- Devel - libvirt List Archives

[PATCH 0/5] Support for launchSecurity type papr-pef
by Daniel Henrique Barboza 02 Jun '21

02 Jun '21

Hi, This is the implementation of ppc64 secure guest execution, known as PAPR-PEF. This implementation is based on the new confidential-guest-support machine option that is available in QEMU 6.0.0. The patches were done on top of Boris s390-pv patches that were sent by review [1] given that Boris provided a common base for new security launches there. The series is also available in gitlab: https://gitlab.com/danielhb/libvirt/-/tree/papr_pef_v1 Note: If someone wishes to give this series a try, considering that you have a host with proper PowerPC PEF hardware, be aware that there is a QEMU bug in which launching a PAPR-PEF guest will trigger an early assert [2]. The fix is not merged upstream at the moment of this posting, so you'll need to apply the patch yourself or build QEMU using David Gibson's ppc-for-6.1 tree [3]. [1] https://listman.redhat.com/archives/libvir-list/2021-May/msg00570.html [2] https://lists.gnu.org/archive/html/qemu-devel/2021-05/msg08491.html [3] https://gitlab.com/dgibson/qemu/-/tree/ppc-for-6.1 Daniel Henrique Barboza (5): tests: Add QEMU and domain ppc64 capapbilities for qemu 6.1 qemu: add 'pef-guest' capability conf, qemu: add 'papr-pef' launch security type tests: add tests for 'papr-pef' support docs/formatdomain.rst: add 'papr-pef' launchSecurity docs docs/formatdomain.rst | 4 + docs/schemas/domaincommon.rng | 1 + src/conf/domain_conf.c | 3 + src/conf/domain_conf.h | 1 + src/qemu/qemu_capabilities.c | 2 + src/qemu/qemu_capabilities.h | 1 + src/qemu/qemu_command.c | 26 + src/qemu/qemu_namespace.c | 1 + src/qemu/qemu_process.c | 1 + src/qemu/qemu_validate.c | 8 + tests/domaincapsdata/qemu_6.1.0.ppc64.xml | 133 + .../launch-security-papr-pef.xml | 22 + tests/genericxml2xmltest.c | 1 + .../caps_6.1.0.ppc64.replies | 30368 ++++++++++++++++ .../qemucapabilitiesdata/caps_6.1.0.ppc64.xml | 1175 + ...default-video-type-ppc64.ppc64-latest.args | 5 +- ...y-papr-pef-ignore-policy.ppc64-latest.args | 30 + ...launch-security-papr-pef-ignore-policy.xml | 19 + ...launch-security-papr-pef.ppc64-latest.args | 30 + .../launch-security-papr-pef.xml | 16 + ...ault-cpu-kvm-pseries-2.7.ppc64-latest.args | 5 +- ...ault-cpu-kvm-pseries-3.1.ppc64-latest.args | 5 +- ...ault-cpu-kvm-pseries-4.2.ppc64-latest.args | 5 +- ...ault-cpu-tcg-pseries-2.7.ppc64-latest.args | 5 +- ...ault-cpu-tcg-pseries-3.1.ppc64-latest.args | 5 +- ...ault-cpu-tcg-pseries-4.2.ppc64-latest.args | 5 +- .../ppc64-pseries-graphics.ppc64-latest.args | 7 +- .../ppc64-pseries-headless.ppc64-latest.args | 7 +- .../ppc64-tpmproxy-single.ppc64-latest.args | 5 +- .../ppc64-tpmproxy-with-tpm.ppc64-latest.args | 5 +- .../tpm-emulator-spapr.ppc64-latest.args | 5 +- tests/qemuxml2argvtest.c | 3 + 32 files changed, 31871 insertions(+), 38 deletions(-) create mode 100644 tests/domaincapsdata/qemu_6.1.0.ppc64.xml create mode 100644 tests/genericxml2xmlindata/launch-security-papr-pef.xml create mode 100644 tests/qemucapabilitiesdata/caps_6.1.0.ppc64.replies create mode 100644 tests/qemucapabilitiesdata/caps_6.1.0.ppc64.xml create mode 100644 tests/qemuxml2argvdata/launch-security-papr-pef-ignore-policy.ppc64-latest.args create mode 100644 tests/qemuxml2argvdata/launch-security-papr-pef-ignore-policy.xml create mode 100644 tests/qemuxml2argvdata/launch-security-papr-pef.ppc64-latest.args create mode 100644 tests/qemuxml2argvdata/launch-security-papr-pef.xml -- 2.31.1

1 5

[PATCH v2 0/3] virsh: Fix logic wrt to --current flag in cmdSetmem
by Michal Privoznik 02 Jun '21

02 Jun '21

v2 of: https://listman.redhat.com/archives/libvir-list/2021-May/msg00484.html diff to v1: - Work in Jano's and Peter's review suggestions in 1/3. - Patches 2/3 and 3/3 are new. I've done 2/3 before 3/3 because it's still worth merging even if 3/3 is disliked. Michal Prívozník (3): virsh: Fix logic wrt to --current flag in cmdSetmem virsh-domain: Fix @ret handling in cmdSetmem and cmdSetmaxmem virsh-domain: Drop support for old APIs in cmdSetmem and cmdSetmaxmem tools/virsh-domain.c | 48 +++++++++++++++++--------------------------- 1 file changed, 18 insertions(+), 30 deletions(-) -- 2.26.3

