January 2025 - Devel - libvirt List Archives

[PATCH] nodedev: udev: Hook up virFileWaitForExist to address uevent, race of pci device
by Guoyi Tu 06 Feb '25

06 Feb '25

this commit addresses the same issue that as commit 1af45804 does. the following message is copying from that commit: If we find ourselves in the situation that the 'add' uevent has been fired earlier than the sysfs tree for a device was created, we should use the best-effort approach and give kernel some predetermined amount of time, thus waiting for the attributes to be ready rather than discarding the device from our device list forever. If those don't appear in the given time frame, we need to move on, since libvirt can't wait indefinitely. Signed-off-by: Guoyi Tu <tugy(a)chinatelecom.cn> --- src/node_device/node_device_udev.c | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/src/node_device/node_device_udev.c b/src/node_device/node_device_udev.c index 1d8486f623..4a1786c21c 100644 --- a/src/node_device/node_device_udev.c +++ b/src/node_device/node_device_udev.c @@ -427,10 +427,19 @@ udevProcessPCI(virNodeDeviceDriverState *driver_state, virPCIEDeviceInfo *pci_express = NULL; virPCIDevice *pciDev = NULL; virPCIDeviceAddress devAddr = { 0 }; + g_autofree char *linkpath = NULL; int ret = -1; char *p; bool privileged = false; + linkpath = g_strdup_printf("%s/config", udev_device_get_syspath(device)); + if (virFileWaitForExists(linkpath, 10, 100) < 0) { + virReportSystemError(errno, + _("failed to wait for file '%1$s' to appear"), + linkpath); + goto cleanup; + } + VIR_WITH_MUTEX_LOCK_GUARD(&driver_state->lock) { privileged = driver_state->privileged; } -- 2.27.0 -- Guoyi

2 2

[PATCH] docs: fix 'virsh domstats --vcpu' measure units and descriptions
by Fabricio Duarte 05 Feb '25

05 Feb '25

The fields are in nanoseconds, not microseconds. Also fixes the description of `vcpu.<num>.wait`, as it does not actually represent the time waiting on I/O. Signed-off-by: Fabricio Duarte <fabricio.duarte.jr(a)gmail.com> --- docs/manpages/virsh.rst | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/docs/manpages/virsh.rst b/docs/manpages/virsh.rst index 2e525d3fac..b0a21e019a 100644 --- a/docs/manpages/virsh.rst +++ b/docs/manpages/virsh.rst @@ -2428,14 +2428,14 @@ When selecting the *--state* group the following fields are returned: * ``vcpu.<num>.state`` - state of the virtual CPU <num>, as number from virVcpuState enum * ``vcpu.<num>.time`` - virtual cpu time spent by virtual - CPU <num> (in microseconds) -* ``vcpu.<num>.wait`` - virtual cpu time spent by virtual - CPU <num> waiting on I/O (in microseconds) + CPU <num> (in nanoseconds) +* ``vcpu.<num>.wait`` - time the vCPU <num> wants to run, but the host + scheduler has something else running ahead of it (in nanoseconds) * ``vcpu.<num>.halted`` - virtual CPU <num> is halted: yes or no (may indicate the processor is idle or even disabled, depending on the architecture) * ``vcpu.<num>.delay`` - time the vCPU <num> thread was enqueued by the - host scheduler, but was waiting in the queue instead of running. + host scheduler, but was waiting in the queue instead of running (in nanoseconds). Exposed to the VM as a steal time. This group of statistics also reports additional hypervisor-originating per-vCPU -- 2.39.2

2 1

Investigating MAC Address Conflict Resolution in libvirt: Log Analysis and Code Location Inquiry
by Xuda Zhang 05 Feb '25

