Bullseye VM network/DHCP fails on ESXi with USB NIC uplink (Bookworm works fine)

### Creating a bug report/issue

[x] I have searched the existing open and closed issues

### Required Information

DietPi version | v8.1.2 (from official DietPi_VMware-x86_64-Bullseye.7z image)

Distro version | bullseye

Kernel version | Linux dietpi 5.10.0-11-amd64 #1 SMP Debian 5.10.92-1 (2022-01-18) x86_64 GNU/Linux

Architecture | amd64

SBC model | Virtual Machine (VMware ESXi 6.7.0)

Power supply used | N/A (VM on Apple Mac hardware host)

SD card used | N/A (ESXi VMFS thin-provisioned datastore)

### Additional Information (if applicable)

Software title | Base system / Networking (dhclient / ifupdown)

Was the software title installed freshly or updated/migrated? | Freshly installed from official VMware Bullseye image

Can this issue be replicated on a fresh installation of DietPi? | Yes, 100% reproducible on fresh install

Bug report ID | N/A (cannot upload bugreport because network is completely down)

### Steps to reproduce

1. Download `DietPi_VMware-x86_64-Bullseye.7z` from the official archive.

2. Extract and convert the virtual disk to ESXi VMFS thin-provisioned format using `vmkfstools -i`.

3. Attach the converted disk to a VM in VMware ESXi 6.7 (running on Mac hardware where the vSwitch uplink is a USB Ethernet adapter - `vusb0`).

4. Boot the VM (with Promiscuous Mode, Forged Transmits, and MAC Address Changes all set to Accept on both the vSwitch and portgroup).

### Expected behaviour

The VM should establish a link state, successfully negotiate a DHCP lease over the vSwitch, and boot to the login prompt. (Note: A DietPi Bookworm VM running on this exact same ESXi host and vSwitch works perfectly and gets an IP immediately).

### Actual behaviour

The VM boots up but hangs indefinitely during the pre-login boot phase at:

`[*** ] A start job is running for ifup for eth0 (36s / 5min 1s)`

Logging into the console shows `dhclient` broadcasting `DHCPDISCOVER on eth0 to 255.255.255.255 port 67` over and over, but receiving `0 DHCPOFFER` replies. Switching the virtual adapter type in ESXi between `vmxnet3`, `e1000`, and `e1000e` makes no difference.

### Extra details

- I specifically need Debian 11 Bullseye (kernel 5.10) for this VM due to legacy compatibility requirements for a Netatalk 4 setup, so upgrading to Bookworm isn’t an option for me.

- I also tried setting a static IP in `/etc/network/interfaces`. When I do that, the VM’s MAC address successfully registers in the ESXi host ARP table (`esxcli network ip neighbor list`), but pinging the gateway or external sites results in 100% packet loss.

- Since Bookworm (kernel 6.1) works flawlessly over this exact same USB uplink (`vusb0`) and vSwitch, but Bullseye (kernel 5.10) drops all incoming traffic regardless of virtual NIC hardware emulation, I’m wondering if there is a known issue or driver quirk with kernel 5.10 and ESXi USB uplinks?

Please use a recent DietPi image rather than a 4 years old one, else we cannot help. We do not support Bullseye anymore, neither does Debian, hence you shouldn’t even attempt to use it. Debian Trixie is the way to go.

And please use either the OVA appliance, or the ISO installer image for ESXi, so no image conversion needed.

There is no official archive. All images we do support are here (which includes all those listed on our download page): Index of /downloads/images
Everything you find anywhere else is not official, and not supported by us.