- Devel - libvirt List Archives

[libvirt] [PATCH] lxc_container: Don't dereference NULL upon failure (x2)
by Jim Meyering 05 Feb '09

05 Feb '09

Spotted these while skimming through all VIR_FREE uses. >From af9e606d97e8321683ba765ba0a4af4c065d3b4b Mon Sep 17 00:00:00 2001 From: Jim Meyering <meyering(a)redhat.com> Date: Wed, 4 Feb 2009 18:44:34 +0100 Subject: [PATCH] lxc_container: Don't dereference NULL upon failure * src/lxc_container.c (lxcContainerPivotRoot): Free and zero oldroot *after* printing the diagnostic, both for virFileMakePath and pivot_root failure. --- src/lxc_container.c | 5 ++--- 1 files changed, 2 insertions(+), 3 deletions(-) diff --git a/src/lxc_container.c b/src/lxc_container.c index ea52eed..90c70d5 100644 --- a/src/lxc_container.c +++ b/src/lxc_container.c @@ -284,20 +284,20 @@ static int lxcContainerPivotRoot(virDomainFSDefPtr root) } if ((rc = virFileMakePath(oldroot)) < 0) { - VIR_FREE(oldroot); virReportSystemError(NULL, rc, _("failed to create %s"), oldroot); + VIR_FREE(oldroot); return -1; } /* The old root directory will live at /.oldroot after * this and will soon be unmounted completely */ if (pivot_root(root->src, oldroot) < 0) { - VIR_FREE(oldroot); virReportSystemError(NULL, errno, _("failed to pivot root %s to %s"), oldroot, root->src); + VIR_FREE(oldroot); return -1; } VIR_FREE(oldroot); @@ -677,4 +677,3 @@ int lxcContainerAvailable(int features) return 0; } - -- 1.6.1.2.467.g081e7

2 1

[libvirt] [PATCH] trivial libvirt example code
by Dave Allan 04 Feb '09

04 Feb '09

The examples directory doesn't have a trivial example of how to connect to a hypervisor, make a few calls, and disconnect, so I put one together. I would appreciate any suggestions on anything that I've done wrong as well as suggestions for other fundamental API calls that should be illustrated. Dave commit 14531c0fa3680890086d61ae4d66aec525f9c9a0 Author: David Allan <dallan(a)redhat.com> Date: Fri Jan 23 14:27:39 2009 -0500 Added a trivial example program to illustrate connecting to the hypervisor and making some simple API calls. diff --git a/examples/hellolibvirt/hellolibvirt.c b/examples/hellolibvirt/hellolibvirt.c new file mode 100644 index 0000000..aae79b8 --- /dev/null +++ b/examples/hellolibvirt/hellolibvirt.c @@ -0,0 +1,137 @@ +/* This file contains trivial example code to connect to the running + * hypervisor and gather a few bits of information. */ + +#include <stdio.h> +#include <stdlib.h> +#include <libvirt/libvirt.h> + +static int +showHypervisorInfo(virConnectPtr conn) +{ + int ret = 0; + unsigned long hvVer, major, minor, release; + const char *hvType; + + /* virConnectGetType returns a pointer to a static string, so no + * allocation or freeing is necessary; it is possible for the call + * to fail if, for example, there is no connection to a + * hypervisor, so check what it returns. */ + hvType = virConnectGetType(conn); + if (NULL == hvType) { + ret = 1; + printf("Failed to get hypervisor type\n"); + goto out; + } + + if (0 != virConnectGetVersion(conn, &hvVer)) { + ret = 1; + printf("Failed to get hypervisor version\n"); + goto out; + } + + major = hvVer / 1000000; + hvVer %= 1000000; + minor = hvVer / 1000; + release = hvVer % 1000; + + printf("Hypervisor: \"%s\" version: %lu.%lu.%lu\n", + hvType, + major, + minor, + release); + +out: + return ret; +} + + +static int +showDomains(virConnectPtr conn) +{ + int ret = 0, i, numNames, numInactiveDomains, numActiveDomains; + char **nameList = NULL; + + numActiveDomains = virConnectNumOfDomains(conn); + numInactiveDomains = virConnectNumOfDefinedDomains(conn); + + printf("There are %d active and %d inactive domains\n", + numActiveDomains, numInactiveDomains); + + nameList = malloc(sizeof(char *) * (unsigned int)numInactiveDomains); + + if (NULL == nameList) { + ret = 1; + printf("Could not allocate memory for list of inactive domains\n"); + goto out; + } + + numNames = virConnectListDefinedDomains(conn, + nameList, + numInactiveDomains); + + if (-1 == numNames) { + ret = 1; + printf("Could not get list of defined domains from hypervisor\n"); + goto out; + } + + if (numNames > 0) { + printf("Inactive domains:\n"); + } + + for (i = 0 ; i < numNames ; i++) { + printf(" %s\n", *(nameList + i)); + /* The API documentation doesn't say so, but the names + * returned by virConnectListDefinedDomains are strdup'd and + * must be freed here. */ + free(*(nameList + i)); + } + +out: + if (NULL != nameList) { + free(nameList); + } + + return ret; +} + + +int +main(void) +{ + int ret = 0; + virConnectPtr conn = NULL; + + printf("Attempting to connect to hypervisor\n"); + + /* virConnectOpenAuth called here with all default parameters */ + conn = virConnectOpenAuth(NULL, virConnectAuthPtrDefault, 0); + + if (NULL == conn) { + ret = 1; + printf("No connection to hypervisor\n"); + goto out; + } + + printf("Connected to hypervisor\n"); + + if (0 != showHypervisorInfo(conn)) { + ret = 1; + goto disconnect; + } + + if (0 != showDomains(conn)) { + ret = 1; + goto disconnect; + } + +disconnect: + if (0 != virConnectClose(conn)) { + printf("Failed to disconnect from hypervisor\n"); + } else { + printf("Disconnected from hypervisor\n"); + } + +out: + return ret; +}

