- Devel - libvirt List Archives

[libvirt] Doubt over Kvm migration support in libvirt 0.5.0
by Shanmuga Rajan 02 Dec '08

02 Dec '08

I am using libvirt(python) to develop a small application which will manage my nodes which running in KVM. But when i tried migration i got this error. >>> vs.migrate(conn, libvirt.VIR_MIGRATE_LIVE,'testtt','192.168.1.82',0) libvir: error : this function is not supported by the hypervisor: virDomainMigrate Traceback (most recent call last): File "<stdin>", line 1, in ? File "/usr/lib64/python2.4/site-packages/libvirt.py", line 301, in migrate if ret is None:raise libvirtError('virDomainMigrate() failed', dom=self) libvirt.libvirtError: virDomainMigrate() failed this function is not supported by the hypervisor: virDomainMigrate i am running kvm-66 taken from Centos-5 Testing repositary and libvirt 0.5.0( 0.5.0: Nov 25 2008 release note says that support for kvm migration included) i am bit confused after reading this... Please any one help me in this. Thanks and Regards, -Shan

3 2

[libvirt] [PATCH] qemud/qemud.c (qemudCleanup): Plug a leak.
by Jim Meyering 01 Dec '08

01 Dec '08

Hi Dan, With your patches, "make check" hits one new leak: 40 bytes in 1 blocks are definitely lost in loss record 9 of 65 at 0x4A05174: calloc (vg_replace_malloc.c:397) by 0x447F29: virAllocN (memory.c:128) by 0x40CD82: qemudRunLoop (qemud.c:1839) by 0x40E713: main (qemud.c:2462) This fixes it: >From 9230c41d3deb00b4a0bb6800126e5cd41a350a54 Mon Sep 17 00:00:00 2001 From: Jim Meyering <meyering(a)redhat.com> Date: Mon, 1 Dec 2008 15:07:40 +0100 Subject: [PATCH] * qemud/qemud.c (qemudCleanup): Plug a leak. --- qemud/qemud.c | 1 + 1 files changed, 1 insertions(+), 0 deletions(-) diff --git a/qemud/qemud.c b/qemud/qemud.c index 94f5918..8518bd5 100644 --- a/qemud/qemud.c +++ b/qemud/qemud.c @@ -1942,6 +1942,7 @@ static void qemudCleanup(struct qemud_server *server) { virStateCleanup(); + VIR_FREE(server->workers); free(server); } -- 1.6.0.4.1044.g77718

