- Devel - libvirt List Archives

Re: [libvirt] [PATCH] libxl: fix domxml-to-native wrong output for qcow2 format
by Bamvor Jian Zhang 13 Jun '14

13 Jun '14

>>>Jim Fehlig wrote: > Bamvor Jian Zhang<bjzhang(a)suse.com> wrote: >> e.g. for these following disk configuration in libvirt. >> <disk type='file' device='disk'> > > For Xen, using <driver name='qemu' .../> (which means qdisk) is only > supported by the libxl driver. The old xm/xend stack does not support qdisk. > >> <driver name='qemu' type='qcow2'/> >> <source file='/var/lib/xen/images/001/disk0.qcow2'/> >> <target dev='hdc'/> >> >> without this patch, it will be >> "qemu:/var/lib/xen/images/001/disk0.qcow2,hdc,w" > > Yeah, that won't work with xend or libxl :) . > >> but it should be(if with this patch) >> "qcow2:/var/lib/xen/images/001/disk0.qcow2,hdc,w" >> >> Signed-off-by: Bamvor Jian Zhang <bjzhang(a)suse.com> >> --- >> src/xenxs/xen_xm.c | 5 ++++- >> 1 file changed, 4 insertions(+), 1 deletion(-) >> >> diff --git a/src/xenxs/xen_xm.c b/src/xenxs/xen_xm.c >> index b2db97d..29835b4 100644 >> --- a/src/xenxs/xen_xm.c >> +++ b/src/xenxs/xen_xm.c >> @@ -1209,7 +1209,10 @@ xenFormatXMDisk(virConfValuePtr list, >> type = "aio"; >> else >> type = virStorageFileFormatTypeToString(format); >> - virBufferAsprintf(&buf, "%s:", driver); >> + if (!STREQ(type, "qcow2")) >> + virBufferAsprintf(&buf, "%s:", driver); >> + else >> + virBufferAsprintf(&buf, "%s:", type); > > But this only handles the specific case of <driver name='qemu' > type='qcow2'/>. > With <driver name='qemu' type='raw'/>, I get > > "qemu:/path/to/disk,hdc,w" > > Unfortunately, this is a case where we are trying to use the xm config > parser to parse something that is xl-specific. With <driver > name='qemu'.../>, > the corresponding xl disk config would look something like > > disk = ['backendtype=qdisk,/var/lib/xen/images/001/disk0.qcow2,hdc,w'] > > With the type also specified, e.g. <driver name='qemu' type='qcow2'/>, the > xl > disk config would be > > disk = > ['backendtype=qdisk,format=qcow2,/var/lib/xen/images/001/disk0.qcow2,hdc,w'] > > As I see it, the options are > > 1. start working on a xen-xl parser > 2. map 'qemu' to 'tap' > > We'll eventually need 1 anyhow. I haven't looked to see how much work that > would be. Certainly a lot of the xm parsing code could be used since xm > config is a subset of xl config. understand. the main difference between xl and xm format is disk and network. and it worth to do it. > > Option 2 is really a no-op in the libxl stack, where qdisk is used in place > of tap anyhow. One could argue that mapping qemu to tap isn't too insane in > the xm/xend stack either, since much of the blktap userspace code was > originally based on qemu. > > The below patch works for me. Can you give it a try? I'd like to hear wha > others think about taking the easy way out with option 2. the following patch works for me. regards bamvor > > Regards, > Jim > > > From 92c476cbe47009c43ebb4a3076a17347c8eea238 Mon Sep 17 00:00:00 2001 > From: Jim Fehlig <jfehlig(a)suse.com> > Date: Thu, 12 Jun 2014 15:28:48 -0600 > Subject: [PATCH] libxl: fix domxml-to-native with qemu disk driver > > Disk config containing <driver name='qemu'.../> is converted to > the native config containg "qemu:/path/to/disk", which is not > understood by xm or xl. > > This patch maps 'qemu' to 'tap'. In xl, tap is mapped to the qemu > backend (aka qdisk), making this change essentially a no-op. In > the xm stack, one could argue that mapping qemu to tap isn't too > insane, since much of the blktap userspace code was originally > based on qemu. > > While at it, noticed that 'driver' wasn't being honored in > xenFormatXMDisk, so change the logic a bit to honor a > user-specified driver. > > Signed-off-by: Jim Fehlig <jfehlig(a)suse.com> > --- > src/xenxs/xen_xm.c | 29 ++++++++++++++++++++--------- > 1 file changed, 20 insertions(+), 9 deletions(-) > > diff --git a/src/xenxs/xen_xm.c b/src/xenxs/xen_xm.c > index b2db97d..ebd525b 100644 > --- a/src/xenxs/xen_xm.c > +++ b/src/xenxs/xen_xm.c > @@ -1201,17 +1201,28 @@ xenFormatXMDisk(virConfValuePtr list, > int format = virDomainDiskGetFormat(disk); > const char *driver = virDomainDiskGetDriver(disk); > > - if (src) { > - if (format) { > - const char *type; > + /* > + * In pre-libxl Xen, 'tap' provided the qemu driver functionality. > + * In libxl Xen, qemu (aka qdisk) is in fact used. For backwards > + * compatibility, tap == qdisk in libxl Xen, so it is safe to use > + * 'tap' in libxl tool. > + */ > + if (STREQ(driver, "qemu")) > + driver = "tap"; > > - if (format == VIR_STORAGE_FILE_RAW) > - type = "aio"; > - else > - type = virStorageFileFormatTypeToString(format); > + if (src) { > + if (driver) { > virBufferAsprintf(&buf, "%s:", driver); > - if (STREQ(driver, "tap")) > - virBufferAsprintf(&buf, "%s:", type); > + if (format) { > + const char *type; > + > + if (format == VIR_STORAGE_FILE_RAW) > + type = "aio"; > + else > + type = virStorageFileFormatTypeToString(format); > + if (STREQ(driver, "tap")) > + virBufferAsprintf(&buf, "%s:", type); > + } > } else { > switch (virDomainDiskGetType(disk)) { > case VIR_STORAGE_TYPE_FILE: > -- 1.8.4.5

