- Devel - libvirt List Archives

[libvirt PATCH] vz: fix vzCapsAddGuestDomain
by Ján Tomko 29 Nov '21

29 Nov '21

There is a stray 'return -1' executed on all code paths. Fixes: c18d9e23fafabcfbb80481e0705931036b8e7331 Signed-off-by: Ján Tomko <jtomko(a)redhat.com> --- src/vz/vz_driver.c | 2 -- 1 file changed, 2 deletions(-) diff --git a/src/vz/vz_driver.c b/src/vz/vz_driver.c index b47266290c..23b7795035 100644 --- a/src/vz/vz_driver.c +++ b/src/vz/vz_driver.c @@ -85,8 +85,6 @@ vzCapsAddGuestDomain(virCaps *caps, guest = virCapabilitiesAddGuest(caps, ostype, arch, emulator, NULL, 0, NULL); - return -1; - virCapabilitiesAddGuestDomain(guest, virt_type, NULL, NULL, 0, NULL); -- 2.31.1

2 1

[PATCH v2] tools: fix iterating over argv when recovering xattr
by Daniel P. Berrangé 29 Nov '21

29 Nov '21

The libvirt_recover_xattrs.sh tool hangs when run. When not flags are provided OPTIND is 1, so the loop expands to 'shift 0' which has not effect. Rewrite to just loop over $@ instead which involves less cleverness. Signed-off-by: Daniel P. Berrangé <berrange(a)redhat.com> --- tools/libvirt_recover_xattrs.sh | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/tools/libvirt_recover_xattrs.sh b/tools/libvirt_recover_xattrs.sh index be6ee84b5f..35c164a9c7 100755 --- a/tools/libvirt_recover_xattrs.sh +++ b/tools/libvirt_recover_xattrs.sh @@ -106,9 +106,9 @@ fix_xattrs() { shift $((OPTIND - 1)) if [ $# -gt 0 ]; then - while [ $# -gt 0 ]; do - fix_xattrs "$1" - shift $((OPTIND - 1)) + for arg in "$@" + do + fix_xattrs "$arg" done else if [ ${UNSAFE} -eq 1 ]; then -- 2.33.1

2 1

[PATCH v3 00/12] Implement detach device related APIs for test driver
by Luke Yue 29 Nov '21

29 Nov '21

v3: - Rebase to current master branch - Split some tests to different functions in PATCH 12/12 and add some tests link to v2: https://listman.redhat.com/archives/libvir-list/2021-August/msg00637.html link to CI: https://gitlab.com/lukedyue/libvirt/-/pipelines/405150749 Luke Yue (12): conf: Introduce virDomainInputDefRemove and fix memory leak domain_driver: extract DetachXXXDeviceConfig related functions and use them test_driver: Implement virDomainDetachDeviceFlags test_driver: Implement virDomainDetachDeviceAlias test_driver: Implement virDomainDetachDevice conf: Add tpm helpers for future use test_driver: add TPM support for testDomainDetachDeviceLiveAndConfig conf: Add a memballoon helper for future use test_driver: add memballoon support for testDomainDetachDeviceLiveAndConfig examples: xml: test: add xml for testing devices related APIs virshtest: add expectError parameter to testCompareOutputLit tests: Test detach-device and detach-device-alias for test driver examples/xml/test/testdevcontroller.xml | 1 + examples/xml/test/testdevdiskcdrom.xml | 5 + examples/xml/test/testdevfs.xml | 6 + examples/xml/test/testdevhostdev.xml | 5 + examples/xml/test/testdevif.xml | 6 + examples/xml/test/testdevinput.xml | 1 + examples/xml/test/testdevlease.xml | 5 + examples/xml/test/testdevmem.xml | 6 + examples/xml/test/testdevmemballoon.xml | 3 + examples/xml/test/testdevrng.xml | 4 + examples/xml/test/testdevshmem.xml | 4 + examples/xml/test/testdevsound.xml | 3 + examples/xml/test/testdevtpm.xml | 5 + examples/xml/test/testdevvsock.xml | 3 + examples/xml/test/testdevwatchdog.xml | 1 + examples/xml/test/testdomfc5.xml | 54 +++++ examples/xml/test/testnodeinline.xml | 54 +++++ src/conf/domain_conf.c | 103 ++++++++ src/conf/domain_conf.h | 12 + src/hypervisor/domain_driver.c | 302 ++++++++++++++++++++++++ src/hypervisor/domain_driver.h | 47 ++++ src/libvirt_private.syms | 20 ++ src/libxl/libxl_driver.c | 41 +--- src/lxc/lxc_driver.c | 37 +-- src/qemu/qemu_driver.c | 124 ++-------- src/test/test_driver.c | 239 +++++++++++++++++++ tests/virshtest.c | 157 +++++++++--- 27 files changed, 1054 insertions(+), 194 deletions(-) create mode 100644 examples/xml/test/testdevcontroller.xml create mode 100644 examples/xml/test/testdevdiskcdrom.xml create mode 100644 examples/xml/test/testdevfs.xml create mode 100644 examples/xml/test/testdevhostdev.xml create mode 100644 examples/xml/test/testdevif.xml create mode 100644 examples/xml/test/testdevinput.xml create mode 100644 examples/xml/test/testdevlease.xml create mode 100644 examples/xml/test/testdevmem.xml create mode 100644 examples/xml/test/testdevmemballoon.xml create mode 100644 examples/xml/test/testdevrng.xml create mode 100644 examples/xml/test/testdevshmem.xml create mode 100644 examples/xml/test/testdevsound.xml create mode 100644 examples/xml/test/testdevtpm.xml create mode 100644 examples/xml/test/testdevvsock.xml create mode 100644 examples/xml/test/testdevwatchdog.xml -- 2.33.1

2 26

[PATCH 0/2] qemu: Fix update of persistent confing on blockcommit
by Peter Krempa 29 Nov '21

29 Nov '21

When commiting a image which is configured via a 'volume' storage source we fail to update the persistent definition. Peter Krempa (2): qemuBlockJobRewriteConfigDiskSource: Add debug statements when skipping disk update virStorageSourceIsSameLocation: Special-case storage sources of type 'volume' src/conf/storage_source_conf.c | 7 +++++++ src/qemu/qemu_blockjob.c | 15 ++++++++++++--- 2 files changed, 19 insertions(+), 3 deletions(-) -- 2.31.1

2 6

[PATCH 0/2] wireshark: Fix build with newer wireshark
by Michal Privoznik 29 Nov '21

29 Nov '21

I've started seeing a build failure on rawhide because of new wireshark (3.6.0). The problem is that wmem_alloc() function which we use was removed from public symbols starting with the 3.6.0 release in favor of wmem_new(). However, as I was looking at the code I realized that wireshark already has a function for what our code does and decided to drop our implementation in favor of wireshark's. Michal Prívozník (2): wireshark: Switch to tvb_bytes_to_str() wireshark: Drop needless comment in dissect_xdr_bytes() tools/wireshark/src/packet-libvirt.c | 32 +++++++--------------------- 1 file changed, 8 insertions(+), 24 deletions(-) -- 2.32.0

2 3

Entering freeze for libvirt-7.10.0
by Jiri Denemark 29 Nov '21

29 Nov '21

I have just tagged v7.10.0-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

1 1

[libvirt PATCH 0/4] virnetserver: Make pool job name less generic
by Jiri Denemark 27 Nov '21

27 Nov '21

Jiri Denemark (4): virthreadpool: Copy job name virnetserver: Format functions consistently virnetserver: Use autoptr for virNetServer and virNetServerClient virnetserver: Make pool job name less generic src/rpc/virnetserver.c | 247 ++++++++++++++++++++--------------- src/rpc/virnetserverclient.h | 2 + src/util/virthreadpool.c | 5 +- 3 files changed, 145 insertions(+), 109 deletions(-) -- 2.34.0

2 5

[PATCH] tools: fix iterating over argv when recovering xattr
by Daniel P. Berrangé 26 Nov '21

26 Nov '21

The libvirt_recover_xattrs.sh tool hangs when run. It appears that calling 'shift' is not modifying the '$#' value, so the loop never terminates. Rewrite to just loop over $@ instead which involves less cleverness. Signed-off-by: Daniel P. Berrangé <berrange(a)redhat.com> --- tools/libvirt_recover_xattrs.sh | 7 +++---- 1 file changed, 3 insertions(+), 4 deletions(-) diff --git a/tools/libvirt_recover_xattrs.sh b/tools/libvirt_recover_xattrs.sh index be6ee84b5f..ae9a18bad8 100755 --- a/tools/libvirt_recover_xattrs.sh +++ b/tools/libvirt_recover_xattrs.sh @@ -104,11 +104,10 @@ fix_xattrs() { } -shift $((OPTIND - 1)) if [ $# -gt 0 ]; then - while [ $# -gt 0 ]; do - fix_xattrs "$1" - shift $((OPTIND - 1)) + for arg in "$@" + do + fix_xattrs "$arg" done else if [ ${UNSAFE} -eq 1 ]; then -- 2.33.1