1 0

[libvirt] avoid "make check" hang on daemon-conf
by Jim Meyering 01 Dec '08

01 Dec '08

Hi Dan, With your 28-part patch, I was getting a daemon-conf failure in "make check". This fixes it: Unless you tell me otherwise, I will apply most of this right away, (and let you remove the space-before-max_clients in your patch) since no other part of this change conflicts with your series. >From 57842865f8845c788408e18033f739cf02987ca3 Mon Sep 17 00:00:00 2001 From: Jim Meyering <meyering(a)redhat.com> Date: Mon, 1 Dec 2008 14:12:44 +0100 Subject: [PATCH] tests: daemon-conf: accommodate new numeric-valued config params: This test would hang when failing to perturb the new numeric (and non-boolean) valued parameters, max_clients, max_workers, min_workers. * tests/daemon-conf: Require that all commented out settings in libvirtd.conf have the same form. Before, two parameters were not being tested, since a space had snuck between the leading "#" and the "param = value" parts. Apply each RHS-value-perturbing transformation separately, not in series. Let VERBOSE=yes turn on debugging. Be more verbose by default, since this is a relatively long-running test. * qemud/libvirtd.conf: Normalize the spacing around each setting that is to be perturbed by tests/daemon-conf. --- qemud/libvirtd.conf | 10 ++++++---- tests/daemon-conf | 32 ++++++++++++++++++++++++++------ 2 files changed, 32 insertions(+), 10 deletions(-) diff --git a/qemud/libvirtd.conf b/qemud/libvirtd.conf index f269766..b0962f2 100644 --- a/qemud/libvirtd.conf +++ b/qemud/libvirtd.conf @@ -1,7 +1,10 @@ # Master libvirt daemon configuration file # # For further information consult http://libvirt.org/format.html - +# +# NOTE: the tests/daemon-conf regression test script requires +# that each "PARAMETER = VALUE" line in this file have the parameter +# name just after a leading "#". ################################################################# # @@ -45,7 +48,7 @@ # Override the default configuration which binds to all network # interfaces. This can be a numeric IPv4/6 address, or hostname # -# listen_addr = "192.168.0.1" +#listen_addr = "192.168.0.1" # Flag toggling mDNS advertizement of the libvirt service. @@ -233,7 +236,7 @@ # The maximum number of concurrent client connections to allow # over all sockets combined. -# max_clients = 20 +#max_clients = 20 # The minimum limit sets the number of workers to start up @@ -243,4 +246,3 @@ # of clients allowed #min_workers = 5 #max_workers = 20 - diff --git a/tests/daemon-conf b/tests/daemon-conf index db1f0d3..03189d5 100755 --- a/tests/daemon-conf +++ b/tests/daemon-conf @@ -1,6 +1,11 @@ #!/bin/sh # Get coverage of libvirtd's config-parsing code. +if test "$VERBOSE" = yes; then + set -x + libvirtd --version +fi + # Boilerplate code to set up a test directory, cd into it, # and to ensure we remove it upon completion. this_test_() { echo "./$0" | sed 's,.*/,,'; } @@ -12,8 +17,14 @@ trap '(exit $?); exit $?' 1 2 13 15 mkdir "$t_" || fail=1 cd "$t_" || fail=1 +conf="$abs_top_srcdir/qemud/libvirtd.conf" + +# Ensure that each commented out PARAMETER = VALUE line has the expected form. +grep '[a-z_] *= *[^ ]' "$conf" | grep -vE '^#[a-z_]+ = ' \ + && { echo "$0: found unexpected lines (above) in $conf" 1>&2; exit 1; } + # Start with the sample libvirtd.conf file, uncommenting all real directives. -sed -n 's/^#$[^ #]$/\1/p' $abs_top_srcdir/qemud/libvirtd.conf > tmp.conf +sed -n 's/^#$[^ #]$/\1/p' "$conf" > tmp.conf # Iterate through that list of directives, corrupting one RHS at a # time and running libvirtd with the resulting config. Each libvirtd @@ -22,16 +33,23 @@ n=$(wc -l < tmp.conf) i=1 while :; do param_name=$(sed -n "$i"'s/ = .*//p' tmp.conf) + printf "testing with corrupted config: $param_name\n" 1>&2 rhs=$(sed -n "$i"'s/.* = $.*$/\1/p' tmp.conf) f=in$i.conf - # Change an RHS that starts with '"' or '[' to "3". - # Change an RHS that starts with 0 or 1 to the string '"foo"'. - sed "$i"'s/ = [["].*/ = 3/;'"$i"'s/ = [01].*/ = "foo"/' tmp.conf > $f + case $rhs in + # Change an RHS that starts with '"' or '[' to "3". + [[\"]*) sed "$i"'s/ = [["].*/ = 3/' tmp.conf > $f;; + # Change an RHS that starts with a digit to the string '"foo"'. + [0-9]*) sed "$i"'s/ = [0-9].*/ = "foo"/' tmp.conf > $f;; + esac + + # Run libvirtd, expecting it to fail. libvirtd --config=$f 2> err && fail=1 + case $rhs in # '"'*) msg='should be a string';; '"'*) msg='invalid type: got long; expected string';; - [01]*) msg='invalid type: got string; expected long';; + [0-9]*) msg='invalid type: got string; expected long';; '['*) msg='must be a string or list of strings';; *) echo "unexpected RHS: $rhs" 1>&2; fail=1;; esac @@ -48,8 +66,10 @@ while :; do done # Run with the unmodified config file. +sleep_secs=2 +printf "running libvirtd with a valid config file ($sleep_secs seconds)\n" 1>&2 libvirtd --config=tmp.conf > log 2>&1 & pid=$! -sleep 2 +sleep $sleep_secs kill $pid # Expect an orderly shut-down and successful exit. -- 1.6.0.4.1044.g77718

1 0

[libvirt] KVM/qemu: problems with autostart of vms with non-bridged nets
by Gerd von Egidy 01 Dec '08

