify qemudNodeGetSecurityModel * Add "test" security driver * Fix security driver probe logic * Ensure VIR_CONNECT_RO checked where appropriate * Move security model/doi to last fields in virsh dominfo * Change virDomainSecLabel -> virSecurityLabel * similar change to API calls * Change virDomainSecModel -> virSecurityModel * similar change to API calls * general renaming then of seclabel to security * Use CHECK_LIB/HEADER to detect libselinux (and fix build system in general) * Rename virDomainGetSecModel to virNodeGetSecurityModel * Change -2 error returns to -1 * Use remoteDispatchOOMError() for OOM errors * Create virXPathStringLimit() from virDomainSecLabelDefParseXMLString() Comments welcome. - James -- James Morris <jmorris(a)namei.org>

5 9

[libvirt] [PATCH] Improve xend_get error message
by john.levon＠sun.com 15 Jan '09

15 Jan '09

# HG changeset patch # User john.levon(a)sun.com # Date 1231982481 28800 # Node ID 082e0f7d5de236e69bea177c8d4c4204350144c1 # Parent 1a16d4b76232845091c1f800569ec462860cb1fb Improve xend_get error message Signed-off-by: John Levon <john.levon(a)sun.com> diff --git a/src/xend_internal.c b/src/xend_internal.c --- a/src/xend_internal.c +++ b/src/xend_internal.c @@ -394,7 +394,8 @@ xend_get(virConnectPtr xend, const char if (((ret < 0) || (ret >= 300)) && ((ret != 404) || (!STRPREFIX(path, "/xend/domain/")))) { virXendError(xend, VIR_ERR_GET_FAILED, - _("xend_get: error from xen daemon: %s"), content); + _("%d status from xen daemon: %s:%s"), + ret, path, content); } return ret;

2 1

[libvirt] [PATCH] Invoke PortableXDR's rpcgen if available
by Richard W.M. Jones 15 Jan '09

15 Jan '09

We discovered over the new year that the XDR code in glibc, which was derived from some original Sun code, doesn't have an unambiguously free license[1]. Furthermore the rpcgen in glibc is unmaintained and produces such bad code that we have to run a Perl script over it in order to correct some egregious errors. Sun/glibc rpcgen code also contains all sorts of strange micro-optimizations that were probably good back in 1988, but don't make so much sense now. Also rpcgen was implemented directly in C, so we took this chance to use flex/bison for more accurate and faster parsing. For these reasons we are rewriting the XDR code and the rpcgen tool with a free license[2]. I'm pleased to say that as a result of this effort we've rewritten rpcgen already: http://git.et.redhat.com/?p=portablexdr.git;a=summary To avoid conflicts, the new rpcgen will be installed as 'portable-rpcgen'. The attached patch changes libvirt's configure so that if it finds 'portable-rpcgen' on the path, it uses it in preference to 'rpcgen'. Also attached are the updated remote_protocol.[ch] files (ie. the files which are generated from remote_protocol.x by portable-rpcgen). I'm not suggesting we use these quite yet, because (a) they still generate some warnings, and (b) they need a lot more testing. However I think the patch is safe enough. Rich. [1] For more about this, see this message and the reply: https://www.redhat.com/archives/fedora-legal-list/2008-December/msg00022.ht… [2] LGPLv2+ for the library parts and GPLv2+ for the rpcgen program. -- Richard Jones, Emerging Technologies, Red Hat http://et.redhat.com/~rjones virt-p2v converts physical machines to virtual machines. Boot with a live CD or over the network (PXE) and turn machines into Xen guests. http://et.redhat.com/~rjones/virt-p2v

2 2

[libvirt] xenUnifiedDomainLookupByName error handling
by John Levon 15 Jan '09

15 Jan '09

We do: 700 /* Not found. */ 701 xenUnifiedError (conn, VIR_ERR_NO_DOMAIN, __FUNCTION__); But: # virsh start domu-defefe 12:09:30.581: error : Domain not found: xenUnifiedDomainLookupByName libvir: Xen error : Domain not found: xenUnifiedDomainLookupByName libvir: Xen error : Domain not found: xenUnifiedDomainLookupByName error: failed to get domain 'domu-defefe' It doesn't seem right to be reporting errors like this. What's the intent? Most of xenUnifiedError() seems to be internal errors, but many of them are just clients doing the wrong thing... regards john

2 2

[libvirt] [PATCH] Fix activeDomainList handling
by john.levon＠sun.com 14 Jan '09

14 Jan '09

# HG changeset patch # User john.levon(a)sun.com # Date 1231958879 28800 # Node ID 1fca4b0dd4c6ab3c0030b4986de6fe5cb7c94631 # Parent 097f5b4497d7ecc5cce14d10559244313d26ce46 Fix activeDomainList handling It must be NULLed out after free()ing. Signed-off-by: John Levon <john.levon(a)sun.com> diff --git a/src/xs_internal.c b/src/xs_internal.c --- a/src/xs_internal.c +++ b/src/xs_internal.c @@ -388,8 +388,10 @@ xenStoreClose(virConnectPtr conn) } xenStoreWatchListFree(priv->xsWatchList); + priv->xsWatchList = NULL; #ifndef PROXY xenUnifiedDomainInfoListFree(activeDomainList); + activeDomainList = NULL; #endif if (priv->xshandle == NULL) return(-1);

4 5

[libvirt] [PATCH] Fix xs_unwatch() to correctly pass the token
by john.levon＠sun.com 14 Jan '09

14 Jan '09

# HG changeset patch # User john.levon(a)sun.com # Date 1231958879 28800 # Node ID 672cdb87a32fb0b275bb7cdc9abb9c2d506ec1b0 # Parent 46dc909bda5b20d11f08d1516f21c05f27b6648f Fix xs_unwatch() to correctly pass the token Signed-off-by: John Levon <john.levon(a)sun.com> diff --git a/src/xs_internal.c b/src/xs_internal.c --- a/src/xs_internal.c +++ b/src/xs_internal.c @@ -1106,7 +1106,7 @@ int xenStoreRemoveWatch(virConnectPtr co if (!xs_unwatch(priv->xshandle, list->watches[i]->path, - list->watches[i]->path)) + list->watches[i]->token)) { DEBUG0("WARNING: Could not remove watch"); /* Not fatal, continue */

2 1

[libvirt] [PATCH] Fix ref-counting for Xen driver event registration
by john.levon＠sun.com 14 Jan '09

14 Jan '09

# HG changeset patch # User john.levon(a)sun.com # Date 1231958879 28800 # Node ID 46dc909bda5b20d11f08d1516f21c05f27b6648f # Parent 1fca4b0dd4c6ab3c0030b4986de6fe5cb7c94631 Fix ref-counting for Xen driver event registration Signed-off-by: John Levon <john.levon(a)sun.com> diff --git a/src/xen_unified.c b/src/xen_unified.c --- a/src/xen_unified.c +++ b/src/xen_unified.c @@ -1359,15 +1359,21 @@ xenUnifiedDomainEventRegister (virConnec void *opaque, void (*freefunc)(void *)) { + int ret; + GET_PRIVATE (conn); if (priv->xsWatch == -1) { xenUnifiedError (conn, VIR_ERR_NO_SUPPORT, __FUNCTION__); return -1; } - conn->refs++; - return virDomainEventCallbackListAdd(conn, priv->domainEventCallbacks, - callback, opaque, freefunc); + ret = virDomainEventCallbackListAdd(conn, priv->domainEventCallbacks, + callback, opaque, freefunc); + + if (ret == 0) + conn->refs++; + + return (ret); } static int @@ -1382,8 +1388,10 @@ xenUnifiedDomainEventDeregister (virConn } ret = virDomainEventCallbackListRemove(conn, priv->domainEventCallbacks, - callback); - virUnrefConnect(conn); + callback); + + if (ret == 0) + virUnrefConnect(conn); return ret; }

3 2

[libvirt] libvirt-qpid and migrate issue
by Ian Main 14 Jan '09

14 Jan '09

Howdy all! So in working with libvirt-qpid, I've come to realize that it imposes a new difficulty in doing migrations. Traditionally you would connect out to each individual libvirt instance from a central location and be able to pass the destination connection pointer to the virDomainMigrate function. However in libvirt-qpids case, since it's running on each physical host, it does not necessarily have access to the destination host due to authentication requirements. So I'd like to discuss possible ways of resolving this issue. I'm not sure how much support libvirt proper wants to extend to allowing this case to work, but I'm hoping we can figure something out that will make everyone happy. I see that in the implementation of virDomainMigrate, it is already split up into prepare and preform functions. I'm thinking that, if agreeable by the libvirt team, it'd be nice to have an API which allowed us to call into destination and source nodes separately. Something like: char * virConnectMigratePrepare(virConnectPtr conn, unsigned long flags, const char *dname); which would set up the destination node to receive a migrating domain and return an xml document containing the information required to perform the migration (host name, port, maybe even capabilities and/or encryption keys in the future etc.), coupled with: int virDomainMigratePerform(virDomainPtr dom, const char *prepare_description, unsigned long bandwidth); which would initiate the transfer of the domain to the destination host. Or something along those lines.. this is just off the top of my head but I wanted to get the discussion going. Note I'm using the same internal names prepare/perform but we can rename them if appropriate. Anyway this would allow us to hook these up in qpid where we have access to both systems and call them in sequence. Let me know what you think. Thanks, Ian

