- Devel - libvirt List Archives

[PATCH 0/2] disk iomode validation cleanups
by Peter Krempa 07 Dec '21

07 Dec '21

Peter Krempa (2): qemuValidateDomainDeviceDefDiskFrontend: Fix error message if io='native' is unsupported qemuValidateDomainDeviceDefDiskFrontend: Aggregate disk iomode validation src/qemu/qemu_validate.c | 38 ++++++++++++++++++++++---------------- 1 file changed, 22 insertions(+), 16 deletions(-) -- 2.31.1

2 4

[libvirt PATCH v2] qemu: Enable unprivileged userfaultfd for post-copy migration
by Jiri Denemark 07 Dec '21

07 Dec '21

Userfaultfd is by default allowed only for privileged processes. Since libvirt runs QEMU unprivileged, we need to enable unprivileged access to userfaultfd before starting post-copy migration. Rather than providing a static sysctl configuration file, we set the sysctl knob in runtime once post-copy migration is requested. This way unprivileged_userfaultfd is only enabled once actually used. https://bugzilla.redhat.com/show_bug.cgi?id=1945420 Signed-off-by: Jiri Denemark <jdenemar(a)redhat.com> --- Notes: Version 2: - setting unprivileged_userfaultfd only when it is not already enabled - virReportSystemError replaced with VIR_WARN src/qemu/qemu_migration_params.c | 42 ++++++++++++++++++++++++++++++++ 1 file changed, 42 insertions(+) diff --git a/src/qemu/qemu_migration_params.c b/src/qemu/qemu_migration_params.c index dbc3219826..9ba4811242 100644 --- a/src/qemu/qemu_migration_params.c +++ b/src/qemu/qemu_migration_params.c @@ -804,6 +804,41 @@ qemuMigrationCapsToJSON(virBitmap *caps, } +/** + * qemuMigrationParamsEnableUserfaultfd + * + * Try to enable unprivileged userfaultfd unless it's missing or already + * enabled. Only a warning is logged when we cannot enable it, QEMU will + * report an error when enabling post-copy migration capability. + */ +static void +qemuMigrationParamsEnableUserfaultfd(void) +{ + const char *sysctl = "/proc/sys/vm/unprivileged_userfaultfd"; + g_autofree char *buf = NULL; + + if (!virFileExists(sysctl)) + return; + + if (virFileReadAll(sysctl, 10, &buf) < 0) { + VIR_WARN("Cannot read unprivileged userfaultfd state"); + return; + } + + if (STREQ(buf, "1\n")) { + VIR_DEBUG("Unprivileged userfaultfd already enabled"); + return; + } + + VIR_DEBUG("Enabling unprivileged userfaultfd for post-copy migration"); + + if (virFileWriteStr(sysctl, "1", 0) < 0) { + VIR_WARN("Failed to enable unprivileged userfaultfd: %s", + g_strerror(errno)); + } +} + + /** * qemuMigrationParamsApply * @driver: qemu driver @@ -839,6 +874,13 @@ qemuMigrationParamsApply(virQEMUDriver *driver, goto cleanup; } } else { + /* userfaultfd may only be enabled for privileged processes by default, + * we need to make sure QEMU can use it before enabling post-copy + * migration */ + if (virBitmapIsBitSet(priv->migrationCaps, QEMU_MIGRATION_CAP_POSTCOPY) && + virBitmapIsBitSet(migParams->caps, QEMU_MIGRATION_CAP_POSTCOPY)) + qemuMigrationParamsEnableUserfaultfd(); + if (!(caps = qemuMigrationCapsToJSON(priv->migrationCaps, migParams->caps))) goto cleanup; -- 2.34.1

3 3

[PATCH 0/3] Add virDomainSetLaunchSecurityState API
by Jim Fehlig 06 Dec '21

06 Dec '21

V1 series based on initial RFC found here https://listman.redhat.com/archives/libvir-list/2021-November/msg00460.html This series is compile tested only. Functional testing will require reserving the test machine, configuring it, and writing some form of functional test :-). Is anyone aware of functional tests for the existing SEV-related APIs? Also, it has been a while since I added a new API, so not sure about the current state of the language bindings. Are there any non-automated tasks to expose the API in the various bindings? Jim Fehlig (3): libvirt: Introduce virDomainSetLaunchSecurityState public API remote: Add RPC support for the virDomainSetLaunchSecurityState API qemu: Implement the virDomainSetLaunchSecurityState API include/libvirt/libvirt-domain.h | 35 ++++++++++++++ src/driver-hypervisor.h | 7 +++ src/libvirt-domain.c | 52 +++++++++++++++++++++ src/libvirt_public.syms | 5 ++ src/qemu/qemu_driver.c | 78 ++++++++++++++++++++++++++++++++ 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_driver.c | 1 + src/remote/remote_protocol.x | 17 ++++++- src/remote_protocol-structs | 9 ++++ 12 files changed, 260 insertions(+), 1 deletion(-) -- 2.33.0