01 Dec '08

Hi, I'm using libvirt 0.5.0 to manage my kvm domains. Everything works fine except autostarting vms which use a network device not directly bridged to hw. I have the following net defined (and set to autostart): <network> <name>local</name> <uuid>7ec7a457-0b8e-4670-3b4a-2d0914288daa</uuid> <bridge stp='on' forwardDelay='0' /> <ip address='192.168.100.1' netmask='255.255.255.0'> <dhcp> <range start='192.168.100.128' end='192.168.100.254' /> </dhcp> </ip> </network> When a vm using this net is autostarted I get the following error: libvir: error : invalid connection pointer in virNetworkLookupByName libvir: QEMU error : internal error Network 'local' not found Failed to autostart VM 'testlocalnet': internal error Network 'local' not found I took a dive into the code and found two problems causing this: 1. initialization order: qemuStateDriver is registered (and thus initialized) before networkStateDriver. This means the network is not up yet when qemu does it's autostart. I could fix this with the attached patch. 2. missing virConnectPtr during autostart of the vm: in src/qemu_driver.c:145, qemudAutostartConfigs() you can find this call: int ret = qemudStartVMDaemon(NULL, driver, vm, NULL); This means virConnectPtr conn is NULL within qemudStartVMDaemon(). Following the calls: qemudStartVMDaemon() -> qemudBuildCommandLine() -> qemudNetworkIfaceConnect() -> virNetworkLookupByName() virNetworkLookupByName bails out if called with invalid virConnectPtr. This means starting this vm fails. Currently I'm not familiar enough with the libvirt structures to come up with an idea how to fix this. Could anybody with a deeper knowledge of the code take a look at this? Thank you very much. Kind regards, Gerd diff -r -u libvirt-0.5.0.orig/qemud/qemud.c libvirt-0.5.0/qemud/qemud.c --- libvirt-0.5.0.orig/qemud/qemud.c 2008-11-21 13:47:32.000000000 +0100 +++ libvirt-0.5.0/qemud/qemud.c 2008-11-30 21:27:06.000000000 +0100 @@ -761,22 +761,13 @@ * If they try to use a open a connection for a module that * is not loaded they'll get a suitable error at that point */ - virDriverLoadModule("qemu"); - virDriverLoadModule("lxc"); - virDriverLoadModule("uml"); virDriverLoadModule("network"); virDriverLoadModule("storage"); virDriverLoadModule("nodedev"); + virDriverLoadModule("qemu"); + virDriverLoadModule("lxc"); + virDriverLoadModule("uml"); #else -#ifdef WITH_QEMU - qemuRegister(); -#endif -#ifdef WITH_LXC - lxcRegister(); -#endif -#ifdef WITH_UML - umlRegister(); -#endif #ifdef WITH_NETWORK networkRegister(); #endif @@ -786,6 +777,15 @@ #if defined(HAVE_HAL) || defined(HAVE_DEVKIT) nodedevRegister(); #endif +#ifdef WITH_QEMU + qemuRegister(); +#endif +#ifdef WITH_LXC + lxcRegister(); +#endif +#ifdef WITH_UML + umlRegister(); +#endif #endif virEventRegisterImpl(virEventAddHandleImpl, -- Address (better: trap) for people I really don't want to get mail from: james(a)cactusamerica.com

2 2

[libvirt] [PATCH] avoid format string warnings
by Jim Meyering 01 Dec '08

01 Dec '08

Hi Dan, I've just applied all of your changes, built, and saw some new warnings: uml_driver.c:1742: warning: format not a string literal and no format arguments uml_driver.c:1748: warning: format not a string literal and no format arguments uml_driver.c:1785: warning: format not a string literal and no format arguments openvz_conf.c:468: warning: format not a string literal and no format arguments But they were introduced before your changes. Here's the patch, on top of your changes. Note that as is, the patch does not apply to the latest in CVS, so my checking it in there would induce merge conflicts for you. So I'll hold off until after your 28-part series is committed. Jim >From fb32deb5dfb2a3c6b598de619f25c0bae4314a05 Mon Sep 17 00:00:00 2001 From: Jim Meyering <meyering(a)redhat.com> Date: Mon, 1 Dec 2008 11:17:08 +0100 Subject: [PATCH] avoid format string warnings * src/uml_driver.c (umlDomainBlockPeek): Put "%s" before each literal format string with no % directive. * src/openvz_conf.c (openvzGetNodeCPUs): Likewise. --- src/openvz_conf.c | 2 +- src/uml_driver.c | 6 +++--- 2 files changed, 4 insertions(+), 4 deletions(-) diff --git a/src/openvz_conf.c b/src/openvz_conf.c index fc4ef20..1361c53 100644 --- a/src/openvz_conf.c +++ b/src/openvz_conf.c @@ -466,7 +466,7 @@ openvzGetNodeCPUs(void) if (virNodeInfoPopulate(NULL, &nodeinfo) < 0) { openvzError(NULL, VIR_ERR_INTERNAL_ERROR, - _("Cound not read nodeinfo")); + "%s", _("Cound not read nodeinfo")); return 0; } diff --git a/src/uml_driver.c b/src/uml_driver.c index d2f2c7b..8d33c70 100644 --- a/src/uml_driver.c +++ b/src/uml_driver.c @@ -1740,13 +1740,13 @@ umlDomainBlockPeek (virDomainPtr dom, if (!vm) { umlReportError (dom->conn, dom, NULL, VIR_ERR_INVALID_DOMAIN, - _("no domain with matching uuid")); + "%s", _("no domain with matching uuid")); goto cleanup; } if (!path || path[0] == '\0') { umlReportError(dom->conn, dom, NULL, VIR_ERR_INVALID_ARG, - _("NULL or empty path")); + "%s", _("NULL or empty path")); goto cleanup; } @@ -1783,7 +1783,7 @@ umlDomainBlockPeek (virDomainPtr dom, ret = 0; } else { umlReportError (dom->conn, dom, NULL, VIR_ERR_INVALID_ARG, - _("invalid path")); + "%s", _("invalid path")); } cleanup: -- 1.6.0.4.1044.g77718

