Re: USB: Lexar E300 21c4:b083 fails SuperSpeedPlus negotiation due to missing BOS descriptor — request for NO_BOS quirk

[Gregkh <gregkh@linuxfoundation.org>]

On Fri, May 08, 2026 at 09:18:22PM +0200, c2mismo wrote:
> 
> Dear USB subsystem maintainers,
> 
> I have a Lexar E300 M.2 SSD enclosure (idVendor=0x21c4, idProduct=0xb083)
> that supports USB 3.2 Gen 2 (10 Gbps). Under Linux it always negotiates
> down to 5 Gbps, while on Windows the same hardware (cable, port) works
> correctly at 10 Gbps (CrystalDiskMark ~1087 MB/s vs ~450 MB/s on Linux).
> 
> The kernel log shows that the device initially attempts SuperSpeed Plus
> (Gen 2x1) but fails to provide a valid BOS descriptor, causing an error
> -71 and a forced fallback to SuperSpeed (5 Gbps):
> 
> [  179.146227] usb 2-1: new SuperSpeed Plus Gen 2x1 USB device number 4 using xhci_hcd
> [  182.237405] usb 2-1: USB disconnect, device number 4
> [  182.420843] sd 12:0:0:0: [sda] Synchronize Cache(10) failed: Result: hostbyte=DID_ERROR driverbyte=DRIVER_OK
> [  182.552343] usb usb2-port1: config error
> [  183.131966] usb 2-1: new SuperSpeed USB device number 5 using xhci_hcd
> 
> "lsusb -t" always reports the device at 5000M, even when connected to a
> 20 Gbps port.
> 
> I have reliable reports that other users on Ubuntu experience the
> identical issue with the same enclosure.
> 
> The likely cause is that the enclosure's firmware does not handle the
> BOS descriptor request correctly. The fix is to mark it with the
> USB_QUIRK_NO_BOS flag, already existsting in previous kernels
> (since commit 7b2f6c5b1b1c, "USB: core: Add quirk for devices that
> can't handle BOS"). That quirk would prevent the kernel from trying to
> read the BOS descriptor and thus avoid the downgrade.
> 
> i did try to add the following entry to drivers/usb/core/quirks.c
> in the usb_quirk_list[]:
> 
> { USB_DEVICE(0x21c4, 0xb083), .driver_info = USB_QUIRK_NO_BOS },
> 
> I am running Arch Linux with kernel 7.0.3 (in the recent version does not contain
> the NO_BOS definition in its source),

7.0.3 has that quirk flag, so this should work.

> so I have been unable to test the
> patch myself. However, I am fully available to test any patch or
> directly apply and verify a fix if a developer can provide one. I can
> also provide additional debug output if needed.

Can you make the patch so you get full credit for it?

thanks,

greg k-h