February 2018 - Devel - libvirt List Archives

[libvirt] [PATCH v3 00/12] qemu: Validate PCI controller options
by Andrea Bolognani 02 Mar '18

02 Mar '18

Applies cleanly on top of f565321b26dfb78c1dc2d6cf5456b9b80f04c7f3. Changes from [v2]: * replace the old implementation bit by bit using a clever trick suggested by pkrempa; * don't move QEMU capability validation; * add a default: label to all switch statements as recommended by danpb. Changes from [v1]: * error out instead of silently accept invalid options; * shave quite a lot of yaks. [v2] https://www.redhat.com/archives/libvir-list/2018-February/msg00813.html [v1] https://www.redhat.com/archives/libvir-list/2018-February/msg00244.html Andrea Bolognani (12): tests: Add some tests for PCI controller options qemu: Create new qemuDomainDeviceDefValidateControllerPCI() qemu: Validate PCI controller options (modelName) qemu: Validate PCI controller options (index) qemu: Validate PCI controller options (targetIndex) qemu: Validate PCI controller options (pcihole64) qemu: Validate PCI controller options (busNr) qemu: Validate PCI controller options (numaNode) qemu: Validate PCI controller options (chassisNr) qemu: Validate PCI controller options (chassis and port) qemu: Validate PCI controllers (QEMU capabilities) qemu: Remove old qemuDomainDeviceDefValidateControllerPCI() src/qemu/qemu_domain.c | 594 +++++++++++++++------ .../i440fx-controllers-pciopts.xml | 36 ++ tests/qemuxml2argvdata/pcie-expander-bus.xml | 3 - .../pseries-controllers-pciopts.xml | 35 ++ tests/qemuxml2argvdata/q35-controllers-pciopts.xml | 60 +++ tests/qemuxml2argvtest.c | 4 + tests/qemuxml2xmloutdata/pcie-expander-bus.xml | 4 +- 7 files changed, 575 insertions(+), 161 deletions(-) create mode 100644 tests/qemuxml2argvdata/i440fx-controllers-pciopts.xml create mode 100644 tests/qemuxml2argvdata/pseries-controllers-pciopts.xml create mode 100644 tests/qemuxml2argvdata/q35-controllers-pciopts.xml -- 2.14.3

2 20

[libvirt] [PATCH v2 0/2] guest crash information on S390
by Bjoern Walk 02 Mar '18

02 Mar '18

v1 -> v2: * refactored error paths according to John and Jiri * added entry to news.xml Bjoern Walk (2): qemu: log the crash information for S390 news: add logging of guest crash information on S390 docs/news.xml | 9 +++++++++ src/qemu/qemu_monitor.c | 19 ++++++++++++++++++- src/qemu/qemu_monitor.h | 12 ++++++++++++ src/qemu/qemu_monitor_json.c | 38 ++++++++++++++++++++++++++++++++++++++ 4 files changed, 77 insertions(+), 1 deletion(-) -- 2.13.4

2 5

[libvirt] [tck PATCH 0/2] set CTRL_IP_LEARNING and DHCPSERVER in filter during no-ip-spoofing test
by Laine Stump 02 Mar '18

02 Mar '18

We've recently discovered two separate bugs that caused libvirt's "DHCP Snooping" feature to not work: https://bugzilla.redhat.com/show_bug.cgi?id=1529338 - libvirt regression https://bugzilla.redhat.com/show_bug.cgi?id=1547237 - libpcap regression Since we didn't have any test suite covering that code, we had the embarrassment of learning of it from someone else's QE (RHV/oVirt QE at Red Hat). This series adds the necessary stuff to the test domain config of libvirt-tck's "no-ip-spoofing" test to exercise the DHCPSnoop thread. (There may be a much better way of dealing with a hash-inside-a-hash; I am an imbecile at perl, and arrived at this code by trial, error, and google searches). Laine Stump (2): new helper function get_network_ip() set CTRL_IP_LEARNING and DHCPSERVER in filter during no-ip-spoofing test lib/Sys/Virt/TCK.pm | 11 ++++++++--- lib/Sys/Virt/TCK/DomainBuilder.pm | 8 +++++++- lib/Sys/Virt/TCK/NetworkHelpers.pm | 10 ++++++++++ scripts/nwfilter/220-no-ip-spoofing.t | 9 ++++++++- 4 files changed, 33 insertions(+), 5 deletions(-) -- 2.14.3