1 0

[libvirt] Memory leak in xen_inotify.c
by Niraj Tolia 30 Nov '08

30 Nov '08

Hi, It looks like xenInotifyOpen() in xen_inotify.c has an opendir() call. However, as there is no corresponding closedir() and opendir() uses malloc internally, valgrind reports this as a memory leak. Adding a closedir() to the end of this function should fix this. Cheers, Niraj

2 1

[libvirt] libvirt + xen 3.2.1 oddities
by Guido Günther 28 Nov '08

28 Nov '08

Hi, I just ran across these oddities when using a bit more libvirt+xen: 1.) virsh setmaxmem: On a running domain: # virsh setmaxmem domain 256000 completes but virsh dumpxml as well as the config.sxp still shows the old amount of memory. Looks as the set_maxmem hypercall simply gets ignored. xm mem-max works as expected. Smells like a bug in the ioctl? 2.) virsh list: Sometimes (didn't find a pattern yet) when shutting down a running domains and restarting it I'm seeing: Id Name State ---------------------------------- 0 Domain-0 running 2 foo idle libvir: Xen Daemon error : GET operation failed: xend_get: error from xen daemon: libvir: Xen Daemon error : GET operation failed: xend_get: error from xen daemon: libvir: Xen Daemon error : GET operation failed: xend_get: error from xen daemon: libvir: Xen Daemon error : GET operation failed: xend_get: error from xen daemon: 7 bar idle Note that the number of errors the corresponds to the number of shutdowns. VirXen_getdomaininfolist returns 7 in the above case. virDomainLookupByID later on fails for these "additional" domains. 3.) virsh list: Duplicate domains: Id Name State ---------------------------------- 0 Domain-0 running libvir: Xen Daemon error : GET operation failed: xend_get: error from xen daemon: libvir: Xen Daemon error : GET operation failed: xend_get: error from xen daemon: libvir: Xen Daemon error : GET operation failed: xend_get: error from xen daemon: 14 bar no state libvir: Xen Daemon error : GET operation failed: xend_get: error from xen daemon: 16 bar idle Domain 14 can't be shut down (xm list only lists domain 16). Could be a similar problem as the above. This is all libvirt 0.4.6 (but the code looks very similar in gurrent CVS) and xen-3.2.1 on Debian. The detected ABI is hypervisor call v2, sys ver6 dom ver5. And from a quick glance at the libxen-dev package the structs seem to match. Cheers, -- Guido

3 7

[libvirt] PATCH: Fix XM driver handling of disks without source file
by Daniel P. Berrange 28 Nov '08