1 0

[libvirt] libvirt issue
by Zvi Dubitzky 13 Jun '14

13 Jun '14

Hi I am looking for libvirt user mailing list (Red Hat ) but could not find one . So I am using this list (or maybe someone can direct me to such a list). The problem: using libvirt under openstack I encounter a problem when launching a VM with multiple partitions (2 G image) While a single partition can be launched successfully . We are using in libvirtd.conf log_level=1 and log_outputs="1:file:/var/log/libvirt/libvirtd.log" we can see that in /var/log/libvirt/libvirtd.log appear the following error message : 2014-06-11 14:47:26.166+0000: 12971: error : qemuMonitorIO:614 : internal error End of file from monitor ( file attached below) >From the internet we could not figure out a solution to the problem The host has enough resources and the XML of the VM is : The libvirt version is: 0.10.2 The qemu-kvm version: 0.12.1.2 Any idea what is causing this error and how to overcome ? Thx Zvi Dubitzky Email:dubi@il.ibm.com

2 1

[libvirt] [PATCH v3 0/5] Active commit
by Eric Blake 13 Jun '14

13 Jun '14

I still don't have qemu capability detection working reliably, but want to post this series right now so that it can be built into a scratch build containing Peter's and my changes. (Or put another way, I was testing what conflict resolutions are required - patch 2/5 (virsh) and 5/5 (qemu_driver) has some conflicts with Peter's addition of relative backing name; and I think the resolutions were fairly simple). These patches are tested on top of: https://www.redhat.com/archives/libvir-list/2014-June/msg00492.html I may still end up posting a v4 and/or pushing my series before Peter's, once I get capability detection working the way I want. Eric Blake (5): virsh: improve blockcopy UI virsh: expose new active commit controls blockcommit: update error messages related to block jobs blockcommit: track job type in xml blockcommit: turn on active commit docs/formatdomain.html.in | 20 +++--- docs/schemas/domaincommon.rng | 6 ++ src/conf/domain_conf.c | 26 ++++++- src/conf/domain_conf.h | 1 + src/qemu/qemu_driver.c | 56 ++++++++++++--- src/qemu/qemu_hotplug.c | 2 +- src/qemu/qemu_process.c | 18 +++-- .../qemuxml2argv-disk-active-commit.xml | 37 ++++++++++ .../qemuxml2argvdata/qemuxml2argv-disk-mirror.xml | 4 +- .../qemuxml2xmlout-disk-mirror-old.xml | 4 +- tests/qemuxml2xmltest.c | 1 + tools/virsh-domain.c | 83 +++++++++++++++++----- tools/virsh.pod | 53 +++++++++----- 13 files changed, 242 insertions(+), 69 deletions(-) create mode 100644 tests/qemuxml2argvdata/qemuxml2argv-disk-active-commit.xml -- 1.9.3

