- Devel - libvirt List Archives

[libvirt] [PATCH] [BUGFIX][PATCH] send-key: fix scan keycode map
by KAMEZAWA Hiroyuki 22 Aug '11

22 Aug '11

Now, bad key-code in send-key can cause segmentation fault in libvirt. (example) % virsh send-key --codeset win32 12 error: End of file while reading data: Input/output error This is caused by overrun at scanning keycode array. Fix it. Signed-off-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu(a)jp.fujitsu.com> --- src/util/virkeycode.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/src/util/virkeycode.c b/src/util/virkeycode.c index 4d54060..0d42767 100644 --- a/src/util/virkeycode.c +++ b/src/util/virkeycode.c @@ -118,7 +118,7 @@ static int __virKeycodeValueTranslate(unsigned int from_offset, { int i; - for (i = 0; ARRAY_CARDINALITY(virKeycodes); i++) { + for (i = 0; i < ARRAY_CARDINALITY(virKeycodes); i++) { if (getfield(virKeycodes + i, unsigned short, from_offset) == key_value) return getfield(virKeycodes + i, unsigned short, to_offset); } -- 1.7.4.1

2 1

[libvirt] portability issue on Centos 6.0
by Daniel Veillard 22 Aug '11

22 Aug '11

I just tred compiling 0.9.4 and git head on Centos 6.0, for the possible libvirt.org server replacement and it fails in both cases with: CC libvirtd-remote.o remote.c: In function 'remoteDispatchAuthPolkit': remote.c:2161: error: invalid initializer This correspond to the following macro PROBE(CLIENT_AUTH_ALLOW, "fd=%d, auth=%d, username=%s", virNetServerClientGetFD(client), REMOTE_AUTH_POLKIT, ident); I'm a bit lost how the expansion could lead to this and why, unless the DTrace support in Centos 6 is slightly different than on RHEL, I will look but if someone has an idea :-) Daniel -- Daniel Veillard | libxml Gnome XML XSLT toolkit http://xmlsoft.org/ daniel(a)veillard.com | Rpmfind RPM search engine http://rpmfind.net/ http://veillard.com/ | virtualization library http://libvirt.org/

2 3

[libvirt] How to avoid the disk still persistent after no persistent hotplug and with restarting libvirtd then.
by Osier Yang 22 Aug '11

22 Aug '11

Libvirt loads the domain conf from the status file (/var/run/libvirt/qemu/$dom.xml) if the domain is running, the problem if one restart libvirtd just after some changing on the domain conf, and then destroy/start the guest. libvirt will never known what the original domain conf is then. (vm->newdef is NULL at this moment, generally libvirt restore the domain conf to the original conf (vm->newDef) when destroying and shutdown). Following is a demo of the problem: 1) hotplug a disk to a running domain without "--persistent" specified. 2) restart libvirtd 3) Destroy/start the guest 4) you will see the hotpluged disk is still there. The solution in my mind is to add some attribute to the status XML, such as "live_updated=1", and do the checking when loading the domain conf from status XML: * if "live_updated=1", then loads the domain conf from the persistent domain XML. But the problem is: For a persistent domain, if one made changes on the domain conf both persistently and transiently, we have no way to known which part are changed. Loading from the persistent domain conf is not correct either. For a transient domain, we still have no way to get the original domain conf, that means one will still be surprised when seeing the the disk is still there. Any idea on this? Thanks Osier

2 3

[libvirt] Proposal: A place to store metadata about guests
by Richard W.M. Jones 22 Aug '11

22 Aug '11

I think we need a place to store general metadata about guests. Here are some uses: - Store data about available installer ISOs, guests which are archived and not known to libvirt. - Replace the guest inspection thread in virt-manager. virt-manager would just collect the data from the store. The inspection data would be generated by another process. - Expand on <description/> field in the libvirt XML, and satisfy various requests we've had for a place to store metadata and extend libvirt XML. - Store data on behalf of applications, esp. virt-manager. I'm not envisaging a database or daemon for this. Just a simple directory per host containing flat files. Perhaps we can add some tools or a library to make creation/querying of these files easy. It would be nice if the database was made available over the network (Avahi?). You could also use this database to answer questions like: - "Do I have a Debian wheezy 32 bit guest anywhere on my network?" - "Does any guest have libpng <= <vulnerable-version> installed?" - "Bring up a Fedora 14 instance" (the tools would have to locate a suitable install ISO and put together the correct virt-install command line) There is some crossover with VDSM and even RHEV-M here. Thoughts? Rich. -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones libguestfs lets you edit virtual machines. Supports shell scripting, bindings from many languages. http://libguestfs.org