05 Feb '25

Dear Team, I am reaching out regarding an issue I encountered with libvirt and MAC address conflicts. Below is a summary of the situation: 1. Initially, the vNIC's MAC address was different from the target VM's MAC address. 2. After modifying the vNIC's MAC address to match the VM's MAC address, the network was interrupted. 3. After rebooting the VM, the vNIC's MAC address was automatically modified again. I have observed the following kernel logs during this process: Dec 24 16:59:40 zstack-manager kernel: br_enp2s0: port 14(vnic43.0) entered disabled stateDec 24 16:59:40 zstack-manager kernel: device vnic43.0 left promiscuous modeDec 24 16:59:40 zstack-manager kernel: br_enp2s0: port 14(vnic43.0) entered disabled stateDec 24 17:00:11 zstack-manager kernel: br_enp2s0: port 14(vnic43.0) entered blocking stateDec 24 17:00:11 zstack-manager kernel: br_enp2s0: port 14(vnic43.0) entered disabled stateDec 24 17:00:11 zstack-manager kernel: device vnic43.0 entered promiscuous modeDec 24 17:00:11 zstack-manager kernel: br_enp2s0: port 14(vnic43.0) entered blocking stateDec 24 17:00:11 zstack-manager kernel: br_enp2s0: port 14(vnic43.0) entered forwarding state I am looking to understand the underlying code that handles the automatic modification of the vNIC's MAC address after the conflict and how the network interruption occurs. Can you help direct me to the relevant code segment or provide any insights into this behavior? Thank you for your assistance. Best regards,

4 7

[PATCH v3 00/17] hw/microblaze: Allow running cross-endian vCPUs
by Philippe Mathieu-Daudé 05 Feb '25

05 Feb '25

Missing review: 4 (new) & 10 Since v2: - Addressed Richard's review comments Since v1: - Make device endianness configurable (Edgar) - Convert more Xilinx devices - Avoid preprocessor #if (Richard) - Add R-b tags Make machines endianness-agnostic, allowing to run a big-endian vCPU on the little-endian 'qemu-system-microblazeel' binary, and a little endian one on the big-endian 'qemu-system-microblaze' binary. Tests added, following combinations covered: - little-endian vCPU using little-endian binary (in-tree) - little-endian vCPU using big-endian binary (new) - big-endian vCPU using little-endian binary (new) - big-endian vCPU using big-endian binary (in-tree) To make a target endian-agnostic we need to remove the MO_TE uses. In order to do that, we propagate the MemOp from earlier in the call stack, or we extract it from the vCPU env (on MicroBlaze the CPU endianness is exposed by the 'ENDI' bit). Next step: Look at unifying binaries. Please review, Phil. Philippe Mathieu-Daudé (17): hw/microblaze: Restrict MemoryRegionOps are implemented as 32-bit hw/microblaze: Propagate CPU endianness to microblaze_load_kernel() hw/intc/xilinx_intc: Make device endianness configurable RFC hw/net/xilinx_ethlite: Simplify by having configurable endianness RFC hw/timer/xilinx_timer: Allow down to 8-bit memory access hw/timer/xilinx_timer: Make device endianness configurable hw/char/xilinx_uartlite: Make device endianness configurable hw/ssi/xilinx_spi: Make device endianness configurable hw/ssi/xilinx_spips: Make device endianness configurable hw/arm/xlnx-zynqmp: Use &error_abort for programming errors target/microblaze: Explode MO_TExx -> MO_TE | MO_xx target/microblaze: Set MO_TE once in do_load() / do_store() target/microblaze: Introduce mo_endian() helper target/microblaze: Consider endianness while translating code hw/microblaze: Support various endianness for s3adsp1800 machines tests/functional: Explicit endianness of microblaze assets tests/functional: Add microblaze cross-endianness tests hw/microblaze/boot.h | 4 +- include/hw/ssi/xilinx_spips.h | 1 + target/microblaze/cpu.h | 7 +++ hw/arm/xilinx_zynq.c | 1 + hw/arm/xlnx-zynqmp.c | 40 +++++-------- hw/char/xilinx_uartlite.c | 31 ++++++---- hw/intc/xilinx_intc.c | 50 ++++++++++++---- hw/microblaze/boot.c | 8 +-- hw/microblaze/petalogix_ml605_mmu.c | 4 +- hw/microblaze/petalogix_s3adsp1800_mmu.c | 58 ++++++++++++++++--- hw/microblaze/xlnx-zynqmp-pmu.c | 2 +- hw/net/xilinx_ethlite.c | 42 +++++++++----- hw/ppc/virtex_ml507.c | 1 + hw/ssi/xilinx_spi.c | 24 +++++--- hw/ssi/xilinx_spips.c | 36 +++++++----- hw/timer/xilinx_timer.c | 33 +++++++---- target/microblaze/translate.c | 49 ++++++++++------ .../functional/test_microblaze_s3adsp1800.py | 27 ++++++++- .../test_microblazeel_s3adsp1800.py | 25 +++++++- 19 files changed, 309 insertions(+), 134 deletions(-) -- 2.45.2

