April 2016 - Devel - libvirt List Archives

[libvirt] [PATCH for 1.3.3 0/5] Fix parsing our own XMLs
by Martin Kletzander 04 Apr '16

04 Apr '16

Laine found out that he can't do 'virsh nodedev-detach pci_dev' because of some strange error message. That was caused by my commit, but also by all the previous ones that skipped adding tests and parsing of new functions. In order for this to work in 1.3.3, we need either a) only the first patch or b) all of them. I specifically created the first one so that it has enough in itself to fix the problem and we don't need to push more and more patches into the release. We can push the rest after release. If someone wants to have all in for 1.3.3, well, I hope I added enough tests for that ;) Martin Kletzander (5): nodedev: Fix parsing of generated XMLs Change virPCIDeviceAddress to virDevicePCIAddress Move capability formatting together schemas: Update nodedev schema to match reality conf: Parse more of our nodedev XML docs/schemas/nodedev.rng | 29 +++-- src/conf/device_conf.h | 11 +- src/conf/node_device_conf.c | 121 ++++++++++++++++++--- src/conf/node_device_conf.h | 6 +- src/libvirt_private.syms | 10 +- src/network/bridge_driver.c | 4 +- src/node_device/node_device_linux_sysfs.c | 6 +- src/util/virhostdev.c | 12 +- src/util/virnetdev.c | 4 +- src/util/virnetdev.h | 2 +- src/util/virpci.c | 80 +++++++------- src/util/virpci.h | 29 ++--- .../pci_0000_00_1c_0_header_type.xml | 2 +- tests/nodedevschemadata/pci_0000_02_10_7_sriov.xml | 23 ++++ .../pci_0000_02_10_7_sriov_pf_vfs_all.xml | 29 +++++ ...i_0000_02_10_7_sriov_pf_vfs_all_header_type.xml | 30 +++++ .../pci_0000_02_10_7_sriov_vfs.xml | 26 +++++ .../pci_0000_02_10_7_sriov_zero_vfs_max_count.xml | 21 ++++ tests/nodedevxml2xmltest.c | 5 + 19 files changed, 333 insertions(+), 117 deletions(-) create mode 100644 tests/nodedevschemadata/pci_0000_02_10_7_sriov.xml create mode 100644 tests/nodedevschemadata/pci_0000_02_10_7_sriov_pf_vfs_all.xml create mode 100644 tests/nodedevschemadata/pci_0000_02_10_7_sriov_pf_vfs_all_header_type.xml create mode 100644 tests/nodedevschemadata/pci_0000_02_10_7_sriov_vfs.xml create mode 100644 tests/nodedevschemadata/pci_0000_02_10_7_sriov_zero_vfs_max_count.xml -- 2.8.0

3 10

[libvirt] [PATCH 0/3] fix regression wrt <interface typ='hostdev'>
by Laine Stump 04 Apr '16

04 Apr '16

The full description of the problem fixed by these patches is in patch 3/3. Basically, a recent patch changed the ordering of things during startup such that hostdev interfaces are given an alias named "netN" rather than the expected "hostdevN". That caused a problem when subsequently trying to hotplug another interface of any kind. This is solved by adding a new function that lets us learn the type of any interface prior to setting it up, then using that function to change the behavior when assigning aliases. This *really* should be pushed before 1.3.3 is released. Laine Stump (3): network: new function networkGetActualType qemu: change args to qemuAssignDeviceHostdevAlias() qemu: fix alias name for <interface type='hostdev'> src/network/bridge_driver.c | 72 +++++++++++++++++++++++++++++++++++++++++++++ src/network/bridge_driver.h | 6 +++- src/qemu/qemu_alias.c | 60 +++++++++++++++++++++---------------- src/qemu/qemu_alias.h | 2 +- src/qemu/qemu_hotplug.c | 10 +++---- 5 files changed, 118 insertions(+), 32 deletions(-) -- 2.5.5