2 9

Re: [PATCH for-7.0] i386: Deprecate the -no-hpet QEMU command line option
by Thomas Huth 06 Dec '21

06 Dec '21

On 06/12/2021 09.40, Thomas Huth wrote: > The HPET setting has been turned into a machine property a while ago > already, so we should finally do the next step and deprecate the > legacy CLI option, too. > While we're at it, add a proper help text for the machine property, too. > > Signed-off-by: Thomas Huth <thuth(a)redhat.com> > --- > docs/about/deprecated.rst | 6 ++++++ > hw/i386/pc.c | 2 ++ > qemu-options.hx | 2 +- > softmmu/vl.c | 1 + > 4 files changed, 10 insertions(+), 1 deletion(-) > > diff --git a/docs/about/deprecated.rst b/docs/about/deprecated.rst > index 5693abb663..1dfe69aa6a 100644 > --- a/docs/about/deprecated.rst > +++ b/docs/about/deprecated.rst > @@ -198,6 +198,12 @@ form is preferred. > Using ``-drive if=none`` to configure the OTP device of the sifive_u > RISC-V machine is deprecated. Use ``-drive if=pflash`` instead. > > +``-no-hpet`` (since 7.0) > +'''''''''''''''''''''''' > + > +The HPET setting has been turned into a machine property. > +Use ``-machine hpet=off`` instead. [...] Forgot to CC: the libvirt folks, doing so now. Seems like libvirt is still using -no-hpet in some few spots, so I guess these would need to be changed first, before we could finally remove this option in QEMU? Thomas

3 5

[libvirt PATCH v2 0/2] Fix failure to find VF netdev names during virtual network start
by Laine Stump 06 Dec '21

06 Dec '21

The first patch resolves https://bugzilla.redhat.com/2025432, the 2nd simplifies lower level code in the same manner. V1 is here: https://listman.redhat.com/archives/libvir-list/2021-December/msg00000.html Change in V2: Rather than adding an extra bool to the arglist of virPCIGetVirtualFunctionsFull() (what I did in V1), just switch to sending the name of the netdev whose phys_port_id we want to match (called physPortNetDevName) (which will always be non-NULL in the cases where we want) rather than the phys_port_id itself (physPortID) (which may or may not be NULL). This way we don't need an extra arg (we can just check for physPortNetDevName != NULL), and the lower level function can call virNetDevGetPhysPortID() as needed. Also added a similar 2nd patch that pushes the call to virNetDevGetPhysPortID() down even further, into virPCIGetNetName(). This simplifies the callers, and concentrates all calls to virNetDevGetPhysPortID() into a single function (virPCIGetNetName(), duh). Laine Stump (2): util: fix erroneous requirement for phys_port_id to get ifname of a VF util: call virNetDevGetPhysPortID() in less places src/util/virnetdev.c | 24 +++----------------- src/util/virpci.c | 52 ++++++++++++++++++++++++++------------------ src/util/virpci.h | 4 ++-- 3 files changed, 36 insertions(+), 44 deletions(-) -- 2.33.1

2 3

[libvirt PATCH v2 00/15] WIP/RFC: add QEMU "-display dbus" support
by marcandre.lureau＠redhat.com 06 Dec '21