2 1

[libvirt PATCH 00/10] Revert <acpi-hotplug-bridge>
by Laine Stump 26 Nov '21

26 Nov '21

Starting with commit 58ba0f6a3d7342fba29edbbf2bb9cb5497c870e5, support for overriding the default hotplug method for a guest using the <acpi-bridge-hotplug> subelement of <features>/<pci> was added to libvirt. This uses the QEMU global commandline switch: ICH9-LPC.acpi-pci-hotplug-with-bridge-support=<off|on> that was added to QEMU in qemu-6.1 (along with QEMU switching the default hotplug type for new Q35-based machinetypes from "native PCIe" to "ACPI"). Unfortunately, soon after we pushed the <acpi-bridge-hotplug> patches to libvirt (2 days after, to be exact), during a regular weekly meeting I attend with some of the QEMU developers, along with bugs that had been found in ACPI hotplug since it was made the default, they were also discussing issues they'd found with the implementation of the QEMU commandline switch to change back to native PCIe hotplug. Apparently the current method used by acpi-pci-hotplug-with-bridge-support is causing "confusion" in guests, so they were talking about the possibility of changing what the switch does (or replacing it), and suggested that libvirt should "hold off" on supporting it for now. (oops) (they didn't know that we had just pushed Ani's patches that used it) Since the current QEMU option is doomed to either changed behavior (which would result in a guest-visible ABI change) or full deprecation and replacement, it would be better for libvirt to not use it at all, and re-implement based on whatever replacement QEMU comes up with. Once a new XML element has appeared in a libvirt upstream release, we are committed to keeping it there "forever". However, since there has not yet been an upstream release since <acpi-bridge-hotplug> was added, there is still time to revert and avoid the endless support/maintenance burden. Not much time though! And that is the purpose of this patch series. They revert, in reverse order, Ani's 4 original patches adding the feature, along with Peter's 6 followup patches that correct a logic error and streamline/beef up the unit tests. (Note that it is still possible that QEMU would figure out a way out of this without modifying their current flag in any significant way, and in that case we could always "re-apply" the original patches. However, if we don't revert before the next release (Andrea has pointed out the freeze for RC1 is next Tuesday, and it would be best to have it done before then), we will no longer have the option to remove it.) There was some conflict resolution necessary after the "git revert" commands, due to other unrelated patches that changed test case data, and due to a couple of Peter's patches also touching up the i440fx pci-root-hotplug disabling test cases (which are *not* being reverted - they work as advertised!). Note that this series involves *re-adding* some things, just to remove them again in a later revert - this is because Peter's patches effectively reverted the addition of a QEMU capability flag that had been added in Ani's original patches. If anyone would prefer, I can squash those patches together so that the extra dance is eliminated from the diffs; it just seemed more mechanical to do it this way though. A more detailed explanation of the issue: Current Behavior of existing QEMU option ======================================== As far as I understand (and keep in mind that I have misunderstood and misinterpreted this *at least* 4 separate times since it was first explained to me!), the effect of this QEMU setting: -global ICH9-LPC.acpi-pci-hotplug-with-bridge-support=on is to: 1) enable ACPI hotplug (i.e. expose it to the guest) 2) disable native PCIe hotplug (don't expose it to the guest) By having only one of the two types of hotplug enabled, the intent is to force the guest to use the still-enabled type of hotplug. Unfortunately, after QEMU 6.1 was released with acpi-pci-hotplug=on as the default for new machinetypes, problems were encountered with ACPI hotplug, which caused more attention to be called to the QEMU switch, and the people looking into that found that enabling ACPI and disabling native PCIe hotplug doesn't necessarily have the desired effect of causing the guest to use ACPI for hotplug (or maybe it was the opposite direction). Instead, it "gets confused" (a very technical term, I know. You can ask Julia or Michael for a definition :-)). One possible fix that they talked about was changing the behavior of ICH9-LPC.acpi-pci-hotplug-with-bridge-support: Potential Change to Behavior of QEMU option =========================================== I know it's more complex than this (again, Julia or Michael can explain), but my basic understanding of the way that they're currently thinking of modifying the acpi-pci-hotplug-with-bridge-support option is to have everything the same, *except* that when acpi-hotplug=off, ACPI hotplug will *still be enabled* along with native PCIe hotplug; but because guest OSes prefer native hotplug over ACPI, native PCIe hotplug will be chosen in that case. (Presumably this change prevents the "confusion" that is seen with the existing behavior of the option). So essentially, the choice of whether to use ACPI is controlled by enabling/disabling native PCIe hotplug (which *kind of* goes against the libvirt philosophy of "the XML describes the virtual machine that is presented to the guest"; instead it becomes "the XML describes how the guest will behave"). Another possibility would be to completely scrap (well, deprecate and later remove) the current QEMU commandline switch in favor of one or more switches that behave differently but result in the desired behavior. In either case, I think the best course of action for libvirt is to revert the current <acpi-bridge-hotplug>, wait until QEMU settles down with a new workable set of switches, and then re-do libvirt support based on that. Laine Stump (10): Revert "qemu: capabilities: Remove QEMU_CAPS_PIIX4_ACPI_HOTPLUG_BRIDGE" Revert "qemuxml2xmltest: Convert all acpi-hotplug control related tests to DO_TEST_CAPS_LATEST" Revert "qemuxml2argvtest: Add '-enable' variants for ACPI-hotplug related cases" Revert "qemuxml2argvtest: Use real-caps testing for 'acpi-hotplug-bridge-disable'" Revert "qemuValidateDomainDefPCIFeature: Fix validation logic" Revert "qemuValidateDomainDefPCIFeature: un-break error messages" Revert "NEWS: document new acpi pci hotplug config option" Revert "qemu: command: add support for acpi-bridge-hotplug feature" Revert "conf: introduce support for acpi-bridge-hotplug feature" Revert "qemu: capablities: detect acpi-pci-hotplug-with-bridge-support" NEWS.rst | 8 -- docs/formatdomain.rst | 29 ------ docs/schemas/domaincommon.rng | 15 ---- src/conf/domain_conf.c | 89 +------------------ src/conf/domain_conf.h | 9 -- src/qemu/qemu_capabilities.c | 4 +- src/qemu/qemu_capabilities.h | 3 +- src/qemu/qemu_command.c | 19 ---- src/qemu/qemu_validate.c | 41 --------- .../caps_6.1.0.x86_64.xml | 1 - .../caps_6.2.0.x86_64.xml | 1 - ...-hotplug-bridge-disable.aarch64-latest.err | 1 - .../aarch64-acpi-hotplug-bridge-disable.xml | 13 --- ...-hotplug-bridge-disable.x86_64-latest.args | 34 ------- .../pc-i440fx-acpi-hotplug-bridge-disable.xml | 33 ------- ...i-hotplug-bridge-enable.x86_64-latest.args | 34 ------- .../pc-i440fx-acpi-hotplug-bridge-enable.xml | 33 ------- ...pi-hotplug-bridge-disable.x86_64-6.0.0.err | 1 - ...-hotplug-bridge-disable.x86_64-latest.args | 37 -------- .../q35-acpi-hotplug-bridge-disable.xml | 47 ---------- ...cpi-hotplug-bridge-enable.x86_64-6.0.0.err | 1 - ...i-hotplug-bridge-enable.x86_64-latest.args | 37 -------- .../q35-acpi-hotplug-bridge-enable.xml | 47 ---------- tests/qemuxml2argvtest.c | 10 --- ...i-hotplug-bridge-disable.x86_64-latest.xml | 36 -------- ...pi-hotplug-bridge-enable.x86_64-latest.xml | 36 -------- ...cpi-root-hotplug-disable.x86_64-latest.xml | 33 ------- .../pc-i440fx-acpi-root-hotplug-disable.xml | 1 + ...acpi-root-hotplug-enable.x86_64-latest.xml | 33 ------- .../pc-i440fx-acpi-root-hotplug-enable.xml | 1 + ...i-hotplug-bridge-disable.x86_64-latest.xml | 53 ----------- ...pi-hotplug-bridge-enable.x86_64-latest.xml | 53 ----------- tests/qemuxml2xmltest.c | 10 +-- 33 files changed, 9 insertions(+), 794 deletions(-) delete mode 100644 tests/qemuxml2argvdata/aarch64-acpi-hotplug-bridge-disable.aarch64-latest.err delete mode 100644 tests/qemuxml2argvdata/aarch64-acpi-hotplug-bridge-disable.xml delete mode 100644 tests/qemuxml2argvdata/pc-i440fx-acpi-hotplug-bridge-disable.x86_64-latest.args delete mode 100644 tests/qemuxml2argvdata/pc-i440fx-acpi-hotplug-bridge-disable.xml delete mode 100644 tests/qemuxml2argvdata/pc-i440fx-acpi-hotplug-bridge-enable.x86_64-latest.args delete mode 100644 tests/qemuxml2argvdata/pc-i440fx-acpi-hotplug-bridge-enable.xml delete mode 100644 tests/qemuxml2argvdata/q35-acpi-hotplug-bridge-disable.x86_64-6.0.0.err delete mode 100644 tests/qemuxml2argvdata/q35-acpi-hotplug-bridge-disable.x86_64-latest.args delete mode 100644 tests/qemuxml2argvdata/q35-acpi-hotplug-bridge-disable.xml delete mode 100644 tests/qemuxml2argvdata/q35-acpi-hotplug-bridge-enable.x86_64-6.0.0.err delete mode 100644 tests/qemuxml2argvdata/q35-acpi-hotplug-bridge-enable.x86_64-latest.args delete mode 100644 tests/qemuxml2argvdata/q35-acpi-hotplug-bridge-enable.xml delete mode 100644 tests/qemuxml2xmloutdata/pc-i440fx-acpi-hotplug-bridge-disable.x86_64-latest.xml delete mode 100644 tests/qemuxml2xmloutdata/pc-i440fx-acpi-hotplug-bridge-enable.x86_64-latest.xml delete mode 100644 tests/qemuxml2xmloutdata/pc-i440fx-acpi-root-hotplug-disable.x86_64-latest.xml create mode 120000 tests/qemuxml2xmloutdata/pc-i440fx-acpi-root-hotplug-disable.xml delete mode 100644 tests/qemuxml2xmloutdata/pc-i440fx-acpi-root-hotplug-enable.x86_64-latest.xml create mode 120000 tests/qemuxml2xmloutdata/pc-i440fx-acpi-root-hotplug-enable.xml delete mode 100644 tests/qemuxml2xmloutdata/q35-acpi-hotplug-bridge-disable.x86_64-latest.xml delete mode 100644 tests/qemuxml2xmloutdata/q35-acpi-hotplug-bridge-enable.x86_64-latest.xml -- 2.31.1