1 3

[libvirt] [PATCH] virtio-scsi: Add virtqueue_size parameter for qemu
by Klim Kireev 01 Mar '18

01 Mar '18

In QEMU commit 5c0919d02066c3d0eb896c33265ad90101a6a84a adds new option virtqueue_size. This Patch allows this option to be set in libvirt. --- docs/formatdomain.html.in | 6 ++++++ src/conf/domain_conf.c | 12 ++++++++++++ src/conf/domain_conf.h | 1 + src/qemu/qemu_command.c | 3 +++ 4 files changed, 22 insertions(+) diff --git a/docs/formatdomain.html.in b/docs/formatdomain.html.in index 6fd2189cd..2f33b5e70 100644 --- a/docs/formatdomain.html.in +++ b/docs/formatdomain.html.in @@ -3867,6 +3867,12 @@ host. <span class="since">Since 1.2.7 (QEMU and KVM only)</span> </dd> + <dt><code>virtqueue_size</code></dt> + <dd> + The optional <code>virtqueue_size</code> attribute specifies the size of + virtio queue. + <span class="since">Since 3.7.0 (QEMU and KVM only)</span> + </dd> <dt><code>max_sectors</code></dt> <dd> The optional <code>max_sectors</code> attribute specifies the maximum diff --git a/src/conf/domain_conf.c b/src/conf/domain_conf.c index d96b012b9..36a4deea2 100644 --- a/src/conf/domain_conf.c +++ b/src/conf/domain_conf.c @@ -9976,6 +9976,7 @@ virDomainControllerDefParseXML(virDomainXMLOptionPtr xmlopt, char *model = NULL; char *queues = NULL; char *cmd_per_lun = NULL; + char *virtqueue_size = NULL; char *max_sectors = NULL; bool processedModel = false; char *modelName = NULL; @@ -10035,6 +10036,7 @@ virDomainControllerDefParseXML(virDomainXMLOptionPtr xmlopt, queues = virXMLPropString(cur, "queues"); cmd_per_lun = virXMLPropString(cur, "cmd_per_lun"); max_sectors = virXMLPropString(cur, "max_sectors"); + virtqueue_size = virXMLPropString(cur, "virtqueue_size"); ioeventfd = virXMLPropString(cur, "ioeventfd"); iothread = virXMLPropString(cur, "iothread"); @@ -10089,6 +10091,12 @@ virDomainControllerDefParseXML(virDomainXMLOptionPtr xmlopt, goto error; } + if (virtqueue_size && virStrToLong_ui(virtqueue_size, NULL, 10, &def->virtqueue_size) < 0) { + virReportError(VIR_ERR_XML_ERROR, + _("Malformed 'virtqueue_size' value '%s'"), virtqueue_size); + goto error; + } + if (max_sectors && virStrToLong_ui(max_sectors, NULL, 10, &def->max_sectors) < 0) { virReportError(VIR_ERR_XML_ERROR, _("Malformed 'max_sectors' value %s"), max_sectors); @@ -10310,6 +10318,7 @@ virDomainControllerDefParseXML(virDomainXMLOptionPtr xmlopt, VIR_FREE(model); VIR_FREE(queues); VIR_FREE(cmd_per_lun); + VIR_FREE(virtqueue_size); VIR_FREE(max_sectors); VIR_FREE(modelName); VIR_FREE(chassisNr); @@ -23293,6 +23302,9 @@ virDomainControllerDriverFormat(virBufferPtr buf, if (def->cmd_per_lun) virBufferAsprintf(&driverBuf, " cmd_per_lun='%u'", def->cmd_per_lun); + if (def->virtqueue_size) + virBufferAsprintf(&driverBuf, " virtqueue_size='%u'", def->virtqueue_size); + if (def->max_sectors) virBufferAsprintf(&driverBuf, " max_sectors='%u'", def->max_sectors); diff --git a/src/conf/domain_conf.h b/src/conf/domain_conf.h index 368f16f3f..141bd1aca 100644 --- a/src/conf/domain_conf.h +++ b/src/conf/domain_conf.h @@ -816,6 +816,7 @@ struct _virDomainControllerDef { int model; /* -1 == undef */ unsigned int queues; unsigned int cmd_per_lun; + unsigned int virtqueue_size; unsigned int max_sectors; int ioeventfd; /* enum virTristateSwitch */ unsigned int iothread; /* unused = 0, > 0 specific thread # */ diff --git a/src/qemu/qemu_command.c b/src/qemu/qemu_command.c index fa0aa5d5c..9b1affd30 100644 --- a/src/qemu/qemu_command.c +++ b/src/qemu/qemu_command.c @@ -2808,6 +2808,9 @@ qemuBuildControllerDevStr(const virDomainDef *domainDef, if (def->queues) virBufferAsprintf(&buf, ",num_queues=%u", def->queues); + if (def->virtqueue_size) + virBufferAsprintf(&buf, ",virtqueue_size=%u", def->virtqueue_size); + if (def->cmd_per_lun) virBufferAsprintf(&buf, ",cmd_per_lun=%u", def->cmd_per_lun); -- 2.16.1