4 27

[PATCH] qemu: introduce load qemu.conf for "virt-qemu-run"
by Adam Julis 31 Jan '25

31 Jan '25

Adding a new option --config (or -c) for specifying a custom qemu.conf file. Previously, virt-qemu-run loaded default configuration values for QEMU via qemuStateInitialize(). The configuration was loaded from a temporary ../etc/ directory using virQEMUDriverConfigLoadFile(), and any qemu.conf file present in that directory was also loaded automatically. This patch allows users to specify a custom configuration file, which is copied into the temporary directory (or a permanent folder if the -r option is used) before loading the configuration. If an existing qemu.conf is present, it is properly backed up and restored in case of a permanent folder. The custom qemu.conf is always removed when the program exits. Resolves: https://gitlab.com/libvirt/libvirt/-/issues/723 Signed-off-by: Adam Julis <ajulis(a)redhat.com> --- docs/manpages/virt-qemu-run.rst | 9 +++ src/qemu/qemu_shim.c | 121 +++++++++++++++++++++++++++++++- 2 files changed, 129 insertions(+), 1 deletion(-) diff --git a/docs/manpages/virt-qemu-run.rst b/docs/manpages/virt-qemu-run.rst index 4d546ff8cc..ba1c90b52a 100644 --- a/docs/manpages/virt-qemu-run.rst +++ b/docs/manpages/virt-qemu-run.rst @@ -72,6 +72,15 @@ whose UUID should match a secret referenced in the guest domain XML. Display verbose information about startup. +``-c`` *QEMU-CONF-FILE*, +``--config``\ =\ *QEMU-CONF-FILE* + +Specify the QEMU configuration file to be used for starting the VM. +*QEMU-CONF-FILE* is the full path to the QEMU configuration file. + +If this parameter is omitted, the default configuration values will +be used. + ``-h``, ``--help`` Display the command line help. diff --git a/src/qemu/qemu_shim.c b/src/qemu/qemu_shim.c index 7fdd69b538..64c87ab264 100644 --- a/src/qemu/qemu_shim.c +++ b/src/qemu/qemu_shim.c @@ -23,6 +23,7 @@ #include <stdio.h> #include <stdbool.h> #include <unistd.h> +#include <fcntl.h> #include "virfile.h" #include "virgettext.h" @@ -132,6 +133,100 @@ qemuShimQuench(void *userData G_GNUC_UNUSED, { } +/* Load specific QEMU config file, is the -c option is used */ +static int +qemuAddConfigFile(const char *source_file, const char *root, bool *config_to_delete, long long deltams) +{ + int ret = -1; + struct stat st; + VIR_AUTOCLOSE srcFD = -1; + VIR_AUTOCLOSE dstFD = -1; + g_autofree char *config_dir = NULL; + g_autofree char *source_file_real = NULL; + g_autofree char *config_path_file = NULL; + g_autofree char *config_path_file_real = NULL; + + if ((srcFD = open(source_file, O_RDONLY)) < 0) { + g_printerr("Couldn't open specific config file\n"); + goto cleanup; + } + + if (fstat(srcFD, &st) != 0) { + g_printerr("Specific config file does not exist\n"); + goto cleanup; + } + if (!S_ISREG(st.st_mode)) { + g_printerr("Specific config is not a regular file\n"); + goto cleanup; + } + + /* Since source file exists, make the destination path or + * validate that it already exists */ + config_dir = g_strdup_printf("%s/etc/", root); + + if (g_mkdir_with_parents(config_dir, 0777) < 0) { + g_printerr("Couldn't make the directory for specific config file\n"); + goto cleanup; + } + + config_path_file = g_strdup_printf("%sqemu.conf", config_dir); + + /* If the source file is same as the destination file, no action needed */ + if ((source_file_real = realpath(source_file, NULL)) && + (config_path_file_real = realpath(config_path_file, NULL))) { + if (STREQ(source_file_real, config_path_file_real)) { + ret = 0; + goto cleanup; + } + } + + /* Check already existing qemu.conf in the subfolder, if so, renamed + * (appended via deltams constant - should be unique). Final cleanup + * at main() will revert this change */ + if (access(config_path_file, R_OK) == 0) { + if (rename(config_path_file, g_strdup_printf("%sqemu_old_%lld.conf", + config_dir, deltams)) != 0) { + g_printerr("Couldn't rename old config file, try delete it\n"); + goto cleanup; + } + } + + if ((dstFD = open(config_path_file, O_WRONLY | O_CREAT | O_TRUNC, 0644)) < 0) { + g_printerr("Couldn't open file for define specific config\n"); + goto cleanup; + } + + /* Set the flag for deleting the new/modified config in main() cleanup, + * for ensure that behaviour without the user config will be always + * consistent and take default value */ + *config_to_delete = true; + + do { + char buffer[1024]; + ssize_t readed = 0; + + readed = saferead(srcFD, buffer, 1024); + + if (readed < 0) { + g_printerr("Couldn't read from specific config\n"); + goto cleanup; + } else if (readed == 0) { + break; + } else { + ssize_t writed = safewrite(dstFD, buffer, readed); + if (writed != readed) { + g_printerr("Couldn't write to file for define specific config\n"); + goto cleanup; + } + } + } while (1); + + ret = 0; + + cleanup: + return ret; +} + int main(int argc, char **argv) { g_autoptr(virIdentity) sysident = NULL; @@ -142,8 +237,11 @@ int main(int argc, char **argv) g_autofree char *uri = NULL; g_autofree char *suri = NULL; const char *root = NULL; + const char *config = NULL; + long long delta_ms = 0; g_autofree char *escaped = NULL; bool tmproot = false; + bool config_to_delete = false; int ret = 1; g_autoptr(GError) error = NULL; g_auto(GStrv) secrets = NULL; @@ -156,6 +254,7 @@ int main(int argc, char **argv) { "root", 'r', 0, G_OPTION_ARG_STRING, &root, "Root directory", "DIR" }, { "debug", 'd', 0, G_OPTION_ARG_NONE, &debug, "Debug output", NULL }, { "verbose", 'v', 0, G_OPTION_ARG_NONE, &verbose, "Verbose output", NULL }, + { "config", 'c', 0, G_OPTION_ARG_STRING, &config, "Load specific QEMU configuration file", "QEMU-CONF-FILE"}, { 0 } }; int quitfd[2] = {-1, -1}; @@ -172,7 +271,7 @@ int main(int argc, char **argv) return 1; } - if (argc != 2) { + if (argc != 2 && argc != 3) { g_autofree char *help = g_option_context_get_help(ctx, TRUE, NULL); g_printerr("%s", help); return 1; @@ -234,6 +333,10 @@ int main(int argc, char **argv) goto cleanup; } + delta_ms = deltams(); + if (config && (qemuAddConfigFile(config, root, &config_to_delete, delta_ms) < 0)) + g_printerr("Specific config file was not loaded, default was used\n"); + escaped = g_uri_escape_string(root, NULL, true); virFileActivateDirOverrideForProg(argv[0]); @@ -402,6 +505,22 @@ int main(int argc, char **argv) VIR_FORCE_CLOSE(quitfd[0]); VIR_FORCE_CLOSE(quitfd[1]); + if (config_to_delete) { + g_autofree char *possible_old_file = g_strdup_printf("%s/etc/qemu_old_%lld.conf", + root, delta_ms); + g_autofree char *to_delete = g_strdup_printf("%s/etc/qemu.conf", root); + + if (remove(to_delete) != 0) + g_printerr("Deleting specific config failed, located in: %s\n", + to_delete); + + if (access(possible_old_file, R_OK) == 0) { + if (rename(possible_old_file, to_delete) != 0) + g_printerr("Renaming your old qemu.conf failed, ups, located in %s\n", + possible_old_file); + } + } + if (dom != NULL) virDomainFree(dom); if (sconn != NULL) -- 2.47.1