3 7

Recommended volume permissions (being created for vagrant-libvirt via fog-libvirt)
by Darragh Bailey 02 Jun '21

02 Jun '21

Hi, A request has come up recently in vagrant-libvirt about changing the permissions used for the VM volume image file. Currently there is a backing image file uploaded that gets 744 as the file permissions, and then the VM domain is created using this as the backing file for any changes. The file containing the changes for the VM gets 600, so accessing what is contained is limited to libvirt and thus to those that can connect to libvirt. The request is to change this to be 744, it appears to have been triggered due to a desire to try and use virt-v2v to create a portable XML and export the disks. However I'm a little hesitant as in general I would default to more secure rather than less secure to avoid creating security concerns down the line. Even though vagrant-libvirt is typically used for development, it wouldn't surprise me to see it being used on CI build infrastructure and given the shared nature of that, making things less secure may cause issues for some users. Of course working out who would be impacted is virtually impossible without making the change and seeing who is concerned. And that might be several months down the line before it's raised. Rather than just merging this, wondering if there are any security guidelines on the file permissions for VM image files? That or something that can outline the risks, or even clarify that it's unnecessary to worry about? -- Darragh Bailey "Nothing is foolproof to a sufficiently talented fool"

2 3

[PATCH 0/4] node_device_udev: Also process ID_TYPE=cd/dvd in udevProcessStorage()
by Michal Privoznik 02 Jun '21

02 Jun '21

See the last one which fixes a real bug. The rest is just cleanups. Michal Prívozník (4): node_device_udev: Make udevGenerateDeviceName() return void node_device_udev: Make udevGetStringProperty() return void node_device_udev: Don't overwrite @ret in udevProcessStorage() node_device_udev: Also process ID_TYPE=cd/dvd in udevProcessStorage() src/node_device/node_device_udev.c | 114 +++++++++++------------------ 1 file changed, 43 insertions(+), 71 deletions(-) -- 2.31.1

2 7

[libvirt PATCH v2] meson: Don't include warning flags in summary
by Andrea Bolognani 01 Jun '21

01 Jun '21

The list takes up half a screen on my machine and it doesn't seem useful enough for that to be warranted, especially when it results in other, arguably more relevant parts of the summary being pushed off-screen. Signed-off-by: Andrea Bolognani <abologna(a)redhat.com> Reviewed-by: Pavel Hrdina <phrdina(a)redhat.com> --- Changes from [v1]: * move out of the series, which has mostly been pushed now. [v1] https://listman.redhat.com/archives/libvir-list/2021-June/msg00002.html meson.build | 1 - 1 file changed, 1 deletion(-) diff --git a/meson.build b/meson.build index 2fa379f47a..b09d766b11 100644 --- a/meson.build +++ b/meson.build @@ -2313,7 +2313,6 @@ else loader_res = '' endif misc_summary = { - 'Warning Flags': supported_cc_flags, 'docs': gen_docs, 'tests': build_tests, 'DTrace': conf.has('WITH_DTRACE_PROBES'), -- 2.31.1

1 0

[libvirt PATCH 0/2] fix disk XML formatting for qemuDomainBlockCopy API
by Pavel Hrdina 01 Jun '21

01 Jun '21

Pavel Hrdina (2): domain_conf: extract disk driver source bits to its own function virDomainDiskDefParseSource: parse source bits from driver element src/conf/domain_conf.c | 52 ++++++++++++++++++++++++++++++------------ 1 file changed, 37 insertions(+), 15 deletions(-) -- 2.31.1

2 3

[libvirt PATCH 00/26] meson: Various fixes and improvements
by Andrea Bolognani 01 Jun '21

01 Jun '21

Test pipeline: https://gitlab.com/abologna/libvirt/-/pipelines/312508056 Andrea Bolognani (26): spec: Be explicit about more features meson: Whitespace tweaks meson: Don't use 'required: true' meson: Drop curl workaround meson: Drop netinet workaround meson: Use get_pkgconfig_variable('cflags') meson: Use built-in pcap detection meson: Make libm a required dependency meson: Drop numactl_version meson: Fix vstorage detection meson: Fix sanlock detection meson: Rewrite polkit check meson: Tweak XDR check meson: Fix disabling netcf meson: Rewrite firewalld check meson: Fix firewalld check meson: Rewrite libacl check meson: Use dependency() when possible meson: Rewrite apparmor_profiles check meson: Turn apparmor_profiles into a feature meson: Switch to autodetection for apparmor_profiles meson: Switch to autodetection for driver_remote meson: Switch to autodetection for driver_test meson: Style tweaks meson: Reorganize summary meson: Don't include warning flags in summary libvirt.spec.in | 3 + meson.build | 301 +++++++++++++++++++------------------ meson_options.txt | 6 +- mingw-libvirt.spec.in | 2 + src/locking/meson.build | 4 +- src/qemu/qemu_namespace.c | 2 +- src/util/virfile.c | 8 +- src/util/virnetdevbridge.c | 18 --- 8 files changed, 166 insertions(+), 178 deletions(-) -- 2.31.1

