The RTL8852BE-VT card at PCI slot 01:00.0 fails to initialize after resuming from suspend. The device completely disappears from lspci and rfkill lists.
Attempts to fix using pci=noaer, pcie_aspm=off, and echo 1 > /sys/bus/pci/rescan have failed. The issue seems specific to the -VT variant of the chipset on the Meteor Lake platform.
Network Controller: Realtek Semiconductor Co., Ltd. RTL8852BE-VT PCIe 802.11ax Wireless Network Controller
PCI Slot: 01:00.0
Device ID: (Please check lspci -nn output for ID like [10ec:b852])
System: HP 15-fd1000 Laptop (Intel Core Ultra 5 125H)
Storage: SK hynix BC901 NVMe SSD
Kernel: Fedora 43 (Rawhide)
I am experiencing the exact same architectural failure on the exact same hardware (HP 15-fd1000, Core Ultra 5, RTL8852BE-VT, Fedora 44). I have also exhausted all sysfs rescans, unbind/bind hooks, and ASPM overrides without success. The motherboard completely cuts power to the M.2 slot.
To add vital telemetry to your report, I managed to extract the kernel panic exactly when the machine wakes up from S3/s2idle. The kernel is reporting that the crystal oscillator (xtal) on the Realtek chip is dead due to the lack of an electrical wake signal from the HP ACPI tables:
[ 100.573505] rtw89_8852bte 0000:01:00.0: xtal si not ready(W): offset=90 val=10 mask=10
[ 100.594452] rtw89_8852bte 0000:01:00.0: mac preinit fail, ret: -110
We need a low-level ACPI override or undocumented kernel parameter to bypass this HP power state bug. Has anyone found a solution besides physically swapping the card for an Intel AX210?