3 18

Re: [libvirt] [PATCH] libxl: fix domxml-to-native wrong output for qcow2 format
by Jim Fehlig 13 Jun '14

13 Jun '14

Bamvor Jian Zhang<bjzhang(a)suse.com> wrote: > e.g. for these following disk configuration in libvirt. > <disk type='file' device='disk'> For Xen, using <driver name='qemu' .../> (which means qdisk) is only supported by the libxl driver. The old xm/xend stack does not support qdisk. > <driver name='qemu' type='qcow2'/> > <source file='/var/lib/xen/images/001/disk0.qcow2'/> > <target dev='hdc'/> > > without this patch, it will be > "qemu:/var/lib/xen/images/001/disk0.qcow2,hdc,w" Yeah, that won't work with xend or libxl :) . > but it should be(if with this patch) > "qcow2:/var/lib/xen/images/001/disk0.qcow2,hdc,w" > > Signed-off-by: Bamvor Jian Zhang <bjzhang(a)suse.com> > --- > src/xenxs/xen_xm.c | 5 ++++- > 1 file changed, 4 insertions(+), 1 deletion(-) > > diff --git a/src/xenxs/xen_xm.c b/src/xenxs/xen_xm.c > index b2db97d..29835b4 100644 > --- a/src/xenxs/xen_xm.c > +++ b/src/xenxs/xen_xm.c > @@ -1209,7 +1209,10 @@ xenFormatXMDisk(virConfValuePtr list, > type = "aio"; > else > type = virStorageFileFormatTypeToString(format); > - virBufferAsprintf(&buf, "%s:", driver); > + if (!STREQ(type, "qcow2")) > + virBufferAsprintf(&buf, "%s:", driver); > + else > + virBufferAsprintf(&buf, "%s:", type); But this only handles the specific case of <driver name='qemu' type='qcow2'/>. With <driver name='qemu' type='raw'/>, I get "qemu:/path/to/disk,hdc,w" Unfortunately, this is a case where we are trying to use the xm config parser to parse something that is xl-specific. With <driver name='qemu'.../>, the corresponding xl disk config would look something like disk = ['backendtype=qdisk,/var/lib/xen/images/001/disk0.qcow2,hdc,w'] With the type also specified, e.g. <driver name='qemu' type='qcow2'/>, the xl disk config would be disk = ['backendtype=qdisk,format=qcow2,/var/lib/xen/images/001/disk0.qcow2,hdc,w'] As I see it, the options are 1. start working on a xen-xl parser 2. map 'qemu' to 'tap' We'll eventually need 1 anyhow. I haven't looked to see how much work that would be. Certainly a lot of the xm parsing code could be used since xm config is a subset of xl config. Option 2 is really a no-op in the libxl stack, where qdisk is used in place of tap anyhow. One could argue that mapping qemu to tap isn't too insane in the xm/xend stack either, since much of the blktap userspace code was originally based on qemu. The below patch works for me. Can you give it a try? I'd like to hear wha others think about taking the easy way out with option 2. Regards, Jim >From 92c476cbe47009c43ebb4a3076a17347c8eea238 Mon Sep 17 00:00:00 2001 From: Jim Fehlig <jfehlig(a)suse.com> Date: Thu, 12 Jun 2014 15:28:48 -0600 Subject: [PATCH] libxl: fix domxml-to-native with qemu disk driver Disk config containing <driver name='qemu'.../> is converted to the native config containg "qemu:/path/to/disk", which is not understood by xm or xl. This patch maps 'qemu' to 'tap'. In xl, tap is mapped to the qemu backend (aka qdisk), making this change essentially a no-op. In the xm stack, one could argue that mapping qemu to tap isn't too insane, since much of the blktap userspace code was originally based on qemu. While at it, noticed that 'driver' wasn't being honored in xenFormatXMDisk, so change the logic a bit to honor a user-specified driver. Signed-off-by: Jim Fehlig <jfehlig(a)suse.com> --- src/xenxs/xen_xm.c | 29 ++++++++++++++++++++--------- 1 file changed, 20 insertions(+), 9 deletions(-) diff --git a/src/xenxs/xen_xm.c b/src/xenxs/xen_xm.c index b2db97d..ebd525b 100644 --- a/src/xenxs/xen_xm.c +++ b/src/xenxs/xen_xm.c @@ -1201,17 +1201,28 @@ xenFormatXMDisk(virConfValuePtr list, int format = virDomainDiskGetFormat(disk); const char *driver = virDomainDiskGetDriver(disk); - if (src) { - if (format) { - const char *type; + /* + * In pre-libxl Xen, 'tap' provided the qemu driver functionality. + * In libxl Xen, qemu (aka qdisk) is in fact used. For backwards + * compatibility, tap == qdisk in libxl Xen, so it is safe to use + * 'tap' in libxl tool. + */ + if (STREQ(driver, "qemu")) + driver = "tap"; - if (format == VIR_STORAGE_FILE_RAW) - type = "aio"; - else - type = virStorageFileFormatTypeToString(format); + if (src) { + if (driver) { virBufferAsprintf(&buf, "%s:", driver); - if (STREQ(driver, "tap")) - virBufferAsprintf(&buf, "%s:", type); + if (format) { + const char *type; + + if (format == VIR_STORAGE_FILE_RAW) + type = "aio"; + else + type = virStorageFileFormatTypeToString(format); + if (STREQ(driver, "tap")) + virBufferAsprintf(&buf, "%s:", type); + } } else { switch (virDomainDiskGetType(disk)) { case VIR_STORAGE_TYPE_FILE: -- 1.8.4.5