5 22

[libvirt] PATCH: Prevent libvirtd 100% CPU usage when --timeout is used
by Daniel P. Berrange 04 Feb '09

04 Feb '09

A couple of bugs conspired to make libvirtd use 100% CPU usage when the --timeout option is used. This is the default for qemu:///session instance which are auto-spawned, so quite annoying! eg libvirt --timeout=30 - The qemudRunLoop() method was registering a timeout on every iteration of the loop, when it considered itself inactive, even if already registered - virEventInteruptLocked() was looking at eventLoop.leader and comparing to pthread_self() before anyone used the event loop. 'leader' was 0 at this point so it thought it had to wake someone up even though no one was waiting in the event loop. The latter bug conspired with the former, so the act of registering the timeout, caused it to immediately see a wakeup signal on the following poll. So it did another loop iteration, adding another timer, getting woken up again, etc 100% cpu follows. Another minor problem was that when completing an event loop iteration, we reset eventloop.leader to '0'. It is not safe to assume that pthread_t is an integer like this. The solution is to track when something is using the event loop explicitly, and not rely on the 'leader' variable, unless we know a thread is active. Finally we never de-registered the shutdown timeout if a new client turned up while we were counting down. This patch addresses all these flaws Daniel Index: qemud/event.c =================================================================== RCS file: /data/cvs/libvirt/qemud/event.c,v retrieving revision 1.17 diff -u -p -u -r1.17 event.c --- qemud/event.c 22 Dec 2008 12:55:47 -0000 1.17 +++ qemud/event.c 4 Feb 2009 14:40:09 -0000 @@ -68,6 +68,7 @@ struct virEventTimeout { /* State for the main event loop */ struct virEventLoop { pthread_mutex_t lock; + int running; pthread_t leader; int wakeupfd[2]; int handlesCount; @@ -521,6 +522,7 @@ int virEventRunOnce(void) { int ret, timeout, nfds; virEventLock(); + eventLoop.running = 1; eventLoop.leader = pthread_self(); if ((nfds = virEventMakePollFDs(&fds)) < 0) { virEventUnlock(); @@ -572,7 +574,7 @@ int virEventRunOnce(void) { return -1; } - eventLoop.leader = 0; + eventLoop.running = 0; virEventUnlock(); return 0; } @@ -611,7 +613,9 @@ int virEventInit(void) static int virEventInterruptLocked(void) { char c = '\0'; - if (pthread_self() == eventLoop.leader) + + if (!eventLoop.running || + pthread_self() == eventLoop.leader) return 0; if (safewrite(eventLoop.wakeupfd[1], &c, sizeof(c)) != sizeof(c)) Index: qemud/qemud.c =================================================================== RCS file: /data/cvs/libvirt/qemud/qemud.c,v retrieving revision 1.138 diff -u -p -u -r1.138 qemud.c --- qemud/qemud.c 28 Jan 2009 11:31:39 -0000 1.138 +++ qemud/qemud.c 4 Feb 2009 14:40:09 -0000 @@ -2048,9 +2048,18 @@ static void qemudFreeClient(struct qemud static int qemudRunLoop(struct qemud_server *server) { int timerid = -1; int ret = -1, i; + int timerActive = 0; virMutexLock(&server->lock); + if (timeout > 0 && + (timerid = virEventAddTimeoutImpl(-1, + qemudInactiveTimer, + server, NULL)) < 0) { + VIR_ERROR0(_("Failed to register shutdown timeout")); + return -1; + } + if (min_workers > max_workers) max_workers = min_workers; @@ -2071,11 +2080,21 @@ static int qemudRunLoop(struct qemud_ser * if any drivers have active state, if not * shutdown after timeout seconds */ - if (timeout > 0 && !virStateActive() && !server->clients) { - timerid = virEventAddTimeoutImpl(timeout*1000, - qemudInactiveTimer, - server, NULL); - DEBUG("Scheduling shutdown timer %d", timerid); + if (timeout > 0) { + if (timerActive) { + if (server->clients) { + DEBUG("Deactivating shutdown timer %d", timerid); + virEventUpdateTimeoutImpl(timerid, -1); + timerActive = 0; + } + } else { + if (!virStateActive() && + !server->clients) { + DEBUG("Activating shutdown timer %d", timerid); + virEventUpdateTimeoutImpl(timerid, timeout * 1000); + timerActive = 1; + } + } } virMutexUnlock(&server->lock); -- |: 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 :|