06 Dec '21

From: Marc-André Lureau <marcandre.lureau(a)redhat.com> Hi, This series implements supports for the upcoming QEMU "-display dbus" support. Development is still in progress, but I hope to land the QEMU support early in 6.3 (last version posted: https://patchew.org/QEMU/20211009210838.2219430-1-marcandre.lureau@redhat.c…) By default, libvirt will start a private VM bus (sharing and reusing the existing "vmstate" VM bus & code). The feature set should cover the needs to replace Spice as local client of choice, including 3daccel/dmabuf, audio, clipboard sharing, usb redirection, and arbitrary chardev/channels (for serial etc). The test Gtk4 client is also in progress, currently in development at https://gitlab.com/marcandre.lureau/qemu-display/. A few dependencies, such as zbus, require an upcoming release. virt-viewer & boxes will need a port to Gtk4 to make use of the shared widget. Comments welcome, as we can still adjust the QEMU side etc. thanks v2: after Michal review - split patches for conf & qemu changes - start D-Bus in qemuExtDevicesStart() - use symlinks for xml2xml tests - teach domdisplay to return a dbus URI - rebase & bug fixes Marc-André Lureau (15): qemu: add chardev-vdagent capability check qemu: add -display dbus capability check (to update to 6.3) conf: add <graphics type='dbus'> qemu: start the D-Bus daemon for the display qemu: add -display dbus support virsh: refactor/split cmdDomDisplay() virsh: report the D-Bus bus URI for domdisplay conf: add <audio type='dbus'> support qemu: add audio type 'dbus' conf: add dbus <clipboard> qemu: add dbus clipboard sharing conf: add <serial type='dbus'> qemu: add -chardev dbus support qemu: add usbredir type 'dbus' docs: document <graphics> type dbus docs/formatdomain.rst | 43 +- docs/schemas/basictypes.rng | 7 + docs/schemas/domaincommon.rng | 71 ++++ src/bhyve/bhyve_command.c | 1 + src/conf/domain_conf.c | 141 ++++++- src/conf/domain_conf.h | 15 + src/conf/domain_validate.c | 41 +- src/libxl/libxl_conf.c | 1 + src/qemu/qemu_capabilities.c | 6 + src/qemu/qemu_capabilities.h | 2 + src/qemu/qemu_command.c | 80 +++- src/qemu/qemu_domain.c | 1 + src/qemu/qemu_driver.c | 10 +- src/qemu/qemu_extdevice.c | 13 + src/qemu/qemu_hotplug.c | 1 + src/qemu/qemu_monitor_json.c | 10 + src/qemu/qemu_process.c | 40 +- src/qemu/qemu_validate.c | 33 ++ src/security/security_dac.c | 2 + src/vmx/vmx.c | 1 + .../domaincapsdata/qemu_6.2.0-q35.x86_64.xml | 1 + .../domaincapsdata/qemu_6.2.0-tcg.x86_64.xml | 1 + tests/domaincapsdata/qemu_6.2.0.x86_64.xml | 1 + .../caps_6.1.0.x86_64.xml | 1 + .../caps_6.2.0.aarch64.xml | 1 + .../caps_6.2.0.x86_64.replies | 10 +- .../caps_6.2.0.x86_64.xml | 2 + .../graphics-dbus-address.args | 30 ++ .../graphics-dbus-address.xml | 35 ++ .../qemuxml2argvdata/graphics-dbus-audio.args | 33 ++ .../qemuxml2argvdata/graphics-dbus-audio.xml | 45 +++ .../graphics-dbus-chardev.args | 32 ++ .../graphics-dbus-chardev.xml | 43 ++ .../graphics-dbus-clipboard.args | 31 ++ .../graphics-dbus-clipboard.xml | 35 ++ tests/qemuxml2argvdata/graphics-dbus-p2p.args | 30 ++ tests/qemuxml2argvdata/graphics-dbus-p2p.xml | 33 ++ .../graphics-dbus-usbredir.args | 34 ++ .../graphics-dbus-usbredir.xml | 30 ++ tests/qemuxml2argvdata/graphics-dbus.args | 30 ++ tests/qemuxml2argvdata/graphics-dbus.xml | 33 ++ tests/qemuxml2argvtest.c | 21 + .../graphics-dbus-address.xml | 1 + .../graphics-dbus-audio.xml | 1 + .../graphics-dbus-chardev.xml | 1 + .../graphics-dbus-clipboard.xml | 1 + .../qemuxml2xmloutdata/graphics-dbus-p2p.xml | 1 + tests/qemuxml2xmloutdata/graphics-dbus.xml | 1 + tests/qemuxml2xmltest.c | 20 + tools/virsh-domain.c | 366 +++++++++--------- 50 files changed, 1231 insertions(+), 192 deletions(-) create mode 100644 tests/qemuxml2argvdata/graphics-dbus-address.args create mode 100644 tests/qemuxml2argvdata/graphics-dbus-address.xml create mode 100644 tests/qemuxml2argvdata/graphics-dbus-audio.args create mode 100644 tests/qemuxml2argvdata/graphics-dbus-audio.xml create mode 100644 tests/qemuxml2argvdata/graphics-dbus-chardev.args create mode 100644 tests/qemuxml2argvdata/graphics-dbus-chardev.xml create mode 100644 tests/qemuxml2argvdata/graphics-dbus-clipboard.args create mode 100644 tests/qemuxml2argvdata/graphics-dbus-clipboard.xml create mode 100644 tests/qemuxml2argvdata/graphics-dbus-p2p.args create mode 100644 tests/qemuxml2argvdata/graphics-dbus-p2p.xml create mode 100644 tests/qemuxml2argvdata/graphics-dbus-usbredir.args create mode 100644 tests/qemuxml2argvdata/graphics-dbus-usbredir.xml create mode 100644 tests/qemuxml2argvdata/graphics-dbus.args create mode 100644 tests/qemuxml2argvdata/graphics-dbus.xml create mode 120000 tests/qemuxml2xmloutdata/graphics-dbus-address.xml create mode 120000 tests/qemuxml2xmloutdata/graphics-dbus-audio.xml create mode 120000 tests/qemuxml2xmloutdata/graphics-dbus-chardev.xml create mode 120000 tests/qemuxml2xmloutdata/graphics-dbus-clipboard.xml create mode 120000 tests/qemuxml2xmloutdata/graphics-dbus-p2p.xml create mode 120000 tests/qemuxml2xmloutdata/graphics-dbus.xml -- 2.34.1.8.g35151cf07204

