January 2024 - Devel - libvirt List Archives

[PATCH 0/4] vmx: A couple of disk related fixes
by Michal Privoznik 26 Jan '24

26 Jan '24

The first patch fixes an issue. The last one - I am not sure if it's not going to break something and thus it's optional. Michal Prívozník (4): vmx: Accept empty fileName for cdrom-image vmx2xmltest: Add another test case vmx: Separate disk target name generation into a function vmx: Ensure unique disk targets when parsing src/vmx/vmx.c | 199 +++++++++++++++-------- tests/vmx2xmldata/esx-in-the-wild-12.vmx | 86 ++++++++++ tests/vmx2xmldata/esx-in-the-wild-12.xml | 39 +++++ tests/vmx2xmldata/esx-in-the-wild-8.xml | 4 +- tests/vmx2xmltest.c | 1 + 5 files changed, 261 insertions(+), 68 deletions(-) create mode 100644 tests/vmx2xmldata/esx-in-the-wild-12.vmx create mode 100644 tests/vmx2xmldata/esx-in-the-wild-12.xml -- 2.41.0

3 9

[PATCH] qemu_hotplug: Don't lose 'created' flag in qemuDomainChangeNet()
by Michal Privoznik 26 Jan '24

26 Jan '24

After v9.1.0-rc1~116 we track whether it's us who created a macvtap or not. But when updating a vNIC its definition might be replaced with a new one (though, ifname is not allowed to change), e.g. to reflect new QoS, link state, etc. Now, the fact whether we created macvtap for given vNIC is stored in net->privateData->created. And replacing definition is done by simply freeing the old definition and making the pointer point to the new one. But this does not preserve the 'created' flag, which in turn means when a domain is shutting off, the macvtap is not removed (see loop inside of qemuProcessStop()). Copy this flag into new definition and leave a note in _qemuDomainNetworkPrivate struct. Fixes: 61d1b9e6592660121aeda66bf7adbcd39de06aa8 Resolves: https://issues.redhat.com/browse/RHEL-22714 Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com> --- I also contemplated idea of preserving whole privateData, e.g.: virObjectUnref(newdev->privateData); newdev->privateData = g_steal_pointer(&olddev->privateData); but then decided against it. Looks like a heavy gun, though in my debugging all members but 'created' were zero/NULL at point of calling qemuDomainChangeNet(). Then there is qemuDomainNetworkPrivateFormat() and qemuDomainNetworkPrivateParse() combo, but in order to use it I'd need to create and then parse an XML doc (to get 'ctxt') and that looks too much work for very little gain. Does somebody have any other idea? src/qemu/qemu_domain.h | 2 ++ src/qemu/qemu_hotplug.c | 5 +++++ 2 files changed, 7 insertions(+) diff --git a/src/qemu/qemu_domain.h b/src/qemu/qemu_domain.h index b4512cc80e..6ba3f10e8d 100644 --- a/src/qemu/qemu_domain.h +++ b/src/qemu/qemu_domain.h @@ -420,6 +420,8 @@ typedef struct _qemuDomainNetworkPrivate qemuDomainNetworkPrivate; struct _qemuDomainNetworkPrivate { virObject parent; + /* Don't forget to possibly copy these members in qemuDomainChangeNet(). */ + /* True if the device was created by us. Otherwise we should * avoid removing it. Currently only used for * VIR_DOMAIN_NET_TYPE_DIRECT. */ diff --git a/src/qemu/qemu_hotplug.c b/src/qemu/qemu_hotplug.c index 0e45bd53e1..31b00e05ca 100644 --- a/src/qemu/qemu_hotplug.c +++ b/src/qemu/qemu_hotplug.c @@ -4166,6 +4166,11 @@ qemuDomainChangeNet(virQEMUDriver *driver, else VIR_WARN("Unable to release network device '%s'", NULLSTR(olddev->ifname)); } + + /* Carry over fact whether we created the device or not. */ + QEMU_DOMAIN_NETWORK_PRIVATE(newdev)->created = + QEMU_DOMAIN_NETWORK_PRIVATE(olddev)->created; + virDomainNetDefFree(olddev); /* move newdev into the nets list, and NULL it out from the * virDomainDeviceDef that we were given so that the caller -- 2.43.0