2 2

[libvirt] virGetLastError() disagrees with virConnGetLastError()
by John Levon 03 Feb '09

03 Feb '09

If you edit virsh sources, you should find that on 'virsh dump Domain-0 /var/tmp/foo', the conn error is correctly set, but the thread-load error is not set. Isn't this a bug? regards john

2 4

[libvirt] strange bug
by Farkas Levente 03 Feb '09

03 Feb '09

hi, this strange bug always happened with me on centos-5 with virt-manager. our setup: - host: - Intel(R) Core(TM)2 Quad CPU Q6600 @ 2.40GHz - Intel S3000AHV - 8GB RAM - CentOS-5.2 - kernel-2.6.18-92.1.22.el5 x86_64 64bit - guest-1: - CentOS-5.2 - 4 vcpu - kernel-2.6.18-92.1.22.el5 i386 32bit - guest-2: - CentOS-5.2 - 4 vcpu - kernel-2.6.18-92.1.22.el5 x86_64 64bit - guest-3: - Mandrake-9 - 1 vcpu - kernel-2.4.19.16mdk-1-1mdk 32bit - guest-4: - Mandrake-10 - 1 vcpu - kernel-2.6.14.2-p4-smp 32bit - guest-5: - Windows XP Professional 32bit - 2 vcpu - guest-7: - Fedora-9 - 4 vcpu - kernel-2.6.25.14-108.fc9.i686 if i restart libvirt i can start as the first guest the guest-4/mandrake-10 but if i start any of the other guest i no longer able to start guest-4. now the most strange thing happened with guest-7 (fedora-9). i use these packages build myself: http://www.lfarkas.org/linux/packages/centos/5/ libvirt-0.6.0-1.x86_64.rpm libvirt-python-0.6.0-1.x86_64.rpm virt-manager-0.6.1-1.x86_64.rpm python-virtinst-0.400.1-1.noarch.rpm kvm-83-1.x86_64.rpm and what've got from virt-manager is not too useful for me: ------------------------------------ Traceback (most recent call last): File "/usr/share/virt-manager/virtManager/engine.py", line 498, in run_domain vm.startup() File "/usr/share/virt-manager/virtManager/domain.py", line 508, in startup self.vm.create() File "/usr/lib64/python2.4/site-packages/libvirt.py", line 287, in create if ret == -1: raise libvirtError ('virDomainCreate() failed', dom=self) libvirtError: internal error unable to start guest: ------------------------------------ is there any way to find the reason? thanks. -- Levente "Si vis pacem para bellum!"

