This goes into the "reset bug" and its many facets. Afaik this work is not yet finished, it certainly wasn't when this video was published. Is this now completed? Can you please link to info on this, e.g. the relevant commits or changelog item?
I'm not going to watch a 35 minute video, create a bug report if you can reproduce it.
I can now confirm that on 15.0p3 this is still an issue.
uname -ar
\
FreeBSD beastie 15.0-RELEASE-p5 FreeBSD 15.0-RELEASE-p5 GENERIC amd64#cd /usr/
#rm -rf /usr/src
#git clone https://github.com/beckhoff/freebsd-src /usr/src
cd /usr/src
#git checkout -f origin/phab/corvink/14.2/nvidia-wip
git checkout -f origin/phab/corvink/15.0/nvidia-wip
cd /usr/src/usr.sbin/bhyve
sudo make && make install
\
...
Error
/usr/src/usr.sbin/bhyve/bootrom.c:123:33: error: use of undeclared identifier 'VM_BOOTROM'
123 | romptr = vm_create_devmem(ctx, VM_BOOTROM, "bootrom", BOOTROM_SIZE);
| ^
/usr/src/usr.sbin/bhyve/bootrom.c:177:31: error: use of undeclared identifier 'VM_BOOTROM'
177 | if (vm_mmap_memseg(ctx, gpa, VM_BOOTROM, segoff, len, prot) != 0) {
| ^
2 errors generated.
*** Error code 1
Stop.
make: stopped making "all" in /usr/src/usr.sbin/bhyvepciconf -l | grep ppt
ppt0@pci0:8:0:0: class=0x030000 rev=0xa1 hdr=0x00 vendor=0x10de device=0x1e87 subvendor=0x3842 subdevice=0x2183
ppt1@pci0:8:0:1: class=0x040300 rev=0xa1 hdr=0x00 vendor=0x10de device=0x10f8 subvendor=0x3842 subdevice=0x2183
ppt2@pci0:8:0:2: class=0x0c0330 rev=0xa1 hdr=0x00 vendor=0x10de device=0x1ad8 subvendor=0x3842 subdevice=0x2183
ppt3@pci0:8:0:3: class=0x0c8000 rev=0xa1 hdr=0x00 vendor=0x10de device=0x1ad9 subvendor=0x3842 subdevice=0x2183cbsd bhyve-ppt mode=list
\
Configured PCI Passthru devices (4):
PCI DRIVER DEVICE CLASS SUBCLASS VM VENDOR
8/0/3 ppt (1) TU104 USB Type-C UCSI Controller serial bus USB - NVIDIA Corporationcmd: env LIB9P_LOGGING=/usr/jails/jails-system/cachyos_linux/cbsd_lib9p.log /usr/bin/nice -n 1 /usr/sbin/bhyve -w -H -S
-A -c cpus=8,sockets=8,cores=1,threads=1 -m 8589934592 -H -A -S -u -w -U dd1c24ca-2c4c-11f1-8a0c-7085c25db4b0 -s 0,hostb
ridge -s 4,virtio-blk,/usr/jails/vm/cachyos_linux/dsk1.vhd,sectorsize=512/4096 -s 5,virtio-net,tap3,mtu=1500,mac=00
:a0:98:c2:1f:00 -s 6,virtio-rnd -s 10:0,passthru,8/0/0 -s 10:1,passthru,8/0/1 -s 10:2,passthru,8/0/2 -s 10:3,passthru,8
/0/3 -s 7,fbuf,tcp=127.0.0.1:5902,w=1920,h=1080 -s 30,xhci,tablet -s 31,lpc -l bootrom,/usr/local/cbsd/upgrade/patch/ef
i.fd,/usr/jails/jails-system/cachyos_linux/BHYVE_UEFI_VARS.fd,fwcfg=qemu -l com1,stdio cachyos_linux
-----
bhyve: PCI device at 8/0/0 is not using the ppt(4) driver
Device emulation initialization error: Device busy
fbuf frame buffer base: 0x2279fe400000 [sz 33554432]
Sleep 1 seconds...
Erase is delete.
119: not found
Warning: unable to determine bhyve pid for: cachyos_linux
bstop done in 1 seconds#BHYVE setting needed for the cbsd bconstruct-tui
vmm_load="YES"
##Passthough
nmdm_load="YES"
#Bhyve Passthrough
pptdevs="8/0/0 8/0/1 8/0/2 8/0/3"
hw.vmm.amdvi.enable="1"
# Force the IOMMU to be more permissive with BAR mapping
hw.vmm.iommu.enable="1"
hw.vmm.iommu.relaxed_mapping="1"
# Disable the host's EFI framebuffer from "holding" the GPU memory
hw.vga.textmode="1"
machdep.disable_mtrr_setup="1"
# Specific for NVIDIA TU104 on FreeBSD 15
hw.pci.honor_msi_blacklist="0"git checkout -f origin/phab/corvink/15.0/nvidia-wipThat's not how it's done, you cant just build bhyve like that, do a buildworld and buildkernel if you have no idea what you're doing. Don't overcomplicate things yet.
It is not really a bug, it is more like a lack of feature in GPU card. Most consumer GPUs do not support Function Level Reset (FLR) properly or do not support it at all. Sometimes it is possible to reset those cards using various hacks.
vm poweroff win11gamesvm my system has Kernel Panic and restarts. I was testing a little bit more by installing fresh Windows 11 but it makes no difference. When I have disabled in /boot/loader.conf T1000 I have managed to start VM, get freeze and kill VM with no Kernel Panic. I have updated FreeBSD 15.0 to 15.0-RELEASE-p9
pciconf -lv
ppt0@pci0:130:0:0: class=0x030000 rev=0xa1 hdr=0x00 vendor=0x10de device=0x2d05 subvendor=0x1458 subdevice=0x41a1
vendor = 'NVIDIA Corporation'
device = 'GB206 [GeForce RTX 5060]'
class = display
subclass = VGA
ppt1@pci0:130:0:1: class=0x040300 rev=0xa1 hdr=0x00 vendor=0x10de device=0x22eb subvendor=0x10de subdevice=0x0000
vendor = 'NVIDIA Corporation'
class = multimedia
subclass = HDA
loader="uefi"
graphics="yes"
graphics_res="1920x1080"
xhci_mouse="yes"
cpu=8
cpu_sockets=1
cpu_cores=4
cpu_threads=2
memory=16G
# put up to 8 disks on a single ahci controller.
# without this, adding a disk pushes the following network devices onto higher slot numbers,
# which causes windows to see them as a new interface
ahci_device_limit="8"
# ideally this should be changed to virtio-net and drivers installed in the guest
# e1000 works out-of-the-box
network0_type="virtio-net"
#network0_type="e1000"
network0_device="tap7"
network0_switch="public"
disk0_type="nvme"
disk0_name="disk0.img"
#disk1_type="nvme"
#disk1_name="gamedrive.img"
# Passthrough
#passthru0="129/0/0=3:0"
#passthru1="129/0/1=3:1"
#RTX 5060
passthru0="130/0/0=3:0"
passthru1="130/0/1=3:1"
# windows expects the host to expose localtime by default, not UTC
utctime="no"
uuid="21280ac9-3329-11ef-bb25-ecf4bbe5c838"
network0_mac="58:9c:fc:03:0b:77"
debug="yes"