28 Nov '08

It is possible for disks to be listed without a source file against them, eg a CDROM device with no media loaded. The XenD driver handles this, but the XM driver incorrectly generates XML with a <source file=''/> element instead of omitting the element entirely. This causes a bogus SXEXPR to be sent to XenD when starting the domain. This patch does three things - Makes the generic domain_conf.c XML parser accept XML docs with a bogus <source file=''/> and convert the source to NULL, instead of passing along the empty string "". This protects against broken apps - Makes the XM driver correctly generate XML in the first place, so omitting the <source> tag entirely. This is the root cause fix - Adds a test case for the XM driver to validate handling of devices with a source file src/domain_conf.c | 8 +++ src/xm_internal.c | 70 ++++++++++++++++------------ tests/xmconfigdata/test-no-source-cdrom.cfg | 23 +++++++++ tests/xmconfigdata/test-no-source-cdrom.xml | 46 ++++++++++++++++++ tests/xmconfigtest.c | 1 5 files changed, 120 insertions(+), 28 deletions(-) Daniel Index: src/xm_internal.c =================================================================== RCS file: /data/cvs/libvirt/src/xm_internal.c,v retrieving revision 1.102 diff -u -p -r1.102 xm_internal.c --- src/xm_internal.c 25 Nov 2008 11:18:08 -0000 1.102 +++ src/xm_internal.c 27 Nov 2008 12:14:57 -0000 @@ -828,7 +828,7 @@ xenXMDomainConfigParse(virConnectPtr con while (list) { char *head; char *offset; - char *tmp, *tmp1; + char *tmp; if ((list->type != VIR_CONF_STRING) || (list->str == NULL)) goto skipdisk; @@ -850,10 +850,15 @@ xenXMDomainConfigParse(virConnectPtr con goto skipdisk; if ((offset - head) >= (PATH_MAX-1)) goto skipdisk; - if (VIR_ALLOC_N(disk->src, (offset - head) + 1) < 0) - goto no_memory; - strncpy(disk->src, head, (offset - head)); - disk->src[(offset-head)] = '\0'; + + if (offset == head) { + disk->src = NULL; /* No source file given, eg CDROM with no media */ + } else { + if (VIR_ALLOC_N(disk->src, (offset - head) + 1) < 0) + goto no_memory; + strncpy(disk->src, head, (offset - head)); + disk->src[(offset-head)] = '\0'; + } head = offset + 1; /* Remove legacy ioemu: junk */ @@ -871,32 +876,41 @@ xenXMDomainConfigParse(virConnectPtr con /* Extract source driver type */ - if (disk->src && - (tmp = strchr(disk->src, ':')) != NULL) { - if (VIR_ALLOC_N(disk->driverName, (tmp - disk->src) + 1) < 0) - goto no_memory; - strncpy(disk->driverName, disk->src, (tmp - disk->src)); - disk->driverName[tmp - disk->src] = '\0'; - } else { - if (!(disk->driverName = strdup("phy"))) - goto no_memory; - tmp = disk->src; - } + if (disk->src) { + /* The main type phy:, file:, tap: ... */ + if ((tmp = strchr(disk->src, ':')) != NULL) { + if (VIR_ALLOC_N(disk->driverName, (tmp - disk->src) + 1) < 0) + goto no_memory; + strncpy(disk->driverName, disk->src, (tmp - disk->src)); + disk->driverName[tmp - disk->src] = '\0'; - /* And the source driver sub-type */ - if (STRPREFIX(disk->driverName, "tap")) { - if (!(tmp1 = strchr(tmp+1, ':')) || !tmp1[0]) - goto skipdisk; - if (VIR_ALLOC_N(disk->driverType, (tmp1-(tmp+1))) < 0) - goto no_memory; - strncpy(disk->driverType, tmp+1, (tmp1-(tmp+1))); - memmove(disk->src, disk->src+(tmp1-disk->src)+1, strlen(disk->src)-(tmp1-disk->src)); - } else { - disk->driverType = NULL; - if (disk->src[0] && tmp) - memmove(disk->src, disk->src+(tmp-disk->src)+1, strlen(disk->src)-(tmp-disk->src)); + /* Strip the prefix we found off the source file name */ + memmove(disk->src, disk->src+(tmp-disk->src)+1, + strlen(disk->src)-(tmp-disk->src)); + } + + /* And the sub-type for tap:XXX: type */ + if (disk->driverName && + STREQ(disk->driverName, "tap")) { + if (!(tmp = strchr(disk->src, ':'))) + goto skipdisk; + if (VIR_ALLOC_N(disk->driverType, (tmp - disk->src) + 1) < 0) + goto no_memory; + strncpy(disk->driverType, disk->src, (tmp - disk->src)); + disk->driverType[tmp - disk->src] = '\0'; + + /* Strip the prefix we found off the source file name */ + memmove(disk->src, disk->src+(tmp-disk->src)+1, + strlen(disk->src)-(tmp-disk->src)); + } } + /* No source, or driver name, so fix to phy: */ + if (!disk->driverName && + !