3 3

[PATCH 0/5] Drop support for VirtualBox-6.1
by Michal Privoznik 31 Jan '25

31 Jan '25

This was initiated by the following issue: https://gitlab.com/libvirt/libvirt/-/issues/681 and I tried to add support for VBOX-7.1 but unfortunately got hit by a bug in VBOX which renders the way we initialize C bindings useless. I've reported the bug here: https://www.virtualbox.org/ticket/22224 but it didn't get any attention. Michal Prívozník (5): vbox: Use g_autofree in tryLoadOne() vbox: Report an error when VBox CAPI initialization fails vbox: Drop support for VirtualBox-6.1.x vbox: Drop code supporting old VBox version NEWS: Document VBOX-6.1 removal NEWS.rst | 5 + src/vbox/meson.build | 1 - src/vbox/vbox_CAPI_v6_1.h | 32896 -------------------------------- src/vbox/vbox_V6_1.c | 13 - src/vbox/vbox_XPCOMCGlue.c | 5 +- src/vbox/vbox_XPCOMCGlue.h | 2 +- src/vbox/vbox_common.h | 4 +- src/vbox/vbox_storage.c | 4 +- src/vbox/vbox_tmpl.c | 30 +- src/vbox/vbox_uniformed_api.h | 1 - 10 files changed, 16 insertions(+), 32945 deletions(-) delete mode 100644 src/vbox/vbox_CAPI_v6_1.h delete mode 100644 src/vbox/vbox_V6_1.c -- 2.45.2

2 6

[PATCH V2 0/3] qemu: Improve opening and verifying save images
by Jim Fehlig 31 Jan '25

31 Jan '25

V2 of https://lists.libvirt.org/archives/list/devel@lists.libvirt.org/message/DAT… Changes in V2: * Move unlinking corrupt save images from qemuSaveImageOpen to the only caller using that functionality * Add a function to read save image header * Correctly position file pointer of save image for QEMU Jim Fehlig (3): qemu: Move unlinking corrupt save image file to caller qemu: Decompose qemuSaveImageOpen qemu: Check for valid save image format when verifying image header src/qemu/qemu_driver.c | 46 ++++--- src/qemu/qemu_saveimage.c | 268 +++++++++++++++++++++++--------------- src/qemu/qemu_saveimage.h | 21 ++- src/qemu/qemu_snapshot.c | 9 +- 4 files changed, 207 insertions(+), 137 deletions(-) -- 2.43.0

2 4

[PATCH 0/2] qemu: Improve opening and verifying save images
by Jim Fehlig 31 Jan '25

31 Jan '25

qemuSaveImageOpen does a lot of stuff. When bypass-cache is specified, it creates a virFileWrapperFd, which is not cleaned up in the subsequent failure paths. This results in errors from the iohelper, which can overwrite the actual error. E.g. consider libvirt attempting to restore an image saved in an unknown format # virsh restore --bypass-cache /data/test.sav.sparse error: Failed to restore domain from /data/test.sav.sparse error: internal error: Child process (LIBVIRT_LOG_OUTPUTS=1:stderr /usr/lib64/libvirt/libvirt_iohelper /data/test.sav.sparse 0) unexpected fatal signal 13 When not using the iohelper, and not creating a virFileWrapperFd, we see the real error: # virsh restore /data/test.sav.sparse error: Failed to restore domain from /data/test.sav.sparse error: operation failed: Invalid compressed save format 6 Although that error highlights a spot I missed when removing the "compression" implications around the 'foo_image_format' settings in qemu.conf with commit bd6d7ebf622 :-). IMO, the error message would be best fixed by checking for valid values when reading the save image metadata. Patch 1 decomposes qemuSaveImageOpen to allow for better error handling and more flexibility. Patch 2 checks for a valid format when checking the other header fields. Jim Fehlig (2): qemu: Decompose qemuSaveImageOpen qemu: Check for valid save image format when verifying image header src/qemu/qemu_driver.c | 37 +++++++-------- src/qemu/qemu_saveimage.c | 95 +++++++++++++++++++++++++-------------- src/qemu/qemu_saveimage.h | 16 ++++--- src/qemu/qemu_snapshot.c | 9 ++-- 4 files changed, 95 insertions(+), 62 deletions(-) -- 2.43.0