3 6

[libvirt] [libvirt-perl][PATCH 0/3] Adapt to new virDomain{Get, Set}PerfEvents API
by Michal Privoznik 04 Apr '16

04 Apr '16

It would be nice to have these in before the release. Michal Privoznik (3): Adapt to new virDomain{Get,Set}PerfEvents() gitignore: Add more files to ignore & reorder Add myself to AUTHORS .gitignore | 17 ++++++++++------- AUTHORS | 1 + Changes | 1 + Virt.xs | 38 ++++++++++++++++++++++++++++++++++++++ lib/Sys/Virt/Domain.pm | 32 ++++++++++++++++++++++++++++++++ lib/Sys/Virt/Error.pm | 4 ++++ 6 files changed, 86 insertions(+), 7 deletions(-) -- 2.7.3

2 7

[libvirt] [PATCH for 1.3.3 0/3] qemu: Regenerate per-domain paths on restart
by Martin Kletzander 04 Apr '16

04 Apr '16

We forgot to clean up after the domain is torn down. And because they are kept, we don't regenerate them upn another start and because of that they contain old IDs in them. Again, the series is structured as with the previous one [1]. The smallest part (firs patch only) is enough to fix it for the release, others can be pushed before or after release. That depends on the reviewer. [1] https://www.redhat.com/archives/libvir-list/2016-April/msg00081.html Martin Kletzander (3): qemu: Clear generated private paths qemu: Simplify calls to qemuDomainSetPrivatePaths qemu: Add qemuDomainClearPrivatePaths and use it src/qemu/qemu_domain.c | 40 ++++++++++++++++++++++++++++------------ src/qemu/qemu_domain.h | 10 ++++------ src/qemu/qemu_process.c | 9 +++------ 3 files changed, 35 insertions(+), 24 deletions(-) -- 2.8.0

3 7

[libvirt] [PATCH] apparmor: QEMU monitor socket moved
by Guido Günther 02 Apr '16

02 Apr '16

The directory name changed in a89f05ba8df095875f5ec8a9065a585af63a010b. --- src/security/virt-aa-helper.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/src/security/virt-aa-helper.c b/src/security/virt-aa-helper.c index a2d7226..0ded671 100644 --- a/src/security/virt-aa-helper.c +++ b/src/security/virt-aa-helper.c @@ -1366,6 +1366,8 @@ main(int argc, char **argv) LOCALSTATEDIR, ctl->def->name); virBufferAsprintf(&buf, " \"%s/lib/libvirt/qemu/domain-%s/monitor.sock\" rw,\n", LOCALSTATEDIR, ctl->def->name); + virBufferAsprintf(&buf, " \"%s/lib/libvirt/qemu/domain-*-%.*s/monitor.sock\" rw,\n", + LOCALSTATEDIR, 20, ctl->def->name); virBufferAsprintf(&buf, " \"%s/run/libvirt/**/%s.pid\" rwk,\n", LOCALSTATEDIR, ctl->def->name); virBufferAsprintf(&buf, " \"/run/libvirt/**/%s.pid\" rwk,\n", -- 2.8.0.rc3

2 4

[libvirt] [PATCH v2 00/10] Introduce worker tuning APIs
by Erik Skultety 02 Apr '16