3 8

[libvirt] [libvirt-snmp][PATCH 1/3] subagent: Avoid dead code
by Michal Privoznik 22 Aug '11

22 Aug '11

Detected by Coverity --- src/libvirtMib_subagent.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/src/libvirtMib_subagent.c b/src/libvirtMib_subagent.c index f5749be..ee99fb9 100644 --- a/src/libvirtMib_subagent.c +++ b/src/libvirtMib_subagent.c @@ -60,7 +60,7 @@ main (int argc, char **argv) { /* Defs for arg-handling code: handles setting of policy-related variables */ int ch; extern char *optarg; - int dont_fork = 0, use_syslog = 0; + int dont_fork = 0, use_syslog = 1; char *agentx_socket = NULL; while ((ch = getopt(argc, argv, "D:fHLMx:")) != EOF) -- 1.7.3.4

1 2

[libvirt] [PATCH v2] qemu: Allow graceful domain destroy
by Michal Privoznik 22 Aug '11

22 Aug '11

This patch introduces support for domain destroying via 'quit' monitor command which gives qemu time to flush caches and therefore prevent disks corruption. However, qemu can be unresponsive and to prevent waiting indefinitely, execute command in a separate thread and wait reasonable time (QEMU_QUIT_WAIT_SECONDS) on a condition. If we hit timeout, qemu is qemuProcessKill'ed which causes monitor close and therefore also thread being terminable. The synchronization between qemu driver and monitor-quit thread is done through mutex and condition. However, this alone is not enough. If a condition is signalized but without anybody listening signal is lost. To prevent this a boolean variable is used that is set iff nobody is listening but condition would be signalized, or iff driver is waiting on given condition. --- diff to v1: - more debug printings - don't test (!flags) but (!(flags & VIR_DOMAIN_DESTROY_FLUSH_CACHE)) include/libvirt/libvirt.h.in | 11 ++- src/qemu/qemu_driver.c | 146 ++++++++++++++++++++++++++++++++++++++++-- src/qemu/qemu_monitor.c | 13 ++++ src/qemu/qemu_monitor.h | 2 + src/qemu/qemu_monitor_json.c | 22 ++++++ src/qemu/qemu_monitor_json.h | 1 + src/qemu/qemu_monitor_text.c | 20 ++++++ src/qemu/qemu_monitor_text.h | 1 + tools/virsh.c | 9 ++- 9 files changed, 213 insertions(+), 12 deletions(-) diff --git a/include/libvirt/libvirt.h.in b/include/libvirt/libvirt.h.in index aa29fb6..0610b53 100644 --- a/include/libvirt/libvirt.h.in +++ b/include/libvirt/libvirt.h.in @@ -919,10 +919,13 @@ virConnectPtr virDomainGetConnect (virDomainPtr domain); * Domain creation and destruction */ -/* - * typedef enum { - * } virDomainDestroyFlagsValues; - */ + +typedef enum { + VIR_DOMAIN_DESTROY_FLUSH_CACHE = 1 << 0, /* In hypervisors supporting this, + try to get cache flushed prior + to destroying the domain */ +} virDomainDestroyFlagsValues; + virDomainPtr virDomainCreateXML (virConnectPtr conn, const char *xmlDesc, unsigned int flags); diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c index c8dda73..44f4bbd 100644 --- a/src/qemu/qemu_driver.c +++ b/src/qemu/qemu_driver.c @@ -1564,6 +1564,47 @@ cleanup: return ret; } +struct qemuDomainDestroyHelperData { + struct qemud_driver *driver; + virDomainObjPtr vm; + qemuDomainObjPrivatePtr priv; + virMutex lock; + virCond cond; + bool guard; +}; + +static void +qemuDomainDestroyHelper(void *opaque) +{ + struct qemuDomainDestroyHelperData *data = opaque; + struct qemud_driver *driver = data->driver; + virDomainObjPtr vm = data->vm; + qemuDomainObjPrivatePtr priv = data->priv; + + /* Job was already obtained by caller */ + VIR_DEBUG("Destroy thread started, sending 'quit' command."); + qemuDomainObjEnterMonitorWithDriver(driver, vm); + qemuMonitorQuit(priv->mon); + qemuDomainObjExitMonitorWithDriver(driver, vm); + VIR_DEBUG("Command finished."); + + /* To avoid loosing signal, we need to remember + * we tried to send one, but nobody was waiting + * for it. + */ + virMutexLock(&data->lock); + if (data->guard) { + VIR_DEBUG("Driver is listening, signalize on condition"); + virCondSignal(&data->cond); + } else { + VIR_DEBUG("Nobody's listening, setting guard variable"); + data->guard = true; + } + virMutexUnlock(&data->lock); + VIR_DEBUG("Destroy thread exiting"); +} + +#define QEMU_QUIT_WAIT_SECONDS 5 static int qemuDomainDestroyFlags(virDomainPtr dom, @@ -1574,8 +1615,13 @@ qemuDomainDestroyFlags(virDomainPtr dom, int ret = -1; virDomainEventPtr event = NULL; qemuDomainObjPrivatePtr priv; + bool use_thread = false; + bool use_kill = false; + virThread destroy_thread; + struct qemuDomainDestroyHelperData data; + unsigned long long then; - virCheckFlags(0, -1); + virCheckFlags(VIR_DOMAIN_DESTROY_FLUSH_CACHE, -1); qemuDriverLock(driver); vm = virDomainFindByUUID(&driver->domains, dom->uuid); @@ -1590,12 +1636,14 @@ qemuDomainDestroyFlags(virDomainPtr dom, priv = vm->privateData; priv->fakeReboot = false; - /* Although qemuProcessStop does this already, there may - * be an outstanding job active. We want to make sure we - * can kill the process even if a job is active. Killing - * it now means the job will be released - */ - qemuProcessKill(vm); + if (!(flags & VIR_DOMAIN_DESTROY_FLUSH_CACHE)) { + /* Although qemuProcessStop does this already, there may + * be an outstanding job active. We want to make sure we + * can kill the process even if a job is active. Killing + * it now means the job will be released + */ + qemuProcessKill(vm); + } if (qemuDomainObjBeginJobWithDriver(driver, vm, QEMU_JOB_DESTROY) < 0) goto cleanup; @@ -1606,6 +1654,78 @@ qemuDomainDestroyFlags(virDomainPtr dom, goto endjob; } + if (flags & VIR_DOMAIN_DESTROY_FLUSH_CACHE) { + /* Try to shutdown domain gracefully. + * Send 'quit' to qemu. However, qemu can be unresponsive. + * Therefore create a separate thread in which we execute + * that monitor comand. Wait max QEMU_QUIT_WAIT_SECONDS. + */ + VIR_DEBUG("Trying to gracefully shutdown %s", vm->def->name); + if (virMutexInit(&data.lock) < 0) { + virReportOOMError(); + goto endjob; + } + + if (virCondInit(&data.cond) < 0) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, "%s", + _("cannot initialize thread