4 19

[PATCH 00/17] qemu: Add support for blocking writes for the blockdev-mirror job
by Peter Krempa 03 Dec '21

03 Dec '21

Blocking writes ensure that the mirroring converges even when the guest is I/O intensive with a fast local storage and slow mirror. This patchset does it by introducing flags which use the blocking mode as it will have performance impact, and for guests which do I/O in bursts it might be detrimental to their performance. One could argue that both the copy job and migration have the expectations of actually using the copy so ensuring that it converges might be also something we'd want to do by default. This would obviously greatly simplify this series, but I didn't want to change the default. Peter Krempa (17): qemu: monitor: Avoid ternary operators in helpers for drive/blockdev-mirror qemuMonitorJSONHandleShutdown: Use virTristateBoolFromBool qemuMonitorJSONEjectMedia: Use a bool directly for constructing JSON with 'b' modifier qemuMonitorJSONMigrate: Extract flags prior to constructing command qemuMonitorJSONGraphicsRelocate: Clean up command argument construction qemu: monitor: Add support for 'write-blocking' copy mode for blockdev-mirror include: virDomainBlockCopyFlags: Convert to prefix comments virDomainBlockCopy: Introduce VIR_DOMAIN_BLOCK_COPY_SYNCHRONOUS_WRITES flag qemuDomainBlockCopy: Implement VIR_DOMAIN_BLOCK_COPY_SYNCHRONOUS_WRITES include: virDomainMigrateFlags: Remove "block alignment" whitespace man: virsh: Separate paragrapsh describing distinct flags VIR_REQUIRE_FLAG_(GOTO|RET): Add parens around arguments in expansion virDomainMigrate: Introduce VIR_MIGRATE_NON_SHARED_SYNCHRONOUS_WRITES flag qemu: migration: Implement VIR_MIGRATE_NON_SHARED_SYNCHRONOUS_WRITES docs: Convert 'migration' doc to RST docs: migration: Add a paragraph about non-shared storage migration NEWS: Mention synchronous copy job additions NEWS.rst | 12 + docs/manpages/virsh.rst | 45 +- docs/meson.build | 2 +- docs/migration.html.in | 688 ------------------------------- docs/migration.rst | 490 ++++++++++++++++++++++ include/libvirt/libvirt-domain.h | 63 +-- src/internal.h | 4 +- src/libvirt-domain.c | 26 ++ src/qemu/qemu_driver.c | 17 +- src/qemu/qemu_migration.c | 17 +- src/qemu/qemu_migration.h | 1 + src/qemu/qemu_monitor.c | 10 +- src/qemu/qemu_monitor.h | 3 +- src/qemu/qemu_monitor_json.c | 67 ++- src/qemu/qemu_monitor_json.h | 3 +- tests/qemumonitorjsontest.c | 2 +- tools/virsh-domain.c | 22 +- 17 files changed, 707 insertions(+), 765 deletions(-) delete mode 100644 docs/migration.html.in create mode 100644 docs/migration.rst -- 2.31.1

2 21

[PATCH 00/16] Clean up of virJSONValueFree usage and few related refactors
by Peter Krempa 03 Dec '21