3 18

[PATCH for 7.10] qemu: Fix validation of PCI option rom settings on hotplug
by Peter Krempa 25 Nov '21

25 Nov '21

Commit 24be92b8e moved the option rom settings validation code to the validation callbacks, but that doesn't work properly with device hotplug as we assign addresses only after parsing the whole XML. The check is too strict for that and caused failures when hotplugging devices such as: <interface type='network'> <source network='default'/> <model type='virtio'/> <rom enabled='no'/> </interface> This patch relaxes the check in the validation callback to accept also _NONE and _UNASSIGNED address types and returns the check to 'qemuBuildRomProps' so that we preserve the full validation as we've used to. Fixes: 24be92b8e38762e9ba13e Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=2021437 Signed-off-by: Peter Krempa <pkrempa(a)redhat.com> --- src/qemu/qemu_command.c | 6 ++++++ src/qemu/qemu_validate.c | 4 +++- 2 files changed, 9 insertions(+), 1 deletion(-) diff --git a/src/qemu/qemu_command.c b/src/qemu/qemu_command.c index 7a185061d8..c47998aabd 100644 --- a/src/qemu/qemu_command.c +++ b/src/qemu/qemu_command.c @@ -1147,6 +1147,12 @@ qemuBuildRomProps(virJSONValue *props, !info->romfile) return 0; + if (info->type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI) { + virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s", + _("ROM tuning is only supported for PCI devices")); + return -1; + } + if (info->romenabled == VIR_TRISTATE_BOOL_NO) { romfile = ""; } else { diff --git a/src/qemu/qemu_validate.c b/src/qemu/qemu_validate.c index 397eea5ede..1de6e05101 100644 --- a/src/qemu/qemu_validate.c +++ b/src/qemu/qemu_validate.c @@ -1485,7 +1485,9 @@ qemuValidateDomainDeviceInfo(const virDomainDeviceDef *dev, } if (info->romenabled || info->rombar || info->romfile) { - if (info->type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI) { + if (info->type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_PCI && + info->type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_NONE && + info->type != VIR_DOMAIN_DEVICE_ADDRESS_TYPE_UNASSIGNED) { virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s", _("ROM tuning is only supported for PCI devices")); return -1; -- 2.31.1

2 1

[PATCH 0/6] virXMLPropStringLimit removal follow-up
by Peter Krempa 25 Nov '21

25 Nov '21

Tim pointed out that some of the code could be simplified a bit more by using virTristateBool parsing functions and also that strnlen could be used instead of strlen. This series does those cleanups as well as one further refactor of character verification in virDomainDeviceLoadparmIsValid. Peter Krempa (6): util: enum: Add helpers for converting virTristate* to a plain bool conf: seclabel: Parse booleans using virXMLPropTristateBool instead of virStringParseYesNo internal: Add STRLIM macro for checking string length using strnlen() virDomainDeviceLoadparmIsValid: Simplify value lenght check virDomainDeviceLoadparmIsValid: Use 'strspn' instead of a loop conf: domain: Convert all string length checks to STRLIM src/conf/domain_conf.c | 83 +++++++------------ src/internal.h | 9 ++ src/libvirt_private.syms | 2 + src/util/virenum.c | 54 ++++++++++++ src/util/virenum.h | 2 + .../machine-loadparm-s390-char-invalid.err | 2 +- .../machine-loadparm-s390-len-invalid.err | 2 +- 7 files changed, 99 insertions(+), 55 deletions(-) -- 2.31.1

2 9

[libvirt PATCH 00/11] qemu snapshot create cleanup and refactor
by Pavel Hrdina 25 Nov '21

25 Nov '21

Pavel Hrdina (11): qemu_snapshot: create: move XML parsing to separate function qemu_snapshot: create: move XML def validation to separate function qemu_snapshot: create: move disk align to separate function qemu_snapshot: create: move virDomainSnapshotAssignDef to both code paths qemu_snapshot: create: virDomainSnapshotGetCurrent is not used with redefine qemu_snapshot: create: drop update_current variable qemu_snapshot: create: refactor endjob in qemuSnapshotCreateXML qemu_snapshot: create: move saving metadata to separate function qemu_snapshot: create: move snapshot redefine to separate function qemu_snapshot: create: move snapshot create to separate function qemu_snapshot: create: drop redefine boolean src/qemu/qemu_snapshot.c | 409 ++++++++++++++++++++++++--------------- 1 file changed, 251 insertions(+), 158 deletions(-) -- 2.31.1

2 13

[PATCH for 7.10] qemu: monitor: Fix usage of 'query-blockstats'
by Peter Krempa 25 Nov '21

25 Nov '21

Commit bc24810c2cab modified code querying blockstats to use the 'query-nodes' parameter so that we can fetch stats also for images which are not attached to a frontend such as block copy and backup scratch images. Unfortunately that broke the old blockstats because if 'query-nodes' is enabled qemu doesn't output the 'qdev' parameter which our code used for matching to the disk and also qemu neglects to populate the frontend stats at all so we can't even switch to using nodenema for matching. To fix this we need to do two calls, one with 'query-nodes' disabled using the old logic to populate everything and then an aditional one which populates all the remaining images. Closes: https://gitlab.com/libvirt/libvirt/-/issues/246 Signed-off-by: Peter Krempa <pkrempa(a)redhat.com> --- src/qemu/qemu_monitor_json.c | 52 +++++++++++++++++++++++++++++++++--- tests/qemumonitorjsontest.c | 2 ++ 2 files changed, 50 insertions(+), 4 deletions(-) diff --git a/src/qemu/qemu_monitor_json.c b/src/qemu/qemu_monitor_json.c index 3030cff664..56f0b22b2a 100644 --- a/src/qemu/qemu_monitor_json.c +++ b/src/qemu/qemu_monitor_json.c @@ -2446,6 +2446,30 @@ qemuMonitorJSONGetOneBlockStatsInfo(virJSONValue *dev, } +static int +qemuMonitorJSONGetOneBlockStatsNodeInfo(virJSONValue *dev, + GHashTable *hash) +{ + qemuBlockStats *bstats = NULL; + int nstats = 0; + const char *nodename = NULL; + + if (!(nodename = virJSONValueObjectGetString(dev, "node-name"))) + return 0; + + /* we already have the stats */ + if (g_hash_table_contains(hash, nodename)) + return 0; + + if (!(bstats = qemuMonitorJSONBlockStatsCollectData(dev, &nstats))) + return -1; + + g_hash_table_insert(hash, g_strdup(nodename), bstats); + + return nstats; +} + + virJSONValue * qemuMonitorJSONQueryBlockstats(qemuMonitor *mon, bool queryNodes) @@ -2475,13 +2499,14 @@ qemuMonitorJSONGetAllBlockStatsInfo(qemuMonitor *mon, int nstats = 0; int rc; size_t i; - g_autoptr(virJSONValue) devices = NULL; + g_autoptr(virJSONValue) blockstatsDevices = NULL; + g_autoptr(virJSONValue) blockstatsNodes = NULL; - if (!(devices = qemuMonitorJSONQueryBlockstats(mon, true))) + if (!(blockstatsDevices = qemuMonitorJSONQueryBlockstats(mon, false))) return -1; - for (i = 0; i < virJSONValueArraySize(devices); i++) { - virJSONValue *dev = virJSONValueArrayGet(devices, i); + for (i = 0; i < virJSONValueArraySize(blockstatsDevices); i++) { + virJSONValue *dev = virJSONValueArrayGet(blockstatsDevices, i); const char *dev_name; if (!dev || virJSONValueGetType(dev) != VIR_JSON_TYPE_OBJECT) { @@ -2505,6 +2530,25 @@ qemuMonitorJSONGetAllBlockStatsInfo(qemuMonitor *mon, nstats = rc; } + if (!(blockstatsNodes = qemuMonitorJSONQueryBlockstats(mon, true))) + return -1; + + for (i = 0; i < virJSONValueArraySize(blockstatsNodes); i++) { + virJSONValue *dev = virJSONValueArrayGet(blockstatsNodes, i); + + if (!dev || virJSONValueGetType(dev) != VIR_JSON_TYPE_OBJECT) { + virReportError(VIR_ERR_INTERNAL_ERROR, "%s", + _("blockstats device entry was not in expected format")); + return -1; + } + + if ((rc = qemuMonitorJSONGetOneBlockStatsNodeInfo(dev, hash)) < 0) + return -1; + + if (rc > nstats) + nstats = rc; + } + return nstats; } diff --git a/tests/qemumonitorjsontest.c b/tests/qemumonitorjsontest.c index 0de523007b..bcf5caa9a4 100644 --- a/tests/qemumonitorjsontest.c +++ b/tests/qemumonitorjsontest.c @@ -1615,6 +1615,8 @@ testQemuMonitorJSONqemuMonitorJSONGetAllBlockStatsInfo(const void *opaque) if (qemuMonitorTestAddItem(test, "query-blockstats", reply) < 0) return -1; + if (qemuMonitorTestAddItem(test, "query-blockstats", reply) < 0) + return -1; #define CHECK0FULL(var, value, varformat, valformat) \ if (stats->var != value) { \ -- 2.31.1