2 2

[libvirt] [PATCH] Allow no <source> for bridge interfaces
by john.levon＠sun.com 03 Feb '09

03 Feb '09

# HG changeset patch # User john.levon(a)sun.com # Date 1233586907 28800 # Node ID 68087e3caf39c74797720bde71a8df524ec036c8 # Parent aa50ca252e1419f30eb0e16204b48db42d92c785 Allow no <source> for bridge interfaces Historically, it was allowed to specify a bridge interface without a specific bridge source, in which case xend would choose one. Keep compatible with this behaviour. 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 @@ -962,13 +962,10 @@ virDomainNetDefParseXML(virConnectPtr co break; case VIR_DOMAIN_NET_TYPE_BRIDGE: - if (bridge == NULL) { - virDomainReportError(conn, VIR_ERR_INTERNAL_ERROR, "%s", - _("No <source> 'dev' attribute specified with <interface type='bridge'/>")); - goto error; - } - def->data.bridge.brname = bridge; - bridge = NULL; + if (bridge != NULL) { + def->data.bridge.brname = bridge; + bridge = NULL; + } if (script != NULL) { def->data.bridge.script = script; script = NULL; @@ -2920,8 +2917,9 @@ virDomainNetDefFormat(virConnectPtr conn break; case VIR_DOMAIN_NET_TYPE_BRIDGE: - virBufferEscapeString(buf, " <source bridge='%s'/>\n", - def->data.bridge.brname); + if (def->data.bridge.brname) + virBufferEscapeString(buf, " <source bridge='%s'/>\n", + def->data.bridge.brname); if (def->data.bridge.ipaddr) virBufferVSprintf(buf, " <ip address='%s'/>\n", def->data.bridge.ipaddr); diff --git a/src/xend_internal.c b/src/xend_internal.c --- a/src/xend_internal.c +++ b/src/xend_internal.c @@ -5149,7 +5149,8 @@ xenDaemonFormatSxprNet(virConnectPtr con switch (def->type) { case VIR_DOMAIN_NET_TYPE_BRIDGE: - virBufferVSprintf(buf, "(bridge '%s')", def->data.bridge.brname); + if (def->data.bridge.brname) + virBufferVSprintf(buf, "(bridge '%s')", def->data.bridge.brname); if (def->data.bridge.script) script = def->data.bridge.script;

3 3

[libvirt] Change of CDROM source via virsh
by Radek Hladik 03 Feb '09

03 Feb '09

Hi, I would like to ask what is the simplest way, how to change ISO mounted in VM's drive. I am using libvirt+kvm+qemu and only way I found is via virt-manager, but there are some problems with it - it is not working properly on some Fedora testing guests and it disables the VNC password for the VM. I'll try to look to the sources of virt-manager if I could find the cause (I am thinking using dumpxml without SECURE flag) but I would like to change the ISO via script (i.e. via virsh). There must be some simple way but I can not find it... Radek

1 0

[libvirt] Sendkey after shutdown
by Mario Wolff 03 Feb '09

03 Feb '09

Hello all, first post, first contact! I hope my english is at least understandable. I'm using libvirt to manage some kvm hosts with virtual Windows-Systems. If i call 'shutdown <system>' in virsh, i get "Domain <system> is being shutdown" but on the system a window is displayed telling my about remote users. Without libvirt i had the kvm-monitor on an ip-port and used 'sendkey ret' to kill the dialog. Is there a hook/script/extention-api to implement something like this? Thanks a lot! Regards, Mario Wolff PS: Is someone working on a drbd-storage-driver? Are there interests? Is there an interface-description for storage-drivers? Should one wait until the drbd is moved to lvm (i've heard there are plans?!?)? Thanks again!