1 0

[libvirt] [PATCH] Implement capabilities for Solaris
by john.levon＠sun.com 14 Jan '09

14 Jan '09

# HG changeset patch # User John Levon <john.levon(a)sun.com> # Date 1231940906 28800 # Node ID a0d98d39955f4f304d318c7c780742ab929eb351 # Parent ddfcba6b4181ab433dce345d66dc399a81c92a3b Implement capabilities for Solaris Use Solaris interfaces to derive the hypervisor capabilities. Signed-off-by: John Levon <john.levon(a)sun.com> diff --git a/src/xen_internal.c b/src/xen_internal.c --- a/src/xen_internal.c +++ b/src/xen_internal.c @@ -35,6 +35,10 @@ #ifdef HAVE_XEN_SYS_PRIVCMD_H #include <xen/sys/privcmd.h> #endif +#endif + +#ifdef __sun +#include <sys/systeminfo.h> #endif /* required for shutdown flags */ @@ -662,11 +666,8 @@ typedef struct xen_op_v2_dom xen_op_v2_d #ifdef __linux__ #define XEN_HYPERVISOR_SOCKET "/proc/xen/privcmd" #define HYPERVISOR_CAPABILITIES "/sys/hypervisor/properties/capabilities" -#define CPUINFO "/proc/cpuinfo" #elif defined(__sun) #define XEN_HYPERVISOR_SOCKET "/dev/xen/privcmd" -#define HYPERVISOR_CAPABILITIES "" -#define CPUINFO "/dev/cpu/self/cpuid" #else #error "unsupported platform" #endif @@ -2123,7 +2124,7 @@ xenHypervisorBuildCapabilities(virConnec xenHypervisorBuildCapabilities(virConnectPtr conn, const char *hostmachine, int host_pae, - char *hvm_type, + const char *hvm_type, struct guest_arch *guest_archs, int nr_guest_archs) { virCapsPtr caps; @@ -2227,8 +2228,126 @@ xenHypervisorBuildCapabilities(virConnec return NULL; } -/** - * xenHypervisorGetCapabilities: +#ifdef __sun + +static int +get_cpu_flags(virConnectPtr conn, const char **hvm, int *pae, int *longmode) +{ + struct { + uint32_t r_eax, r_ebx, r_ecx, r_edx; + } regs; + + char tmpbuf[20]; + int fd = -1; + int ret = 0; + + /* returns -1, errno 22 if in 32-bit mode */ + *longmode = (sysinfo(SI_ARCHITECTURE_64, tmpbuf, 20) != -1); + + if ((fd = open("/dev/cpu/self/cpuid", O_RDONLY)) == -1 || + pread(fd, &regs, sizeof(regs), 0) != sizeof(regs)) { + virXenError(conn, VIR_ERR_SYSTEM_ERROR, + "couldn't read CPU flags: %s", strerror(errno)); + goto out; + } + + *pae = 0; + *hvm = ""; + + if (strncmp((const char *)&regs.r_ebx, "AuthcAMDenti", 12) == 0) { + if (pread(fd, &regs, sizeof (regs), 0x80000001) == sizeof (regs)) { + /* Read secure virtual machine bit (bit 2 of ECX feature ID) */ + if ((regs.r_ecx >> 2) & 1) { + *hvm = "svm"; + } + if ((regs.r_edx >> 6) & 1) + *pae = 1; + } + } else if (strncmp((const char *)&regs.r_ebx, "GenuntelineI", 12) == 0) { + if (pread(fd, &regs, sizeof (regs), 0x00000001) == sizeof (regs)) { + /* Read VMXE feature bit (bit 5 of ECX feature ID) */ + if ((regs.r_ecx >> 5) & 1) + *hvm = "vmx"; + if ((regs.r_edx >> 6) & 1) + *pae = 1; + } + } + + ret = 1; + +out: + if (fd != -1) + close(fd); + return ret; +} + +static virCapsPtr +xenHypervisorMakeCapabilitiesSunOS(virConnectPtr conn) +{ + struct guest_arch guest_arches[32]; + int i = 0; + virCapsPtr caps = NULL; + struct utsname utsname; + int pae, longmode; + const char *hvm; + + if (!get_cpu_flags(conn, &hvm, &pae, &longmode)) + return NULL; + + /* Really, this never fails - look at the man-page. */ + uname (&utsname); + + guest_arches[i].model = "i686"; + guest_arches[i].bits = 32; + guest_arches[i].hvm = 0; + guest_arches[i].pae = pae; + guest_arches[i].nonpae = 1; + guest_arches[i].ia64_be = 0; + i++; + + if (longmode) { + guest_arches[i].model = "x86_64"; + guest_arches[i].bits = 64; + guest_arches[i].hvm = 0; + guest_arches[i].pae = 0; + guest_arches[i].nonpae = 1; + guest_arches[i].ia64_be = 0; + i++; + } + + if (hvm[0] != '\0') { + guest_arches[i].model = "i686"; + guest_arches[i].bits = 32; + guest_arches[i].hvm = 1; + guest_arches[i].pae = pae; + guest_arches[i].nonpae = 1; + guest_arches[i].ia64_be = 0; + i++; + + if (longmode) { + guest_arches[i].model = "x86_64"; + guest_arches[i].bits = 64; + guest_arches[i].hvm = 1; + guest_arches[i].pae = 0; + guest_arches[i].nonpae = 1; + guest_arches[i].ia64_be = 0; + i++; + } + } + + if ((caps = xenHypervisorBuildCapabilities(conn, + utsname.machine, + pae, hvm, + guest_arches, i)) == NULL) + virXenError(NULL, VIR_ERR_NO_MEMORY, NULL); + + return caps; +} + +#endif /* __sun */ + +/** + * xenHypervisorMakeCapabilitiesInternal: * @conn: pointer to the connection block * @cpuinfo: file handle containing /proc/cpuinfo data, or NULL * @capabilities: file handle containing /sys/hypervisor/properties/capabilities data, or NULL @@ -2394,6 +2513,9 @@ virCapsPtr virCapsPtr xenHypervisorMakeCapabilities(virConnectPtr conn) { +#ifdef __sun + return xenHypervisorMakeCapabilitiesSunOS(conn); +#else virCapsPtr caps; FILE *cpuinfo, *capabilities; struct utsname utsname; @@ -2432,6 +2554,7 @@ xenHypervisorMakeCapabilities(virConnect fclose(capabilities); return caps; +#endif /* __sun */ }

4 3

[libvirt] [PATCH] Remove references to non-existent files
by john.levon＠sun.com 14 Jan '09

14 Jan '09

# HG changeset patch # User John Levon <john.levon(a)sun.com> # Date 1231946129 28800 # Node ID 9d8302140e58e20c7fb22b3242a52b06467b08e6 # Parent b614f5780261464e8f10e48306e67d43d56931b0 Remove references to non-existent files Some docs files don't exist. Signed-off-by: John Levon <john.levon(a)sun.com> diff --git a/docs/Makefile.am b/docs/Makefile.am --- a/docs/Makefile.am +++ b/docs/Makefile.am @@ -130,9 +130,8 @@ rebuild: api all install-data-local: $(mkinstalldirs) $(DESTDIR)$(HTML_DIR) - -@INSTALL@ -m 0644 $(srcdir)/libvir.html $(srcdir)/FAQ.html \ - $(srcdir)/redhat.gif $(srcdir)/Libxml2-Logo-90x34.gif \ - $(DESTDIR)$(HTML_DIR) + -@INSTALL@ -m 0644 $(srcdir)/FAQ.html \ + $(srcdir)/Libxml2-Logo-90x34.gif $(DESTDIR)$(HTML_DIR) $(mkinstalldirs) $(DESTDIR)$(HTML_DIR)/html for h in $(apihtml); do \ $(INSTALL) -m 0644 $(srcdir)/$$h $(DESTDIR)$(HTML_DIR)/html; done diff --git a/docs/examples/Makefile.am b/docs/examples/Makefile.am --- a/docs/examples/Makefile.am +++ b/docs/examples/Makefile.am @@ -20,7 +20,7 @@ install-data-local: install-data-local: $(mkinstalldirs) $(DESTDIR)$(HTML_DIR) -@INSTALL@ -m 0644 $(srcdir)/*.html $(srcdir)/*.c $(srcdir)/*.xml \ - $(srcdir)/*.xsl $(srcdir)/*.res $(DESTDIR)$(HTML_DIR) + $(srcdir)/*.xsl $(DESTDIR)$(HTML_DIR) EXTRA_DIST=examples.xsl index.py examples.xml diff --git a/docs/examples/index.py b/docs/examples/index.py --- a/docs/examples/index.py +++ b/docs/examples/index.py @@ -218,8 +218,6 @@ def dump_Makefile(): def dump_Makefile(): for file in glob.glob('*.xml'): extras.append(file) - for file in glob.glob('*.res'): - extras.append(file) Makefile="""# -*- buffer-read-only: t -*- vi: set ro: # Beware this is autogenerated by index.py SUBDIRS=python

