Re: [PATCH 23/23] usb: xhci: Add debugfs support for xHCI Port Link Info (PORTLI) register.

[Michal Pecio <michal.pecio@gmail.com>]

On Wed, 19 Nov 2025 16:24:17 +0200, Mathias Nyman wrote:
> From: "Rai, Amardeep" <amardeep.rai@intel.com>
> 
> Each xHCI roothub port has a Port Link Info (PORTLI) register that is
> used by USB3 and eUSB2V2 ports.
> 
> USB3 ports show link error count, rx lane count, and tx lane count.
> 
> eUSB2V2 ports show Rx Data Rate (RDR) and Tx Data Rate (TDR).
> 
> Rx/Tx Data Rate is a multiple of USB2 2.0 HS 480 Mb/s data rates,
> and is only valid if a eUSB2V2 device is connected (CCS=1).
> 
> 0 = "USB 2.0 HS" normal HS 480 Mb/s, no eUSB2V2 in use
> 1 = "HS1" Assymetric eUSB2V2 where this direction runs normal 480Mb/s
> 2 = "HS2" 960Mb/s
> ...
> 10 = "HS10" 4.8 Gb/s, max eUSB2V2 rate
> 
> PORTLI is Reserved and preserve "RsvdP" for normal USB2 ports
> 
> Sample output of USB3 port PORTLI:
> cat /sys/kernel/debug/usb/xhci/0000:00:14.0/ports/port14/portli
> 0x00000000 LEC=0 RLC=0 TLC=0
> 
> Signed-off-by: Rai, Amardeep <amardeep.rai@intel.com>
> Co-developed-by: Mathias Nyman <mathias.nyman@linux.intel.com>
> Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>

Hi,

This patch causes an oops when there are more port registers counted in
xhci->max_ports than ports reported by Supported Protocol capabilities.
On my HW it's due to max_ports being more than maximum port number, but
it seems that gaps between ports of different speeds are also possible.

> +static int xhci_portli_show(struct seq_file *s, void *unused)
> +{
> +	struct xhci_port	*port = s->private;
> +	struct xhci_hcd		*xhci = hcd_to_xhci(port->rhub->hcd);

In such cases port->rhub will be NULL so we can't reach xhci. One
obvious solution (which works for me) is an explicit NULL check here
and another seq_printf just for this case, followed by early return.

Or maybe not bother printing anything and just return?
Or make this check before creating the files in the first place?

> +	u32			portli;
> +
> +	portli = readl(&port->port_reg->portli);
> +
> +	/* PORTLI fields are valid if port is a USB3 or eUSB2V2 port */
> +	if (port->rhub == &xhci->usb3_rhub)
> +		seq_printf(s, "0x%08x LEC=%u RLC=%u TLC=%u\n", portli,
> +			   PORT_LEC(portli), PORT_RX_LANES(portli), PORT_TX_LANES(portli));
> +	else if (xhci->hcc_params2 & HCC2_E2V2C)
> +		seq_printf(s, "0x%08x RDR=%u TDR=%u\n", portli,
> +			   PORTLI_RDR(portli), PORTLI_TDR(portli));

Or I suppose we could alternatively use PORTSC to determine speed
and remove the dependency on hcc_params2 by always decoding eUSB2v2
fields whether the HW supports it or not. It's debugfs, who cares.

> +	else
> +		seq_printf(s, "0x%08x RsvdP\n", portli);
> +
> +	return 0;
> +}

Regards,
Michal