02 Apr '16

NOTE: patch 2/10 moves typed params definition from libvirt-host.h to libvirt-common.h.in to enable it for admin as well -> therefore libvirt-common.h must be regenerated with config.status - also I'd like to open a discussion about virt-admin commands naming, since I was sort of of out ideas when I copied numatune,memtune,etc. design into "workertune"; I thought about srv-threadpool-info and srv-threadpool-set to be a little consistent with srv-list, but for some reason I found it quite long-ish so I dropped that... v2: - repost of v1 due to rebase conflicts - due to rebase conflicts and overall changes since the original v1 po/POTFILES.in (patch 1/10) had to be updated Erik Skultety (10): po: Fix record ordering in POTFILES.in libvirt-host: Move virTypedParam* to libvirt-common admin: Enable usage of typed parameters util: Refactor thread creation by introducing virThreadPoolExpand util: Report system error when virThreadCreateFull fails util: Add more getters to threadpool parameters admin: Prepare admin protocol for future worker related procedures admin: Introduce virAdmServerGethreadPoolParameters admin: Introduce virAdmServerSetThreadPoolParameters virt-admin: Introduce srv-workertune command cfg.mk | 2 +- daemon/admin.c | 102 +++++++++++++++++++ daemon/admin_server.c | 112 +++++++++++++++++++++ daemon/admin_server.h | 11 ++ include/libvirt/libvirt-admin.h | 71 +++++++++++++ include/libvirt/libvirt-common.h.in | 185 ++++++++++++++++++++++++++++++++++ include/libvirt/libvirt-host.h | 186 ---------------------------------- po/POTFILES.in | 4 +- src/admin/admin_protocol.x | 54 +++++++++- src/admin/admin_remote.c | 77 ++++++++++++++ src/admin_protocol-structs | 45 +++++++++ src/libvirt-admin.c | 83 +++++++++++++++ src/libvirt_admin_private.syms | 3 + src/libvirt_admin_public.syms | 2 + src/libvirt_private.syms | 4 + src/rpc/virnetserver.c | 37 +++++++ src/rpc/virnetserver.h | 13 +++ src/util/virthreadpool.c | 196 ++++++++++++++++++++++++------------ src/util/virthreadpool.h | 8 ++ tools/virt-admin.c | 132 ++++++++++++++++++++++++ 20 files changed, 1073 insertions(+), 254 deletions(-) -- 2.4.3

2 15

[libvirt] [PATCH] Revert "hostdev: Use actual device when reattaching"
by Andrea Bolognani 02 Apr '16

02 Apr '16

This reverts commit ee4cfb56436b50345b072c706b87aff82e06d760. Since we're still not persisting our bookkeeping lists across daemon restarts, we might have lost some information virPCIDeviceReattach() relies on, for example whether the device needs to be unbound from the stub driver. As a result, if the daemon has been restarted in the meantime, the device might end up remaining bound to the stub driver even after 'virsh nodedev-reattach' or similar has been called, with no way of giving it back to the host short of messing with sysfs behind libvirt's back. Revert back to the previous behavior of always trying to bind the device to the host driver, regardless of its status when it was detached, until persistent bookkeeping lists have been implemented. --- src/util/virhostdev.c | 10 ++++------ 1 file changed, 4 insertions(+), 6 deletions(-) diff --git a/src/util/virhostdev.c b/src/util/virhostdev.c index b397b79..933c942 100644 --- a/src/util/virhostdev.c +++ b/src/util/virhostdev.c @@ -1658,7 +1658,6 @@ virHostdevPCINodeDeviceReAttach(virHostdevManagerPtr mgr, virPCIDevicePtr pci) { struct virHostdevIsPCINodeDeviceUsedData data = { mgr, NULL, false }; - virPCIDevicePtr actual; int ret = -1; virObjectLock(mgr->activePCIHostdevs); @@ -1667,12 +1666,11 @@ virHostdevPCINodeDeviceReAttach(virHostdevManagerPtr mgr, if (virHostdevIsPCINodeDeviceUsed(virPCIDeviceGetAddress(pci), &data)) goto cleanup; - /* We need to look up the actual device because that's what - * virPCIDeviceReattach() expects as its argument */ - if (!(actual = virPCIDeviceListFind(mgr->inactivePCIHostdevs, pci))) - goto cleanup; + virPCIDeviceSetUnbindFromStub(pci, true); + virPCIDeviceSetRemoveSlot(pci, true); + virPCIDeviceSetReprobe(pci, true); - if (virPCIDeviceReattach(actual, mgr->activePCIHostdevs, + if (virPCIDeviceReattach(pci, mgr->activePCIHostdevs, mgr->inactivePCIHostdevs) < 0) goto cleanup; -- 2.5.5