2 1

[PATCH] Add explanation about the attribute "delay"
by Yalan Zhang 26 Jan '24

26 Jan '24

The libvirt created linux bridge has a configurable value "delay", the default value is "0", but it will not take effect. That's because kernel has a minimum value for linux bridge. Add some explanation about it in the document. Signed-off-by: Yalan Zhang <yalzhang(a)redhat.com> --- docs/formatnetwork.rst | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/docs/formatnetwork.rst b/docs/formatnetwork.rst index 5335da5e11..aee70bbd0d 100644 --- a/docs/formatnetwork.rst +++ b/docs/formatnetwork.rst @@ -106,7 +106,8 @@ to the physical LAN (if at all). prefix "virbr" is recommended (and that is what is auto-generated), but not enforced. Attribute ``stp`` specifies if Spanning Tree Protocol is 'on' or 'off' (default is 'on'). Attribute ``delay`` sets the bridge's forward delay - value in seconds (default is 0). :since:`Since 0.3.0` + value in seconds (default is 0. As the kernel has a minimum delay, values + below it may not be counted). :since:`Since 0.3.0` The ``macTableManager`` attribute of the bridge element is used to tell libvirt how the bridge's MAC address table (used to determine the correct -- 2.43.0

2 1

[libvirt PATCH] build: Make daemons depend on generated *_protocol.[ch]
by Jiri Denemark 26 Jan '24

26 Jan '24

This should fix build failures when a daemon code is compiled before the included *_protocol.h headers are ready, such as: FAILED: src/virtqemud.p/remote_remote_daemon_config.c.o ../src/remote/remote_daemon_config.c: In function ‘daemonConfigNew’: ../src/remote/remote_daemon_config.c:111:30: error: ‘REMOTE_AUTH_POLKIT’ undeclared (first use in this function) 111 | data->auth_unix_rw = REMOTE_AUTH_POLKIT; | ^~~~~~~~~~~~~~~~~~ ../src/remote/remote_daemon_config.c:111:30: note: each undeclared identifier is reported only once for each function it appears in ../src/remote/remote_daemon_config.c:115:30: error: ‘REMOTE_AUTH_NONE’ undeclared (first use in this function) 115 | data->auth_unix_rw = REMOTE_AUTH_NONE; | ^~~~~~~~~~~~~~~~ ../src/remote/remote_daemon_config.c: In function ‘daemonConfigLoadOptions’: ../src/remote/remote_daemon_config.c:252:31: error: ‘REMOTE_AUTH_POLKIT’ undeclared (first use in this function) 252 | if (data->auth_unix_rw == REMOTE_AUTH_POLKIT) { | ^~~~~~~~~~~~~~~~~~ or FAILED: src/virtqemud.p/remote_remote_daemon_dispatch.c.o In file included from ../src/remote/remote_daemon.h:28, from ../src/remote/remote_daemon_dispatch.c:26: src/remote/lxc_protocol.h:13:5: error: unknown type name ‘remote_nonnull_domain’ 13 | remote_nonnull_domain dom; | ^~~~~~~~~~~~~~~~~~~~~ In file included from ../src/remote/remote_daemon.h:29, from ../src/remote/remote_daemon_dispatch.c:26: src/remote/qemu_protocol.h:13:5: error: unknown type name ‘remote_nonnull_domain’ 13 | remote_nonnull_domain dom; | ^~~~~~~~~~~~~~~~~~~~~ src/remote/qemu_protocol.h:14:5: error: unknown type name ‘remote_nonnull_string’ 14 | remote_nonnull_string cmd; | ^~~~~~~~~~~~~~~~~~~~~ ... Signed-off-by: Jiri Denemark <jdenemar(a)redhat.com> --- Notes: I couldn't check this patch actually fixes the error because the build failure is non-deterministic and I cannot reproduce the failure now. po/meson.build | 1 + src/meson.build | 6 +++++- src/remote/meson.build | 15 +++++++++++---- 3 files changed, 17 insertions(+), 5 deletions(-) diff --git a/po/meson.build b/po/meson.build index a20877ad34..592b254447 100644 --- a/po/meson.build +++ b/po/meson.build @@ -20,6 +20,7 @@ potfiles_dep = [ access_gen_sources, admin_client_generated, admin_driver_generated, + remote_protocol_generated, remote_driver_generated, remote_daemon_generated, ] diff --git a/src/meson.build b/src/meson.build index 6538c43628..f52d2d5994 100644 --- a/src/meson.build +++ b/src/meson.build @@ -616,7 +616,11 @@ foreach daemon : virt_daemons bin = executable( daemon['name'], [ - daemon.get('sources', [ remote_daemon_sources, remote_daemon_generated ]), + daemon.get('sources', [ + remote_protocol_generated, + remote_daemon_sources, + remote_daemon_generated + ]), dtrace_gen_objects, ], c_args: [ diff --git a/src/remote/meson.build b/src/remote/meson.build index 681c709823..831acaaa01 100644 --- a/src/remote/meson.build +++ b/src/remote/meson.build @@ -7,8 +7,6 @@ remote_driver_generated = [] foreach name : [ 'remote', 'qemu', 'lxc' ] client_bodies_h = '@0@_client_bodies.h'.format(name) - protocol_c = '@0@_protocol.c'.format(name) - protocol_h = '@0@_protocol.h'.format(name) protocol_x = '@0@_protocol.x'.format(name) remote_driver_generated += custom_target( @@ -20,8 +18,16 @@ foreach name : [ 'remote', 'qemu', 'lxc' ] ], capture: true, ) +endforeach - remote_driver_generated += custom_target( +remote_protocol_generated = [] + +foreach name : [ 'remote', 'qemu', 'lxc' ] + protocol_c = '@0@_protocol.c'.format(name) + protocol_h = '@0@_protocol.h'.format(name) + protocol_x = '@0@_protocol.x'.format(name) + + remote_protocol_generated += custom_target( protocol_h, input: protocol_x, output: protocol_h, @@ -32,7 +38,7 @@ foreach name : [ 'remote', 'qemu', 'lxc' ] ], ) - remote_driver_generated += custom_target( + remote_protocol_generated += custom_target( protocol_c, input: protocol_x, output: protocol_c, @@ -143,6 +149,7 @@ if conf.has('WITH_REMOTE') remote_driver_lib = static_library( 'virt_remote_driver', [ + remote_protocol_generated, remote_driver_sources, remote_driver_generated, ], -- 2.43.0