2 1

[libvirt] [PATCH] Separate objdir build fixes
by john.levon＠sun.com 14 Jan '09

14 Jan '09

# HG changeset patch # User John Levon <john.levon(a)sun.com> # Date 1231946129 28800 # Node ID b614f5780261464e8f10e48306e67d43d56931b0 # Parent 53c621a65055f4752abef748c0ad15cbb1d8013d Separate objdir build fixes Let devhelp build in a separate objdir. Signed-off-by: John Levon <john.levon(a)sun.com> diff --git a/docs/devhelp/Makefile.am b/docs/devhelp/Makefile.am --- a/docs/devhelp/Makefile.am +++ b/docs/devhelp/Makefile.am @@ -1,7 +1,8 @@ DEVHELP_DIR=$(datadir)/gtk-doc/html/libv DEVHELP_DIR=$(datadir)/gtk-doc/html/libvirt HTML_FILES=index.html general.html $(HTML_MODULES) HTML_MODULES= \ - libvirt-libvirt.html + libvirt-libvirt.html \ + libvirt-virterror.html EXTRA_FORMAT= \ home.png \ @@ -17,11 +18,12 @@ libvirt.devhelp $(HTML_FILES): $(srcdir) libvirt.devhelp $(HTML_FILES): $(srcdir)/devhelp.xsl html.xsl $(top_srcdir)/docs/libvirt-api.xml -@(echo Rebuilding devhelp files) -@(if [ -x $(XSLTPROC) ] ; then \ - $(XSLTPROC) --nonet -o $(srcdir)/libvirt.devhelp $(srcdir)/devhelp.xsl $(top_srcdir)/docs/libvirt-api.xml ; fi ); + $(XSLTPROC) --nonet -o libvirt.devhelp $(srcdir)/devhelp.xsl $(top_srcdir)/docs/libvirt-api.xml ; fi ); install-data-local: $(mkinstalldirs) $(DESTDIR)$(DEVHELP_DIR) -@INSTALL@ -m 0644 libvirt.devhelp $(DESTDIR)$(DEVHELP_DIR) - -@INSTALL@ -m 0644 $(EXTRA_FORMAT) $(DESTDIR)$(DEVHELP_DIR) -@INSTALL@ -m 0644 $(HTML_FILES) $(DESTDIR)$(DEVHELP_DIR) - + -for file in $(EXTRA_FORMAT); do \ + @INSTALL@ -m 0644 $(srcdir)/$${file} $(DESTDIR)$(DEVHELP_DIR) ; \ + done

2 1

[libvirt] [PATCH] Fix devhelp build
by john.levon＠sun.com 14 Jan '09

14 Jan '09

# HG changeset patch # User John Levon <john.levon(a)sun.com> # Date 1231946129 28800 # Node ID c08132809fb6e060b1301e97954a1b0ababb384c # Parent 52dcd17a9fd4bf4a6171386f083d235eebd5aa23 Fix devhelp build For devhelp to actually be created during a build, we need to explicitly depend on the generated files. Signed-off-by: John Levon <john.levon(a)sun.com> diff --git a/docs/devhelp/Makefile.am b/docs/devhelp/Makefile.am --- a/docs/devhelp/Makefile.am +++ b/docs/devhelp/Makefile.am @@ -20,7 +20,7 @@ libvirt.devhelp $(HTML_FILES): $(srcdir) -@(if [ -x $(XSLTPROC) ] ; then \ $(XSLTPROC) --nonet -o libvirt.devhelp $(srcdir)/devhelp.xsl $(top_srcdir)/docs/libvirt-api.xml ; fi ); -install-data-local: +install-data-local: libvirt.devhelp $(HTML_FILES) $(mkinstalldirs) $(DESTDIR)$(DEVHELP_DIR) -@INSTALL@ -m 0644 libvirt.devhelp $(DESTDIR)$(DEVHELP_DIR) -@INSTALL@ -m 0644 $(HTML_FILES) $(DESTDIR)$(DEVHELP_DIR)

2 1

[libvirt] [PATCH] Add copy-on-write image support
by Miloslav Trmač 14 Jan '09

14 Jan '09

Hello, the attached patch adds support for copy-on-write images: creating them, deleting them, identifying them. Mirek

3 5

[libvirt] [PATCH] Mark Xen PV CD-ROM devices as such
by john.levon＠sun.com 14 Jan '09

14 Jan '09

# HG changeset patch # User John Levon <john.levon(a)sun.com> # Date 1231946129 28800 # Node ID 53c621a65055f4752abef748c0ad15cbb1d8013d # Parent bde7017c447cf36451194e64179b60e1c8d9f039 Mark Xen PV CD-ROM devices as such Add a ':cdrom' marker for PV as well as HVM devices. Signed-off-by: John Levon <john.levon(a)sun.com> diff --git a/src/xend_internal.c b/src/xend_internal.c --- a/src/xend_internal.c +++ b/src/xend_internal.c @@ -5025,6 +5025,8 @@ xenDaemonFormatSxprDisk(virConnectPtr co virBufferVSprintf(buf, "(dev '%s:%s')", def->dst, def->device == VIR_DOMAIN_DISK_DEVICE_CDROM ? "cdrom" : "disk"); + } else if (def->device == VIR_DOMAIN_DISK_DEVICE_CDROM) { + virBufferVSprintf(buf, "(dev '%s:cdrom')", def->dst); } else { virBufferVSprintf(buf, "(dev '%s')", def->dst); }

2 1

[libvirt] [PATCH] Fix oversized stack allocation
by john.levon＠sun.com 14 Jan '09