3 2

[libvirt PATCH v2 00/12] Expose TPM availability in domain capabilities
by Daniel P. Berrangé 25 Nov '21

25 Nov '21

If we can report whuether TPM is available, then mgmt apps can enable it by default for new VMs. This is important because OS like Win11 consider TPM to be mandatory. In v2: - Avoid triggering reporting of errors when querying caps - Only skip reporting of emulator backend when swtpm is missing, not entire tpm suppport - Lots of refactoring to support point 1. Daniel P. Berrangé (12): util: use consistent naming for swtpm global variables util: rename typedef for parsing swtpm capabilities util: refactor TPM helper methods to reduce duplicationm util: replace TPM global variables with a struct array util: don't export virTPMEmulatorInit method util: ensure all TPM global vars access is protected by lock util: pull TPM capabilities probing out of main init method util: add a method for checking if swtpm is available conf: add TPM devices to domain capabilities qemu: fill in domain capabilities for TPMs qemu: mock swtpm initialization in tests run: detect daemons when run via wrapper commands docs/schemas/domaincaps.rng | 10 + run.in | 22 +- src/conf/domain_capabilities.c | 14 ++ src/conf/domain_capabilities.h | 10 + src/libvirt_private.syms | 2 +- src/qemu/qemu_capabilities.c | 34 +++ src/qemu/qemu_capabilities.h | 3 + src/qemu/qemu_tpm.c | 6 - src/util/virtpm.c | 232 ++++++++++-------- src/util/virtpm.h | 3 +- .../domaincapsdata/qemu_2.11.0-q35.x86_64.xml | 9 + .../domaincapsdata/qemu_2.11.0-tcg.x86_64.xml | 9 + tests/domaincapsdata/qemu_2.11.0.s390x.xml | 1 + tests/domaincapsdata/qemu_2.11.0.x86_64.xml | 9 + .../domaincapsdata/qemu_2.12.0-q35.x86_64.xml | 10 + .../domaincapsdata/qemu_2.12.0-tcg.x86_64.xml | 10 + .../qemu_2.12.0-virt.aarch64.xml | 1 + tests/domaincapsdata/qemu_2.12.0.aarch64.xml | 1 + tests/domaincapsdata/qemu_2.12.0.ppc64.xml | 1 + tests/domaincapsdata/qemu_2.12.0.s390x.xml | 1 + tests/domaincapsdata/qemu_2.12.0.x86_64.xml | 10 + .../domaincapsdata/qemu_3.0.0-q35.x86_64.xml | 10 + .../domaincapsdata/qemu_3.0.0-tcg.x86_64.xml | 10 + tests/domaincapsdata/qemu_3.0.0.ppc64.xml | 1 + tests/domaincapsdata/qemu_3.0.0.s390x.xml | 1 + tests/domaincapsdata/qemu_3.0.0.x86_64.xml | 10 + .../domaincapsdata/qemu_3.1.0-q35.x86_64.xml | 10 + .../domaincapsdata/qemu_3.1.0-tcg.x86_64.xml | 10 + tests/domaincapsdata/qemu_3.1.0.ppc64.xml | 1 + tests/domaincapsdata/qemu_3.1.0.x86_64.xml | 10 + .../domaincapsdata/qemu_4.0.0-q35.x86_64.xml | 10 + .../domaincapsdata/qemu_4.0.0-tcg.x86_64.xml | 10 + .../qemu_4.0.0-virt.aarch64.xml | 1 + tests/domaincapsdata/qemu_4.0.0.aarch64.xml | 1 + tests/domaincapsdata/qemu_4.0.0.ppc64.xml | 1 + tests/domaincapsdata/qemu_4.0.0.s390x.xml | 1 + tests/domaincapsdata/qemu_4.0.0.x86_64.xml | 10 + .../domaincapsdata/qemu_4.1.0-q35.x86_64.xml | 10 + .../domaincapsdata/qemu_4.1.0-tcg.x86_64.xml | 10 + tests/domaincapsdata/qemu_4.1.0.x86_64.xml | 10 + .../domaincapsdata/qemu_4.2.0-q35.x86_64.xml | 10 + .../domaincapsdata/qemu_4.2.0-tcg.x86_64.xml | 10 + .../qemu_4.2.0-virt.aarch64.xml | 1 + tests/domaincapsdata/qemu_4.2.0.aarch64.xml | 1 + tests/domaincapsdata/qemu_4.2.0.ppc64.xml | 6 + tests/domaincapsdata/qemu_4.2.0.s390x.xml | 1 + tests/domaincapsdata/qemu_4.2.0.x86_64.xml | 10 + .../domaincapsdata/qemu_5.0.0-q35.x86_64.xml | 10 + .../domaincapsdata/qemu_5.0.0-tcg.x86_64.xml | 10 + .../qemu_5.0.0-virt.aarch64.xml | 9 + tests/domaincapsdata/qemu_5.0.0.aarch64.xml | 9 + tests/domaincapsdata/qemu_5.0.0.ppc64.xml | 10 + tests/domaincapsdata/qemu_5.0.0.x86_64.xml | 10 + .../domaincapsdata/qemu_5.1.0-q35.x86_64.xml | 10 + .../domaincapsdata/qemu_5.1.0-tcg.x86_64.xml | 10 + tests/domaincapsdata/qemu_5.1.0.sparc.xml | 1 + tests/domaincapsdata/qemu_5.1.0.x86_64.xml | 10 + .../domaincapsdata/qemu_5.2.0-q35.x86_64.xml | 10 + .../domaincapsdata/qemu_5.2.0-tcg.x86_64.xml | 10 + .../qemu_5.2.0-virt.aarch64.xml | 9 + tests/domaincapsdata/qemu_5.2.0.aarch64.xml | 9 + tests/domaincapsdata/qemu_5.2.0.ppc64.xml | 10 + tests/domaincapsdata/qemu_5.2.0.s390x.xml | 1 + tests/domaincapsdata/qemu_5.2.0.x86_64.xml | 10 + .../domaincapsdata/qemu_6.0.0-q35.x86_64.xml | 10 + .../domaincapsdata/qemu_6.0.0-tcg.x86_64.xml | 10 + .../qemu_6.0.0-virt.aarch64.xml | 9 + tests/domaincapsdata/qemu_6.0.0.aarch64.xml | 9 + tests/domaincapsdata/qemu_6.0.0.s390x.xml | 1 + tests/domaincapsdata/qemu_6.0.0.x86_64.xml | 10 + .../domaincapsdata/qemu_6.1.0-q35.x86_64.xml | 10 + .../domaincapsdata/qemu_6.1.0-tcg.x86_64.xml | 10 + tests/domaincapsdata/qemu_6.1.0.x86_64.xml | 10 + .../domaincapsdata/qemu_6.2.0-q35.x86_64.xml | 10 + .../domaincapsdata/qemu_6.2.0-tcg.x86_64.xml | 10 + .../qemu_6.2.0-virt.aarch64.xml | 9 + tests/domaincapsdata/qemu_6.2.0.aarch64.xml | 9 + tests/domaincapsdata/qemu_6.2.0.ppc64.xml | 10 + tests/domaincapsdata/qemu_6.2.0.x86_64.xml | 10 + tests/domaincapstest.c | 9 + 80 files changed, 731 insertions(+), 127 deletions(-) -- 2.33.1

