Re: thunderbolt: AMD Phoenix USB4 host advertises Gen3 Support=0 in SB LINK_CONF with TB5 device, link stuck at Gen2

[Mika Westerberg <mika.westerberg@linux.intel.com>]

Hi,

On Mon, Nov 09, 2026 at 07:41:13AM -0800, James R Taylor wrote:
> Hi Mika,
> 
> On Mon, Nov 09, 2026 at 04:39:55PM +0100, Mika Westerberg wrote:
> > We could still look at the PORT_CS_18 but the driver log already tells that
> > it is in TBT3 mode. What kind of cable you are using? Is it short TB4/5
> > cable or something else?
> 
> It is a short Thunderbolt 5 cable: StarTech TBLT5MM50CM240W, 0.5 m.

Okay so definitely it should enter USB4 mode.

> One retimer enumerates on the link, 1-0:2.1 (vendor 0x7fea, device
> 0x1032, "retimer NVM format of vendor 0x7fea unknown"), and it goes
> away when the dock is unplugged. I haven't been able to tell from
> Linux whether it sits in the cable or on the DEG2 board.

That's the host side retimer.

> In case it is useful, here is PORT_CS_18 from both ends, read just now
> (same boot as my last mail):
> 
>     $ tbdump -d 1 -r 0 -vv -a 2 -N 1 PORT_CS_18
>     0x00d2 0x00002f00 0b00000000 00000000 00101111 00000000 ../. PORT_CS_18
>       [00:07]        0x0 Cable USB4 Version
>       [08:08]        0x1 Bonding Enabled (BE)
>       [09:09]        0x1 TBT3-Compatible Mode (TCM)
>       [10:10]        0x1 CLx Protocol Support (CPS)
>       [11:11]        0x1 RS-FEC Enabled (Gen 2) (RE2)
>       [12:12]        0x0 RS-FEC Enabled (Gen 3) (RE3)
>       [13:13]        0x1 Router Detected (RD)
>       [20:20]        0x0 Cable Gen 3 Support (CG3)
>       [21:21]        0x0 Cable Gen 4 Support (CG4)
>       [22:22]        0x0 Cable Asymmetric Support (CSA)
>       [23:23]        0x0 Cable CLx Support (CSC)
> 
>     $ tbdump -d 1 -r 2 -vv -a 1 -N 1 PORT_CS_18
>     0x00d2 0x00902f01 0b00000000 10010000 00101111 00000001 ../. PORT_CS_18
>       [00:07]        0x1 Cable USB4 Version
>       [08:08]        0x1 Bonding Enabled (BE)
>       [09:09]        0x1 TBT3-Compatible Mode (TCM)
>       [10:10]        0x1 CLx Protocol Support (CPS)
>       [11:11]        0x1 RS-FEC Enabled (Gen 2) (RE2)
>       [12:12]        0x0 RS-FEC Enabled (Gen 3) (RE3)
>       [13:13]        0x1 Router Detected (RD)
>       [20:20]        0x1 Cable Gen 3 Support (CG3)
>       [21:21]        0x0 Cable Gen 4 Support (CG4)
>       [22:22]        0x0 Cable Asymmetric Support (CSA)
>       [23:23]        0x1 Cable CLx Support (CSC)
> 
> (The wake status bits and TIP, 16-19 and 24, are 0 on both sides.)
> 
> So both ends are in TBT3-compatible mode, but the host side reports
> Cable USB4 Version 0 and Cable Gen 3 Support 0, while the DEG2 side
> reports Cable USB4 Version 1 and Cable Gen 3 Support 1 for the same
> cable.

Agree it looks like the AMD host only gets the link up Gen2 for some
reason. It could be retimer/PD firmware issue. Anyways I'm adding the AMD
experts if they have any ideas.