14 Jan '09

# HG changeset patch # User John Levon <john.levon(a)sun.com> # Date 1231946129 28800 # Node ID c7538cb7fab4e6f3aa6714fdd4126dcf2f0371b4 # Parent cfe04caba5de61620fa6d1c2bff85cfd62255c1d Fix oversized stack allocation REMOTE_MESSAGE_MAX is a whopping 262Kb and shouldn't be allocated on the stack. Signed-off-by: Ryan Scott <ryan.scott(a)sun.com> diff --git a/src/remote_internal.c b/src/remote_internal.c --- a/src/remote_internal.c +++ b/src/remote_internal.c @@ -4797,10 +4797,11 @@ call (virConnectPtr conn, struct private xdrproc_t args_filter, char *args, xdrproc_t ret_filter, char *ret) { - char buffer[REMOTE_MESSAGE_MAX]; + char *buffer = NULL; char buffer2[4]; struct remote_message_header hdr; XDR xdr; + int err = -1; int len; struct remote_error rerror; @@ -4814,18 +4815,23 @@ call (virConnectPtr conn, struct private hdr.serial = serial; hdr.status = REMOTE_OK; + if ((buffer = malloc(REMOTE_MESSAGE_MAX)) == NULL) { + error (conn, VIR_ERR_SYSTEM_ERROR, strerror (errno)); + goto out; + } + /* Serialise header followed by args. */ - xdrmem_create (&xdr, buffer, sizeof buffer, XDR_ENCODE); + xdrmem_create (&xdr, buffer, REMOTE_MESSAGE_MAX, XDR_ENCODE); if (!xdr_remote_message_header (&xdr, &hdr)) { error (flags & REMOTE_CALL_IN_OPEN ? NULL : conn, VIR_ERR_RPC, _("xdr_remote_message_header failed")); - return -1; + goto out; } if (!(*args_filter) (&xdr, args)) { error (flags & REMOTE_CALL_IN_OPEN ? NULL : conn, VIR_ERR_RPC, _("marshalling args")); - return -1; + goto out; } /* Get the length stored in buffer. */ @@ -4842,25 +4848,25 @@ call (virConnectPtr conn, struct private if (!xdr_int (&xdr, &len)) { error (flags & REMOTE_CALL_IN_OPEN ? NULL : conn, VIR_ERR_RPC, _("xdr_int (length word)")); - return -1; + goto out; } xdr_destroy (&xdr); /* Send length word followed by header+args. */ if (really_write (conn, priv, flags & REMOTE_CALL_IN_OPEN, buffer2, sizeof buffer2) == -1 || really_write (conn, priv, flags & REMOTE_CALL_IN_OPEN, buffer, len-4) == -1) - return -1; + goto out; retry_read: /* Read and deserialise length word. */ if (really_read (conn, priv, flags & REMOTE_CALL_IN_OPEN, buffer2, sizeof buffer2) == -1) - return -1; + goto out; xdrmem_create (&xdr, buffer2, sizeof buffer2, XDR_DECODE); if (!xdr_int (&xdr, &len)) { error (flags & REMOTE_CALL_IN_OPEN ? NULL : conn, VIR_ERR_RPC, _("xdr_int (length word, reply)")); - return -1; + goto out; } xdr_destroy (&xdr); @@ -4870,20 +4876,25 @@ retry_read: if (len < 0 || len > REMOTE_MESSAGE_MAX) { error (flags & REMOTE_CALL_IN_OPEN ? NULL : conn, VIR_ERR_RPC, _("packet received from server too large")); - return -1; } /* Read reply header and what follows (either a ret or an error). */ if (really_read (conn, priv, flags & REMOTE_CALL_IN_OPEN, buffer, len) == -1) - return -1; + goto out; /* Deserialise reply header. */ xdrmem_create (&xdr, buffer, len, XDR_DECODE); if (!xdr_remote_message_header (&xdr, &hdr)) { error (flags & REMOTE_CALL_IN_OPEN ? NULL : conn, VIR_ERR_RPC, _("invalid header in reply")); - return -1; - } + goto out; + } + + /* + * Although the previous lines were the last references to buffer in this + * function, we can't free it yet, as the xdr struct still has some + * pointers to it. + */ /* Check program, version, etc. are what we expect. */ if (hdr.prog != REMOTE_PROGRAM) { @@ -4892,7 +4903,7 @@ retry_read: VIR_ERR_RPC, VIR_ERR_ERROR, NULL, NULL, NULL, 0, 0, _("unknown program (received %x, expected %x)"), hdr.prog, REMOTE_PROGRAM); - return -1; + goto out; } if (hdr.vers != REMOTE_PROTOCOL_VERSION) { virRaiseError (flags & REMOTE_CALL_IN_OPEN ? NULL : conn, @@ -4900,7 +4911,7 @@ retry_read: VIR_ERR_RPC, VIR_ERR_ERROR, NULL, NULL, NULL, 0, 0, _("unknown protocol version (received %x, expected %x)"), hdr.vers, REMOTE_PROTOCOL_VERSION); - return -1; + goto out; } if (hdr.proc == REMOTE_PROC_DOMAIN_EVENT && @@ -4923,7 +4934,7 @@ retry_read: VIR_ERR_RPC, VIR_ERR_ERROR, NULL, NULL, NULL, 0, 0, _("unknown procedure (received %x, expected %x)"), hdr.proc, proc_nr); - return -1; + goto out; } if (hdr.direction != REMOTE_REPLY) { virRaiseError (flags & REMOTE_CALL_IN_OPEN ? NULL : conn, @@ -4931,14 +4942,14 @@ retry_read: VIR_ERR_RPC, VIR_ERR_ERROR, NULL, NULL, NULL, 0, 0, _("unknown direction (received %x, expected %x)"), hdr.direction, REMOTE_REPLY); - return -1; + goto out; } if (hdr.serial != serial) { virRaiseError (flags & REMOTE_CALL_IN_OPEN ? NULL : conn, NULL, NULL, VIR_FROM_REMOTE, VIR_ERR_RPC, VIR_ERR_ERROR, NULL, NULL, NULL, 0, 0, _("unknown serial (received %x, expected %x)"), hdr.serial, serial); - return -1; + goto out; } /* Status is either REMOTE_OK (meaning that what follows is a ret @@ -4950,17 +4961,18 @@ retry_read: if (!(*ret_filter) (&xdr, ret)) { error (flags & REMOTE_CALL_IN_OPEN ? NULL : conn, VIR_ERR_RPC, _("unmarshalling ret")); - return -1; + goto out; } xdr_destroy (&xdr); - return 0; + err = 0; + goto out; case REMOTE_ERROR: memset (&rerror, 0, sizeof rerror); if (!xdr_remote_error (&xdr, &rerror)) { error (flags & REMOTE_CALL_IN_OPEN ? NULL : conn, VIR_ERR_RPC, _("unmarshalling remote_error")); - return -1; + goto out; } xdr_destroy (&xdr); /* See if caller asked us to keep quiet about missing RPCs @@ -4970,11 +4982,12 @@ retry_read: rerror.code == VIR_ERR_RPC && rerror.level == VIR_ERR_ERROR && STRPREFIX(*rerror.message, "unknown procedure")) { - return -2; + err = -2; + goto out; } server_error (flags & REMOTE_CALL_IN_OPEN ? NULL : conn, &rerror); xdr_free ((xdrproc_t) xdr_remote_error, (char *) &rerror); - return -1; + goto out; default: virRaiseError (flags & REMOTE_CALL_IN_OPEN ? NULL : conn, NULL, NULL, VIR_FROM_REMOTE, @@ -4982,8 +4995,12 @@ retry_read: _("unknown status (received %x)"), hdr.status); xdr_destroy (&xdr); - return -1; - } + goto out; + } + +out: + VIR_FREE (buffer); + return err; } static int @@ -5551,7 +5568,7 @@ remoteDomainEventFired(int watch, int event, void *opaque) { - char buffer[REMOTE_MESSAGE_MAX]; + char *buffer = NULL; char buffer2[4]; struct remote_message_header hdr; XDR xdr; @@ -5566,22 +5583,22 @@ remoteDomainEventFired(int watch, DEBUG("%s : VIR_EVENT_HANDLE_HANGUP or " "VIR_EVENT_HANDLE_ERROR encountered", __FUNCTION__); virEventRemoveHandle(watch); - return; + goto out; } if (fd != priv->sock) { virEventRemoveHandle(watch); - return; + goto out; } /* Read and deserialise length word. */ if (really_read (conn, priv, 0, buffer2, sizeof buffer2) == -1) - return; + goto out; xdrmem_create (&xdr, buffer2, sizeof buffer2, XDR_DECODE); if (!xdr_int (&xdr, &len)) { error (conn, VIR_ERR_RPC, _("xdr_int (length word, reply)")); - return; + goto out; } xdr_destroy (&xdr); @@ -5590,20 +5607,25 @@ remoteDomainEventFired(int watch, if (len < 0 || len > REMOTE_MESSAGE_MAX) { error (conn, VIR_ERR_RPC, _("packet received from server too large")); - return; + goto out; + } + + if ((buffer = malloc(REMOTE_MESSAGE_MAX)) == NULL) { + error (conn, VIR_ERR_SYSTEM_ERROR, strerror (errno)); + goto out; } /* Read reply header and what follows (either a ret or an error). */ if (really_read (conn, priv, 0, buffer, len) == -1) { error (conn, VIR_ERR_RPC, _("error reading buffer from memory")); - return; + goto out; } /* Deserialise reply header. */ xdrmem_create (&xdr, buffer, len, XDR_DECODE); if (!xdr_remote_message_header (&xdr, &hdr)) { error (conn, VIR_ERR_RPC, _("invalid header in event firing")); - return; + goto out; } if (hdr.proc == REMOTE_PROC_DOMAIN_EVENT && @@ -5614,6 +5636,9 @@ remoteDomainEventFired(int watch, DEBUG0("invalid proc in event firing"); error (conn, VIR_ERR_RPC, _("invalid proc in event firing")); } + +out: + VIR_FREE (buffer); } void

2 1

[libvirt] [PATCH] Improve compiler flag checking
by john.levon＠sun.com 14 Jan '09

14 Jan '09

# HG changeset patch # User john.levon(a)sun.com # Date 1229789393 28800 # Node ID a158bbc6df6fb8b15af875d08e73cfd3ca75f907 # Parent 34706075e6c801cd9532490c11b20b64dfcf464b Improve compiler flag checking Some compilers (including GCC) don't set the return value consistently if an erroneous option is passed on the command line. Account for that. Signed-off-by: John Levon <john.levon(a)sun.com> diff --git a/acinclude.m4 b/acinclude.m4 --- a/acinclude.m4 +++ b/acinclude.m4 @@ -45,21 +45,9 @@ AC_DEFUN([LIBVIRT_COMPILE_WARNINGS],[ ;; esac - compiler_flags= + COMPILER_FLAGS= for option in $try_compiler_flags; do - SAVE_CFLAGS="$CFLAGS" - CFLAGS="$CFLAGS $option" - AC_MSG_CHECKING([whether gcc understands $option]) - AC_TRY_LINK([], [], - has_option=yes, - has_option=no,) - CFLAGS="$SAVE_CFLAGS" - AC_MSG_RESULT($has_option) - if test $has_option = yes; then - compiler_flags="$compiler_flags $option" - fi - unset has_option - unset SAVE_CFLAGS + gl_COMPILER_FLAGS($option) done unset option unset try_compiler_flags @@ -85,7 +73,7 @@ AC_DEFUN([LIBVIRT_COMPILE_WARNINGS],[ fi AC_MSG_RESULT($complCFLAGS) - WARN_CFLAGS="$compiler_flags $complCFLAGS" + WARN_CFLAGS="$COMPILER_FLAGS $complCFLAGS" AC_SUBST(WARN_CFLAGS) dnl Needed to keep compile quiet on python 2.4 diff --git a/m4/compiler-flags.m4 b/m4/compiler-flags.m4 --- a/m4/compiler-flags.m4 +++ b/m4/compiler-flags.m4 @@ -25,10 +25,16 @@ AC_DEFUN([gl_COMPILER_FLAGS], AC_SUBST(COMPILER_FLAGS) ac_save_CFLAGS="$CFLAGS" CFLAGS="$CFLAGS $1" - AC_TRY_COMPILE(, - [int x;], - COMPILER_FLAGS="$COMPILER_FLAGS $1" - AC_MSG_RESULT(yes), - AC_MSG_RESULT(no)) - CFLAGS="$ac_save_CFLAGS" + AC_TRY_LINK([], [], has_option=yes, has_option=no,) + echo 'int x;' >conftest.c + $CC $CFLAGS -c conftest.c 2>conftest.err + ret=$? + if test $ret != 0 -o -s conftest.err -o $has_option = "no"; then + AC_MSG_RESULT(no) + else + AC_MSG_RESULT(yes) + COMPILER_FLAGS="$COMPILER_FLAGS $1" + fi + CFLAGS="$ac_save_CFLAGS" + rm -f conftest* ])