condition")); + virMutexDestroy(&data.lock); + goto endjob; + } + + data.driver = driver; + data.vm = vm; + data.priv = priv; + data.guard = false; + + VIR_DEBUG("Spawning thread to run 'quit' command"); + if (virThreadCreate(&destroy_thread, true, + qemuDomainDestroyHelper, &data) < 0) { + virReportSystemError(errno, "%s", + _("unable to create destroy thread")); + ignore_value(virCondDestroy(&data.cond)); + virMutexDestroy(&data.lock); + goto endjob; + } + use_thread = true; + + if (virTimeMs(&then) < 0) + goto endjob; + + then += QEMU_QUIT_WAIT_SECONDS * 1000; + + /* How synchronization with destroy thread works: + * Basically wait on data.cond. But because conditions + * does not 'remember' that they have been signalized + * data.guard is used. Practically, data.guard says + * to destroy thread if we are waiting on condition + * and to us whether we should even try. + */ + virMutexLock(&data.lock); + if (!data.guard) { + data.guard = true; + VIR_DEBUG("Waiting for thread to exit"); + if (virCondWaitUntil(&data.cond, &data.lock, then) < 0) { + if (errno == ETIMEDOUT) { + VIR_DEBUG("Condition timeout"); + use_kill = true; + data.guard = false; + } else { + virReportSystemError(errno, "%s", _("unable to wait " + "on condition")); + virMutexUnlock(&data.lock); + goto endjob; + } + } + } else { + VIR_DEBUG("Guard variable set, thread exited"); + } + virMutexUnlock(&data.lock); + + if (use_kill) + qemuProcessKill(vm); + } + qemuProcessStop(driver, vm, 0, VIR_DOMAIN_SHUTOFF_DESTROYED); event = virDomainEventNewFromObj(vm, VIR_DOMAIN_EVENT_STOPPED, @@ -1626,6 +1746,18 @@ endjob: vm = NULL; cleanup: + if (use_thread) { + VIR_DEBUG("Thread used, clean it up"); + virMutexLock(&data.lock); + if (!data.guard) { + data.guard = true; + ignore_value(virCondWait(&data.cond, &data.lock)); + } + virMutexUnlock(&data.lock); + ignore_value(virCondDestroy(&data.cond)); + virMutexDestroy(&data.lock); + virThreadJoin(&destroy_thread); + } if (vm) virDomainObjUnlock(vm); if (event) diff --git a/src/qemu/qemu_monitor.c b/src/qemu/qemu_monitor.c index db6107c..41b9c5c 100644 --- a/src/qemu/qemu_monitor.c +++ b/src/qemu/qemu_monitor.c @@ -2475,3 +2475,16 @@ int qemuMonitorBlockJob(qemuMonitorPtr mon, ret = qemuMonitorTextBlockJob(mon, device, bandwidth, info, mode); return ret; } + +int qemuMonitorQuit(qemuMonitorPtr mon) +{ + int ret; + + VIR_DEBUG("mon=%p", mon); + + if (mon->json) + ret = qemuMonitorJSONQuit(mon); + else + ret = qemuMonitorTextQuit(mon); + return ret; +} diff --git a/src/qemu/qemu_monitor.h b/src/qemu/qemu_monitor.h index f241c9e..3fe6bb9 100644 --- a/src/qemu/qemu_monitor.h +++ b/src/qemu/qemu_monitor.h @@ -475,6 +475,8 @@ int qemuMonitorBlockJob(qemuMonitorPtr mon, virDomainBlockJobInfoPtr info, int mode); +int qemuMonitorQuit(qemuMonitorPtr mon); + /** * When running two dd process and using <> redirection, we need a * shell that will not truncate files. These two strings serve that diff --git a/src/qemu/qemu_monitor_json.c b/src/qemu/qemu_monitor_json.c index 