(disk->driverName = strdup("phy"))) + goto no_memory; + + /* phy: type indicates a block device */ disk->type = STREQ(disk->driverName, "phy") ? VIR_DOMAIN_DISK_TYPE_BLOCK : VIR_DOMAIN_DISK_TYPE_FILE; Index: src/domain_conf.c =================================================================== RCS file: /data/cvs/libvirt/src/domain_conf.c,v retrieving revision 1.39 diff -u -p -r1.39 domain_conf.c --- src/domain_conf.c 21 Nov 2008 11:42:51 -0000 1.39 +++ src/domain_conf.c 27 Nov 2008 12:15:06 -0000 @@ -546,6 +546,14 @@ virDomainDiskDefParseXML(virConnectPtr c source = virXMLPropString(cur, "file"); else source = virXMLPropString(cur, "dev"); + + /* People sometimes pass a bogus '' source path + when they mean to omit the source element + completely. eg CDROM without media. This is + just a little compatability check to help + those broken apps */ + if (source && STREQ(source, "")) + VIR_FREE(source); } else if ((target == NULL) && (xmlStrEqual(cur->name, BAD_CAST "target"))) { target = virXMLPropString(cur, "dev"); Index: tests/xmconfigtest.c =================================================================== RCS file: /data/cvs/libvirt/tests/xmconfigtest.c,v retrieving revision 1.24 diff -u -p -r1.24 xmconfigtest.c --- tests/xmconfigtest.c 24 Nov 2008 19:23:39 -0000 1.24 +++ tests/xmconfigtest.c 27 Nov 2008 12:15:13 -0000 @@ -229,6 +229,7 @@ mymain(int argc, char **argv) DO_TEST("fullvirt-sound", 2); DO_TEST("escape-paths", 2); + DO_TEST("no-source-cdrom", 2); virCapabilitiesFree(caps); Index: tests/xmconfigdata/test-no-source-cdrom.cfg =================================================================== RCS file: tests/xmconfigdata/test-no-source-cdrom.cfg diff -N tests/xmconfigdata/test-no-source-cdrom.cfg --- /dev/null 1 Jan 1970 00:00:00 -0000 +++ tests/xmconfigdata/test-no-source-cdrom.cfg 27 Nov 2008 12:15:27 -0000 @@ -0,0 +1,23 @@ +name = "test" +uuid = "cc2315e7-d26a-307a-438c-6d188ec4c09c" +maxmem = 382 +memory = 350 +vcpus = 1 +builder = "hvm" +kernel = "/usr/lib/xen/boot/hvmloader" +boot = "c" +pae = 1 +acpi = 1 +apic = 1 +localtime = 0 +on_poweroff = "destroy" +on_reboot = "destroy" +on_crash = "destroy" +device_model = "/usr/lib/xen/bin/qemu-dm" +sdl = 0 +vnc = 1 +vncunused = 1 +disk = [ "phy:/dev/sda8,hda,w", ",hdc:cdrom,r" ] +vif = [ "mac=00:16:3e:0a:7b:39,bridge=xenbr0,type=ioemu" ] +parallel = "none" +serial = "pty" Index: tests/xmconfigdata/test-no-source-cdrom.xml =================================================================== RCS file: tests/xmconfigdata/test-no-source-cdrom.xml diff -N tests/xmconfigdata/test-no-source-cdrom.xml --- /dev/null 1 Jan 1970 00:00:00 -0000 +++ tests/xmconfigdata/test-no-source-cdrom.xml 27 Nov 2008 12:15:27 -0000 @@ -0,0 +1,46 @@ +<domain type='xen'> + <name>test</name> + <uuid>cc2315e7-d26a-307a-438c-6d188ec4c09c</uuid> + <memory>391168</memory> + <currentMemory>358400</currentMemory> + <vcpu>1</vcpu> + <os> + <type arch='i686' machine='xenfv'>hvm</type> + <loader>/usr/lib/xen/boot/hvmloader</loader> + <boot dev='hd'/> + </os> + <features> + <acpi/> + <apic/> + <pae/> + </features> + <clock offset='utc'/> + <on_poweroff>destroy</on_poweroff> + <on_reboot>destroy</on_reboot> + <on_crash>destroy</on_crash> + <devices> + <emulator>/usr/lib/xen/bin/qemu-dm</emulator> + <disk type='block' device='disk'> + <driver name='phy'/> + <source dev='/dev/sda8'/> + <target dev='hda' bus='ide'/> + </disk> + <disk type='block' device='cdrom'> + <driver name='phy'/> + <target dev='hdc' bus='ide'/> + <readonly/> + </disk> + <interface type='bridge'> + <mac address='00:16:3e:0a:7b:39'/> + <source bridge='xenbr0'/> + </interface> + <serial type='pty'> + <target port='0'/> + </serial> + <console type='pty'> + <target port='0'/> + </console> + <input type='mouse' bus='ps2'/> + <graphics type='vnc' port='-1' autoport='yes'/> + </devices> +</domain> -- |: 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] PATCH: Fix QEMU/UML use of libnuma
by Daniel P. Berrange 28 Nov '08