03 Dec '21

Peter Krempa (16): qemu: hotplug: Use automatic freeing for virJSONValue qemuDomainHotplugAddIOThread: Automatically free virJSONValue qemuMonitorBlockdevCreate: Use double pointer instead of always consuming '@props' virLockDaemonClientPreExecRestart: Modernize JSON object construction virLockDaemonPostExecRestart: Automatically free temporary variables virLockDaemonPostExecRestart: Refactor cleanup virLogDaemonPostExecRestart: Use automatic freeing for variables virLogDaemonPostExecRestart: Refactor cleanup virCHProcessUpdateInfo: Automatically free virJSONValue tests/virnetdaemontest.c: testExecRestart: Automatically free virJSONValue-s qemuAgentGuestSync: Don't use goto for looping qemuMonitorJSONGetCPUModelExpansion: Don't use goto for looping qemuMonitorAddObject: Use g_clear_pointer for a free and reset operation qemuAgentIOProcessLine: refactor cleanup qemu: agent: Automatically free virJSONValue-s qemu: agent: Remove unneeded cleanup sections src/ch/ch_process.c | 4 +- src/locking/lock_daemon.c | 85 ++++------ src/logging/log_daemon.c | 42 ++--- src/qemu/qemu_agent.c | 304 ++++++++++++++++------------------- src/qemu/qemu_block.c | 3 +- src/qemu/qemu_driver.c | 3 +- src/qemu/qemu_hotplug.c | 13 +- src/qemu/qemu_monitor.c | 20 +-- src/qemu/qemu_monitor.h | 2 +- src/qemu/qemu_monitor_json.c | 79 +++++---- src/qemu/qemu_monitor_json.h | 2 +- tests/virnetdaemontest.c | 6 +- 12 files changed, 254 insertions(+), 309 deletions(-) -- 2.31.1

2 17

[libvirt PATCH] libxl: libxlLoggerFileFree: remove redundant NULL assignment
by Ján Tomko 03 Dec '21

03 Dec '21

The 'file' pointer is already zeroed in virFileFclose and never read again. Signed-off-by: Ján Tomko <jtomko(a)redhat.com> --- Pushed as trivial. src/libxl/libxl_logger.c | 1 - 1 file changed, 1 deletion(-) diff --git a/src/libxl/libxl_logger.c b/src/libxl/libxl_logger.c index 44ee5a95da..4708553f29 100644 --- a/src/libxl/libxl_logger.c +++ b/src/libxl/libxl_logger.c @@ -53,7 +53,6 @@ libxlLoggerFileFree(void *payload) { FILE *file = payload; VIR_FORCE_FCLOSE(file); - file = NULL; } G_GNUC_PRINTF(5, 0) static void -- 2.31.1

1 0

[libvirt PATCH] meson: Improve network driver enablement logic
by Andrea Bolognani 03 Dec '21

03 Dec '21

The Homebrew package explicitly enables this driver despite us disabling it by default on macOS, so it must be functional to at least some extent and certainly can't be causing any build failures. Additionally, if the user has explicitly asked for the network driver to be enabled but libvirtd is disabled for whatever reason, we should error out instead of silently disabling the network driver. Signed-off-by: Andrea Bolognani <abologna(a)redhat.com> --- Homebrew formula showing that the network driver is explicitly enabled in the package: https://github.com/Homebrew/homebrew-core/blob/3aff61292a3ed16bc715ba23fc2f… Test pipeline showing that enabling the network driver by default on macOS doesn't result in build failures: https://gitlab.com/abologna/libvirt/-/pipelines/421956249 meson.build | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/meson.build b/meson.build index ba2f0d21de..cea8bbfa0c 100644 --- a/meson.build +++ b/meson.build @@ -1645,10 +1645,10 @@ if not get_option('driver_ch').disabled() and host_machine.system() == 'linux' a endif endif -# there's no use compiling the network driver without the libvirt -# daemon, nor compiling it for macOS, where it breaks the compile -if not get_option('driver_network').disabled() and conf.has('WITH_LIBVIRTD') and host_machine.system() != 'darwin' +if not get_option('driver_network').disabled() and conf.has('WITH_LIBVIRTD') conf.set('WITH_NETWORK', 1) +elif get_option('driver_network').enabled() + error('libvirtd must be enabled to build the network driver') endif if udev_dep.found() and conf.has('WITH_LIBVIRTD') -- 2.31.1

2 1