7adfb26..1f078cd 100644 --- a/src/qemu/qemu_monitor_json.c +++ b/src/qemu/qemu_monitor_json.c @@ -2956,3 +2956,25 @@ int qemuMonitorJSONBlockJob(qemuMonitorPtr mon, virJSONValueFree(reply); return ret; } + + +int qemuMonitorJSONQuit(qemuMonitorPtr mon) +{ + int ret = -1; + virJSONValuePtr cmd = NULL; + virJSONValuePtr reply = NULL; + + cmd = qemuMonitorJSONMakeCommand("quit", NULL); + + if (!cmd) + return -1; + + ret =qemuMonitorJSONCommand(mon, cmd, &reply); + + if (ret == 0) + ret = qemuMonitorJSONCheckError(cmd, reply); + + virJSONValueFree(cmd); + virJSONValueFree(reply); + return ret; +} diff --git a/src/qemu/qemu_monitor_json.h b/src/qemu/qemu_monitor_json.h index 9512793..2a7df76 100644 --- a/src/qemu/qemu_monitor_json.h +++ b/src/qemu/qemu_monitor_json.h @@ -231,4 +231,5 @@ int qemuMonitorJSONBlockJob(qemuMonitorPtr mon, virDomainBlockJobInfoPtr info, int mode); +int qemuMonitorJSONQuit(qemuMonitorPtr mon); #endif /* QEMU_MONITOR_JSON_H */ diff --git a/src/qemu/qemu_monitor_text.c b/src/qemu/qemu_monitor_text.c index 52d924a..ca9c21b 100644 --- a/src/qemu/qemu_monitor_text.c +++ b/src/qemu/qemu_monitor_text.c @@ -3063,3 +3063,23 @@ cleanup: VIR_FREE(reply); return ret; } + + +int qemuMonitorTextQuit(qemuMonitorPtr mon) +{ + const char *cmd = "quit"; + char *reply = NULL; + int ret = -1; + + if (qemuMonitorHMPCommand(mon, cmd, &reply) < 0) { + qemuReportError(VIR_ERR_INTERNAL_ERROR, + "%s", _("cannot run monitor command")); + goto cleanup; + } + + ret = 0; + +cleanup: + VIR_FREE(reply); + return ret; +} diff --git a/src/qemu/qemu_monitor_text.h b/src/qemu/qemu_monitor_text.h index b250738..9ade938 100644 --- a/src/qemu/qemu_monitor_text.h +++ b/src/qemu/qemu_monitor_text.h @@ -224,4 +224,5 @@ int qemuMonitorTextBlockJob(qemuMonitorPtr mon, virDomainBlockJobInfoPtr info, int mode); +int qemuMonitorTextQuit(qemuMonitorPtr mon); #endif /* QEMU_MONITOR_TEXT_H */ diff --git a/tools/virsh.c b/tools/virsh.c index 7d849ec..4c7962b 100644 --- a/tools/virsh.c +++ b/tools/virsh.c @@ -2567,6 +2567,8 @@ static const vshCmdInfo info_destroy[] = { static const vshCmdOptDef opts_destroy[] = { {"domain", VSH_OT_DATA, VSH_OFLAG_REQ, N_("domain name, id or uuid")}, + {"flush", VSH_OT_BOOL, 0, N_("try to flush hypervisor's cache prior to " + "destroying the domain")}, {NULL, 0, 0, NULL} }; @@ -2576,6 +2578,7 @@ cmdDestroy(vshControl *ctl, const vshCmd *cmd) virDomainPtr dom; bool ret = true; const char *name; + int flags = 0; if (!vshConnectionUsability(ctl, ctl->conn)) return false; @@ -2583,7 +2586,11 @@ cmdDestroy(vshControl *ctl, const vshCmd *cmd) if (!(dom = vshCommandOptDomain(ctl, cmd, &name))) return false; - if (virDomainDestroy(dom) == 0) { + if (vshCommandOptBool(cmd, "flush")) + flags |= VIR_DOMAIN_DESTROY_FLUSH_CACHE; + + if ((!flags && virDomainDestroy(dom) == 0) || + virDomainDestroyFlags(dom, flags) == 0) { vshPrint(ctl, _("Domain %s destroyed\n"), name); } else { vshError(ctl, _("Failed to destroy domain %s"), name); -- 1.7.3.4