3 2

[PATCH v2 0/2] qemu_process: Skip over non-virtio non-TAP NIC models when refreshing rx-filter
by Michal Privoznik 26 Jan '24

26 Jan '24

This was reported here: https://lists.libvirt.org/archives/list/users@lists.libvirt.org/thread/QZ43… and v1 was posted here: https://lists.libvirt.org/archives/list/devel@lists.libvirt.org/thread/AVRY… diff to v1: - Expanded commit message in 1/2 - Added NEWS entry - Expanded list of vNIC types for which query-rx-filter is issued to basically every TUN/TAP based type. Might be an overkill, but should keep us covered for foreseeable future (e.g. when we decide to report changed MAC address in domain XML). Michal Prívozník (2): qemu_process: Skip over non-virtio non-TAP NIC models when refreshing rx-filter NEWS: Document recent rx-filter bugfix NEWS.rst | 9 +++++++++ src/qemu/qemu_process.c | 27 +++++++++++++++++++++++++++ 2 files changed, 36 insertions(+) -- 2.43.0

2 3

[PATCH 0/3] qemu: migration: Fix allocation and release of NBD port
by Peter Krempa 25 Jan '24

25 Jan '24

See 2/3 for the bugfix Peter Krempa (3): util: virtportallocator: Add VIR_DEBUG statements for port allocations and release qemu: migration: Properly handle reservation of manually specified NBD port qemuMigrationDstStartNBDServer: Refactor cleanup src/qemu/qemu_migration.c | 40 +++++++++++++------------------------ src/util/virportallocator.c | 9 +++++++++ 2 files changed, 23 insertions(+), 26 deletions(-) -- 2.43.0