28 Nov '08

The 2.x series of numactl releases changed the ABI/API for certain libnuma.so functions we use. Fortunately it turns out that they provide back-compatability of both ABI and API with a combo of linker script magic, and header file inline compat functions. So, all we need todo is #define NUMA_VERSION1_COMPATIBILITY before include numa.h in our code. This new code did, however, expose a bug in our use of the existing API. We were calculating size in bytes, instead of size in longs, so were passing a buffer that was 8 times to large. Harmless, but the new libnuma validates that the buffer is expected size. So this patch also fixes us to pass & allocate correct buffer size. Finally, also add a missing BuildRequire on numactl-devel, and fixes the libvirtd automake build dependancy Daniel Index: libvirt.spec.in =================================================================== RCS file: /data/cvs/libvirt/libvirt.spec.in,v retrieving revision 1.103 diff -u -p -r1.103 libvirt.spec.in --- libvirt.spec.in 26 Nov 2008 14:46:49 -0000 1.103 +++ libvirt.spec.in 27 Nov 2008 13:20:24 -0000 @@ -124,6 +124,8 @@ BuildRequires: lvm2 BuildRequires: iscsi-initiator-utils # For disk driver BuildRequires: parted-devel +# For QEMU/LXC numa info +BuildRequires: numactl-devel Obsoletes: libvir # Fedora build root suckage Index: qemud/Makefile.am =================================================================== RCS file: /data/cvs/libvirt/qemud/Makefile.am,v retrieving revision 1.61 diff -u -p -r1.61 Makefile.am --- qemud/Makefile.am 21 Nov 2008 12:27:11 -0000 1.61 +++ qemud/Makefile.am 27 Nov 2008 13:20:24 -0000 @@ -87,7 +87,6 @@ libvirtd_LDFLAGS = \ $(COVERAGE_LDFLAGS) \ $(POLKIT_LIBS) -libvirtd_DEPENDENCIES = ../src/libvirt.la libvirtd_LDADD = \ ../gnulib/lib/libgnu.la @@ -129,6 +128,9 @@ libvirtd_CFLAGS += $(AVAHI_CFLAGS) libvirtd_LDADD += $(AVAHI_LIBS) endif +libvirtd_DEPENDENCIES = $(libvirtd_LDADD) + + default_xml_dest = libvirt/qemu/networks/default.xml install-data-local: install-init install-data-sasl install-data-polkit mkdir -p $(DESTDIR)$(sysconfdir)/libvirt/qemu/networks/autostart Index: src/qemu_conf.c =================================================================== RCS file: /data/cvs/libvirt/src/qemu_conf.c,v retrieving revision 1.108 diff -u -p -r1.108 qemu_conf.c --- src/qemu_conf.c 19 Nov 2008 16:58:23 -0000 1.108 +++ src/qemu_conf.c 27 Nov 2008 13:20:24 -0000 @@ -37,6 +37,7 @@ #include <sys/utsname.h> #if HAVE_NUMACTL +#define NUMA_VERSION1_COMPATIBILITY 1 #include <numa.h> #endif @@ -300,11 +301,11 @@ qemudCapsInitGuest(virCapsPtr caps, #if HAVE_NUMACTL #define MAX_CPUS 4096 #define MAX_CPUS_MASK_SIZE (sizeof(unsigned long)) -#define MAX_CPUS_MASK_LEN (MAX_CPUS / MAX_CPUS_MASK_SIZE) -#define MAX_CPUS_MASK_BYTES (MAX_CPUS / 8) +#define MAX_CPUS_MASK_BITS (MAX_CPUS_MASK_SIZE * 8) +#define MAX_CPUS_MASK_LEN (MAX_CPUS / (MAX_CPUS_MASK_BITS)) #define MASK_CPU_ISSET(mask, cpu) \ - (((mask)[((cpu) / MAX_CPUS_MASK_SIZE)] >> ((cpu) % MAX_CPUS_MASK_SIZE)) & 1) + (((mask)[((cpu) / MAX_CPUS_MASK_BITS)] >> ((cpu) % MAX_CPUS_MASK_BITS)) & 1) static int qemudCapsInitNUMA(virCapsPtr caps) @@ -322,7 +323,8 @@ qemudCapsInitNUMA(virCapsPtr caps) goto cleanup; for (n = 0 ; n <= numa_max_node() ; n++) { - if (numa_node_to_cpus(n, mask, MAX_CPUS_MASK_BYTES) < 0) + + if (numa_node_to_cpus(n, mask, MAX_CPUS_MASK_LEN) < 0) goto cleanup; for (ncpus = 0, i = 0 ; i < MAX_CPUS ; i++) Index: src/qemu_driver.c =================================================================== RCS file: /data/cvs/libvirt/src/qemu_driver.c,v retrieving revision 1.160 diff -u -p -r1.160 qemu_driver.c --- src/qemu_driver.c 21 Nov 2008 12:16:08 -0000 1.160 +++ src/qemu_driver.c 27 Nov 2008 13:20:25 -0000 @@ -45,6 +45,7 @@ #include <sys/ioctl.h> #if HAVE_NUMACTL +#define NUMA_VERSION1_COMPATIBILITY 1 #include <numa.h> #endif Index: src/uml_conf.c =================================================================== RCS file: /data/cvs/libvirt/src/uml_conf.c,v retrieving revision 1.1 diff -u -p -r1.1 uml_conf.c --- src/uml_conf.c 19 Nov 2008 16:58:23 -0000 1.1 +++ src/uml_conf.c 27 Nov 2008 13:20:25 -0000 @@ -37,6 +37,7 @@ #include <sys/utsname.h> #if HAVE_NUMACTL +#define NUMA_VERSION1_COMPATIBILITY 1 #include <numa.h> #endif @@ -56,11 +57,11 @@ #if HAVE_NUMACTL #define MAX_CPUS 4096 #define MAX_CPUS_MASK_SIZE (sizeof(unsigned long)) -#define MAX_CPUS_MASK_LEN (MAX_CPUS / MAX_CPUS_MASK_SIZE) -#define MAX_CPUS_MASK_BYTES (MAX_CPUS / 8) +#define MAX_CPUS_MASK_BITS (MAX_CPUS_MASK_SIZE * 8) +#define MAX_CPUS_MASK_LEN (MAX_CPUS / (MAX_CPUS_MASK_BITS)) #define MASK_CPU_ISSET(mask, cpu) \ - (((mask)[((cpu) / MAX_CPUS_MASK_SIZE)] >> ((cpu) % MAX_CPUS_MASK_SIZE)) & 1) + (((mask)[((cpu) / MAX_CPUS_MASK_BITS)] >> ((cpu) % MAX_CPUS_MASK_BITS)) & 1) static int umlCapsInitNUMA(virCapsPtr caps) @@ -78,7 +79,8 @@ umlCapsInitNUMA(virCapsPtr caps) goto cleanup; for (n = 0 ; n <= numa_max_node() ; n++) { - if (numa_node_to_cpus(n, mask, MAX_CPUS_MASK_BYTES) < 0) + + if (numa_node_to_cpus(n, mask, MAX_CPUS_MASK_LEN) < 0) goto cleanup; for (ncpus = 0, i = 0 ; i < MAX_CPUS ; i++) Index: src/uml_driver.c =================================================================== RCS file: /data/cvs/libvirt/src/uml_driver.c,v retrieving revision 1.2 diff -u -p -r1.2 uml_driver.c --- src/uml_driver.c 21 Nov 2008 10:06:28 -0000 1.2 +++ src/uml_driver.c 27 Nov 2008 13:20:25 -0000 @@ -46,6 +46,7 @@ #include <sys/inotify.h> #if HAVE_NUMACTL +#define NUMA_VERSION1_COMPATIBILITY 1 #include <numa.h> #endif -- |: 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 3