2 32

Release of libvirt-7.4.0
by Jiri Denemark 01 Jun '21

01 Jun '21

The 7.4.0 release of both libvirt and libvirt-python is tagged and signed tarballs and source RPMs are available at https://libvirt.org/sources/ https://libvirt.org/sources/python/ Thanks everybody who helped with this release by sending patches, reviewing, testing, or providing any other feedback. Your work is greatly appreciated. * Removed features * qemu: Remove support for QEMU < 2.11 In accordance with our platform support policy, the oldest supported QEMU version is now bumped from 1.5 to 2.11. * New features * qemu: Add support for hotplugging ``<transient/>`` disks The disk hotplug code in the qemu driver now can handle hotplug of disks with automatically added overlay. * qemu: Add support for sharing base image of ``<transient/>`` disks Users can use ``<transient shareBacking='yes'/>`` to tell the qemu driver to never open the base image in write mode thus multiple VMs can share the same image. Note that the disk will be hotplugged during startup. * Improvements * Add win-dmp crashdump format New ``win-dmp`` format for ``virDomainCoreDumpWithFormat`` API and/or virsh ``dump --format`` was introduced. * Bug fixes * Allow 0 offset in XML schema for ``<slice type='storage' offset='0' size='321'/>`` Having a 0 offset so that the size of the image can be limited is a valid configuration so it was allowed in the XML schema. Enjoy. Jirka

1 0

[libvirt PATCH] docs: introduce stubs for new libvirt Go packages
by Daniel P. Berrangé 01 Jun '21

01 Jun '21

Currently we expose libvirt Go packages at libvirt.org/libvirt-go libvirt.org/libvirt-go-xml These packages have not supported Go modules historically and when we tried to introduce modules, we hit the problem that we're not using semver for versioning. The only way around this is to introduce new packages under a different namespace, that will have the exact same code, but be tagged with a different version numbering scheme. This change proposes: libvirt.org/go/libvirt libvirt.org/go/libvirtxml Note the hyphen is removed so that the import basename matches the Go package name. Signed-off-by: Daniel P. Berrangé <berrange(a)redhat.com> --- The corresponding repos exist https://gitlab.com/libvirt/libvirt-go-module https://gitlab.com/libvirt/libvirt-go-xml-module but I'm leaving them empty until we've completed this release, then I'll populate with content matching 7.4.0, minus the existing tags. This change should also wait until after the release. docs/go/libvirt.rst | 13 +++++++++++ docs/go/libvirtxml.rst | 10 ++++++++ docs/go/meson.build | 53 ++++++++++++++++++++++++++++++++++++++++++ docs/meson.build | 1 + 4 files changed, 77 insertions(+) create mode 100644 docs/go/libvirt.rst create mode 100644 docs/go/libvirtxml.rst create mode 100644 docs/go/meson.build diff --git a/docs/go/libvirt.rst b/docs/go/libvirt.rst new file mode 100644 index 0000000000..46250691c1 --- /dev/null +++ b/docs/go/libvirt.rst @@ -0,0 +1,13 @@ +======================= +Libvirt Go Language API +======================= + +The `Go <https://golang.org/>`__ package ``libvirt.org/go/libvirt`` provides +`CGo <https://golang.org/cmd/cgo/>`__ binding from the OS native Libvirt API. + +In general the Go representation is a direct 1-1 mapping from native API +concepts to Go, so the native API documentation should serve as a reference +for most behaviour. + +For details of Go specific behaviour consult the +`Go package documentation <https://pkg.go.dev/libvirt.org/go/libvirt>`__. diff --git a/docs/go/libvirtxml.rst b/docs/go/libvirtxml.rst new file mode 100644 index 0000000000..7583433f56 --- /dev/null +++ b/docs/go/libvirtxml.rst @@ -0,0 +1,10 @@ +========================== +Libvirt Go XML parsing API +========================== + +The `Go <https://golang.org/>`__ package ``libvirt.org/go/libvirtxml`` provides +annotated Go struct definitions for parsing (and formatting) XML documents used +with libvirt APIs. + +For details of Go specific behaviour consult the +`Go package documentation <https://pkg.go.dev/libvirt.org/go/libvirtxml>`__. diff --git a/docs/go/meson.build b/docs/go/meson.build new file mode 100644 index 0000000000..99fa1b074a --- /dev/null +++ b/docs/go/meson.build @@ -0,0 +1,53 @@ +docs_go_files = [ + 'libvirt', + 'libvirtxml', +] + +html_xslt_gen_xslt = subsite_xsl +html_xslt_gen_install_dir = docs_html_dir / 'go' +html_xslt_gen = [] + +foreach name : docs_go_files + rst_file = '@0@.rst'.format(name) + + html_xslt_gen += { + 'name': name, + 'file': docs_rst2html_gen.process(rst_file), + 'source': 'docs/go' / rst_file, + } +endforeach + +# keep the XSLT processing code block in sync with docs/meson.build + +# --- begin of XSLT processing --- + +foreach data : html_xslt_gen + html_filename = data['name'] + '.html' + + html_file = custom_target( + html_filename, + input: data.get('file', data['name'] + '.html.in'), + output: html_filename, + command: [ + xsltproc_prog, + '--stringparam', 'pagesrc', data.get('source', ''), + '--stringparam', 'builddir', meson.build_root(), + '--stringparam', 'timestamp', docs_timestamp, + '--nonet', + html_xslt_gen_xslt, + '@INPUT@', + ], + depends: data.get('depends', []), + depend_files: [ page_xsl ], + capture: true, + install: true, + install_dir: html_xslt_gen_install_dir, + ) + + install_web_deps += html_file + install_web_files += html_file.full_path() + ':' + html_xslt_gen_install_dir +endforeach + +html_xslt_gen = [] + +# --- end of XSLT processing --- diff --git a/docs/meson.build b/docs/meson.build index f550629d8e..d7afdbd323 100644 --- a/docs/meson.build +++ b/docs/meson.build @@ -296,6 +296,7 @@ html_xslt_gen = [] # --- end of XSLT processing --- subdir('fonts') +subdir('go') subdir('html') subdir('internals') subdir('js') -- 2.31.1

