February 2023 - Devel - libvirt List Archives

[PULL 0/5] Misc next patches
by Daniel P. Berrangé 16 Feb '23

16 Feb '23

The following changes since commit 6a50f64ca01d0a7b97f14f069762bfd88160f31e: Merge tag 'pull-request-2023-02-14' of https://gitlab.com/thuth/qemu into staging (2023-02-14 14:46:10 +0000) are available in the Git repository at: https://gitlab.com/berrange/qemu tags/misc-next-pull-request for you to fetch changes up to 36debafddd788066be10b33c5f11b984a08e5c85: ui: remove deprecated 'password' option for SPICE (2023-02-15 11:14:58 -0500) ---------------------------------------------------------------- * Document 'password-secret' option for -iscsi * Deprecate iSCSI 'password' in favour of 'password-secret' * Remove deprecated 'password' option for SPICE * Fix handling of cached read buffers with TLS ---------------------------------------------------------------- Antoine Damhet (2): crypto: TLS: introduce `check_pending` io/channel-tls: fix handling of bigger read buffers Daniel P. Berrangé (3): block: mention 'password-secret' option for -iscsi block: deprecate iSCSI 'password' in favour of 'password-secret' ui: remove deprecated 'password' option for SPICE block/iscsi.c | 3 ++ crypto/tlssession.c | 14 +++++++ docs/about/deprecated.rst | 16 ++++---- docs/about/removed-features.rst | 7 ++++ include/crypto/tlssession.h | 11 ++++++ io/channel-tls.c | 66 ++++++++++++++++++++++++++++++++- qemu-options.hx | 13 ++----- ui/spice-core.c | 15 -------- 8 files changed, 111 insertions(+), 34 deletions(-) -- 2.39.1

2 6

[PATCH 0/4] qemu_passt: Don't let passt fork off
by Michal Privoznik 16 Feb '23

16 Feb '23

Here are some cleanup, as promised here: https://listman.redhat.com/archives/libvir-list/2023-February/237721.html Now, there are still some patches missing. Firstly, we still don't really capture error from passt. My suggestion was to wait for socket to show up with errfd open. But active wait is viewed as undesirable [1]. Secondly, Laine reported SELinux issues. Yeah, I don't see us setting SELinux label on nor socket that passt and QEMU talk to each other, nor on the log file. Speaking of which - we usually have per domain (or per helper daemon instance even) log file, while for passt we have a global one (/var/log/passt.log). I don't think that will fly if two or more SELinux enabled domains want to use passt. Thirdly, Stefano suggested a graceful shutdown for passt: have libvirt connect to the socket and close it. Since we pass --one-off, this should singal passt to exit. But I haven't implemented that because it's redundant. We can't rely on passt quitting itself and thus use the big gun (virPidFileForceCleanupPath()) at which point, the socket way is just an optimization. I might look into the first two, at some point. But not today. 1: https://listman.redhat.com/archives/libvir-list/2023-February/237663.html Michal Prívozník (4): Revert "qemu: allow passt to self-daemonize" qemu_extdevice: Make qemuExtDevicesHasDevice() check def->nets qemu_passt: Report error when getting passt PID failed qemu_passt: Don't let passt fork off src/qemu/qemu_extdevice.c | 11 +++++++++++ src/qemu/qemu_passt.c | 15 ++++++++++----- 2 files changed, 21 insertions(+), 5 deletions(-) -- 2.39.1

5 18

[PATCH v4 14/16] qapi: deprecate "device" field of DEVICE_* events
by Vladimir Sementsov-Ogievskiy 16 Feb '23