2 2

[libvirt] [PATCH 0/4] Add support for an XKCD mode
by Daniel P. Berrange 01 Apr '16

01 Apr '16

The XKCD website primarily publishes comics, but many of them are in fact design blueprints showing best practice for software developers. A number of these designs are applicable to libvirt and will provide valuable features to our users. This series introduces a new framework for enabling these enhanced features via a so called "XKCD mode". They are not enabled by default since we want to retain backwards compatibility for existing users. Users of libvirt though are encouraged to enable as many of the features as possible for their deployments. Daniel P. Berrange (4): xkcd: add initial framework to enabling XKCD mode xkcd: enable more efficient random number generator xkcd: enabling pre-processing of XML with perl xkcd: generate secure rememberable qcow2 passwords src/Makefile.am | 2 + src/libvirt_private.syms | 4 ++ src/util/virrandom.c | 26 ++++++++----- src/util/virstorageencryption.c | 47 ++++++++++++---------- src/util/virxkcd.c | 86 +++++++++++++++++++++++++++++++++++++++++ src/util/virxkcd.h | 31 +++++++++++++++ src/util/virxml.c | 35 ++++++++++++++--- 7 files changed, 194 insertions(+), 37 deletions(-) create mode 100644 src/util/virxkcd.c create mode 100644 src/util/virxkcd.h -- 2.5.5

6 9

[libvirt] [PATCH] docs: Remove useless p:first-line CSS rule
by Andrea Bolognani 01 Apr '16

01 Apr '16

According to MDN[1], the 'margin-left' CSS property cannot be applied to the '::first-line' pseudo-element, so this rule will never have any effect and can be safely removed. [1] https://developer.mozilla.org/en/docs/Web/CSS/::first-line --- docs/generic.css | 4 ---- 1 file changed, 4 deletions(-) diff --git a/docs/generic.css b/docs/generic.css index 4f98b26..208e31e 100644 --- a/docs/generic.css +++ b/docs/generic.css @@ -7,10 +7,6 @@ body { background: #ffffff; } -p:first-line { - margin-right: 1em; -} - p, ul, ol, dl { padding: 0px; margin: 0px; -- 2.5.5

2 4

[libvirt] [PATCH 00/11] vz: change vz driver to be stateful driver and other enhancements
by Maxim Nestratov 01 Apr '16

01 Apr '16

There is no benefit in providing two ways of connecting to vz driver: by connecting via daemon and directly from client. Both ways finally come to a host where vz daemon sits. Always connecting via daemon allows us to have a single list of domains and share it among all connections. Maxim Nestratov (11): virsh: report when vz driver is compiled vz: change the order of capabilities reported vz: remove drivername field from vzConn structure vz: build driver as module and don't register it on client's side vz: pass vzConnPtr to prlsdkXxx functions instead of virConnectPtr vz: make vzConn structure be a new lockable object and use it vz: implement connectGetSysinfo hypervisor callback vz: remove close callback implementations vz: remove vzDriverLock/Unlock function vz: minor cleanup vz: change vzConnectIsAlive behavior daemon/Makefile.am | 4 + daemon/libvirtd.c | 9 + src/Makefile.am | 21 ++- src/libvirt.c | 7 - src/vz/vz_driver.c | 479 ++++++++++++++++++++++++++++++----------------------- src/vz/vz_sdk.c | 36 ++-- src/vz/vz_sdk.h | 6 +- src/vz/vz_utils.c | 18 +- src/vz/vz_utils.h | 15 +- tools/virsh.c | 3 + 10 files changed, 341 insertions(+), 257 deletions(-) -- 2.4.3

2 16