2 1

[PATCH] NEWS: disks: Mention <transient> improvements and <slice> XML fix
by Peter Krempa 01 Jun '21

01 Jun '21

Signed-off-by: Peter Krempa <pkrempa(a)redhat.com> --- NEWS.rst | 15 +++++++++++++++ 1 file changed, 15 insertions(+) diff --git a/NEWS.rst b/NEWS.rst index 8358cbd369..66ce95d7cb 100644 --- a/NEWS.rst +++ b/NEWS.rst @@ -20,6 +20,17 @@ v7.4.0 (unreleased) * **New features** + * qemu: Add support for hotplugging ``<transient/>`` disks + + The disk hotplug code in the qemu driver now can handle hotplug of disks + with automatically added overlay. + + * qemu: Add support for sharing base image of ``<transient/>`` disks + + Users can use ``<transient shareBacking='yes'/>`` to tell the qemu driver to + never open the base image in write mode thus multiple VMs can share the + same image. Note that the disk will be hotplugged during startup. + * **Improvements** * Add win-dmp crashdump format @@ -29,6 +40,10 @@ v7.4.0 (unreleased) * **Bug fixes** + * Allow 0 offset in XML schema for ``<slice type='storage' offset='0' size='321'/>`` + + Having a 0 offset so that the size of the image can be limited is a + valid configuration so it was allowd in the XML schema. v7.3.0 (2021-05-03) =================== -- 2.31.1

2 1

[PATCH] qemuProcessSetupDisksTransientSnapshot: Skip enabling transientOverlayCreated flag
by Masayoshi Mizuma 01 Jun '21

01 Jun '21

From: Masayoshi Mizuma <m.mizuma(a)jp.fujitsu.com> QEMU_DOMAIN_DISK_PRIVATE(disk)->transientOverlayCreated flag gets true unexpectedly on qemuProcessSetupDisksTransientSnapshot() when the disk has <transient shareBacking='yes'> option. The flag should be enabled on qemuDomainAttachDiskGeneric() after the overlay setup is completed. Skip enabling transientOverlayCreated for the disk here. Fixes: 75871da0ecb8b552f9e304d0f83e216839bbf82d Signed-off-by: Masayoshi Mizuma <m.mizuma(a)jp.fujitsu.com> --- src/qemu/qemu_process.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c index 9b069fe7ce..c37687f249 100644 --- a/src/qemu/qemu_process.c +++ b/src/qemu/qemu_process.c @@ -7039,7 +7039,8 @@ qemuProcessSetupDisksTransientSnapshot(virDomainObj *vm, for (i = 0; i < vm->def->ndisks; i++) { virDomainDiskDef *domdisk = vm->def->disks[i]; - if (!domdisk->transient) + if (!domdisk->transient || + domdisk->transientShareBacking == VIR_TRISTATE_BOOL_YES) continue; QEMU_DOMAIN_DISK_PRIVATE(domdisk)->transientOverlayCreated = true; -- 2.27.0

2 1

[libvirt PATCH 00/10] remote: various refactoring & bugs for remote driver
by Daniel P. Berrangé 27 May '21

27 May '21

This is in prep for switching to use the modular daemons by default. In testing that setup I hit a couple of minor bugs. Daniel P. Berrangé (10): remote: use absolute path to check for daemons remote: stop checking for errors from socket allocations remote: move open flags enum to sockets header remote: consistently use flags for passing ro/user/autostart props remote: change socket helper to return full daemon path remote: don't populate daemon path if autostart is not required rpc: remove "spawnDaemon" parameter remote: move proxy/mode defaults after URI parsing remote: fix regression connecting to remote session daemon remote: use virtproxyd if no URI is given src/admin/admin_remote.c | 2 +- src/locking/lock_driver_lockd.c | 1 - src/logging/log_manager.c | 1 - src/lxc/lxc_monitor.c | 2 +- src/qemu/qemu_migration.c | 3 +- src/remote/remote_daemon_dispatch.c | 4 +- src/remote/remote_driver.c | 135 ++++++++++++---------------- src/remote/remote_sockets.c | 95 +++++++++++++------- src/remote/remote_sockets.h | 15 ++-- src/remote/remote_ssh_helper.c | 16 ++-- src/rpc/virnetclient.c | 5 +- src/rpc/virnetclient.h | 3 +- src/rpc/virnetsocket.c | 26 ++---- src/rpc/virnetsocket.h | 3 +- tests/virnetsockettest.c | 4 +- 15 files changed, 157 insertions(+), 158 deletions(-) -- 2.31.1