2 24

[libvirt PATCH 0/7] do not check return value of qemuDomainObjExitMonitor
by Ján Tomko 25 Nov '21

25 Nov '21

Ján Tomko (7): qemu: qemuDomainObjExitMonitor: do not warn on unused result qemu: do not propagate return value of qemuDomainObjExitMonitor qemu: do not check return value of qemuDomainObjExitMonitor qemu: do not check return value of qemuDomainObjExitMonitor qemu: remove ignore_value for qemuDomainObjExitMonitor qemu: turn qemuDomainObjExitMonitor into void qemu: absorb qemuDomainObjExitMonitorInternal src/qemu/THREADS.txt | 5 - src/qemu/qemu_backup.c | 22 ++-- src/qemu/qemu_block.c | 28 ++-- src/qemu/qemu_blockjob.c | 31 ++--- src/qemu/qemu_checkpoint.c | 12 +- src/qemu/qemu_domain.c | 54 +++----- src/qemu/qemu_domain.h | 7 +- src/qemu/qemu_driver.c | 173 +++++++++++-------------- src/qemu/qemu_hotplug.c | 214 +++++++++++-------------------- src/qemu/qemu_migration.c | 65 +++++----- src/qemu/qemu_migration_cookie.c | 3 +- src/qemu/qemu_migration_params.c | 12 +- src/qemu/qemu_process.c | 75 +++++------ src/qemu/qemu_snapshot.c | 11 +- 14 files changed, 291 insertions(+), 421 deletions(-) -- 2.31.1

2 9

[PATCH v2 0/2] qemu_alias: Refactor functions
by Kristina Hanicova 25 Nov '21

25 Nov '21

This is v2 of: https://listman.redhat.com/archives/libvir-list/2021-November/msg00748.html Diff to v1: * changed return type of functions if possible - when function always returned success (suggested by Martin) Kristina Hanicova (2): qemu_alias: Rewrite of code pattern qemu_alias: change return type to void if possible src/qemu/qemu_alias.c | 185 +++++++++++++++------------------------- src/qemu/qemu_alias.h | 44 +++++----- src/qemu/qemu_hotplug.c | 43 ++++------ 3 files changed, 105 insertions(+), 167 deletions(-) -- 2.31.1

2 4

[PATCH v2] qemu: Rewrite code to the pattern
by Kristina Hanicova 25 Nov '21

25 Nov '21

I have seen this pattern a lot in the project, so I decided to rewrite code I stumbled upon to the same pattern as well. Signed-off-by: Kristina Hanicova <khanicov(a)redhat.com> --- This is v2 of: https://listman.redhat.com/archives/libvir-list/2021-November/msg00747.html Diff to v1: * adding variable 'rc' to fix buggy code and keep the code equivalent (suggested by Jano) src/qemu/qemu_driver.c | 22 +++++++++------------- 1 file changed, 9 insertions(+), 13 deletions(-) diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c index 1959b639da..5c4b493f64 100644 --- a/src/qemu/qemu_driver.c +++ b/src/qemu/qemu_driver.c @@ -15975,6 +15975,7 @@ qemuDomainSetBlockIoTune(virDomainPtr dom, g_autofree char *drivealias = NULL; const char *qdevid = NULL; int ret = -1; + int rc = 0; size_t i; virDomainDiskDef *conf_disk = NULL; virDomainDiskDef *disk; @@ -16229,13 +16230,10 @@ qemuDomainSetBlockIoTune(virDomainPtr dom, !virStorageSourceIsEmpty(disk->src)) { qemuDomainObjEnterMonitor(driver, vm); - ret = qemuMonitorSetBlockIoThrottle(priv->mon, drivealias, qdevid, - &info); - if (qemuDomainObjExitMonitor(driver, vm) < 0) - ret = -1; - if (ret < 0) + rc = qemuMonitorSetBlockIoThrottle(priv->mon, drivealias, qdevid, &info); + + if (qemuDomainObjExitMonitor(driver, vm) < 0 || rc < 0) goto endjob; - ret = -1; } virDomainDiskSetBlockIOTune(disk, &info); @@ -16310,6 +16308,7 @@ qemuDomainGetBlockIoTune(virDomainPtr dom, g_autofree char *drivealias = NULL; const char *qdevid = NULL; int ret = -1; + int rc = 0; int maxparams; virCheckFlags(VIR_DOMAIN_AFFECT_LIVE | @@ -16361,10 +16360,9 @@ qemuDomainGetBlockIoTune(virDomainPtr dom, goto endjob; } qemuDomainObjEnterMonitor(driver, vm); - ret = qemuMonitorGetBlockIoThrottle(priv->mon, drivealias, qdevid, &reply); - if (qemuDomainObjExitMonitor(driver, vm) < 0) - goto endjob; - if (ret < 0) + rc = qemuMonitorGetBlockIoThrottle(priv->mon, drivealias, qdevid, &reply); + + if (qemuDomainObjExitMonitor(driver, vm) < 0 || rc < 0) goto endjob; } @@ -17375,10 +17373,8 @@ qemuDomainSetTime(virDomainPtr dom, if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_RTC_RESET_REINJECTION)) { qemuDomainObjEnterMonitor(driver, vm); rv = qemuMonitorRTCResetReinjection(priv->mon); - if (qemuDomainObjExitMonitor(driver, vm) < 0) - goto endjob; - if (rv < 0) + if (qemuDomainObjExitMonitor(driver, vm) < 0 || rv < 0) goto endjob; } -- 2.31.1

3 2

[PATCH] util: fix cache invalidation of swtpm capabilities
by Daniel P. Berrangé 24 Nov '21

24 Nov '21

The check for whether the swtpm binary was modified is checking pointers to the mtime field in two distinct structs, so will always compare different. This resulted in re-probing swtpm capabilities every time, as many as 20 times for a single VM launch. Fixes: commit 01cf7a1bb9f1da27ad8bcbaa82c4f7a948c6a793 Author: Stefan Berger <stefanb(a)us.ibm.com> Date: Thu Jul 25 14:22:04 2019 -0400 tpm: Check whether previously found executables were updated Signed-off-by: Daniel P. Berrangé <berrange(a)redhat.com> --- src/util/virtpm.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/util/virtpm.c b/src/util/virtpm.c index 8a77ba98ff..ffec23a535 100644 --- a/src/util/virtpm.c +++ b/src/util/virtpm.c @@ -301,7 +301,7 @@ virTPMEmulatorInit(void) findit = true; if (!findit && - &statbuf.st_mtime != &prgs[i].stat->st_mtime) + statbuf.st_mtime != prgs[i].stat->st_mtime) findit = true; } -- 2.33.1

4 3

[PATCH] qemu: Rewrite code to the pattern
by Kristina Hanicova 24 Nov '21

24 Nov '21

I have seen this pattern a lot in the project, so I decided to rewrite code I stumbled upon to the same pattern as well. Signed-off-by: Kristina Hanicova <khanicov(a)redhat.com> --- src/qemu/qemu_driver.c | 14 ++++---------- 1 file changed, 4 insertions(+), 10 deletions(-) diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c index 1959b639da..b938687189 100644 --- a/src/qemu/qemu_driver.c +++ b/src/qemu/qemu_driver.c @@ -16231,10 +16231,9 @@ qemuDomainSetBlockIoTune(virDomainPtr dom, qemuDomainObjEnterMonitor(driver, vm); ret = qemuMonitorSetBlockIoThrottle(priv->mon, drivealias, qdevid, &info); - if (qemuDomainObjExitMonitor(driver, vm) < 0) - ret = -1; - if (ret < 0) + if (qemuDomainObjExitMonitor(driver, vm) < 0 || ret < 0) goto endjob; + ret = -1; } @@ -16362,9 +16361,7 @@ qemuDomainGetBlockIoTune(virDomainPtr dom, } qemuDomainObjEnterMonitor(driver, vm); ret = qemuMonitorGetBlockIoThrottle(priv->mon, drivealias, qdevid, &reply); - if (qemuDomainObjExitMonitor(driver, vm) < 0) - goto endjob; - if (ret < 0) + if (qemuDomainObjExitMonitor(driver, vm) < 0 || ret < 0) goto endjob; } @@ -17375,10 +17372,7 @@ qemuDomainSetTime(virDomainPtr dom, if (virQEMUCapsGet(priv->qemuCaps, QEMU_CAPS_RTC_RESET_REINJECTION)) { qemuDomainObjEnterMonitor(driver, vm); rv = qemuMonitorRTCResetReinjection(priv->mon); - if (qemuDomainObjExitMonitor(driver, vm) < 0) - goto endjob; - - if (rv < 0) + if (qemuDomainObjExitMonitor(driver, vm) < 0 || rv < 0) goto endjob; } -- 2.31.1

