linux-kernel - Re: [PATCH 0/3] thunderbolt: Update XDomain vendor properties dynamically

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <d079c31f-4785-4a41-8c88-cc56dceee7aa@kernel.org>
Date: Wed, 6 Aug 2025 10:05:11 -0500
From: Mario Limonciello <superm1@...nel.org>
To: Mika Westerberg <mika.westerberg@...ux.intel.com>
Cc: "Rangoju, Raju" <raju.rangoju@....com>, linux-usb@...r.kernel.org,
 linux-pci@...r.kernel.org, linux-kernel@...r.kernel.org,
 andreas.noever@...il.com, michael.jamet@...el.com, westeri@...nel.org,
 YehezkelShB@...il.com, bhelgaas@...gle.com, Sanath.S@....com
Subject: Re: [PATCH 0/3] thunderbolt: Update XDomain vendor properties
 dynamically

On 8/6/2025 10:00 AM, Mika Westerberg wrote:
> On Wed, Aug 06, 2025 at 09:06:30AM -0500, Mario Limonciello wrote:
>> On 8/6/2025 3:51 AM, Mika Westerberg wrote:
>>> On Wed, Aug 06, 2025 at 11:46:04AM +0530, Rangoju, Raju wrote:
>>>>
>>>>
>>>> On 7/28/2025 12:17 PM, Mika Westerberg wrote:
>>>>> Hi,
>>>>>
>>>>> On Tue, Jul 22, 2025 at 11:20:23PM +0530, Raju Rangoju wrote:
>>>>>> This patch series aims to update vendor properties for XDomain
>>>>>> dynamically for vendors like AMD, Intel and ASMedia.
>>>>>
>>>>> The XDomain properties pretty much describe "software" not the underlying
>>>>> hardware so I don't understand why this is needed? We could have some USB
>>>>> IF registered Linux specific ID there but I don't see why this matters at
>>>>> all.
>>>>
>>>> Currently, it is showing up as "Intel" on AMD host controllers during
>>>> inter-domain connection. I suppose an alternative is to just call it "Linux"
>>>> or "Linux Connection Manager" to ensure we accurately represent the
>>>> connections across different systems.
>>>>
>>>> I appreciate your guidance on this and suggestions you might have.
>>>
>>> Yeah, something like that (I prefer "Linux"). The "ID" still is 0x8086
>>> though but I don't think that matters. AFAIK we have other "donated" IDs in
>>> use in Linux. Let me check on our side if that's okay.
>>
>> Having looked through this discussion I personally like "Linux" for this
>> string too.
>>
>> As for the vendor ID doesn't the LF have an ID assigned already of 0x1d6b?
>> Would it make sense to use that?
> 
> AFAIK that's PCI ID, right? It should be USB IF assigned ID and LF is not
> here at least:
> 
>    https://www.usb.org/members
> 
> If it really matters we can sure register one.

I see that it's used by drivers/usb/gadget/legacy.c for a few USB 
devices too.

So it's "already in the wild".

> 
>> I was also thinking about the device ID, should we consider encoding the
>> VERSION, PATCHLEVEL, and SUBLEVEL into the ID?  The reason I'm thinking
>> about that is let's say there is some bug found in the CM on Linux and
>> another implementation decides to work around it.  We get wind of it and fix
>> the bug but in Linux but now what about the other end?  If we give them a
>> hint on the version by putting it in the device ID they can potentially key
>> off that to decide to tear out the workaround.
> 
> I'm not sure that's a good idea. Then we expose also all the known
> vulnerabilities.

Yeah I see your point.  This wasn't a strong feeling on my side, leaving 
this as is is fine.