3 15

Entering freeze for libvirt-7.4.0
by Jiri Denemark 27 May '21

27 May '21

I have just tagged v7.4.0-rc1 in the repository and pushed signed tarballs and source RPMs to https://libvirt.org/sources/ Please give the release candidate some testing and in case you find a serious issue which should have a fix in the upcoming release, feel free to reply to this thread to make sure the issue is more visible. If you have not done so yet, please update NEWS.rst to document any significant change you made since the last release. Thanks, Jirka

2 3

[libvirt PATCH 0/2] Revert mdev autostart support
by Jonathon Jongsma 27 May '21

27 May '21

As Daniel Berrange mentioned, the method of supporting autostart for node devices was not consistent with other libvirt objects. This patch series reverts this feature before it gets into a release. I will submit a follow-up patch soon which adds the feature back with a consistent API. Jonathon Jongsma (2): Partial Revert of "tests: nodedevxml2xmltest: test more mdev files" nodedev: Revert auto-start property for mdevs docs/formatnode.html.in | 10 --------- docs/schemas/nodedev.rng | 11 ---------- src/conf/node_device_conf.c | 21 +------------------ src/conf/node_device_conf.h | 12 ----------- src/libvirt_private.syms | 2 -- src/node_device/node_device_driver.c | 7 +------ .../mdevctl-list-multiple.out.xml | 4 ---- tests/nodedevschemadata/mdev_autostart.xml | 10 --------- tests/nodedevschemadata/mdev_manual.xml | 10 --------- ...v_3627463d_b7f0_4fea_b468_f1da537d301b.xml | 10 +-------- tests/nodedevxml2xmlout/mdev_autostart.xml | 11 ---------- ...v_d069d019_36ea_4111_8f0a_8c9a70e21366.xml | 1 - ...v_d2441d39_495e_4243_ad9f_beb3f14c23d9.xml | 1 - ...v_ee0b88c4_f554_4dc1_809d_b2a01e8e48ad.xml | 11 +--------- ...v_fedc4916_1ca8_49ac_b176_871d16c13076.xml | 1 - tests/nodedevxml2xmlout/mdev_manual.xml | 11 ---------- tests/nodedevxml2xmltest.c | 2 -- 17 files changed, 4 insertions(+), 131 deletions(-) delete mode 100644 tests/nodedevschemadata/mdev_autostart.xml delete mode 100644 tests/nodedevschemadata/mdev_manual.xml mode change 100644 => 120000 tests/nodedevxml2xmlout/mdev_3627463d_b7f0_4fea_b468_f1da537d301b.xml delete mode 100644 tests/nodedevxml2xmlout/mdev_autostart.xml mode change 100644 => 120000 tests/nodedevxml2xmlout/mdev_ee0b88c4_f554_4dc1_809d_b2a01e8e48ad.xml delete mode 100644 tests/nodedevxml2xmlout/mdev_manual.xml -- 2.31.1

3 8

[PATCH for 7.4.0] virCapabilitiesHostNUMAInitReal: Don't jump over cleanup
by Michal Privoznik 27 May '21

27 May '21

In one of my recent commits I've done some renaming. But whilst doing so I also mistakenly replaced 'goto cleanup' with 'return -1' in virCapabilitiesHostNUMAInitReal() which was incorrect. Fixes: fe25224fdaa53bbeceed3ddeef1b3a150665e656 Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com> --- src/conf/capabilities.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/conf/capabilities.c b/src/conf/capabilities.c index 1290c9c15d..2f9a1e7d1f 100644 --- a/src/conf/capabilities.c +++ b/src/conf/capabilities.c @@ -1623,7 +1623,7 @@ virCapabilitiesHostNUMAInitReal(virCapsHostNUMA *caps) } if (virCapabilitiesGetNUMADistances(n, &distances, &ndistances) < 0) - return -1; + goto cleanup; if (virCapabilitiesGetNUMAPagesInfo(n, &pageinfo, &npageinfo) < 0) goto cleanup; -- 2.31.1

2 1

[PATCH] qemuDomainAttachDiskGenericTransient: Add NULL check in case the overlay disk already exists
by Masayoshi Mizuma 27 May '21

27 May '21

From: Masayoshi Mizuma <m.mizuma(a)jp.fujitsu.com> When <transient shareBacking='yes'> is set to a disk and the overlay disk already exists because of something abnormal, libvirt is terminated by Segmentation fault. # virsh start Test0 error: Disconnected from qemu:///system due to end of file error: Failed to start domain 'Test0' error: End of file while reading data: Input/output error Add NULL check for snapdiskdef so that the rollback can work correctly. Signed-off-by: Masayoshi Mizuma <m.mizuma(a)jp.fujitsu.com> --- src/qemu/qemu_hotplug.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/src/qemu/qemu_hotplug.c b/src/qemu/qemu_hotplug.c index 9ff6175dd3..0640cdd9f7 100644 --- a/src/qemu/qemu_hotplug.c +++ b/src/qemu/qemu_hotplug.c @@ -695,7 +695,9 @@ qemuDomainAttachDiskGenericTransient(virDomainObj *vm, g_autoptr(qemuSnapshotDiskContext) snapctxt = NULL; g_autoptr(virDomainSnapshotDiskDef) snapdiskdef = NULL; - snapdiskdef = qemuSnapshotGetTransientDiskDef(disk, vm->def->name); + if (!(snapdiskdef = qemuSnapshotGetTransientDiskDef(disk, vm->def->name))) + return NULL; + snapctxt = qemuSnapshotDiskContextNew(1, vm, asyncJob); if (qemuSnapshotDiskPrepareOne(snapctxt, disk, snapdiskdef, -- 2.27.0