2 1

[libvirt] libvirtd shutdown script patches
by Sir Woody Hackswell 14 Jan '09

14 Jan '09

Here is a small patch for libvirtd init and sysconfig script. Before killing libvirtd, we virsh shutdown any running domains. We also have a maximum time limit for shutdown (300 sec default), just in case the VM will not shut down. Any thoughts as to whether this is the right place to put this functionality? Is the shutdown script proper, or since libvirtd handles autostarts domains internally, should libvirtd shut down active domains upon SIGTERM all by itself? -Richard --- /etc/init.d/libvirtd.orig 2009-01-13 19:31:47.000000000 -0500 +++ /etc/init.d/libvirtd 2009-01-13 19:53:25.000000000 -0500 @@ -47,6 +47,11 @@ LIBVIRTD_CONFIG_ARGS="--config $LIBVIRTD_CONFIG" fi +if [ -z "$LIBVIRTD_SHUTDOWN_TIMEOUT" ] +then + LIBVIRTD_SHUTDOWN_TIMEOUT=300 +fi + RETVAL=0 start() { @@ -62,6 +67,15 @@ stop() { echo -n $"Stopping $SERVICE daemon: " + starttime=`date '+%s'` + for VM in $(virsh list | grep running | gawk '{print $2}'); do virsh -q shutdown $VM; done + while ( virsh list | grep -qc running ); do + sleep 5; + nowtime=`date '+%s'` + elapsed=$(($nowtime-$starttime)) + [[ $elapsed -ge $LIBVIRTD_SHUTDOWN_TIMEOUT ]] && break + done + killproc $PROCESS RETVAL=$? echo --- /etc/sysconfig/libvirtd.orig 2009-01-13 19:56:09.000000000 -0500 +++ /etc/sysconfig/libvirtd 2009-01-13 19:41:27.000000000 -0500 @@ -8,4 +8,7 @@ # Override Kerberos service keytab for SASL/GSSAPI #KRB5_KTNAME=/etc/libvirt/krb5.tab +# Shutdown timeout in seconds - Default is 300 seconds (or 5 minutes) +#LIBVIRTD_SHUTDOWN_TIMEOUT=300 -- -Sir Woody Hackswell

3 3

[libvirt] QEMU link on libvirt.org needs updated
by Michael Schenck 13 Jan '09

13 Jan '09

Maintainer of libvirt.org, Currently the QEMU link under "libvirt supports" on http://libvirt.org/index.html points to http://fabrice.bellard.free.fr/qemu/, which should be updated to http://bellard.org/qemu/ Best Regards, Michael Schenck -- Michael Schenck - www.tek-ops.com Karma police, arrest this man He talks in maths He buzzes like a fridge He's like a detuned radio

2 1

[libvirt] proposal: allow use of "bool": HACKING: discuss C types
by Jim Meyering 13 Jan '09

13 Jan '09

Hello, I don't want to make waves, but I do care about certain aspects of code quality, so... I propose to allow (encourage, even) the use of the standard C99 type, bool, in libvirt, but not in public interfaces[1]. There are already uses in cgroup.c and xmlrpc.c, as well as those in the gnulib "c-ctype.h" header which is included from many of libvirt's .c files. The motivation is to make the code as readable as possible. When you see the declaration of an "int" variable, member, or function, that type gives you no clue whether it is used as a boolean. If you see a few uses that treat it as boolean, that's still no guarantee, and it may be non-trivial to ensure that there isn't some non-boolean value with a special meaning. However, if you see a "bool" variable, you do have that guarantee. This is one reason to using the tighter, more descriptive type. In addition, if ever you need to go back to using "int", it's trivial to automate the substitution, e.g., -Dint=bool or sed s/int/bool/. On the other hand, it is very hard to automatically identify "int" variables that may safely be converted to "bool". Portability is no problem whatsoever, thanks to gnulib's stdbool module, which ensures that there is a working <stdbool.h> header in the compiler's include path. libvirt even includes/runs a test to ensure that the <stdbool.h> in use (from gnulib or from the system) works properly. Of course, if you agree on principle, we should update HACKING to reflect the policy, so I've taken a shot at that. But if we start talking about using the right type for boolean variables, we can't avoid the more general question of scalars, signedness, etc. And once you mention scalars, you have to mention pointers and const-correctness, so it got longer still. Patch below. Jim [1]: We can use "bool" reliably in libvirt-internal code, due to the fact that it is guaranteed to be compiled with a usable <stdbool.h> header. We cannot make such a guarantee for all applications that compile code including libvirt's public headers, <libvirt.h> and <virterror.h>. P.S. I confess I have a vested interested. I wrote a patch recently that declares a local to be of type "bool", and would like to ensure that policy permits doing that. >From 3c6a9bc0fd043ce9f84e5aaba3cfbc1db0359eb7 Mon Sep 17 00:00:00 2001 From: Jim Meyering <meyering(a)redhat.com> Date: Fri, 12 Dec 2008 09:45:31 +0100 Subject: [PATCH] HACKING: use the right type --- HACKING | 46 ++++++++++++++++++++++++++++++++++++++++++++++ 1 files changed, 46 insertions(+), 0 deletions(-) diff --git a/HACKING b/HACKING index 3945833..511c8cc 100644 --- a/HACKING +++ b/HACKING @@ -91,6 +91,52 @@ Usually they're in macro definitions or strings, and should be converted anyhow. +C types +======= +Use the right type. + +Scalars +------- +If you're using "int" or "long", odds are good that there's a better type. +If a variable is counting something, be sure to declare it with an +unsigned type. +If it's memory-size-related, use size_t (use ssize_t only if required). +If it's file-size related, use uintmax_t, or maybe off_t. +If it's file-offset related (i.e., signed), use off_t. +If it's just counting small numbers use "unsigned int"; +(on all but oddball embedded systems, you can assume that that +type is at least four bytes wide). +If a variable has boolean semantics, give it the "bool" type +and use the corresponding "true" and "false" macros. It's ok +to include <stdbool.h>, since libvirt's use of gnulib ensures +that it exists and is usable. +In the unusual event that you require a specific width, use a +standard type like int32_t, uint32_t, uint64_t, etc. + +Of course, take all of the above with a grain of salt. If you're about +to use some system interface that requires a type like int or ssize_t, +use types for your own variables that match. + +Also, if you try to use e.g., "unsigned int" as a type, and that +conflicts with the signedness of a related variable, sometimes +it's best just to use the *wrong* type, if "pulling the thread" +and fixing all related variables would be too invasive. + +Finally, while using descriptive types is important, be careful not to +go overboard. If whatever you're doing causes warnings, or requires +casts, then reconsider or ask for help. + +Pointers +-------- +Ensure that all of your pointers are "const-correct". +Unless a pointer is used to modify the pointed-to storage, +give it the "const" attribute. That way, the reader knows +up-front that this is a read-only pointer. Perhaps more +importantly, if we're diligent about this, when you see a non-const +pointer, you're guaranteed that it is used to modify the storage +it points to, or it is aliased to another pointer that is. + + Low level memory management =========================== -- 1.6.1.rc2.299.gead4c.dirty

3 8

[libvirt] Bridge interface and script
by John Levon 12 Jan '09

12 Jan '09

With current libvirt, it's not possible to specify a script for VIR_DOMAIN_NET_TYPE_BRIDGE. Is this just an oversight, or intentional? We use the bridge type with Solaris virtual NICs, and need to specify the correct script to use. Suggestions on how to fix this? Should I just add script handling to this net type? thanks john

2 4

[libvirt] [PATCH] Fix remote_protocol header for Solaris
by john.levon＠sun.com 12 Jan '09

