Same Problems here but on my side with a Linux machine.
Beelink EQ12 — USB 3.2 write failure report
HARDWARE
Model (DMI) AZW / EQ (Beelink EQ12)
CPU Intel N100
BIOS AMI N95V104, 2023-03-17
BIOS Version: N95V104
USB controller Intel Alder Lake-N PCH USB 3.2 xHCI [8086:54ed]
OS Debian 13, kernel 6.12.90-amd64
PROBLEM
Device: Samsung Portable SSD T7, USB ID 04e8:4001, firmware FXG42P2Q
On the USB 3.2 Gen2 port (link trains at 10 Gbit/s):
- Reads work fine: 677 MB/s sustained, no errors
- Every write fails: 0 bytes transferred
Kernel log during any write attempt:
usb 2-4: reset SuperSpeed Plus Gen 2x1 USB device number N using xhci_hcd
(repeats 1/s)
sd 2:0:0:0: [sda] tag#0 FAILED Result: hostbyte=DID_ERROR
driverbyte=DRIVER_OK
sd 2:0:0:0: [sda] tag#0 CDB: Write(10) 2a 00 00 00 08 00 00 08 00 00
I/O error, dev sda, sector 2048 op 0×1🙁WRITE)
(with uas driver: “data cmplt err -71”, commands time out after 180 s)
TESTED AND RULED OUT
- Both storage drivers: uas, and usb-storage via quirks=04e8:4001:u
- Two different USB-C cables
- USB3 link power management disabled (usb3_lpm_permit=0)
- Reduced transfer size (max_sectors_kb 1024 / 64 / 32)
- Latest T7 firmware (Samsung Magician reports FXG42P2Q as current)
WORKING REFERENCE CASES
- Same port, same cable, different enclosure (Realtek RTL9210 bridge):
620 MB/s sustained write, zero errors
- Same T7, same cable, connected to a Mac:
519 MB/s sustained write over 20 GB, zero errors
- Same T7 on this machine’s USB 2.0 ports:
42 MB/s, fully stable
SUSPECTED CAUSE
Interoperability problem between the Alder Lake-N xHCI host controller
and the Samsung T7 USB bridge at Gen2 (10 Gbit/s) link speed, affecting
only the host-to-device (write) direction. The drive, the cable and the
port each work correctly in other combinations, so the fault appears to
be specific to this host/device pairing.
QUESTION
Is there a BIOS newer than N95V104 for the EQ12 containing USB/xHCI
fixes? Our BIOS dates from March 2023.