Re: TB 5 Dock DP-Out non-functional

[Imre Deak <imre.deak@intel.com>]

Hi Frederik,

On Mon, Nov 03, 2025 at 12:26:48PM +0100, Mika Westerberg wrote:
> Hi,
> 
> On Mon, Nov 03, 2025 at 12:01:06PM +0100, Frederik Schwan wrote:
> > Whole log attached.
> 
> Great, thanks!
> 
> From that log the last BW allocation done by i915 is 12.5G:
> 
> [   51.572497] thunderbolt 0000:8a:00.0: 0:13 <-> 1:12 (DP): bandwidth allocation changed to 0/12500 Mb/s
> 
> > > Regarding the script, that's odd too but it could be that you have more
> > > than one host router there (it is not clear from the truncated log). You
> > > need to build the tbtools binaries though.
> > 
> > I do package tbtools for Arch so these are available.
> > 
> > tblist gives me:
> > Domain 0 Route 0: 8087:7eb2 Intel Gen14
> > Domain 1 Route 0: 17ef:a583 Lenovo ThinkPad P1 Gen8 / T1g Gen8
> > Domain 1 Route 1: 0108:234d Lenovo ThinkPad Thunderbolt 5 Smart Dock 7500 - 40BA
> 
> Okay this has discrete Barlow Ridge TB5 host router.
> 
> > tbadapters --domain 1 --route 1:
> >  1: Lane 0 (upstream)             CL0
> >  2: Lane 1                        CL0
> >  3: Lane 0                        CLd
> >  4: Lane 1                        CLd
> >  5: Lane 0                        CLd
> >  6: Lane 1                        CLd
> >  7: Lane 0                        CLd
> >  8: Lane 1                        CLd
> >  9: PCIe Up                       L1
> > 10: PCIe Down                     Disabled
> > 11: DisplayPort OUT               Disabled
> > 12: DisplayPort OUT               Enabled
> 
> And this confirms that the DP tunnel is up.
> 
> That tb-bandwidth.sh script should work too (or tbtunnels -d1 -vv).
> 
> Anyways, from TB perspective this looks okay. Not sure about graphics. Imre
> do you see anything?

Could you please try booting with 'log_buf_len=20M drm.debug=0x10e
i915.enable_panel_replay=0', provide a dmesg log after booting and
connecting the same dock+monitor?

If the above works around the issue, could you please try if applying
the 7 patches of the [1] patchset fixes the issue (booting with those
patches applied and without the i915.panel_panel_replay=0 kernel
parameter)?

These same patches rebased on v6.17.7 are also available at [2] or in
the latest v6.18.0-rc3 based drm-tip tree at [3].

Thanks,
Imre

[1] https://lore.kernel.org/all/20251015161934.262108-1-imre.deak@intel.com
[2] https://github.com/ideak/linux/commits/fix-mst-dsc-panel-replay
[3] https://gitlab.freedesktop.org/drm/tip/-/commits/drm-tip