Re: [PATCH net-next v2 1/2] tls: remove tls_toe and the related driver

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Medium] The cxgb4 driver continues to reserve MSI-X vectors and hardware queues for inline TLS offload even though the chtls driver has been removed.
--

commit fe56f0d87d512a0eea1f91c98873fd22553c6faf
Author: Sabrina Dubroca <sd@queasysnail.net>

tls: remove tls_toe and the related driver

This commit removes the tls_toe feature and its single user, the chelsio
chtls driver. These components have been unmaintained and bypass most of
the networking stack.

> diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c
> index 6df98fca932f7..9e2c2fa16d7a5 100644
> --- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c
> +++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c
> @@ -6795,8 +6795,7 @@ static int init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
>  						   NETIF_F_TSO | NETIF_F_TSO6;
>  
>  			netdev->hw_features |= NETIF_F_GSO_UDP_TUNNEL |
> -					       NETIF_F_GSO_UDP_TUNNEL_CSUM |
> -					       NETIF_F_HW_TLS_RECORD;
> +					       NETIF_F_GSO_UDP_TUNNEL_CSUM;

[Severity: Medium]
Since the chtls driver is being removed, should we also remove the 
FW_CAPS_CONFIG_TLS_INLINE capability check in get_caps() within the same file?

drivers/net/ethernet/chelsio/cxgb4/cxgb4_main.c:get_caps() {
    ...
		if (ntohs(caps_cmd.cryptocaps) &
		    FW_CAPS_CONFIG_TLS_INLINE) {
			params[0] = FW_PARAM_PFVF(TLS_START);
			params[1] = FW_PARAM_PFVF(TLS_END);
			ret = t4_query_params(adap, adap->mbox, adap->pf, 0,
					      2, params, val);
			if (ret < 0)
				goto bye;
			adap->vres.key.start = val[0];
			adap->vres.key.size = val[1] - val[0] + 1;
			adap->num_uld += 1;
		}
    ...
}

Leaving this check in place causes the driver to increment adap->num_uld when
the firmware advertises TLS inline offload. This unconditionally reserves MSI-X
vectors and offload queues in cfg_queues() for a ULD that no longer exists in
the kernel.

Does this waste hardware queues and MSI-X vectors, potentially reducing the
number of queues available for the main Ethernet interface?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/cover.1781165969.git.sd@queasysnail.net?part=1