3 3

[libvirt PATCH 0/3] Expose TPM availability in domain capabilities
by Daniel P. Berrangé 24 Nov '21

24 Nov '21

If we can report whuether TPM is available, then mgmt apps can enable it by default for new VMs. This is important because OS like Win11 consider TPM to be mandatory. Daniel P. Berrangé (3): conf: add TPM devices to domain capabilities qemu: fill in domain capabilities for TPMs qemu: mock swtpm initialization in tests docs/schemas/domaincaps.rng | 10 ++++++ src/conf/domain_capabilities.c | 14 ++++++++ src/conf/domain_capabilities.h | 10 ++++++ src/qemu/qemu_capabilities.c | 32 +++++++++++++++++++ src/qemu/qemu_capabilities.h | 3 ++ .../domaincapsdata/qemu_2.11.0-q35.x86_64.xml | 9 ++++++ .../domaincapsdata/qemu_2.11.0-tcg.x86_64.xml | 9 ++++++ tests/domaincapsdata/qemu_2.11.0.s390x.xml | 6 ++++ tests/domaincapsdata/qemu_2.11.0.x86_64.xml | 9 ++++++ .../domaincapsdata/qemu_2.12.0-q35.x86_64.xml | 10 ++++++ .../domaincapsdata/qemu_2.12.0-tcg.x86_64.xml | 10 ++++++ .../qemu_2.12.0-virt.aarch64.xml | 6 ++++ tests/domaincapsdata/qemu_2.12.0.aarch64.xml | 6 ++++ tests/domaincapsdata/qemu_2.12.0.ppc64.xml | 6 ++++ tests/domaincapsdata/qemu_2.12.0.s390x.xml | 6 ++++ tests/domaincapsdata/qemu_2.12.0.x86_64.xml | 10 ++++++ .../domaincapsdata/qemu_3.0.0-q35.x86_64.xml | 10 ++++++ .../domaincapsdata/qemu_3.0.0-tcg.x86_64.xml | 10 ++++++ tests/domaincapsdata/qemu_3.0.0.ppc64.xml | 6 ++++ tests/domaincapsdata/qemu_3.0.0.s390x.xml | 6 ++++ tests/domaincapsdata/qemu_3.0.0.x86_64.xml | 10 ++++++ .../domaincapsdata/qemu_3.1.0-q35.x86_64.xml | 10 ++++++ .../domaincapsdata/qemu_3.1.0-tcg.x86_64.xml | 10 ++++++ tests/domaincapsdata/qemu_3.1.0.ppc64.xml | 6 ++++ tests/domaincapsdata/qemu_3.1.0.x86_64.xml | 10 ++++++ .../domaincapsdata/qemu_4.0.0-q35.x86_64.xml | 10 ++++++ .../domaincapsdata/qemu_4.0.0-tcg.x86_64.xml | 10 ++++++ .../qemu_4.0.0-virt.aarch64.xml | 4 +++ tests/domaincapsdata/qemu_4.0.0.aarch64.xml | 4 +++ tests/domaincapsdata/qemu_4.0.0.ppc64.xml | 4 +++ tests/domaincapsdata/qemu_4.0.0.s390x.xml | 4 +++ tests/domaincapsdata/qemu_4.0.0.x86_64.xml | 10 ++++++ .../domaincapsdata/qemu_4.1.0-q35.x86_64.xml | 10 ++++++ .../domaincapsdata/qemu_4.1.0-tcg.x86_64.xml | 10 ++++++ tests/domaincapsdata/qemu_4.1.0.x86_64.xml | 10 ++++++ .../domaincapsdata/qemu_4.2.0-q35.x86_64.xml | 10 ++++++ .../domaincapsdata/qemu_4.2.0-tcg.x86_64.xml | 10 ++++++ .../qemu_4.2.0-virt.aarch64.xml | 4 +++ tests/domaincapsdata/qemu_4.2.0.aarch64.xml | 4 +++ tests/domaincapsdata/qemu_4.2.0.ppc64.xml | 6 ++++ tests/domaincapsdata/qemu_4.2.0.s390x.xml | 4 +++ tests/domaincapsdata/qemu_4.2.0.x86_64.xml | 10 ++++++ .../domaincapsdata/qemu_5.0.0-q35.x86_64.xml | 10 ++++++ .../domaincapsdata/qemu_5.0.0-tcg.x86_64.xml | 10 ++++++ .../qemu_5.0.0-virt.aarch64.xml | 9 ++++++ tests/domaincapsdata/qemu_5.0.0.aarch64.xml | 9 ++++++ tests/domaincapsdata/qemu_5.0.0.ppc64.xml | 10 ++++++ tests/domaincapsdata/qemu_5.0.0.x86_64.xml | 10 ++++++ .../domaincapsdata/qemu_5.1.0-q35.x86_64.xml | 10 ++++++ .../domaincapsdata/qemu_5.1.0-tcg.x86_64.xml | 10 ++++++ tests/domaincapsdata/qemu_5.1.0.sparc.xml | 4 +++ tests/domaincapsdata/qemu_5.1.0.x86_64.xml | 10 ++++++ .../domaincapsdata/qemu_5.2.0-q35.x86_64.xml | 10 ++++++ .../domaincapsdata/qemu_5.2.0-tcg.x86_64.xml | 10 ++++++ .../qemu_5.2.0-virt.aarch64.xml | 9 ++++++ tests/domaincapsdata/qemu_5.2.0.aarch64.xml | 9 ++++++ tests/domaincapsdata/qemu_5.2.0.ppc64.xml | 10 ++++++ tests/domaincapsdata/qemu_5.2.0.s390x.xml | 4 +++ tests/domaincapsdata/qemu_5.2.0.x86_64.xml | 10 ++++++ .../domaincapsdata/qemu_6.0.0-q35.x86_64.xml | 10 ++++++ .../domaincapsdata/qemu_6.0.0-tcg.x86_64.xml | 10 ++++++ .../qemu_6.0.0-virt.aarch64.xml | 9 ++++++ tests/domaincapsdata/qemu_6.0.0.aarch64.xml | 9 ++++++ tests/domaincapsdata/qemu_6.0.0.s390x.xml | 4 +++ tests/domaincapsdata/qemu_6.0.0.x86_64.xml | 10 ++++++ .../domaincapsdata/qemu_6.1.0-q35.x86_64.xml | 10 ++++++ .../domaincapsdata/qemu_6.1.0-tcg.x86_64.xml | 10 ++++++ tests/domaincapsdata/qemu_6.1.0.x86_64.xml | 10 ++++++ .../domaincapsdata/qemu_6.2.0-q35.x86_64.xml | 10 ++++++ .../domaincapsdata/qemu_6.2.0-tcg.x86_64.xml | 10 ++++++ .../qemu_6.2.0-virt.aarch64.xml | 9 ++++++ tests/domaincapsdata/qemu_6.2.0.aarch64.xml | 9 ++++++ tests/domaincapsdata/qemu_6.2.0.ppc64.xml | 10 ++++++ tests/domaincapsdata/qemu_6.2.0.x86_64.xml | 10 ++++++ tests/domaincapstest.c | 9 ++++++ 75 files changed, 661 insertions(+) -- 2.33.1

4 14

[PATCH] util: virExec may blocked by reading pipe if grandchild prematurely exit
by Yi Wang 24 Nov '21

24 Nov '21

From: Xu Chao <xu.chao6(a)zte.com.cn> When VIR_EXEC_DAEMON is set, if virPidFileAcquirePath/virSetInherit failed, then pipesync[0] can not be closed when granchild process exit, because pipesync[1] still opened in child process. and then saferead in child process may blocked forever, and left grandchild process in defunct state. Signed-off-by: Xu Chao <xu.chao6(a)zte.com.cn> Signed-off-by: Yi Wang <wang.yi59(a)zte.com.cn> --- src/util/vircommand.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/src/util/vircommand.c b/src/util/vircommand.c index fead373729..fa71f40d81 100644 --- a/src/util/vircommand.c +++ b/src/util/vircommand.c @@ -782,6 +782,9 @@ virExec(virCommand *cmd) } if (pid > 0) { + /* close pipe[1], then the pipe can be closed if grandchild + * died prematurely */ + VIR_FORCE_CLOSE(pipesync[1]); /* The parent expect us to have written the pid file before * exiting. Wait here for the child to write it and signal us. */ if (cmd->pidfile && -- 2.27.0

2 1

[PATCH 0/4] virXXXListAdd: Transfer definition ownership
by Michal Privoznik 24 Nov '21