16 Feb '23

The device field is redundant, because QOM path always include device ID when this ID exist. Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov(a)yandex-team.ru> --- docs/about/deprecated.rst | 9 +++++++++ qapi/qdev.json | 12 ++++++++++-- 2 files changed, 19 insertions(+), 2 deletions(-) diff --git a/docs/about/deprecated.rst b/docs/about/deprecated.rst index da2e6fe63d..b389934691 100644 --- a/docs/about/deprecated.rst +++ b/docs/about/deprecated.rst @@ -171,6 +171,15 @@ accepted incorrect commands will return an error. Users should make sure that all arguments passed to ``device_add`` are consistent with the documented property types. +QEMU Machine Protocol (QMP) events +---------------------------------- + +``DEVICE_DELETED`` & ``DEVICE_UNPLUG_GUEST_ERROR`` field ``device`` (since 8.0) +''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''' + +Devices that has ``ID`` always has QOM path `/machine/peripheral/ID`. So, the +``device`` field is redundant and deprecated. Use the ``path`` field instead. + Host Architectures ------------------ diff --git a/qapi/qdev.json b/qapi/qdev.json index 2708fb4e99..325ef554f9 100644 --- a/qapi/qdev.json +++ b/qapi/qdev.json @@ -124,6 +124,9 @@ # # @path: the device's QOM path # +# Features: +# @deprecated: Member @device is deprecated as redundant. Use @path instead. +# # Since: 1.5 # # Example: @@ -135,7 +138,8 @@ # ## { 'event': 'DEVICE_DELETED', - 'data': { '*device': 'str', 'path': 'str' } } + 'data': { '*device': { 'type': 'str', 'features': [ 'deprecated' ] }, + 'path': 'str' } } ## # @DEVICE_UNPLUG_GUEST_ERROR: @@ -146,6 +150,9 @@ # # @path: the device's QOM path # +# Features: +# @deprecated: Member @device is deprecated as redundant. Use @path instead. +# # Since: 6.2 # # Example: @@ -157,4 +164,5 @@ # ## { 'event': 'DEVICE_UNPLUG_GUEST_ERROR', - 'data': { '*device': 'str', 'path': 'str' } } + 'data': { '*device': { 'type': 'str', 'features': [ 'deprecated' ] }, + 'path': 'str' } } -- 2.34.1

5 17

[PATCH] qemu_extdevice: Add a comment into qemuExtDevicesSetupCgroup()
by Michal Privoznik 15 Feb '23

15 Feb '23

The way setting up CGroups for external helpers work, is: qemuExtDevicesHasDevice() is called first to determine whether there is a helper process running, the CGroup controller is created and then qemuExtDevicesSetupCgroup() is called to place helpers into the CGroup. But when one reads just qemuExtDevicesSetupCgroup() it's easy to miss this hidden logic. Therefore, add a warning at the beginning of the function. Signed-off-by: Michal Privoznik <mprivozn(a)redhat.com> --- src/qemu/qemu_extdevice.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/src/qemu/qemu_extdevice.c b/src/qemu/qemu_extdevice.c index 47e97f3565..1c397972e4 100644 --- a/src/qemu/qemu_extdevice.c +++ b/src/qemu/qemu_extdevice.c @@ -331,6 +331,9 @@ qemuExtDevicesSetupCgroup(virQEMUDriver *driver, virDomainDef *def = vm->def; size_t i; + /* Don't forget to adjust qemuExtDevicesHasDevice() accordingly. + * Otherwise, this function might not be called at all. */ + if (qemuDBusSetupCgroup(driver, vm, cgroup) < 0) return -1; -- 2.39.1