1 0

[libvirt] Qemu guest agent to install RPMs in guest VM from host machine
by Puneet Bakshi 12 Jun '14

12 Jun '14

Hi, I want to be able to install RPM packages (available in host system at some path) to the guest VM and want this facility to be available as a tool. I am thinking of having a gemu guest agent (qemu-ga) running inside guest VM. I did not find any available command ("virsh qemu-agent-command <guest_vm> ...") which can do the same. I am planning to implement a command in qemu guest agent, which I can invoke from virsh like below. "virsh qemu-agent-command vm_01 \ '{"execute":"guest-rpm-install", \ "arguments":{"path":"/usr/local/bin/ABC.rpm"}} I am able to pass arguments from host to guest VM but how am I supposed to pass the whole RPM image from host to guest (which the guest agent can receive and install)? Regards, ~Puneet

3 8

[libvirt] [PATCH 0/2] Expose PCI Express capabilities
by Michal Privoznik 12 Jun '14

12 Jun '14

*** BLURB HERE *** Michal Privoznik (2): virpci: Introduce virPCIDeviceIsPCIExpress and friends nodedev: Introduce <pci-express/> to PCI devices docs/formatnode.html.in | 19 ++++ docs/schemas/nodedev.rng | 26 +++++ src/conf/node_device_conf.c | 123 ++++++++++++++++++++- src/conf/node_device_conf.h | 31 +++++- src/libvirt_private.syms | 3 + src/node_device/node_device_udev.c | 31 ++++++ src/util/virpci.c | 81 +++++++++++++- src/util/virpci.h | 8 ++ .../pci_8086_4238_pcie_wireless.xml | 26 +++++ tests/nodedevxml2xmltest.c | 1 + 10 files changed, 345 insertions(+), 4 deletions(-) create mode 100644 tests/nodedevschemadata/pci_8086_4238_pcie_wireless.xml -- 2.0.0

2 6

[libvirt] [PATCH] blockcommit: fix regression with explicit top argument
by Eric Blake 12 Jun '14