24 Nov '21

At a lot of places we have the following pattern: virXXXDef *def = parseDef(); if (!(obj = virXXXObjListAdd(def))) goto clenaup; def = NULL; cleanup: virXXXDefFree(def); The 'def = NULL' step is necessary because the ownership of the definition was transferred onto the object. Well, this approach is fragile as it relies on developers remembering to set the variable explicitly. If the virXXXObjListAdd() would take address of @def then the explicit set to NULL can be left out. Please note, I've reworked only a few virXXXObjListAdd() functions to see whether these are desired or not. If merged, I can post patches for the rest. Michal Prívozník (4): virInterfaceObjListAssignDef: Transfer definition ownership virSecretObjListAdd: Transfer definition ownership virStoragePoolObjListAdd: Transfer definition ownership virDomainObjListAdd: Transfer definition ownership src/bhyve/bhyve_driver.c | 6 ++---- src/ch/ch_driver.c | 7 ++----- src/conf/domain_conf.c | 8 ++++---- src/conf/domain_conf.h | 2 +- src/conf/virdomainobjlist.c | 22 +++++++++++++--------- src/conf/virdomainobjlist.h | 2 +- src/conf/virinterfaceobj.c | 25 +++++++++++++++++++------ src/conf/virinterfaceobj.h | 2 +- src/conf/virsecretobj.c | 26 ++++++++++++++------------ src/conf/virsecretobj.h | 2 +- src/conf/virstorageobj.c | 29 ++++++++++++++++------------- src/conf/virstorageobj.h | 2 +- src/libxl/libxl_domain.c | 3 +-- src/libxl/libxl_driver.c | 30 ++++++++++-------------------- src/libxl/libxl_migration.c | 6 ++---- src/lxc/lxc_driver.c | 24 ++++++++---------------- src/openvz/openvz_conf.c | 3 +-- src/openvz/openvz_driver.c | 10 ++++------ src/qemu/qemu_driver.c | 30 ++++++++++-------------------- src/qemu/qemu_migration.c | 3 +-- src/qemu/qemu_snapshot.c | 9 +++------ src/secret/secret_driver.c | 4 ++-- src/storage/storage_driver.c | 5 ++--- src/test/test_driver.c | 31 ++++++++++++------------------- src/vmware/vmware_conf.c | 6 ++---- src/vmware/vmware_driver.c | 10 ++++------ 26 files changed, 137 insertions(+), 170 deletions(-) -- 2.32.0

2 5

[PATCH 0/2] libxl: Remove dead code
by Jim Fehlig 24 Nov '21

24 Nov '21

I was too lazy to dig through history, but I think 1/2 was the result of too much copy and paste when originally implementing jobs in the libxl driver. For 2/2, I think there was some ill-conceived locking of the libxlDomainObjPrivate in the past. Its use was killed but the object remained virObjectLockable. Jim Fehlig (2): libxl: Remove unused macros libxl: Don't derive libxlDomainObjPrivate from virObjectLockable src/libxl/libxl_domain.c | 38 ++++++-------------------------------- src/libxl/libxl_domain.h | 7 ------- 2 files changed, 6 insertions(+), 39 deletions(-) -- 2.33.0

2 3

[PATCH] qemu: Remove 'else' branches after 'return' or 'goto'
by Kristina Hanicova 24 Nov '21