2 1

wiki.libvirt.org replacement
by Peter Krempa 15 Feb '23

15 Feb '23

https://gitlab.com/libvirt/libvirt-wiki/-/merge_requests/1 Hi, in order to ease editing of the libvirt wiki, remove the need for registering users and remove the need to run PHP in an openshift instance I've decided to contents of the libvirt wiki into a staticaly generated page using (almost) the same approach we use to generate the libvirt web pages from rST documents in the repo. All articles were converted into rST files [1] and images linked from currently existing articles were dowloaded. Advantages: - same workflow as with editing the libvirt pages - gitlab still provides a web editor - local editing for users who hate web - no need to deal with user registration - no need to run PHP in openshift - we still keep separate space for docs which don't really belong into the main repo - even if we decide to kill-off the wiki eventually the valuable content will be easier to port to the kbase as it'll be in rST - the conversion fixed many orphaned pages Disadvantages: - all links will be broken [2] - changes will need to be reviewed/approved - low quality/obsolete content is forward-ported as I didn't review anything - the build script is in bash (This obviously can be changed if somebody cares more than I do.) The generator is based on a cleaned up page.xsl and other assets from libvirt's repo and the check-html-references script [3] to validate linking. The new wiki can for now be browsed from artifacts of the pipeline job that I've used to test it: https://pipo.sk.gitlab.io/-/libvirt-wiki/-/jobs/3771076975/artifacts/websit… (I know my username is unfortunate to contain a dot which breaks the certificate) If somebody wishes to see the extremely ugly conversion scripts: https://gitlab.com/pipo.sk/libvirt-wiki/-/commit/fe36e37d3580a76ca18e512354… [1] The conversion was done using a collection of ugly scripts and pandoc. Unfortunately markdown-eque formats have the issue of not really having strict rules and in certain cases the tools used to process them not implementing them correctly. This meant that the least painful was actually the coversion from HTML!!! and not the internal mediawiki format. [2] If we really want to preserve links I can modify the script to generate a list of redirects for the webserver, but I really doubt that preserving links has value. Additionally some content was moved to the knowledge base, so I really want to just delete it from the wiki, so breaking links is a feature. [3] It uses the version after the last patchset: https://listman.redhat.com/archives/libvir-list/2023-February/237785.html

2 4

[PATCH 0/9] Improve docs web reference checker
by Peter Krempa 15 Feb '23

15 Feb '23

Improve the checker tool to also track orphaned pages and check presence and usage of images. In addition to that this series fixes handful of instances of the problems that the script caught. This series adds two documents: - docs/logos/README.rst (converted into docs/logos/index.html) The page shows all logos which libvirt has. It's now linked from the docs page. https://pipo.sk.gitlab.io/-/libvirt/-/jobs/3770134533/artifacts/website/log… - docs/golang.rst The page ties together the 4 go module pages we have which were not linked and is linked from docs.html https://pipo.sk.gitlab.io/-/libvirt/-/jobs/3770134533/artifacts/website/gol… Peter Krempa (9): docs: XSL: Add source document name as custom data attribute for <html> scripts: check-html-references: Rename --prefix to --webroot and make it mandatory scripts: check-html-references: Improve error messages and don't mess with relative paths docs: manpages: Add missing manpages to index docs: Add sub-page for all golang modules scripts: check-html-references: Detect pages that are not linked to kbase: eventloop: Fix path to referenced images docs: logos: Turn 'README' into rST, generate an index and link to images scripts: check-html-refernces: Add checking for image file usage docs/docs.rst | 5 +- docs/golang.rst | 21 +++ docs/kbase/internals/eventloop.rst | 4 +- docs/logos/README | 98 -------------- docs/logos/README.rst | 166 ++++++++++++++++++++++++ docs/logos/meson.build | 47 +++++++ docs/manpages/index.rst | 3 + docs/meson.build | 3 +- docs/page.xsl | 2 +- scripts/check-html-references.py | 198 +++++++++++++++++++++++------ 10 files changed, 406 insertions(+), 141 deletions(-) create mode 100644 docs/golang.rst delete mode 100644 docs/logos/README create mode 100644 docs/logos/README.rst -- 2.39.1

2 18

[PATCH 0/3] Virtlogd fixes
by Peter Krempa 15 Feb '23

15 Feb '23

Please see individual patches for justification. Peter Krempa (3): rpc: client: Don't check return value of virNetMessageNew rpc: Don't warn about "max_client_requests" in single-threaded daemons virLogCleanerShutdown: Don't call g_regex_unref on NULL regex src/logging/log_cleaner.c | 3 +-- src/remote/libvirtd.conf.in | 1 + src/rpc/virnetserverclient.c | 14 ++++++-------- 3 files changed, 8 insertions(+), 10 deletions(-) -- 2.39.1

2 6

there is no PCIe device after command "virsh nodedev-list"
by Martin Bayern 14 Feb '23