2 1

[PATCH V5] libxl: adjust handling of libxl_device_nic objects
by Jim Fehlig 26 May '21

26 May '21

libxl objects are supposed to be initialized and disposed. Adjust libxlMakeNic to use an already initialized object owned by the caller. Adjust libxlMakeNicList to initialize the list of objects, before they are filled by libxlMakeNic. The libxl_domain_config object passed to libxlMakeNicList is owned by the caller and will be disposed with libxl_domain_config_dispose, which also disposes embedded objects such as libxl_device_nic. Signed-off-by: Jim Fehlig <jfehlig(a)suse.com> --- src/libxl/libxl_conf.c | 16 ++++++---------- 1 file changed, 6 insertions(+), 10 deletions(-) diff --git a/src/libxl/libxl_conf.c b/src/libxl/libxl_conf.c index 91cebf1498..fc4268db01 100644 --- a/src/libxl/libxl_conf.c +++ b/src/libxl/libxl_conf.c @@ -1313,8 +1313,6 @@ libxlMakeNic(virDomainDef *def, return -1; } - libxl_device_nic_init(x_nic); - virMacAddrGetRaw(&l_nic->mac, x_nic->mac); /* @@ -1502,6 +1500,7 @@ libxlMakeNicList(virDomainDef *def, libxl_domain_config *d_config) size_t nnics = def->nnets; libxl_device_nic *x_nics; size_t i, nvnics = 0; + int ret = -1; x_nics = g_new0(libxl_device_nic, nnics); @@ -1509,8 +1508,9 @@ libxlMakeNicList(virDomainDef *def, libxl_domain_config *d_config) if (virDomainNetGetActualType(l_nics[i]) == VIR_DOMAIN_NET_TYPE_HOSTDEV) continue; + libxl_device_nic_init(&x_nics[nvnics]); if (libxlMakeNic(def, l_nics[i], &x_nics[nvnics], false)) - goto error; + goto out; /* * The devid (at least right now) will not get initialized by * libxl in the setup case but is required for starting the @@ -1521,18 +1521,14 @@ libxlMakeNicList(virDomainDef *def, libxl_domain_config *d_config) nvnics++; } + ret = 0; + out: VIR_SHRINK_N(x_nics, nnics, nnics - nvnics); d_config->nics = x_nics; d_config->num_nics = nvnics; - return 0; - - error: - for (i = 0; i < nnics; i++) - libxl_device_nic_dispose(&x_nics[i]); - VIR_FREE(x_nics); - return -1; + return ret; } int -- 2.31.1

2 2

[PATCH v4] libxl: adjust handling of libxl_device_nic objects
by Olaf Hering 26 May '21

26 May '21

libxl objects are supposed to be initialized and disposed. Adjust libxlMakeNic to use an already initialized object, it is owned by the caller. Adjust libxlMakeNicList to initialize the list of objects, before they are filled by libxlMakeNic. In case of error the objects are disposed by libxl_domain_config_dispose. Signed-off-by: Olaf Hering <olaf(a)aepfle.de> --- src/libxl/libxl_conf.c | 11 ++--------- 1 file changed, 2 insertions(+), 9 deletions(-) diff --git a/src/libxl/libxl_conf.c b/src/libxl/libxl_conf.c index 2ecbcf6911..c672bafbe9 100644 --- a/src/libxl/libxl_conf.c +++ b/src/libxl/libxl_conf.c @@ -1335,8 +1335,6 @@ libxlMakeNic(virDomainDef *def, return -1; } - libxl_device_nic_init(x_nic); - virMacAddrGetRaw(&l_nic->mac, x_nic->mac); /* @@ -1531,8 +1529,9 @@ libxlMakeNicList(virDomainDef *def, libxl_domain_config *d_config) if (virDomainNetGetActualType(l_nics[i]) == VIR_DOMAIN_NET_TYPE_HOSTDEV) continue; + libxl_device_nic_init(&x_nics[nvnics]); if (libxlMakeNic(def, l_nics[i], &x_nics[nvnics], false)) - goto error; + return -1; /* * The devid (at least right now) will not get initialized by * libxl in the setup case but is required for starting the @@ -1549,12 +1548,6 @@ libxlMakeNicList(virDomainDef *def, libxl_domain_config *d_config) d_config->num_nics = nvnics; return 0; - - error: - for (i = 0; i < nnics; i++) - libxl_device_nic_dispose(&x_nics[i]); - VIR_FREE(x_nics); - return -1; } int

2 2

[PATCH RESEND v3 0/8] cleanup meson checks for runtime binaries
by Pavel Hrdina 26 May '21