2 8

[PATCH 0/2] finish sysusers support
by Daniel P. Berrangé 31 Jan '25

31 Jan '25

This fixes another missing sysusers file and changes the RPM spec to take account of new RPM processing of sysusers files at install time. Daniel P. Berrangé (2): tools: add sysusers file to create 'virtlogin' group rpm: disable account creation for Fedora >= 42 libvirt.spec.in | 19 +++++++++++++++++++ tools/libvirt-login-shell.sysusers.conf | 1 + tools/meson.build | 7 +++++++ 3 files changed, 27 insertions(+) create mode 100644 tools/libvirt-login-shell.sysusers.conf -- 2.47.1

2 6

[RFC PATCH 0/5] qemu: Route hostdevs to multiple nested SMMUs
by Nathan Chen 30 Jan '25

30 Jan '25

Hi, This is a draft solution for supporting multiple vSMMU instances in a qemu VM. Based on discussions/suggestions received for a previous RFC by Nicolin here[0], the association of vSMMUs to VFIO devices in VM PCIe topology should be moved out of qemu into libvirt. In addition, the nested SMMU nodes should be passed to qemu as pluggable devices. To address these changes, this patch series introduces a new "nestedSmmuv3" IOMMU model and "nestedSmmuv3" device type. Upon specifying the nestedSmmuv3 IOMMU model, nestedSmmuv3 devices will be auto-added to the VM definition based on the available SMMU nodes in the host's sysfs. The nestedSmmuv3 devices will each be attached to a separate PXB controller, and VFIO devices will be routed to PXBs based on their association with host SMMU nodes. This will maintain a VM PCIe topology that allows for multiple nested SMMUs per Nicolin's original qemu patch series in [0] and Shameer's work in [1] to remove VM topology changes from qemu and allow the nested SMMUs to be specified as pluggable devices. For instance, if we specify the nestedSmmuv3 IOMMU model and a hostdev for passthrough: <devices> <hostdev mode='subsystem' type='pci' managed='no'> <source> <address domain='0x0009' bus='0x01' slot='0x00' function='0x0'/> </source> </hostdev> <iommu model='nestedSmmuv3'/> </devices> Libvirt will scan sysfs and populate the VM definition with controllers and nestedSmmuv3 devices based on host config. So if /sys/bus/pci/devices/0009:01:00.0/iommu is a symlink to the host SMMU node represented by /sys/devices/platform/arm-smmu-v3.8.auto/iommu/smmu3.0x0000000016000000 and there are 3 host SMMU nodes under /sys/class/iommu/, we'll see three auto-added nestedSmmuv3 devices, each routed to a pcie-expander-bus controller. Then the hostdev will be routed to a PXB controller that has a matching host SMMU node associated with it: <devices> ... <controller type='pci' index='1' model='pcie-expander-bus'> <model name='pxb-pcie'/> <target busNr='254'/> <address type='pci' domain='0x0000' bus='0x00' slot='0x01' function='0x0'/> </controller> <controller type='pci' index='2' model='pcie-expander-bus'> <model name='pxb-pcie'/> <target busNr='251'/> <address type='pci' domain='0x0000' bus='0x00' slot='0x02' function='0x0'/> </controller> <controller type='pci' index='3' model='pcie-expander-bus'> <model name='pxb-pcie'/> <target busNr='249'/> <address type='pci' domain='0x0000' bus='0x00' slot='0x03' function='0x0'/> </controller> <controller type='pci' index='4' model='pcie-root-port'> <model name='pcie-root-port'/> <target chassis='7' port='0x8'/> <address type='pci' domain='0x0000' bus='0x02' slot='0x01' function='0x0'/> </controller> <hostdev mode='subsystem' type='pci' managed='no'> <source> <address domain='0x0009' bus='0x01' slot='0x00' function='0x0'/> </source> <address type='pci' domain='0x0000' bus='0x04' slot='0x00' function='0x0'/> </hostdev> <iommu model='nestedSmmuv3'/> <nestedSmmuv3> <name>smmu3.0x0000000012000000</name> <address