12 Jun '14

Commit f586965 accidentally changed the semantics of the virDomainBlockCommit command; where it previously looked for an explicit top argument from the top of the chain, it now starts from the backing file of the top. Of course, until we allow active commits, the only difference it makes is in the quality of the error message, but with code for active commit coming soon, we need to support an explicit mention of the active layer. * src/qemu/qemu_driver.c (qemuDomainBlockCommit): Start looking from top of chain. Signed-off-by: Eric Blake <eblake(a)redhat.com> --- src/qemu/qemu_driver.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c index ec72d9d..bf55f67 100644 --- a/src/qemu/qemu_driver.c +++ b/src/qemu/qemu_driver.c @@ -15528,8 +15528,7 @@ qemuDomainBlockCommit(virDomainPtr dom, if (!top) topSource = disk->src; else if (virStorageFileParseChainIndex(disk->dst, top, &topIndex) < 0 || - !(topSource = virStorageFileChainLookup(disk->src, - disk->src->backingStore, + !(topSource = virStorageFileChainLookup(disk->src, NULL, top, topIndex, &top_parent))) goto endjob; -- 1.9.3

2 2

[libvirt] [PATCH] virNodeDevCapPCIDevParseXML: Initialize numa_node variable
by Michal Privoznik 12 Jun '14

12 Jun '14

With one of my recent patches (1c70277) libvirt's capable of reporting NUMA node locality for PCI devices. The node ID is stored in pci_dev.numa_node variable. However, since zero is valid NUMA node ID, the default is -1 as it is in kernel too. So, if the PCI device is not tied to any specific NUMA node, the default is then NOT printed into XML. Therefore, when parsing node device XML, the <node/> element is optional. But currently, if it's not there, we must set sane default, otherwise after parsing the in memory representation doesn't match the XML. We are already doing this in other place: udevProcessPCI(). Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com> --- src/conf/node_device_conf.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/src/conf/node_device_conf.c b/src/conf/node_device_conf.c index 6153aa1..12e40e6 100644 --- a/src/conf/node_device_conf.c +++ b/src/conf/node_device_conf.c @@ -1145,6 +1145,8 @@ virNodeDevCapPCIDevParseXML(xmlXPathContextPtr ctxt, } } + /* The default value is -1 since zero is valid NUMA node number */ + data->pci_dev.numa_node = -1; if (virNodeDevCapsDefParseIntOptional("number(./numa[1]/@node)", ctxt, &data->pci_dev.numa_node, def, _("invalid NUMA node ID supplied for '%s'")) < 0) -- 1.8.5.5

2 1

[libvirt] [PATCHv4 0/4] virsh: Clean unsigned value parsing
by Peter Krempa 12 Jun '14

12 Jun '14

Jincheng Miao (1): virsh: forbid negative vcpu argument to vcpupin Peter Krempa (3): virsh: Reject negative numbers in vshCommandOptUInt virsh: Reject negative numbers in vshCommandOptUL virsh: Reject negative numbers in vshCommandOptULongLong tests/vcpupin | 29 ++++++++++- tools/virsh-domain.c | 39 ++++++++------- tools/virsh-volume.c | 8 +-- tools/virsh.c | 137 +++++++++++++++++++++++++++++++++++++++++---------- tools/virsh.h | 9 ++++ 5 files changed, 172 insertions(+), 50 deletions(-) -- 1.9.3

2 6

[libvirt] [PATCH 0/3] Resolve Coverity warnings
by John Ferlan 12 Jun '14

12 Jun '14

Resolve the "lower hanging fruit" Coverity issues from recent commits. Still left outstanding is rework of virVBoxSnapshotConfAllChildren() code and callers. John Ferlan (3): vbox_temp: Resolve Coverity warnings vbox_snapshot_conf: Resolve Coverity warnings libxl: Resolve Coverity warnings src/libxl/libxl_migration.c | 6 ++---- src/vbox/vbox_snapshot_conf.c | 27 ++++++++++++++++++--------- src/vbox/vbox_tmpl.c | 10 +++++++++- 3 files changed, 29 insertions(+), 14 deletions(-) -- 1.9.3

3 8