2 1

[libvirt] [PATCH] docs: remove legacy XHTML <!DOCTYPE> declaration
by Daniel P. Berrangé 28 Feb '18

28 Feb '18

We have switched the docs to using the HTML5 doctype declaration in commit b1c81567c7172bc9dcd701cf46ea3f87725d62c7 Author: Daniel P. Berrange <berrange(a)redhat.com> Date: Wed Jul 26 18:01:25 2017 +0100 docs: switch to using HTML5 doctype declaration Signed-off-by: Daniel P. Berrangé <berrange(a)redhat.com> --- Pushed as a trivial fix docs/pci-hotplug.html.in | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/docs/pci-hotplug.html.in b/docs/pci-hotplug.html.in index 4ac87d644f..9a26645aa4 100644 --- a/docs/pci-hotplug.html.in +++ b/docs/pci-hotplug.html.in @@ -1,5 +1,5 @@ <?xml version="1.0" encoding="UTF-8"?> -<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd"> +<!DOCTYPE html> <html xmlns="http://www.w3.org/1999/xhtml"> <body> <h1>PCI topology and hotplug</h1> -- 2.14.3

1 0

[libvirt] [PATCH v2 0/5] Misc improvements & fixes to Travis
by Daniel P. Berrangé 28 Feb '18

28 Feb '18

Changed in v2: - Dropped VPATH patch, since distcheck is doing that already - Skip pdwtags when building with clang - Run install & dist on OS-X - Test systemd as well as upstart - Pass initscript configure flags to distcheck, not autogen.sh Daniel P. Berrangé (5): make: skip pdwtags when building with CLang travis: drop precise distro jobs travis: run "make distcheck" instead of just "make check" travis: test "make install" and "make dist" on OS-X travis: test upstart/systemd init script handling .travis.yml | 22 +++++++++++----------- src/Makefile.am | 10 +++++++++- 2 files changed, 20 insertions(+), 12 deletions(-) -- 2.14.3

2 13

[libvirt] [PATCH 0/3] qemu: Fix updating device with boot order
by Jiri Denemark 28 Feb '18

28 Feb '18

Commit v3.7.0-14-gc57f3fd2f8 prevented adding a <boot order='x'/> element to an inactive domain with global <boot dev='...'/> element. However, as a result of that change updating any device with boot order would fail with 'boot order X is already used by another device', where "another device" is in fact the device which is being updated. To fix this we have to ignore the device which we're about to update when checking for boot order conflicts. https://bugzilla.redhat.com/show_bug.cgi?id=1546971 Jiri Denemark (3): lxc: Drop useless check in live device update Pass oldDev to virDomainDefCompatibleDevice on device update qemu: Fix updating device with boot order src/conf/domain_conf.c | 30 ++++++++++++++++++++++------- src/conf/domain_conf.h | 3 ++- src/lxc/lxc_driver.c | 18 +++++++++--------- src/qemu/qemu_driver.c | 51 ++++++++++++++++++++++++++++++++++++++++---------- 4 files changed, 75 insertions(+), 27 deletions(-) -- 2.16.2