[libvirt] Hide fully-virtualized network devices to domU with PV drivers
by Iain Watson 28 Nov '08

28 Nov '08

Hello, I have a problem with libvirt which I hope someone on this list might be able to help with. I have many DomU images which all have PV drivers loaded for network. I want to start these DomU guests with only the PV network device being visible from within the DomU. In testing I have been able to create the required system with the 'xm' command. The important part of the network line is "type=none". The complete line looks like: vif = [ "mac=00:16:3e:00:a5:57,bridge=eth0,script=vif-bridge,type=none" ] With libvirt I have tried adding "<model type='none'/>" into the interface section but this errors with "qemu: Unsupported NIC: none" so I'm obviously missing something obvious. The complete interface section looks like: <interface type='bridge'> <mac address='00:16:3e:00:a5:57'/> <source bridge='eth0'/> <target dev='vif9.0'/> <model type='none'/> </interface> I have been able to achieve the required configuration if I completely omit an 'interface' section and once the DomU has booted I use virsh to add an interface to the domain. Unfortunately if I then reboot the DomU the guest comes back up with two network interfaces visible to it and the qemu-dm line has added a "-net ...,model=rtl8139" option. Having hunted around I am yet to find a way to configure a domU in libvirt so the networking available to the guest is the way I require. As I can do this with xm, I'm hoping someone on this list can point me in the right direction to also do this with libvirt. Thanks in advance, Iain Watson

1 0