12 Jan '09

# HG changeset patch # User john.levon(a)sun.com # Date 1229367889 28800 # Node ID 46a487775d4b35c32455a5ca8eb3009d10a02c5b # Parent a1f24e9f8a1fbf191c71172ae986110b7e98fc32 Fix remote_protocol header for Solaris quad_t is not a portable type, but rather than force rpcgen every build, we just patch in the fixes needed. Signed-off-by: John Levon <john.levon(a)sun.com> diff --git a/qemud/Makefile.am b/qemud/Makefile.am --- a/qemud/Makefile.am +++ b/qemud/Makefile.am @@ -47,6 +47,7 @@ if GLIBC_RPCGEN chmod 444 $@-t mv $@-t $@ endif + patch < rpcgen-solaris.diff endif remote_protocol.c: remote_protocol.h diff --git a/qemud/rpcgen-solaris.diff b/qemud/rpcgen-solaris.diff new file mode 100644 --- /dev/null +++ b/qemud/rpcgen-solaris.diff @@ -0,0 +1,19 @@ +diff ./remote_protocol.h ./remote_protocol.h +--- ./remote_protocol.h ++++ ./remote_protocol.h +@@ -11,6 +11,15 @@ + + #ifdef __cplusplus + extern "C" { ++#endif ++ ++#if defined(__sun) ++#include <rpc/xdr.h> ++#define u_quad_t uint64_t ++#define quad_t int64_t ++#define xdr_u_quad_t xdr_uint64_t ++#define xdr_quad_t xdr_int64_t ++#define IXDR_GET_LONG IXDR_GET_INT32 + #endif + + #include <config.h>

5 9

[libvirt] [PATCH] Fix invocation of rpcgen
by Richard W.M. Jones 12 Jan '09

12 Jan '09

Currently if rpcgen != glibc's rpcgen, then it gets a bit confused with the temporary files. Also it's best to call the rpcgen that configure detected and not some random rpcgen that happens to be in the path. The attached patch fixes this. Rich. -- Richard Jones, Emerging Technologies, Red Hat http://et.redhat.com/~rjones virt-p2v converts physical machines to virtual machines. Boot with a live CD or over the network (PXE) and turn machines into Xen guests. http://et.redhat.com/~rjones/virt-p2v

2 1

[libvirt] [PATCH] Avoid GCC extensions
by john.levon＠sun.com 12 Jan '09

12 Jan '09

# HG changeset patch # User john.levon(a)sun.com # Date 1229367890 28800 # Node ID 6a8e82d7d2e166880fed8d7ad860a3e2e93d62be # Parent c324c231c6a50be9f970f0f6c6d1629a7c09ab3b Avoid GCC extensions Anonymous unions are not portable, nor are zero-sizes structures. Signed-off-by: John Levon <john.levon(a)sun.com> diff --git a/src/domain_conf.c b/src/domain_conf.c --- a/src/domain_conf.c +++ b/src/domain_conf.c @@ -1437,7 +1437,7 @@ virDomainHostdevSubsysUsbDefParseXML(vir if (vendor) { if (virStrToLong_ui(vendor, NULL, 0, - &def->source.subsys.usb.vendor) < 0) { + &def->source.subsys.u.usb.vendor) < 0) { virDomainReportError(conn, VIR_ERR_INTERNAL_ERROR, _("cannot parse vendor id %s"), vendor); VIR_FREE(vendor); @@ -1454,7 +1454,7 @@ virDomainHostdevSubsysUsbDefParseXML(vir if (product) { if (virStrToLong_ui(product, NULL, 0, - &def->source.subsys.usb.product) < 0) { + &def->source.subsys.u.usb.product) < 0) { virDomainReportError(conn, VIR_ERR_INTERNAL_ERROR, _("cannot parse product %s"), product); VIR_FREE(product); @@ -1472,7 +1472,7 @@ virDomainHostdevSubsysUsbDefParseXML(vir bus = virXMLPropString(cur, "bus"); if (bus) { if (virStrToLong_ui(bus, NULL, 0, - &def->source.subsys.usb.bus) < 0) { + &def->source.subsys.u.usb.bus) < 0) { virDomainReportError(conn, VIR_ERR_INTERNAL_ERROR, _("cannot parse bus %s"), bus); VIR_FREE(bus); @@ -1488,7 +1488,7 @@ virDomainHostdevSubsysUsbDefParseXML(vir device = virXMLPropString(cur, "device"); if (device) { if (virStrToLong_ui(device, NULL, 0, - &def->source.subsys.usb.device) < 0) { + &def->source.subsys.u.usb.device) < 0) { virDomainReportError(conn, VIR_ERR_INTERNAL_ERROR, _("cannot parse device %s"), device); @@ -1510,14 +1510,14 @@ virDomainHostdevSubsysUsbDefParseXML(vir cur = cur->next; } - if (def->source.subsys.usb.vendor == 0 && - def->source.subsys.usb.product != 0) { + if (def->source.subsys.u.usb.vendor == 0 && + def->source.subsys.u.usb.product != 0) { virDomainReportError(conn, VIR_ERR_INTERNAL_ERROR, "%s", _("missing vendor")); goto out; } - if (def->source.subsys.usb.vendor != 0 && - def->source.subsys.usb.product == 0) { + if (def->source.subsys.u.usb.vendor != 0 && + def->source.subsys.u.usb.product == 0) { virDomainReportError(conn, VIR_ERR_INTERNAL_ERROR, "%s", _("missing product")); goto out; @@ -2951,15 +2951,15 @@ virDomainHostdevDefFormat(virConnectPtr virBufferVSprintf(buf, " <hostdev mode='%s' type='%s'>\n", mode, type); virBufferAddLit(buf, " <source>\n"); - if (def->source.subsys.usb.vendor) { + if (def->source.subsys.u.usb.vendor) { virBufferVSprintf(buf, " <vendor id='0x%.4x'/>\n", - def->source.subsys.usb.vendor); + def->source.subsys.u.usb.vendor); virBufferVSprintf(buf, " <product id='0x%.4x'/>\n", - def->source.subsys.usb.product); + def->source.subsys.u.usb.product); } else { virBufferVSprintf(buf, " <address bus='%d' device='%d'/>\n", - def->source.subsys.usb.bus, - def->source.subsys.usb.device); + def->source.subsys.u.usb.bus, + def->source.subsys.u.usb.device); } virBufferAddLit(buf, " </source>\n"); diff --git a/src/domain_conf.h b/src/domain_conf.h --- a/src/domain_conf.h +++ b/src/domain_conf.h @@ -307,12 +307,13 @@ struct _virDomainHostdevDef { unsigned slot; unsigned function; } pci; - }; + } u; } subsys; struct { /* TBD: struct capabilities see: * https://www.redhat.com/archives/libvir-list/2008-July/msg00429.html */ + int dummy; } caps; } source; char* target; diff --git a/src/qemu_conf.c b/src/qemu_conf.c --- a/src/qemu_conf.c +++ b/src/qemu_conf.c @@ -1264,15 +1264,15 @@ int qemudBuildCommandLine(virConnectPtr if (hostdev->mode == VIR_DOMAIN_HOSTDEV_MODE_SUBSYS && hostdev->source.subsys.type == VIR_DOMAIN_HOSTDEV_SUBSYS_TYPE_USB) { - if(hostdev->source.subsys.usb.vendor) { + if(hostdev->source.subsys.u.usb.vendor) { ret = asprintf(&usbdev, "host:%.4x:%.4x", - hostdev->source.subsys.usb.vendor, - hostdev->source.subsys.usb.product); + hostdev->source.subsys.u.usb.vendor, + hostdev->source.subsys.u.usb.product); } else { ret = asprintf(&usbdev, "host:%.3d.%.3d", - hostdev->source.subsys.usb.bus, - hostdev->source.subsys.usb.device); + hostdev->source.subsys.u.usb.bus, + hostdev->source.subsys.u.usb.device); } if (ret < 0) { usbdev = NULL; diff --git a/src/qemu_driver.c b/src/qemu_driver.c --- a/src/qemu_driver.c +++ b/src/qemu_driver.c @@ -2790,14 +2790,14 @@ static int qemudDomainAttachHostDevice(v return -1; } - if (dev->data.hostdev->source.subsys.usb.vendor) { + if (dev->data.hostdev->source.subsys.u.usb.vendor) { ret = asprintf(&cmd, "usb_add host:%.4x:%.4x", - dev->data.hostdev->source.subsys.usb.vendor, - dev->data.hostdev->source.subsys.usb.product); + dev->data.hostdev->source.subsys.u.usb.vendor, + dev->data.hostdev->source.subsys.u.usb.product); } else { ret = asprintf(&cmd, "usb_add host:%.3d.%.3d", - dev->data.hostdev->source.subsys.usb.bus, - dev->data.hostdev->source.subsys.usb.device); + dev->data.hostdev->source.subsys.u.usb.bus, + dev->data.hostdev->source.subsys.u.usb.device); } if (ret == -1) { qemudReportError(dom->conn, NULL, NULL, VIR_ERR_NO_MEMORY, NULL); diff --git a/src/remote_internal.c b/src/remote_internal.c --- a/src/remote_internal.c +++ b/src/remote_internal.c @@ -649,7 +649,7 @@ doRemoteOpen (virConnectPtr conn, if (username) nr_args += 2; /* For -l username */ if (no_tty) nr_args += 5; /* For -T -o BatchMode=yes -e none */ - command = command ? : strdup ("ssh"); + command = command ? command : strdup ("ssh"); if (command == NULL) goto out_of_memory;