2 6

[PATCH] qemu_process: Skip over non-virtio NIC models when refreshing rx-filter
by Michal Privoznik 25 Jan '24

25 Jan '24

After guest is started, or we are reconnecting to already running one (after daemon restart), qemuProcessRefreshRxFilters() is called to refresh rx-filters (basically MAC addresses of guest NICs) as they might have changed while we were not running (for the case when reconnecting to an already running guest), or we need to enable them by running a command (for freshly started guest - see processNicRxFilterChangedEvent()). Now, our XML parser allowed trustGuestRxFilters attribute for all types and models of <interface/> while in reality, only virtio model can see MAC address changes. Fixes: 060d4c83ef436cf56abfad51a4d64c39448e199d Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com> --- src/qemu/qemu_process.c | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c index 3563ad215c..a736846588 100644 --- a/src/qemu/qemu_process.c +++ b/src/qemu/qemu_process.c @@ -7958,6 +7958,12 @@ qemuProcessRefreshRxFilters(virDomainObj *vm, if (!virDomainNetGetActualTrustGuestRxFilters(def)) continue; + /* rx-filters are supported only for virtio macvtaps */ + if (def->model != VIR_DOMAIN_NET_MODEL_VIRTIO || + virDomainNetGetActualType(def) != VIR_DOMAIN_NET_TYPE_DIRECT) { + continue; + } + if (qemuDomainSyncRxFilter(vm, def, asyncJob) < 0) return -1; } -- 2.43.0

3 3

[PATCH 0/3] Allow reserving more memory for PCI controllers
by Michal Privoznik 25 Jan '24

25 Jan '24

*** BLURB HERE *** Michal Prívozník (3): conf: Introduce @memReserve to <controller/> qemu_validate: Restrict setting @memReserve only to some controllers qemu_command: Generate mem-reserve for controllers docs/formatdomain.rst | 6 +++++ src/conf/domain_conf.c | 9 +++++++ src/conf/domain_conf.h | 3 +++ src/conf/schemas/domaincommon.rng | 5 ++++ src/qemu/qemu_command.c | 3 +++ src/qemu/qemu_validate.c | 25 +++++++++++++++++++ .../q35-usb2.x86_64-latest.args | 2 +- tests/qemuxml2argvdata/q35-usb2.xml | 2 +- .../q35-usb2.x86_64-latest.xml | 2 +- 9 files changed, 54 insertions(+), 3 deletions(-) -- 2.41.0

3 5

[PATCH] Explicitly convert type to double to avoid losing precision
by Egor Makrushin 24 Jan '24

24 Jan '24

Division between integers will also be integer. Thus, to preserve fractional part explicitly convert first operand to double. Found by Linux Verification Center (linuxtesting.org) with SVACE. Fixes: 28d54aab05 ("examples: Introduce domtop") Signed-off-by: Egor Makrushin <emakrushin(a)astralinux.ru> --- examples/c/domain/domtop.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/examples/c/domain/domtop.c b/examples/c/domain/domtop.c index dd35ec8661..65663a1c39 100644 --- a/examples/c/domain/domtop.c +++ b/examples/c/domain/domtop.c @@ -224,7 +224,7 @@ print_cpu_usage(size_t cpu, * performed has a bad effect on the precision, so instead of dividing * @now_params and @then_params by 1000 and then multiplying again by * 100, we divide only once by 10 and get the same result. */ - usage = (now_params[pos].value.ul - then_params[pos].value.ul) / + usage = (double)(now_params[pos].value.ul - then_params[pos].value.ul) / (now - then) / 10; if (delim) -- 2.30.2

2 1

[PATCH 0/2] remoteDispatchAuthPolkit: Fix lock ordering deadlock if client closes connection during auth
by Peter Krempa 24 Jan '24

24 Jan '24

Peter Krempa (2): reproducer remoteDispatchAuthPolkit: Fix lock ordering deadlock if client closes connection during auth src/remote/remote_daemon_dispatch.c | 76 +++++++++++++++-------------- src/rpc/virnetserver.c | 26 +++++++++- src/rpc/virnetserverclient.c | 2 + 3 files changed, 65 insertions(+), 39 deletions(-) -- 2.43.0

2 3