2 1

[libvirt] [PATCH] Fix more printf("%s", NULL) usage
by john.levon＠sun.com 03 Feb '09

03 Feb '09

# HG changeset patch # User john.levon(a)sun.com # Date 1233580871 28800 # Node ID e61a2e486a59394ea422d90e1104f805ceabd0be # Parent 430096efc257a704fdd6294829686d05f38da68e Fix more printf("%s", NULL) usage Signed-off-by: John Levon <john.levon(a)sun.com> diff --git a/src/libvirt.c b/src/libvirt.c --- a/src/libvirt.c +++ b/src/libvirt.c @@ -2871,7 +2871,9 @@ virDomainMigratePrepare (virConnectPtr d const char *dname, unsigned long bandwidth) { - DEBUG("dconn=%p, cookie=%p, cookielen=%p, uri_in=%s, uri_out=%p, flags=%lu, dname=%s, bandwidth=%lu", dconn, cookie, cookielen, uri_in, uri_out, flags, dname, bandwidth); + VIR_DEBUG("dconn=%p, cookie=%p, cookielen=%p, uri_in=%s, uri_out=%p, " + "flags=%lu, dname=%s, bandwidth=%lu", dconn, cookie, cookielen, + NULLSTR(uri_in), uri_out, flags, NULLSTR(dname), bandwidth); virResetLastError(); @@ -2917,7 +2919,9 @@ virDomainMigratePerform (virDomainPtr do unsigned long bandwidth) { virConnectPtr conn; - DEBUG("domain=%p, cookie=%p, cookielen=%d, uri=%s, flags=%lu, dname=%s, bandwidth=%lu", domain, cookie, cookielen, uri, flags, dname, bandwidth); + VIR_DEBUG("domain=%p, cookie=%p, cookielen=%d, uri=%s, flags=%lu, " + "dname=%s, bandwidth=%lu", domain, cookie, cookielen, uri, flags, + NULLSTR(dname), bandwidth); virResetLastError(); @@ -2962,7 +2966,8 @@ virDomainMigrateFinish (virConnectPtr dc const char *uri, unsigned long flags) { - DEBUG("dconn=%p, dname=%s, cookie=%p, cookielen=%d, uri=%s, flags=%lu", dconn, dname, cookie, cookielen, uri, flags); + VIR_DEBUG("dconn=%p, dname=%s, cookie=%p, cookielen=%d, uri=%s, " + "flags=%lu", dconn, NULLSTR(dname), cookie, cookielen, uri, flags); virResetLastError(); @@ -3010,7 +3015,9 @@ virDomainMigratePrepare2 (virConnectPtr unsigned long bandwidth, const char *dom_xml) { - DEBUG("dconn=%p, cookie=%p, cookielen=%p, uri_in=%s, uri_out=%p, flags=%lu, dname=%s, bandwidth=%lu, dom_xml=%s", dconn, cookie, cookielen, uri_in, uri_out, flags, dname, bandwidth, dom_xml); + VIR_DEBUG("dconn=%p, cookie=%p, cookielen=%p, uri_in=%s, uri_out=%p," + "flags=%lu, dname=%s, bandwidth=%lu, dom_xml=%s", dconn, cookie, + cookielen, uri_in, uri_out, flags, NULLSTR(dname), bandwidth, dom_xml); virResetLastError(); @@ -3056,7 +3063,9 @@ virDomainMigrateFinish2 (virConnectPtr d unsigned long flags, int retcode) { - DEBUG("dconn=%p, dname=%s, cookie=%p, cookielen=%d, uri=%s, flags=%lu, retcode=%d", dconn, dname, cookie, cookielen, uri, flags, retcode); + VIR_DEBUG("dconn=%p, dname=%s, cookie=%p, cookielen=%d, uri=%s, " + "flags=%lu, retcode=%d", dconn, NULLSTR(dname), cookie, + cookielen, uri, flags, retcode); virResetLastError();

4 3

[libvirt] [PATCH]: don't hardcode default port 22 for ssh protocol
by Guido Günther 03 Feb '09

03 Feb '09

Hi, attached patch removes the hardcoded port 22 when going through ssh, ssh uses it by default. This makes it easier to override this via .ssh/config on a per host basis instead of having to add this to the connection URI explicitly. While at that I cleaned up some free vs. VIR_FREE usage and replaced pointer intializations with 0 by NULL. Cheers, -- Guido P.S.: this originated from Debian Bug #513605

4 6