24 Nov '21

I think it makes no sense to have else branches after return or goto as it will never reach them in cases it should not. This patch makes the code more readable (at least to me). Signed-off-by: Kristina Hanicova <khanicov(a)redhat.com> --- src/libvirt.c | 3 ++- src/qemu/qemu_agent.c | 6 +++--- src/qemu/qemu_alias.c | 9 +++++---- src/qemu/qemu_capabilities.c | 4 ++-- src/qemu/qemu_cgroup.c | 9 +++------ src/qemu/qemu_conf.c | 11 +++++------ src/qemu/qemu_domain_address.c | 11 +++++------ src/qemu/qemu_domainjob.c | 3 +-- src/qemu/qemu_driver.c | 32 ++++++++++++++++---------------- 9 files changed, 42 insertions(+), 46 deletions(-) diff --git a/src/libvirt.c b/src/libvirt.c index 80bdcd1db3..ef9fc403d0 100644 --- a/src/libvirt.c +++ b/src/libvirt.c @@ -663,7 +663,8 @@ virStateInitialize(bool privileged, virStateDriverTab[i]->name, virGetLastErrorMessage()); return -1; - } else if (ret == VIR_DRV_STATE_INIT_SKIPPED && mandatory) { + } + if (ret == VIR_DRV_STATE_INIT_SKIPPED && mandatory) { VIR_ERROR(_("Initialization of mandatory %s state driver skipped"), virStateDriverTab[i]->name); return -1; diff --git a/src/qemu/qemu_agent.c b/src/qemu/qemu_agent.c index c3b02569cd..75eb1620c0 100644 --- a/src/qemu/qemu_agent.c +++ b/src/qemu/qemu_agent.c @@ -989,8 +989,7 @@ qemuAgentCommandName(virJSONValue *cmd) const char *name = virJSONValueObjectGetString(cmd, "execute"); if (name) return name; - else - return "<unknown>"; + return "<unknown>"; } static int @@ -1029,7 +1028,8 @@ qemuAgentCheckError(virJSONValue *cmd, qemuAgentStringifyError(error)); return -1; - } else if (!virJSONValueObjectHasKey(reply, "return")) { + } + if (!virJSONValueObjectHasKey(reply, "return")) { g_autofree char *cmdstr = virJSONValueToString(cmd, false); g_autofree char *replystr = virJSONValueToString(reply, false); diff --git a/src/qemu/qemu_alias.c b/src/qemu/qemu_alias.c index a36f346592..5b525415de 100644 --- a/src/qemu/qemu_alias.c +++ b/src/qemu/qemu_alias.c @@ -137,7 +137,8 @@ qemuAssignDeviceControllerAlias(virDomainDef *domainDef, */ controller->info.alias = g_strdup("pci"); return 0; - } else if (controller->model == VIR_DOMAIN_CONTROLLER_MODEL_PCIE_ROOT) { + } + if (controller->model == VIR_DOMAIN_CONTROLLER_MODEL_PCIE_ROOT) { /* The pcie-root controller on Q35 machinetypes uses a * different naming convention ("pcie.0"), because it is * hardcoded that way in qemu. @@ -151,7 +152,8 @@ qemuAssignDeviceControllerAlias(virDomainDef *domainDef, */ controller->info.alias = g_strdup_printf("pci.%d", controller->idx); return 0; - } else if (controller->type == VIR_DOMAIN_CONTROLLER_TYPE_IDE) { + } + if (controller->type == VIR_DOMAIN_CONTROLLER_TYPE_IDE) { /* for any machine based on e.g. I440FX or G3Beige, the * first (and currently only) IDE controller is an integrated * controller hardcoded with id "ide" @@ -821,8 +823,7 @@ qemuAliasForSecret(const char *parentalias, { if (obj) return g_strdup_printf("%s-%s-secret0", parentalias, obj); - else - return g_strdup_printf("%s-secret0", parentalias); + return g_strdup_printf("%s-secret0", parentalias); } /* qemuAliasTLSObjFromSrcAlias diff --git a/src/qemu/qemu_capabilities.c b/src/qemu/qemu_capabilities.c index a4c492dde2..3b63996170 100644 --- a/src/qemu/qemu_capabilities.c +++ b/src/qemu/qemu_capabilities.c @@ -773,9 +773,9 @@ const char *virQEMUCapsArchToString(virArch arch) { if (arch == VIR_ARCH_I686) return "i386"; - else if (arch == VIR_ARCH_ARMV6L || arch == VIR_ARCH_ARMV7L) + if (arch == VIR_ARCH_ARMV6L || arch == VIR_ARCH_ARMV7L) return "arm"; - else if (arch == VIR_ARCH_OR32) + if (arch == VIR_ARCH_OR32) return "or32"; return virArchToString(arch); diff --git a/src/qemu/qemu_cgroup.c b/src/qemu/qemu_cgroup.c index 1e7b562b33..bcdfa90759 100644 --- a/src/qemu/qemu_cgroup.c +++ b/src/qemu/qemu_cgroup.c @@ -608,9 +608,8 @@ qemuSetupBlkioCgroup(virDomainObj *vm) virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s", _("Block I/O tuning is not available on this host")); return -1; - } else { - return 0; } + return 0; } return virDomainCgroupSetupBlkio(priv->cgroup, vm->def->blkio); @@ -629,9 +628,8 @@ qemuSetupMemoryCgroup(virDomainObj *vm) virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s", _("Memory cgroup is not available on this host")); return -1; - } else { - return 0; } + return 0; } return virDomainCgroupSetupMemtune(priv->cgroup, vm->def->mem); @@ -891,9 +889,8 @@ qemuSetupCpuCgroup(virDomainObj *vm) virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s", _("CPU tuning is not available on this host")); return -1; - } else { - return 0; } + return 0; } if (vm->def->cputune.sharesSpecified) { diff --git a/src/qemu/qemu_conf.c b/src/qemu/qemu_conf.c index 0451bc70ac..bbb05a4e81 100644 --- a/src/qemu/qemu_conf.c +++ b/src/qemu/qemu_conf.c @@ -1001,12 +1001,12 @@ virQEMUDriverConfigLoadMemoryEntry(virQEMUDriverConfig *cfg, g_autofree char *dir = NULL; int rc; - if ((rc = virConfGetValueString(conf, "memory_backing_dir", &dir)) < 0) { + if ((rc = virConfGetValueString(conf, "memory_backing_dir", &dir)) < 0) return -1; - } else if (rc > 0) { + + if (rc > 0) { VIR_FREE(cfg->memoryBackingDir); cfg->memoryBackingDir = g_strdup_printf("%s/libvirt/qemu", dir); - return 0; } return 0; @@ -1777,11 +1777,10 @@ qemuSharedDeviceAddRemoveInternal(virQEMUDriver *driver, if (dev->type == VIR_DOMAIN_DEVICE_DISK) return qemuSharedDiskAddRemoveInternal(driver, dev->data.disk, name, addDevice); - else if (dev->type == VIR_DOMAIN_DEVICE_HOSTDEV) + if (dev->type == VIR_DOMAIN_DEVICE_HOSTDEV) return qemuSharedHostdevAddRemoveInternal(driver, dev->data.hostdev, name, addDevice); - else - return 0; + return 0; } diff --git a/src/qemu/qemu_domain_address.c b/src/qemu/qemu_domain_address.c index 9abe2b84c8..2948630b75 100644 --- a/src/qemu/qemu_domain_address.c +++ b/src/qemu/qemu_domain_address.c @@ -58,13 +58,13 @@ qemuDomainGetSCSIControllerModel(const virDomainDef *def, if (qemuDomainIsPSeries(def)) return VIR_DOMAIN_CONTROLLER_MODEL_SCSI_IBMVSCSI; - else if (ARCH_IS_S390(def->os.arch)) + if (ARCH_IS_S390(def->os.arch)) return VIR_DOMAIN_CONTROLLER_MODEL_SCSI_VIRTIO_SCSI; - else if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_SCSI_LSI)) + if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_SCSI_LSI)) return VIR_DOMAIN_CONTROLLER_MODEL_SCSI_LSILOGIC; - else if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_VIRTIO_SCSI)) + if (virQEMUCapsGet(qemuCaps, QEMU_CAPS_VIRTIO_SCSI)) return VIR_DOMAIN_CONTROLLER_MODEL_SCSI_VIRTIO_SCSI; - else if (qemuDomainHasBuiltinESP(def)) + if (qemuDomainHasBuiltinESP(def)) return VIR_DOMAIN_CONTROLLER_MODEL_SCSI_NCR53C90; virReportError(VIR_ERR_INTERNAL_ERROR, @@ -1589,9 +1589,8 @@ qemuDomainCollectPCIAddress(virDomainDef *def G_GNUC_UNUSED, _("Bus 0 must be PCI for integrated PIIX3 " "USB or IDE controllers")); return -1; - } else { - return 0; } + return 0; } } diff --git a/src/qemu/qemu_domainjob.c b/src/qemu/qemu_domainjob.c index 03a8d9ba5f..cff8d7bb83 100644 --- a/src/qemu/qemu_domainjob.c +++ b/src/qemu/qemu_domainjob.c @@ -1036,8 +1036,7 @@ int qemuDomainObjBeginJob(virQEMUDriver *driver, QEMU_AGENT_JOB_NONE, QEMU_ASYNC_JOB_NONE, false) < 0) return -1; - else - return 0; + return 0; } /** diff --git a/src/qemu/qemu_driver.c b/src/qemu/qemu_driver.c index 16d449913c..437bf63cd4 100644 --- a/src/qemu/qemu_driver.c +++ b/src/qemu/qemu_driver.c @@ -1819,7 +1819,8 @@ static int qemuDomainSuspend(virDomainPtr dom) virReportError(VIR_ERR_OPERATION_INVALID, "%s", _("domain is pmsuspended")); goto endjob; - } else if (state != VIR_DOMAIN_PAUSED) { + } + if (state != VIR_DOMAIN_PAUSED) { if (qemuProcessStopCPUs(driver, vm, reason, QEMU_ASYNC_JOB_NONE) < 0) goto endjob; } @@ -1861,13 +1862,15 @@ static int qemuDomainResume(virDomainPtr dom) virReportError(VIR_ERR_OPERATION_INVALID, "%s", _("domain is pmsuspended")); goto endjob; - } else if (state == VIR_DOMAIN_RUNNING) { + } + if (state == VIR_DOMAIN_RUNNING) { virReportError(VIR_ERR_OPERATION_INVALID, "%s", _("domain is already running")); goto endjob; - } else if ((state == VIR_DOMAIN_CRASHED && - reason == VIR_DOMAIN_CRASHED_PANICKED) || - state == VIR_DOMAIN_PAUSED) { + } + if ((state == VIR_DOMAIN_CRASHED && + reason == VIR_DOMAIN_CRASHED_PANICKED) || + state == VIR_DOMAIN_PAUSED) { if (qemuProcessStartCPUs(driver, vm, VIR_DOMAIN_RUNNING_UNPAUSED, QEMU_ASYNC_JOB_NONE) < 0) { @@ -16358,9 +16361,9 @@ qemuDomainGetBlockIoTune(virDomainPtr dom, *nparams = maxparams; ret = 0; goto endjob; - } else if (*nparams < maxparams) { - maxparams = *nparams; } + if (*nparams < maxparams) + maxparams = *nparams; *nparams = 0; @@ -20428,29 +20431,26 @@ qemuDomainGetGuestInfo(virDomainPtr dom, if (supportedTypes & VIR_DOMAIN_GUEST_INFO_FILESYSTEM) { rc = qemuAgentGetFSInfo(agent, &agentfsinfo, report_unsupported); - if (rc == -1) { + if (rc == -1) goto exitagent; - } else if (rc >= 0) { + if (rc >= 0) nfs = rc; - } } if (supportedTypes & VIR_DOMAIN_GUEST_INFO_DISKS) { rc = qemuAgentGetDisks(agent, &agentdiskinfo, report_unsupported); - if (rc == -1) { + if (rc == -1) goto exitagent; - } else if (rc >= 0) { + if (rc >= 0) ndisks = rc; - } } if (supportedTypes & VIR_DOMAIN_GUEST_INFO_INTERFACES) { rc = qemuAgentGetInterfaces(agent, &ifaces, report_unsupported); - if (rc == -1) { + if (rc == -1) goto exitagent; - } else if (rc >= 0) { + if (rc >= 0) nifaces = rc; - } } qemuDomainObjExitAgent(vm, agent); -- 2.31.1

2 1

[RFC PATCH 0/3] Add virDomainInjectLaunchSecret API
by Jim Fehlig 23 Nov '21

23 Nov '21

Hi All! This series is an RFC impl for the missing API needed for SEV attestation as discussed here https://listman.redhat.com/archives/libvir-list/2021-May/msg00196.html I pinged about the status a few weeks back, found it had stalled, and agreed to work on an impl after returning from vacation https://listman.redhat.com/archives/libvir-list/2021-October/msg01052.html Although the series is only compile tested, I wanted to share it early in case others are considering the task. While discussing the missing API, Daniel suggested virDomainSetLaunchSecurityInfo https://listman.redhat.com/archives/libvir-list/2021-October/msg01074.html but noted the asymmetry with virDomainGetLaunchSecurityInfo. I decided to go with virDomainInjectLaunchSecret, which better describes the function IMO. I also decided to go with an explicit set of parameters, following in the footsteps of virDrvDomainAuthorizedSSHKeys*. It wasn't until patch 3 that I realized virTypedParameter is definitely a better approach for an API that may need future support for other types of secrets. I'll make that change in a V1 after collecting feedback on this RFC. Regards, Jim Jim Fehlig (3): libvirt: Introduce virDomainInjectLaunchSecret public API remote: Implement domain inject launch secret API qemu: Implement the virDomainInjectLaunchSecret API include/libvirt/libvirt-domain.h | 6 ++++ src/driver-hypervisor.h | 8 +++++ src/libvirt-domain.c | 50 +++++++++++++++++++++++++++ src/libvirt_public.syms | 5 +++ src/qemu/qemu_driver.c | 53 +++++++++++++++++++++++++++++ src/qemu/qemu_monitor.c | 12 +++++++ src/qemu/qemu_monitor.h | 6 ++++ src/qemu/qemu_monitor_json.c | 34 ++++++++++++++++++ src/qemu/qemu_monitor_json.h | 5 +++ src/remote/remote_daemon_dispatch.c | 27 +++++++++++++++ src/remote/remote_driver.c | 32 +++++++++++++++++ src/remote/remote_protocol.x | 16 ++++++++- src/remote_protocol-structs | 8 +++++ 13 files changed, 261 insertions(+), 1 deletion(-) -- 2.33.0

3 9