26 May '21

Recent attempt to add a lot of meson options to specify different runtime paths motivated me enough to cleanup this from meson. Changes in v3: - some patches were already pushed - removed patch that moved virFindFileInPath from testutilsqemu.c Changes in v2: - split and rework patch 16/17 to address review comments - added a new patch to cleanup libvirt.spec.in file Pavel Hrdina (8): virfile: introduce virFindFileInPathFull() qemu_conf: use virFindFileInPathFull for runtime binaries meson: drop check for runtime binary dependencies meson: move iscsiadm check into storage_iscsi condition meson: stop setting runtime binaries defines during compilation meson: use runtime binaries to only resolve features with "auto" value meson: optional_programs should be used only for building libvirt libvirt.spec: drop no longer required build dependencies libvirt.spec.in | 31 ---- meson.build | 214 +++++++++---------------- src/bhyve/bhyve_command.c | 4 + src/libvirt_private.syms | 2 +- src/locking/lock_driver_lockd.c | 12 +- src/network/bridge_driver.c | 2 + src/node_device/node_device_driver.c | 2 + src/qemu/qemu_conf.c | 23 ++- src/storage/storage_backend_logical.c | 13 ++ src/storage/storage_backend_sheepdog.c | 2 + src/storage/storage_backend_zfs.c | 3 + src/storage/storage_util.c | 2 + src/storage/storage_util.h | 6 + src/util/virdnsmasq.c | 1 + src/util/virfile.c | 16 +- src/util/virfile.h | 5 +- src/util/virfirewall.h | 4 + src/util/viriscsi.h | 2 + src/util/virkmod.h | 3 + src/util/virnetdevbandwidth.h | 2 + src/util/virnetdevip.c | 2 + src/util/virnetdevmidonet.c | 2 + src/util/virnetdevopenvswitch.c | 2 + src/util/virnuma.c | 1 + src/util/virsysinfo.c | 1 + src/util/virutil.c | 2 + tests/testutilsqemu.c | 3 +- tests/virfirewallmock.c | 3 +- 28 files changed, 173 insertions(+), 192 deletions(-) -- 2.31.1

4 21

[PATCH] meson.build: Fix the -Wvla-larger-than flag
by Thomas Huth 25 May '21

25 May '21

It's "...-than=..." and not "...-then=...". Fixes: 8dd259d0c6 ("meson: add manywarnings") Signed-off-by: Thomas Huth <thuth(a)redhat.com> --- meson.build | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/meson.build b/meson.build index 4f23f9104e..8ee0416700 100644 --- a/meson.build +++ b/meson.build @@ -390,7 +390,7 @@ cc_flags += [ '-Wvariadic-macros', '-Wvector-operation-performance', '-Wvla', - '-Wvla-larger-then=4031', + '-Wvla-larger-than=4031', '-Wvolatile-register-var', '-Wwrite-strings', ] -- 2.27.0

2 1

availability of cluster_size support (re. [libvirt PATCH v2 0/2] storage_driver: add support for cluster_size QCOW2 option)
by Geoff Macartney1 25 May '21

25 May '21

1 0

[PATCH v3] libxl: adjust handling of libxl_device_nic objects
by Olaf Hering 25 May '21

25 May '21

libxl objects are supposed to be initialized and disposed. Adjust libxlMakeNic to use an already initialized object, it is owned by the caller. Adjust libxlMakeNicList to initialize the list of objects, before they are filled by libxlMakeNic. In case of error the objects are disposed by libxl_domain_config_dispose. Signed-off-by: Olaf Hering <olaf(a)aepfle.de> --- src/libxl/libxl_conf.c | 34 +++++++++++++--------------------- 1 file changed, 13 insertions(+), 21 deletions(-) diff --git a/src/libxl/libxl_conf.c b/src/libxl/libxl_conf.c index f29a28a841..8dc7e26cea 100644 --- a/src/libxl/libxl_conf.c +++ b/src/libxl/libxl_conf.c @@ -1339,8 +1339,6 @@ libxlMakeNic(virDomainDef *def, return -1; } - libxl_device_nic_init(x_nic); - virMacAddrGetRaw(&l_nic->mac, x_nic->mac); /* @@ -1526,39 +1524,33 @@ libxlMakeNicList(virDomainDef *def, libxl_domain_config *d_config) { virDomainNetDef **l_nics = def->nets; size_t nnics = def->nnets; - libxl_device_nic *x_nics; size_t i, nvnics = 0; - x_nics = g_new0(libxl_device_nic, nnics); - for (i = 0; i < nnics; i++) { if (virDomainNetGetActualType(l_nics[i]) == VIR_DOMAIN_NET_TYPE_HOSTDEV) continue; + nvnics++; + } + if (!nvnics) + return 0; - if (libxlMakeNic(def, l_nics[i], &x_nics[nvnics], false)) - goto error; + d_config->nics = g_new0(libxl_device_nic, nvnics); + d_config->num_nics = nvnics; + + for (i = 0; i < nvnics; i++) { + libxl_device_nic_init(&d_config->nics[i]); + if (libxlMakeNic(def, l_nics[i], &d_config->nics[i], false)) + return -1; /* * The devid (at least right now) will not get initialized by * libxl in the setup case but is required for starting the * device-model. */ - if (x_nics[nvnics].devid < 0) - x_nics[nvnics].devid = nvnics; - - nvnics++; + if (d_config->nics[i].devid < 0) + d_config->nics[i].devid = i; } - VIR_SHRINK_N(x_nics, nnics, nnics - nvnics); - d_config->nics = x_nics; - d_config->num_nics = nvnics; - return 0; - - error: - for (i = 0; i < nnics; i++) - libxl_device_nic_dispose(&x_nics[i]); - VIR_FREE(x_nics); - return -1; } int