14 Feb '23

Hello, I don't know if this is the right email list for this question, let me try. I experienced a problem when I want to assign PCIe VF to my VMs, I can use "lspci" to list the VF PCIe, even PF PCIe, but the command " virsh nodedev-list" doesn't print any PCIe bus device. I tried using "virsh edit --domain " command and added the PCIe bus to the XML file, but when I start the VM, the error message shows an internal error, there is an invalid link to the device. Meanwhile, I use virt-manager to see if I can add a PCIe host device, but the host device shows "no device available". my system is ARM64 based Nvidia platform, I have enabled IOMMU in the kernel. do you know how can I figure out the root cause? is this the ARM64 kernel issue or something else?? I posted this issue also in Nvidia forum: https://forums.developer.nvidia.com/t/nvidia-orin-pcie-device-there-is-no-p… thanks so much for your suggestions. Martin

1 0

[libvirt PATCH] qemu: allow passt to self-daemonize
by Laine Stump 14 Feb '23

14 Feb '23

I initially had the passt process being started in an identical fashion to the slirp-helper - libvirt was daemonizing the new process and recording its pid in a pidfile. The problem with this is that, since it is daemonized immediately, any startup error in passt happens after the daemonization, and thus isn't seen by libvirt - libvirt believes that the process has started successfully and continues on its merry way. The result was that sometimes a guest would be started, but there would be no passt process for qemu to use for network traffic. Instead, we should be starting passt in the same manner we start dnsmasq - we just exec it as normal (along with a request that passt create the pidfile, which is just another option on the passt commandline) and wait for the child process to exit; passt then has a chance to parse its commandline and complete all the setup prior to daemonizing itself; if it encounters an error and exits with a non-0 code, libvirt will see the code and know about the failure. We can then grab the output from stderr, log that so the "user" has some idea of what went wrong, and then fail the guest startup. Signed-off-by: Laine Stump <laine(a)redhat.com> --- src/qemu/qemu_passt.c | 9 ++++----- 1 file changed, 4 insertions(+), 5 deletions(-) diff --git a/src/qemu/qemu_passt.c b/src/qemu/qemu_passt.c index 0f09bf3db8..f640a69c00 100644 --- a/src/qemu/qemu_passt.c +++ b/src/qemu/qemu_passt.c @@ -141,24 +141,23 @@ qemuPasstStart(virDomainObj *vm, g_autofree char *passtSocketName = qemuPasstCreateSocketPath(vm, net); g_autoptr(virCommand) cmd = NULL; g_autofree char *pidfile = qemuPasstCreatePidFilename(vm, net); + g_autofree char *errbuf = NULL; char macaddr[VIR_MAC_STRING_BUFLEN]; size_t i; pid_t pid = (pid_t) -1; int exitstatus = 0; int cmdret = 0; - VIR_AUTOCLOSE errfd = -1; cmd = virCommandNew(PASST); virCommandClearCaps(cmd); - virCommandSetPidFile(cmd, pidfile); - virCommandSetErrorFD(cmd, &errfd); - virCommandDaemonize(cmd); + virCommandSetErrorBuffer(cmd, &errbuf); virCommandAddArgList(cmd, "--one-off", "--socket", passtSocketName, "--mac-addr", virMacAddrFormat(&net->mac, macaddr), + "--pid", pidfile, NULL); if (net->mtu) { @@ -264,7 +263,7 @@ qemuPasstStart(virDomainObj *vm, if (cmdret < 0 || exitstatus != 0) { virReportError(VIR_ERR_INTERNAL_ERROR, - _("Could not start 'passt'. exitstatus: %d"), exitstatus); + _("Could not start 'passt': %s"), errbuf); goto error; } -- 2.39.1

6 14

Plans for the next release
by Jiri Denemark 14 Feb '23

14 Feb '23

We are getting close to the next release of libvirt. To aim for the release on Mar 01 I suggest entering the freeze on Wednesday Feb 22 and tagging RC2 on Monday Feb 27. I hope this works for everyone. Jirka

1 0