type='pci' domain='0x0000' bus='0x01' slot='0x00' function='0x0'/> </nestedSmmuv3> <nestedSmmuv3> <name>smmu3.0x0000000016000000</name> <address type='pci' domain='0x0000' bus='0x02' slot='0x00' function='0x0'/> </nestedSmmuv3> <nestedSmmuv3> <name>smmu3.0x0000000011000000</name> <address type='pci' domain='0x0000' bus='0x03' slot='0x00' function='0x0'/> </nestedSmmuv3> <iommu model='nestedSmmuv3'/> </devices> TODO: - No DMA mapping can found by UEFI when specifying multiple passthrough devices in the VM definition, and VM boot is subsequently blocked. We need to investigate this for the next revision, but we don't encounter this issue when passing through a single device. We'll include iommufd support in the next revision to narrow down whether the required fix would be outside of libvirt. - Shameer's qemu branch specifies nestedSmmuv3 bus number with "pci-bus" instead of "bus", so the libvirt compilation test args and qemu args in qemuBuildPCINestedSmmuv3DevProps() need to be modified to match this revision of qemu. It will be reverted to using "bus" in the next qemu revision. - This patchset decrements PXB busNr based on how many devices are attached downstream, and the libvirt documentation states we must reserve busNr for the PXB itself in addition to any devices attached downstream. When I launch a VM and a PXB has a pcie-root-port and hostdev attached downstream, busNrs 253, 252, and 251 are reserved. But the PXB itself already has a bus number assigned via the <address/> attribute, and I see 253 and 252 assigned to the hostdev and pcie-root-port in the VM but not 251. Should we decrement busNr based on libvirt documentation or do we only need two busNrs 253 and 252 in the example here? This series is on Github: https://github.com/NathanChenNVIDIA/libvirt/tree/nested-smmuv3-12-05-24 Thanks, Nathan [0] https://lore.kernel.org/qemu-devel/cover.1719361174.git.nicolinc@nvidia.com/ [1] https://lore.kernel.org/qemu-devel/20241108125242.60136-1-shameerali.koloth… Signed-off-by: Nathan Chen <nathanc(a)nvidia.com> Nathan Chen (5): conf: Add a nestedSmmuv3 IOMMU model qemu: Implement and auto-add a nestedSmmuv3 device type qemu: Create PXBs and auto-assign VFIO devs and nested SMMUs qemu: Update PXB busNr for nestedSmmuv3 controllers qemu: Add test case for specifying multiple nested SMMUs docs/formatdomain.rst | 25 ++- src/ch/ch_domain.c | 1 + src/conf/domain_addr.c | 26 ++- src/conf/domain_addr.h | 4 +- src/conf/domain_conf.c | 188 +++++++++++++++++ src/conf/domain_conf.h | 15 ++ src/conf/domain_postparse.c | 1 + src/conf/domain_validate.c | 24 +++ src/conf/schemas/domaincommon.rng | 17 ++ src/conf/virconftypes.h | 2 + src/libvirt_private.syms | 2 + src/lxc/lxc_driver.c | 6 + src/qemu/qemu_command.c | 64 +++++- src/qemu/qemu_command.h | 4 + src/qemu/qemu_domain.c | 2 + src/qemu/qemu_domain_address.c | 193 ++++++++++++++++++ src/qemu/qemu_driver.c | 3 + src/qemu/qemu_hotplug.c | 5 + src/qemu/qemu_postparse.c | 1 + src/qemu/qemu_validate.c | 16 ++ src/test/test_driver.c | 4 + tests/meson.build | 1 + .../iommu-nestedsmmuv3.aarch64-latest.args | 38 ++++ .../iommu-nestedsmmuv3.aarch64-latest.xml | 61 ++++++ tests/qemuxmlconfdata/iommu-nestedsmmuv3.xml | 29 +++ tests/qemuxmlconftest.c | 4 +- tests/schemas/device.rng.in | 1 + tests/virnestedsmmuv3mock.c | 57 ++++++ 28 files changed, 788 insertions(+), 6 deletions(-) create mode 100644 tests/qemuxmlconfdata/iommu-nestedsmmuv3.aarch64-latest.args create mode 100644 tests/qemuxmlconfdata/iommu-nestedsmmuv3.aarch64-latest.xml create mode 100644 tests/qemuxmlconfdata/iommu-nestedsmmuv3.xml create mode 100644 tests/virnestedsmmuv3mock.c -- 2.34.1

4 18