3 5

[libvirt] Entering freeze for libvirt 4.1.0 release
by Daniel Veillard 28 Feb '18

28 Feb '18

I didn't sent a warning mail, but people reminded me that it's time to roll the releases, so I just tagged rc1 in git head and pushed signed tarball and rpms to the usual place: ftp://libvirt.org/libvirt/ Looks fine with my limited testing, amazingly https://ci.centos.org/view/libvirt/ is all green so things seems they should go smoothly, that said the devil is in the details so please give this RC1 some testing especially on other platforms and OSes ! It all goes well I will push rc2 on Wed and the final release early Fri, but in the meantime, please test it :-) thanks, Daniel -- Daniel Veillard | Red Hat Developers Tools http://developer.redhat.com/ veillard(a)redhat.com | libxml Gnome XML XSLT toolkit http://xmlsoft.org/ http://veillard.com/ | virtualization library http://libvirt.org/

1 1

[libvirt] [PATCH] vshCommandOpt: Do more checking if skipChecks is set
by Michal Privoznik 28 Feb '18

28 Feb '18

Currently if cmd->skipChecks is set (done only from completers) some basic checks are skipped because we're working over partially parsed command. See a26ff63ae4 for more detailed explanation. Anyway, the referenced commit was too aggressive in disabling checks and effectively returned success even in clear case of failure. For instance: # domif-getlink --interface <TAB><TAB> causes virshDomainInterfaceCompleter() to be called, which calls virshDomainGetXML() which eventually calls vshCommandOptStringReq(.., name = "domain"); The --domain argument is required for the command and if not present -1 should be returned to tell the caller the argument was not found. Well, zero is returned meaning the argument was not found but it's not required either. Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com> --- tools/vsh.c | 20 ++++++++++---------- 1 file changed, 10 insertions(+), 10 deletions(-) diff --git a/tools/vsh.c b/tools/vsh.c index 37c292a03..73ec007e5 100644 --- a/tools/vsh.c +++ b/tools/vsh.c @@ -817,18 +817,17 @@ vshCommandOpt(const vshCmd *cmd, const char *name, vshCmdOpt **opt, /* See if option is valid and/or required. */ *opt = NULL; - if (!cmd->skipChecks) { - while (valid && valid->name) { - if (STREQ(name, valid->name)) - break; - valid++; - } + while (valid && valid->name) { + if (STREQ(name, valid->name)) + break; + valid++; + } + if (!cmd->skipChecks) assert(valid && (!needData || valid->type != VSH_OT_BOOL)); - if (valid->flags & VSH_OFLAG_REQ) - ret = -1; - } + if (valid && valid->flags & VSH_OFLAG_REQ) + ret = -1; /* See if option is present on command line. */ while (candidate) { @@ -1065,7 +1064,8 @@ vshCommandOptStringReq(vshControl *ctl, error = N_("Option argument is empty"); if (error) { - vshError(ctl, _("Failed to get option '%s': %s"), name, _(error)); + if (!cmd->skipChecks) + vshError(ctl, _("Failed to get option '%s': %s"), name, _(error)); return -1; } -- 2.16.1

2 1

[libvirt] [PATCH] vz: fix ref/unref of domain obj in vzDomainRevertToSnapshot
by Nikolay Shirokovskiy 28 Feb '18

28 Feb '18

--- src/vz/vz_driver.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/vz/vz_driver.c b/src/vz/vz_driver.c index dab1ba8..18f4b4c 100644 --- a/src/vz/vz_driver.c +++ b/src/vz/vz_driver.c @@ -2732,7 +2732,7 @@ vzDomainRevertToSnapshot(virDomainSnapshotPtr snapshot, unsigned int flags) virCheckFlags(VIR_DOMAIN_SNAPSHOT_REVERT_PAUSED, -1); - if (!(dom = vzDomObjFromDomain(snapshot->domain))) + if (!(dom = vzDomObjFromDomainRef(snapshot->domain))) return -1; if (virDomainRevertToSnapshotEnsureACL(snapshot->domain->conn, dom->def) < 0) -- 1.8.3.1

4 5