4 4

[libvirt] [PATCH] fix non-srcdir build failure
by Jim Meyering 12 Jan '09

12 Jan '09

"make distcheck" was failing. This fixes it and adds some quotes. >From 8b6f9ff168dd0ce7318f1947a7766a4b223acdd4 Mon Sep 17 00:00:00 2001 From: Jim Meyering <meyering(a)redhat.com> Date: Thu, 8 Jan 2009 20:53:30 +0100 Subject: [PATCH] fix non-srcdir build failure * qemud/Makefile.am (check-local): Prefix use of test_libvirtd.aug with $(srcdir)/. Add quotes around $(AUGPARSE), in case it expands to something unusual. --- qemud/Makefile.am | 3 ++- 1 files changed, 2 insertions(+), 1 deletions(-) diff --git a/qemud/Makefile.am b/qemud/Makefile.am index 8983416..ad4ccf1 100644 --- a/qemud/Makefile.am +++ b/qemud/Makefile.am @@ -239,7 +239,8 @@ libvirtd.init: libvirtd.init.in mv $@-t $@ check-local: - if [ -x $(AUGPARSE) ]; then $(AUGPARSE) -I $(srcdir) test_libvirtd.aug ; fi + test -x '$(AUGPARSE)' \ + && '$(AUGPARSE)' -I $(srcdir) $(srcdir)/test_libvirtd.aug || : else -- 1.6.1.141.gfe98e

2 2

[libvirt] [PATCH 5/5] read saved vm status on libvirtd startup
by Guido Günther 12 Jan '09

12 Jan '09

The earlier patches didn't change libvirtd behaviour (except for kvm/qemu not being teared down on an unexpected daemon crash), all vms were still being shutoff at daemon shutdown. This patch now adds the code to read back the vm status after on daemon startup and reconnects all running vms found in stateDir which are left over from a daemon restart or crash and also disables shutting down of VMs when libvirt quits. --- src/qemu_driver.c | 165 ++++++++++++++++++++++++++++++++++++++++++++++++++-- 1 files changed, 158 insertions(+), 7 deletions(-) diff --git a/src/qemu_driver.c b/src/qemu_driver.c index d8b87e4..57a396c 100644 --- a/src/qemu_driver.c +++ b/src/qemu_driver.c @@ -183,6 +183,46 @@ qemudAutostartConfigs(struct qemud_driver *driver) { } +static int +qemudGetProcFD(pid_t pid, int fdnum) +{ + int fd; + +#ifdef __linux__ + char *path = NULL; + + if (!asprintf(&path, "/proc/%d/fd/%d", pid, fdnum)) { + qemudReportError(NULL, NULL, NULL, VIR_ERR_NO_MEMORY, NULL); + return -1; + } + + if((fd = open(path, O_RDONLY)) < 0) { + qemudReportError(NULL, NULL, NULL, VIR_ERR_INTERNAL_ERROR, + _("Unable to open %s"), path); + return -1; + } + if (qemudSetCloseExec(fd) < 0) { + qemudReportError(NULL, NULL, NULL, VIR_ERR_INTERNAL_ERROR, + _("Unable to set close-on-exec flag on %s"), path); + goto error; + } + + if (qemudSetNonBlock(fd) < 0) { + qemudReportError(NULL, NULL, NULL, VIR_ERR_INTERNAL_ERROR, + _("Unable to put %s into non-blocking mode"), path); + goto error; + } + + VIR_FREE(path); + return fd; +error: + VIR_FREE(path); + close(fd); +#endif + return -1; +} + + /** * qemudRemoveDomainStatus * @@ -220,6 +260,104 @@ cleanup: } +static int qemudOpenMonitor(virConnectPtr conn, + 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 < qemu_driver->domains.count ; i++) { + virDomainObjPtr vm = qemu_driver->domains.objs[i]; + qemudDomainStatusPtr status = NULL; + char *config = NULL; + int rc; + + virDomainObjLock(vm); + /* Read pid */ + 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, 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; + + vm->stdin_fd = qemudGetProcFD(vm->pid, 0); + vm->stdout_fd = qemudGetProcFD(vm->pid, 1); + vm->stderr_fd = qemudGetProcFD(vm->pid, 2); + if (vm->stdin_fd == -1 || vm->stdout_fd == -1 || vm->stderr_fd == -1) + goto next_error; + + 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)) { + + qemudLog(QEMUD_ERR, _("Failed to reconnect to stdout/stderr\n")); + goto next_error; + } + + 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: * @@ -316,6 +454,7 @@ qemudStartup(void) { qemu_driver->autostartDir, NULL, NULL) < 0) goto error; + qemudReconnectVMs(qemu_driver); qemudAutostartConfigs(qemu_driver); qemuDriverUnlock(qemu_driver); @@ -418,11 +557,14 @@ qemudShutdown(void) { /* shutdown active VMs */ for (i = 0 ; i < qemu_driver->domains.count ; i++) { + /* FIXME: don't shutdown VMs on daemon shutdown for now */ +#if 0 virDomainObjPtr dom = qemu_driver->domains.objs[i]; virDomainObjLock(dom); if (virDomainIsActive(dom)) qemudShutdownVMDaemon(NULL, qemu_driver, dom); virDomainObjUnlock(dom); +#endif } virDomainObjListFree(&qemu_driver->domains); @@ -543,7 +685,8 @@ qemudCheckMonitorPrompt(virConnectPtr conn ATTRIBUTE_UNUSED, static int qemudOpenMonitor(virConnectPtr conn, virDomainObjPtr vm, - const char *monitor) { + const char *monitor, + int reconnect) { int monfd; char buf[1024]; int ret = -1; @@ -564,11 +707,19 @@ 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, @@ -669,7 +820,7 @@ qemudFindCharDevicePTYs(virConnectPtr conn, } /* Got them all, so now open the monitor console */ - ret = qemudOpenMonitor(conn, vm, monitor); + ret = qemudOpenMonitor(conn, vm, monitor, 0); cleanup: VIR_FREE(monitor); -- 1.6.0.2

4 13

[libvirt] [PATCH] PCI passthrough docs, tests and minor change
by Daniel Veillard 12 Jan '09

12 Jan '09

The following add the documentation for the PCI passthrough, extends the RNG and add an extra test for the QEmu parsing to args and serialization back to XML. I also noticed that we parse (and save) and extra PCI domain argument, but it's not actually used when calling qemu, so I assume it's a missing feature for QEmu and just decided to patch the code to not save the extra value when not defined (i.e. 0). Whether it should just be removed at this point is IMHO a separate decision, which is why I took the way with minimal changes. Note that the domain attribute is not documented nor part or the RNG either at this point. 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/

3 7

[libvirt] text console install
by Michael Kress 12 Jan '09

12 Jan '09

Hi, I'm not able to get a text console install of any distribution (CentOS-5.2 for example). I'm using the following command: virt-install -n myname -f myimage.img -s 10 -c /iso/CentOS-5.2-x86_64-bin-DVD.iso --accelerate -r 1024 --hvm --nographics The vnc version of this command works flawlessly, but I'd like this host to be a headless one. The installation seems to start as I can see the instance with state "running" with the following command: virsh -c qemu:///system list But all I see is the following messages: ======================== Starting install... Creating domain... Connecting to uri: qemu:///system ======================== After a while there's one strange character showing up in the console. Pressing Ctrl-Alt Gr-] I can exit the screen and there's a message: ======================== Domain installation still in progress. You can reconnect to the console to complete the installation process. ======================== BTW, I got the following line in /etc/libvirt/qemu.conf vnc_listen="192.168.2.3" But also tried without that line. How can I get the text console to show up? Thanks Michael -- Michael Kress, kress(a)hal.saar.de http://www.michael-kress.de / http://kress.net P E N G U I N S A R E C O O L

2 1

[libvirt] Question, new libvirt driver - OpenNebula.
by "Abel Míguez Rodríguez" 12 Jan '09

12 Jan '09

Hello all, We have updated the libvirt implementation for OpenNebula to libvirt- 0.5.1 release, made use of the internal XML handling API and the existing infrastructure for error, data structures, etc... Now we are facing the problem of needing several libraries to compile the driver, the OpenNebula API (liboneapi) and the xmlrpc library that the API makes use of. The problem is that the xmlrpc library itself depends on: libxmlrpc_client++ libxmlrpc_client libxmlrpc++ libxmlrpc libxmlrpc_util libxmlrpc_xmlparse libxmlrpc_xmltok We would like to link just against less libraries to make the driver compilation clear and simple, any idea? Cheers Abel

