linux-kernel - Re: 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 for Android: free password hash cracker in your pocket

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

Message-ID: <Yu4Gmkous7asng6h@lahna>
Date:   Sat, 6 Aug 2022 09:13:46 +0300
From:   Mika Westerberg <mika.westerberg@...ux.intel.com>
To:     Brad Campbell <lists2009@...rfbargle.com>
Cc:     linux-kernel@...r.kernel.org
Subject: Re: Apple Thunderbolt Display chaining

Hi Brad,

On Fri, Aug 05, 2022 at 10:43:54PM +0800, Brad Campbell wrote:
> On 5/8/22 22:21, Mika Westerberg wrote:
> 
> > They are pretty standard so I suspect myself the display side of things.
> > Not sure if it is possible (I think it is from sysfs /sys/class/drm/*)
> > to disable the tunneled DP connections and see if that makes it not
> > hang. Alternatively you can try to comment out the call to
> > tb_tunnel_dp() from the driver. Let me know if you want me to make hack
> > patch that does it for you.
> > 
> 
> I used this :
> 
> iff --git a/drivers/thunderbolt/tb.c b/drivers/thunderbolt/tb.c
> index 9a3214fb5038..eae7523af78b 100644
> --- a/drivers/thunderbolt/tb.c
> +++ b/drivers/thunderbolt/tb.c
> @@ -865,7 +865,7 @@ static void tb_tunnel_dp(struct tb *tb)
>         struct tb_cm *tcm = tb_priv(tb);
>         struct tb_port *port, *in, *out;
>         struct tb_tunnel *tunnel;
> -
> +       return;
>         if (!tb_acpi_may_tunnel_dp()) {
>                 tb_dbg(tb, "DP tunneling disabled, not creating tunnel\n");
>                 return;
> 
> I'm now using Linus GIT head. No change in behaviour really.
> 
> No tunnels were created. None of the TB displays light up in Linux or BIOS. System still locks
> up on reboot and the first time I attempted to compose this reply it locked up hard before I
> had a chance to finish it because I attempted to verify the devices were present with an lspci -tv.
> 
> If I do an lspci -tv after the module load, it locks hard instantly and reproducibly.

Okay, let's try to deal with one issue at the time so first the hang
that appears after the OS is booted up. Do you still have the
"pcie_port_pm=off" in the kernel command line? The hang that happens
afterwards sounds exactly like that the runtime PM would kick in but the
parameter should prevent that from happening. Since you are connecting
Thunderbolt 1 devices there is really no PM we can do for them at all
and the TBT driver should block it too.

Is it possible to narrow this down to a single device connected and then
get me the full dmesg output (with CONFIG_PCI_DEBUG=y) and output of
'sudo lspci -vv'?