Re: [PATCH net] net: thunderbolt: do not enable E2E flow control on the Tx ring

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

Hi,

On Thu, Jul 23, 2026 at 02:56:49AM +0000, faliye wrote:
> From: FAN YE <fy15309206903@gmail.com>
> 
> Commit a8065af3346e ("net: thunderbolt: Enable end-to-end flow control
> also in transmit") started passing RING_FLAG_E2E to tb_ring_alloc_tx()
> in addition to the Rx ring.
> 
> Per the USB4 spec, a Transmit Descriptor Ring with E2E flow control
> disabled does not require any credits to be available before the Host
> Interface Adapter Layer transmits a tunneled packet from it. Once E2E is
> enabled on that ring the controller must first obtain end-to-end
> credits.
> 
> The ASMedia ASM4242 USB4 host router (PCI 1b21:2425) never delivers
> those credits. The controller does accept the configuration: reading the
> ring OPTIONS register back right after tb_ring_start() returns exactly
> what was written, including RING_FLAG_E2E_FLOW_CONTROL (bit 28) and the
> E2E HopID field. No credit ever arrives though, so the Tx ring's
> hardware consumer index never advances and the link carries no traffic
> at all.
> 
> Measured on two hosts connected point to point, onboard ASM4242 on MSI
> X870E and X870, v6.17, stock drivers/net/thunderbolt/main.c with only
> this patch applied on top:
> 
>   before: 100% packet loss to the peer; thunderbolt0 is up and the
>           XDomain handshake completes ("new host found"), but iperf3
>           fails with "No route to host" once the neighbour entry
>           expires
>   after:  0% packet loss, 0.28 ms RTT; iperf3 4.21 Gb/s one way and
>           5.17 Gb/s the other (5 runs each, stddev <= 0.02), 1
>           retransmit in 10 s
> 
> An instrumented build additionally showed a frozen-Tx-consumer watchdog
> firing ~30k times in a 10 s window before this change.
> 
> Rx-side E2E is left untouched, so peers that do return credits keep
> receive-side flow control.
> 
> ASMedia does not look like an isolated case. The out-of-tree
> thunderbolt-ibverbs project disables native E2E on AMD NHI by default,
> noting that "Strix Halo has reproduced TX completion wedges with
> multiple native E2E rings active" -- the same failure mode, on a
> different vendor. Since the driver has no way to tell in advance which
> host router returns the credits, defaulting Tx-side E2E to off seems
> safer than adding a quirk per affected part; it can be turned back on
> for controllers that are known to implement the credit return.
> 
> Fixes: a8065af3346e ("net: thunderbolt: Enable end-to-end flow control also in transmit")
> Signed-off-by: FAN YE <fy15309206903@gmail.com>
> ---
>  drivers/net/thunderbolt/main.c | 10 +++++++++-
>  1 file changed, 9 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/net/thunderbolt/main.c b/drivers/net/thunderbolt/main.c
> index 02a916505..121a42a4f 100644
> --- a/drivers/net/thunderbolt/main.c
> +++ b/drivers/net/thunderbolt/main.c
> @@ -930,7 +930,15 @@ static int tbnet_open(struct net_device *dev)
>  	if (tbnet_e2e && net->svc->prtcstns & TBNET_E2E)
>  		flags |= RING_FLAG_E2E;
>  
> -	ring = tb_ring_alloc_tx(xd->tb->nhi, -1, TBNET_RING_SIZE, flags);
> +	/*
> +	 * Leave E2E off the Tx ring. Per the USB4 spec a Transmit Descriptor
> +	 * Ring with E2E enabled must obtain end-to-end credits before it may
> +	 * transmit, and not every host router delivers them. On ASMedia
> +	 * ASM4242 the credits never arrive, which parks the Tx consumer index
> +	 * forever and makes the link unusable.
> +	 */
> +	ring = tb_ring_alloc_tx(xd->tb->nhi, -1, TBNET_RING_SIZE,
> +				RING_FLAG_FRAME);

Why not simply send revert for a8065af3346e?

>  	if (!ring) {
>  		netdev_err(dev, "failed to allocate Tx ring\n");
>  		return -ENOMEM;
> -- 
> 2.43.0