2 1

[libvirt] interface.script not launched (KVM/qemu)
by Daniel Schwager 12 Jan '09

12 Jan '09

Hi, I use kvm and libvirt for running windows on a fedora-core9. Problem: I defined a interface script which should run, if the KVM/qemu starts, but libvirt does not tell kvm to use the script: Part of my domain.xml: <interface type="bridge"> <source bridge="br-guest"/> <script path='bridge-kvm.sh'/> <model type='e1000'/> </interface> Part of the logfile of qemu in /var/log/libvirt/qemu LC_ALL=C PATH=/sbin:/usr/sbin:/bin:/usr/bin / /usr/bin/qemu-kvm \ -S -M pc -m 250 -smp 1 -name winxp100 -monitor pty -boot c \ -drive file=/tmp/winxp100,if=ide,index=0,boot=on \ -net nic,macaddr=52:54:00:79:d5:c9,vlan=0,model=e1000 \ -net tap,fd=14,script=,vlan=0,ifname=vnet0 \ -serial pty -parallel none -usb -usbdevice tablet -vnc 0.0.0.0:0 There, you can see the empty script directive "...,script=,...." root@kvm ~]$ rpm -qa | grep libvirt libvirt-0.5.1-2.fc9.x86_64 How can I fix this problem ? regards Danny

2 1

[libvirt] [PATCH] split out logfile opening
by Guido Günther 11 Jan '09

11 Jan '09

Hi, attached patch splits out the qemu logfile opening into a separate function which makes the code a bit more readable and I'll need this for the libvirtd restart code. Cheers, -- Guido

3 10

[libvirt] Libvirt - Xen Enterprise - Koan
by Gerhardus.Geldenhuis＠gta-travel.com 09 Jan '09

09 Jan '09

Hi I hope me "replying" this way does not break anyone's mail sorting, I only now subscribed to the list. > Thanks for this - this sounds like the most likely issue. The problem is > that Citrix seems to move things around somewhat as that config file > does not exist. There is a xensource directory that contains the > following but other than that i am a bit stuck. They might have overriden the config file to point somewhere else. Try and find the file "XendOptions.py" and see where it loads the config from. Or see if anything sets $XEND_CONFIG env variable at startup. In regular upstream xen code, it XendOptions.py has: class XendOptionsFile(XendOptions): """Default path to the config file.""" config_default = "/etc/xen/xend-config.sxp" I did a find . -name *.py > temp.gg and then 141 cat temp.gg | xargs grep -i xendoptions 142 cat temp.gg | xargs grep -i xend 144 cat temp.gg | xargs grep -i XEND_CONFIG 148 cat temp.gg | xargs grep -i config_default All of these searches gave me absolutely no results. I am sure that the config files are in : /etc/xensource/ xapi.conf xapi-ssl.conf the contents of xapi.conf is: schema_filename = /etc/xensource/db_schema.sql license_filename = /etc/xensource/license http-port = 80 use-ssl = true ssl-cert = /etc/ssl/certs/xensource-rio.pem ssl-port = 443 # dom0-mem = 128 and of xapi-ssl.conf is: pid = /var/run/xapi.ssl.pid socket = r:TCP_NODELAY=1 socket = a:TCP_NODELAY=1 socket = l:TCP_NODELAY=1 [xapi] accept = 443 connect = 80 cert = /etc/xensource/xapi-ssl.pem I am thinking that maybe only ssl connections are allowed ... which is why virsh can't connect? we are running "XenServer release 4.1.0-7843p (xenenterprise)" and I am trying to get spacewalk to get the necessary domain details when registering a xen host. ______________________________________________________________________ This email has been scanned by the MessageLabs Email Security System. For more information please visit http://www.messagelabs.com/email ______________________________________________________________________

9 21

[libvirt] [PATCH] Fix non-debug compile
by john.levon＠sun.com 09 Jan '09

09 Jan '09

# HG changeset patch # User john.levon(a)sun.com # Date 1230005985 28800 # Node ID 60090fd4e447795742265241b39f2759d9530514 # Parent d7c3bd669a72371127852aefb3ba2cc0443da8da Fix non-debug compile Remove duplicate defines, and stub out virLogMessage() for non-debug builds. Signed-off-by: John Levon <john.levon(a)sun.com> diff --git a/src/logging.h b/src/logging.h --- a/src/logging.h +++ b/src/logging.h @@ -47,12 +47,6 @@ do { } while (0) #define VIR_ERROR(category, f, l, fmt,...) \ do { } while (0) -#define VIR_INFO(category, fmt,...) \ - do { } while (0) -#define VIR_WARN(category, fmt,...) \ - do { } while (0) -#define VIR_ERROR(category, fmt,...) \ - do { } while (0) #endif /* !ENABLE_DEBUG */ #define DEBUG(fmt,...) \ @@ -82,6 +76,8 @@ typedef enum { VIR_LOG_WARN, VIR_LOG_ERROR, } virLogPriority; + +#ifdef ENABLE_DEBUG /** * virLogOutputFunc: @@ -114,10 +110,6 @@ extern int virLogDefineOutput(virLogOutp extern int virLogDefineOutput(virLogOutputFunc f, virLogCloseFunc c, void *data, int priority, int flags); -#if 0 -extern char *virLogGetDump(int flags); -#endif - /* * Internal logging API */ @@ -127,8 +119,23 @@ extern void virLogShutdown(void); extern void virLogShutdown(void); extern int virLogParseFilters(const char *filters); extern int virLogParseOutputs(const char *output); + extern void virLogMessage(const char *category, int priority, const char *funcname, long long linenr, int flags, const char *fmt, ...) ATTRIBUTE_FORMAT(printf, 6, 7); +#else /* ENABLE_DEBUG */ + +#define virLogSetDefaultPriority(p) +#define virLogDefineFilter(m, p, f) +#define virLogDefineOutput(func, c, d, p, f) +#define virLogStartup() +#define virLogReset() +#define virLogShutdown() +#define virLogParseFilters(f) +#define virLogParseOutputs(o) +#define virLogMessage(c, p, func, l, f, fmt, __VA_ARGS__) + +#endif /* ENABLE_DEBUG */ + #endif

2 1

[libvirt] [PATCH] Fix Solaris compile after gnulib refresh
by john.levon＠sun.com 09 Jan '09

09 Jan '09

# HG changeset patch # User john.levon(a)sun.com # Date 1231444798 28800 # Node ID bf47245a1fecebe59ef30371a4465d713d37af06 # Parent b0dbefc87cb8b0194ad2bd518c2e8261b6bb0a6c Fix Solaris compile after gnulib refresh As documented in NEWS, we now need to explicitly specify usage of libsocket. Signed-off-by: John Levon <john.levon(a)sun.com> diff --git a/configure.in b/configure.in --- a/configure.in +++ b/configure.in @@ -79,9 +79,9 @@ AC_CHECK_HEADERS([pwd.h paths.h regex.h dnl Where are the XDR functions? dnl If portablexdr is installed, prefer that. -dnl Otherwise try -lrpc (Cygwin) -lxdr (some MinGW) or none (most Unix) +dnl Otherwise try -lrpc (Cygwin) -lxdr (some MinGW), nsl (Solaris) or none (most Unix) AC_CHECK_LIB([portablexdr],[xdrmem_create],[],[ - AC_SEARCH_LIBS([xdrmem_create],[rpc xdr]) + AC_SEARCH_LIBS([xdrmem_create],[rpc xdr nsl]) ]) AC_CHECK_LIB([intl],[gettext],[]) diff --git a/qemud/Makefile.am b/qemud/Makefile.am --- a/qemud/Makefile.am +++ b/qemud/Makefile.am @@ -248,7 +248,7 @@ endif # DBUS_INIT_SCRIPTS_RED_HAT # This must be added last, since functions it provides/replaces # are used by nearly every other library. -libvirtd_LDADD += ../gnulib/lib/libgnu.la +libvirtd_LDADD += ../gnulib/lib/libgnu.la $(LIBSOCKET) endif # WITH_LIBVIRTD diff --git a/tests/Makefile.am b/tests/Makefile.am --- a/tests/Makefile.am +++ b/tests/Makefile.am @@ -33,6 +33,7 @@ LDADDS = \ $(WARN_CFLAGS) \ ../src/libvirt_test.la \ ../gnulib/lib/libgnu.la \ + $(LIBSOCKET) \ $(COVERAGE_LDFLAGS) EXTRA_DIST = \

3 2

Re: [libvirt] [PATCH]: pass flags to all virDomain*DefParse* functions
by Guido Günther 08 Jan '09

08 Jan '09

On Thu, Jan 08, 2009 at 05:02:14PM +0100, Jim Meyering wrote: [..snip..] > Hi Guido, > > libvirt doesn't build right now, because of this: > > xm_internal.c:2288: error: too few arguments to function 'virDomainDefParseString' > xm_internal.c:2530: error: too few arguments to function 'virDomainDeviceDefParse' > xm_internal.c:2619: error: too few arguments to function 'virDomainDeviceDefParse' > xend_internal.c:3860: error: too few arguments to function 'virDomainDeviceDefParse' > xend_internal.c:3950: error: too few arguments to function 'virDomainDeviceDefParse' Thought I had tested this with the Xen driver enabled, but seems I didn't. Sorry! Attached patch fixes the build and passes "make check". Thanks for pointing this out! -- Guido

1 1