Re: Inquiry about tb/usb4 driver

["mika.westerberg@linux.intel.com" <mika.westerberg@linux.intel.com>]

Hi,

On Thu, Feb 29, 2024 at 12:18:31PM +0000, mohamed.rayan@siemens.com wrote:
>    Hello,
> 
>     
> 
>    Hope this mail finds you well,

It does but sending HTML has some issues if the receiver such as me uses
text based mail client ;-)

>    I am sending this mail asking you for some help in the usage of tb/usb4
>    driver,
> 
>    I have downloaded this driver for kernel v6.5 to install it with a certain
>    USB4 host controller, Here is the driver link:
>    [1]https://elixir.bootlin.com/linux/v6.5/source/drivers/thunderbol
> 
>     
> 
>    I have found something not clear to me in “switch.c”, and exactly at line
>    #2260: “[2]tb_switch_alloc”, which is responsible for allocating and
>    initialization a switch:
> 
>  1-    This function has a local variable called “upstream_port”
> 
>  2-    A “tb_cfg_get_upstream_port” is first called to get the value of “upstream_port” from Switch config space
> 
>  3-    When I checked “tb_cfg_get_upstream_port” function, I found that it reads first DW (DW#0) of the router configuration space
> 
>  4-    In USB4 standard specifications, the first DW of the router config space is Vendor/Product IDs, not the Upstream Adapter, please check the below snapshot from USB4 standard
> 
>  5-    Later in the code of the “[3]tb_switch_alloc” function, at line#2297, this value of “upstream_port”  is used to overwrite another variable in the struct as follow:
> 
>                  
> 
>   
> 
>   
> 
>   
> 
>  6-    So at the end, “sw->config.upstream_port_number” will get updated with incorrect settings, So I am wondering if this is an expected behavior, Am I missing something?

This is historic to support Thunderbolt 1-3 devices where the upstream
port could change. In USB4 routers it is read-only. Do you see any real
issue with this?