1 0

[libvirt] USB2 controllers & domain XML
by Marc-André Lureau 22 Aug '11

22 Aug '11

Hi I have started looking at: https://bugzilla.redhat.com/show_bug.cgi?id=725670 My initial thought was to just add a new type of controller: <controller type='usb2' index='0'/> But it also needs the 3 companion UHCI controllers (see bug). We can either include them implicitely (not really elegant in libvirt, but doable), or with more flexibility list all the controllers in the XML: <controller type='ich9-usb-ehci1' index='0'/> <controller type='ich9-usb-uhci1' index='0' masterbus='0' firstport='0'> <controller type='ich9-usb-uhci2' index='0' masterbus='0' firstport='2'> <controller type='ich9-usb-uhci3' index='0' masterbus='0' firstport='4'> Any other idea? Daniel, have you also started working on USB2 & usb-redir support? regards -- Marc-André Lureau

5 7

[libvirt] virt-manager, debug error, question
by Jason Helfman 22 Aug '11

22 Aug '11

Hi, I am getting this error when I run virt-manager in FreeBSD: [Fri, 19 Aug 2011 09:32:14 virt-manager 24762] DEBUG (engine:555) No inspection thread because libguestfs is too old, not available, or libvirt is not thread safe. libguestfs isn't ported to FreeBSD, however I was wondering about the thread safe for libvirt. Is there a way to "enable" this, or is the software telling me it really is not thread safe, and can't be. [jhelfman@eggman ~/ports/devel/libvirt/work/libvirt-0.9.4]$ ./configure --help |grep thread --enable-threads={posix|solaris|pth|win32} specify multithreading API --disable-threads build without multithread safety --enable-test-locking thread locking tests using CIL [default=no] Thanks, Jason -- Jason Helfman System Administrator experts-exchange.com http://www.experts-exchange.com/M_4830110.html E4AD 7CF1 1396 27F6 79DD 4342 5E92 AD66 8C8C FBA5

3 2

[libvirt] status of ocaml for libvirt
by Jason Helfman 20 Aug '11

20 Aug '11

Hello All, I was looking at porting virt-top to FreeBSD, and noticed that it needed ocaml-libvirt, and I know that isn't ported. What is the status of this library, and it is up-to-date? Thanks, Jason -- Jason Helfman System Administrator experts-exchange.com http://www.experts-exchange.com/M_4830110.html E4AD 7CF1 1396 27F6 79DD 4342 5E92 AD66 8C8C FBA5

1 0

[libvirt] [PATCH] virsh: properly interleave shared stdout and stderr
by Eric Blake 20 Aug '11

20 Aug '11

Without this patch, invoking 'virsh >file 2>&1' results in error messages appearing before normal output, even if they occurred later in time than the normal output (since stderr is unbuffered, but stdout waits until a full buffer). * tools/virsh.c (print_job_progress, vshError): Flush between stream transitions. * tests/undefine: Test it. --- This fixes the bug I mentioned here: https://www.redhat.com/archives/libvir-list/2011-August/msg00891.html tests/undefine | 5 +---- tools/virsh.c | 8 ++++++++ 2 files changed, 9 insertions(+), 4 deletions(-) diff --git a/tests/undefine b/tests/undefine index 6c821ad..22d6c14 100755 --- a/tests/undefine +++ b/tests/undefine @@ -58,17 +58,14 @@ compare exp out || fail=1 # Succeed, now: first shut down, then undefine, both via name. $abs_top_builddir/tools/virsh -q -c test:///default \ - 'shutdown test; undefine test; dominfo test' > out 2> err + 'shutdown test; undefine test; dominfo test' > out 2>&1 test $? = 1 || fail=1 cat <<\EOF > expout || fail=1 Domain test is being shutdown Domain test has been undefined -EOF -cat <<\EOF > experr || fail=1 error: failed to get domain 'test' error: Domain not found EOF compare expout out || fail=1 -compare experr err || fail=1 (exit $fail); exit $fail diff --git a/tools/virsh.c b/tools/virsh.c index 1c67150..7d849ec 100644 --- a/tools/virsh.c +++ b/tools/virsh.c @@ -4956,7 +4956,10 @@ print_job_progress(const char *label, unsigned long long remaining, } } + /* see comments in vshError about why we must flush */ + fflush(stdout); fprintf(stderr, "\r%s: [%3d %%]", label, progress); + fflush(stderr); } static bool @@ -14336,6 +14339,10 @@ vshError(vshControl *ctl, const char *format, ...) va_end(ap); } + /* Most output is to stdout, but if someone ran virsh 2>&1, then + * printing to stderr will not interleave correctly with stdout + * unless we flush between every transition between streams. */ + fflush(stdout); fputs(_("error: "), stderr); va_start(ap, format); @@ -14345,6 +14352,7 @@ vshError(vshControl *ctl, const char *format, ...) va_end(ap); fprintf(stderr, "%s\n", NULLSTR(str)); + fflush(stderr); VIR_FREE(str); } -- 1.7.4.4

2 2