2 2

[PATCH] spec: Fix %endif indentation
by Michal Privoznik 25 May '21

25 May '21

In recent commit f772c1fd2a a misaligned %endif sneaked in which upsets syntax-check. Align it properly. Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com> --- Pushed as trivial and build breaker fix. libvirt.spec.in | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/libvirt.spec.in b/libvirt.spec.in index cdca7d8db6..7920599498 100644 --- a/libvirt.spec.in +++ b/libvirt.spec.in @@ -73,7 +73,7 @@ %else %ifnarch %{arches_qemu_kvm} %define with_storage_gluster 0 - %endif + %endif %endif %endif -- 2.26.3

1 0

[PATCH for 7.4.0 v2 0/3] Avoid double indentation of <metadata/> element
by Michal Privoznik 25 May '21

25 May '21

v2 of: https://listman.redhat.com/archives/libvir-list/2021-May/msg00767.html diff to v1: - Deduplicated the code per Pavel's suggestion Michal Prívozník (3): virxml: Introduce and use virXMLFormatMetadata() virxml: Report error if virXMLFormatMetadata() fails virxml: Avoid double indentation of <metadata/> element src/conf/domain_conf.c | 23 ++---------------- src/conf/network_conf.c | 23 ++---------------- src/libvirt_private.syms | 1 + src/util/virxml.c | 51 ++++++++++++++++++++++++++++++++++++++++ src/util/virxml.h | 3 +++ 5 files changed, 59 insertions(+), 42 deletions(-) -- 2.26.3

3 6

[PATCH for 7.4.0] conf: Avoid double indentation of <metadata/> element
by Michal Privoznik 25 May '21

25 May '21

There was a recent change in libxml2 that caused a trouble for us. To us, <metadata/> in domain or network XMLs are just opaque value where management application can store whatever data it finds fit. At XML parser/formatter level, we just make a copy of the element during parsing and then format it back. For formatting we use xmlNodeDump() which allows caller to specify level of indentation. Previously, the indentation was not applied onto the very first line, but as of v2.9.12-2-g85b1792e libxml2 is applying indentation also on the first line. This does not work well with out virBuffer because as soon as we call virBufferAsprintf() to append <metadata/> element, virBufferAsprintf() will apply another level of indentation. Instead of version checking, let's skip any indentation added by libxml2 before virBufferAsprintf() is called. Note, the problem is only when telling xmlNodeDump() to use indentation, i.e. level argument is not zero. Therefore, virXMLNodeToString() which also calls xmlNodeDump() is safe as it passes zero. Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com> --- I've raised this issue with libxml2 team: https://gitlab.gnome.org/GNOME/libxml2/-/commit/85b1792e37b131e7a51af98a37f… but I'm not sure it'll be fixed. Unfortunately, the patch that's causing us trouble is being backported all over the place, because it's supposedly fixing a regression. src/conf/domain_conf.c | 9 ++++++++- src/conf/network_conf.c | 9 ++++++++- 2 files changed, 16 insertions(+), 2 deletions(-) diff --git a/src/conf/domain_conf.c b/src/conf/domain_conf.c index 4f78b7b43d..84a8c269be 100644 --- a/src/conf/domain_conf.c +++ b/src/conf/domain_conf.c @@ -27808,6 +27808,7 @@ virDomainDefFormatInternalSetRootName(virDomainDef *def, if (def->metadata) { g_autoptr(xmlBuffer) xmlbuf = NULL; + const char *xmlbufContent = NULL; int oldIndentTreeOutput = xmlIndentTreeOutput; /* Indentation on output requires that we previously set @@ -27824,7 +27825,13 @@ virDomainDefFormatInternalSetRootName(virDomainDef *def, xmlIndentTreeOutput = oldIndentTreeOutput; return -1; } - virBufferAsprintf(buf, "%s\n", (char *) xmlBufferContent(xmlbuf)); + + /* After libxml2-v2.9.12-2-g85b1792e even the first line is indented. + * But virBufferAsprintf() also adds indentation. Skip one of them. */ + xmlbufContent = (const char *) xmlBufferContent(xmlbuf); + virSkipSpaces(&xmlbufContent); + + virBufferAsprintf(buf, "%s\n", xmlbufContent); xmlIndentTreeOutput = oldIndentTreeOutput; } diff --git a/src/conf/network_conf.c b/src/conf/network_conf.c index a9eadff29c..20c6dc091a 100644 --- a/src/conf/network_conf.c +++ b/src/conf/network_conf.c @@ -2488,6 +2488,7 @@ virNetworkDefFormatBuf(virBuffer *buf, if (def->metadata) { g_autoptr(xmlBuffer) xmlbuf = NULL; + const char *xmlbufContent = NULL; int oldIndentTreeOutput = xmlIndentTreeOutput; /* Indentation on output requires that we previously set @@ -2504,7 +2505,13 @@ virNetworkDefFormatBuf(virBuffer *buf, xmlIndentTreeOutput = oldIndentTreeOutput; return -1; } - virBufferAsprintf(buf, "%s\n", (char *) xmlBufferContent(xmlbuf)); + + /* After libxml2-v2.9.12-2-g85b1792e even the first line is indented. + * But virBufferAsprintf() also adds indentation. Skip one of them. */ + xmlbufContent = (const char *) xmlBufferContent(xmlbuf); + virSkipSpaces(&xmlbufContent); + + virBufferAsprintf(buf, "%s\n", xmlbufContent); xmlIndentTreeOutput = oldIndentTreeOutput; } -- 2.26.3

5 6