linux-kernel - Apple Thunderbolt Display chaining

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: Windows password security audit tool. GUI, reports in PDF.

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

Message-ID: <acbb3a86-ea15-47ec-90fa-72fbd94921b1@fnarfbargle.com>
Date:   Tue, 29 Mar 2022 19:09:36 +0800
From:   Brad Campbell <lists2009@...rfbargle.com>
To:     linux-kernel@...r.kernel.org,
        Mika Westerberg <mika.westerberg@...ux.intel.com>
Subject: Apple Thunderbolt Display chaining

G'day Mika,

Back in 2019 you assisted me with an issue on an iMac with a "Light 
Ridge" controller running a pair of Apple 27" Thunderbolt displays. At 
the time I commented they worked when plugged into a port each, however 
they don't work when chained.

Back then things crashed horribly. Now they don't crash, and the PCI 
devices all work but there is an issue with the display.

I'm currently testing on vanilla git-head commit 
ae085d7f9365de7da27ab5c0d16b12d51ea7fca9 (Sunday March 27th). A recent 
kernel with all the "for 5.18" patches.

On both the iMac with the "Light Ridge" controller and a MacBookPro with 
a "Falcon Ridge" controller the result is the same.

Plugged into a port each, they work perfectly.

If I chain them, the first one plugged in works and when the second is 
plugged in it's almost like the link is being starved of bandwidth. 
Graphical displays like a desktop, or a terminal break up, tear or 
exhibit almost white noise/snow on both displays. Solid colours with no 
gradient sometimes displays cleanly until something else is displayed 
and then it breaks up.

I've attached 3 clean-boot dmesg from the Laptop.

001 is "Boot up, then plug one display into each port".
002 is "Boot up, plug a display into one port and then plug the second 
display into the first".
003 is "Chain the displays together, boot up then plug the first chained 
display in".

002 and 003 wind up with the same result and the same corruption. The 
corruption is also identical on the iMac with a different GPU and the 
older "Light Ridge" controller.

All of these configurations work under OSX (10.13 on the iMac and 10.15 
on the Laptop).

This isn't important or urgent, however any insight you can provide 
would be appreciated.

Regards,
Brad
View attachment "001" of type "text/plain" (134620 bytes)

View attachment "002" of type "text/plain" (138634 bytes)

View attachment "003" of type "text/plain" (135035 bytes)