APPLIED: [SRU][R][PATCH 0/1] thunderbolt: Defer DP tunnel teardown until display driver is ready (LP: 2155096)

[Mehmet Basaran]

Applied to resolute:linux master-next branch.

-------------- next part --------------
ChunAn Wu via kernel-team <kernel-team at lists.ubuntu.com> writes:

> BugLink: https://bugs.launchpad.net/bugs/2155096
>
> [Impact]
> On systems with Full Disk Encryption (FDE/LUKS) and a Thunderbolt dock
> providing the external display, the Thunderbolt driver loads early from
> initramfs and discovers the BIOS-established DisplayPort tunnel. It starts
> DPRX polling with a 12-second timeout (TB_DPRX_TIMEOUT), expecting the GPU
> driver to read DPCD from the connected monitor within that window.
>
> However, the GPU driver (i915, xe, amdgpu, etc.) resides on the encrypted
> root filesystem and cannot load until the user enters the LUKS passphrase.
> This creates a driver load ordering issue: the DPRX timeout fires before
> the GPU driver has had a chance to initialize, causing the Thunderbolt
> driver to permanently tear down the DP tunnel and remove the DP IN adapter
> from available resources. Once torn down, the display never recovers --
> recovery requires a physical re-plug of the dock with a full power cycle.
>
> Affected platform: Dell Pro Max 14 MC14250 with Dell SD25TB5 TB5 Dock.
> Other Thunderbolt dock + FDE combinations are likely affected as well.
>
> [Fix]
> Register a PCI bus notifier that watches for display class
> (PCI_BASE_CLASS_DISPLAY) driver bind events. When the DPRX timeout fires:
>
>  - If no display driver is bound: tear down the tunnel but keep the DP IN
>    adapter in the available resources list, allowing a retry later.
>  - If a display driver is already bound: proceed with the existing behavior
>    of permanently removing the DP IN resource.
>
> When a display driver eventually binds, the notifier triggers a DP tunnel
> retry via a scheduled work item, re-establishing the tunnel.
>
> This is an UBUNTU: SAUCE patch -- there is no upstream commit to cherry-pick.
> The fix has been submitted upstream for review but is carried here as SAUCE
> to address the immediate regression for Ubuntu users.
>
> [Test Plan]
> Tested on Dell Pro Max 14 MC14250 with Dell SD25TB5 Thunderbolt 5 Dock
> and LUKS full disk encryption enabled:
>
> 1. Boot with TB dock connected providing external display
> 2. Enter LUKS passphrase at ~T+60s (well beyond the 12s DPRX timeout)
> 3. Without patch: external display never comes up after login
> 4. With patch: dmesg shows "not active, deferring" at T+14s, then
>    "display driver bound, retrying DP tunnels" when i915 loads,
>    and the display comes up successfully
>
> Also verified no regression on non-FDE systems:
> - Dell Pro Max 16 MA16250 (6.17-oem, i915) -- no regression
> - Dell XPS 14 DA14260 (7.0-generic, xe) -- no regression
>
> [Where problems could occur]
> - The PCI bus notifier adds minimal overhead (a single class check per
>   driver bind event). Risk of performance regression is negligible.
> - If a display driver binds but the DP tunnel retry fails for unrelated
>   reasons, the behavior is no worse than without the patch (tunnel stays
>   torn down).
> - The fix covers all GPU vendors via the generic PCI base class check
>   (0x03xx), so vendor-specific regressions are unlikely.
>
> ChunAn Wu (1):
>   UBUNTU: SAUCE: thunderbolt: Defer DP tunnel teardown until display
>     driver is ready
>
>  drivers/thunderbolt/tb.c | 96 ++++++++++++++++++++++++++++++++++++----
>  1 file changed, 88 insertions(+), 8 deletions(-)
>
>
> base-commit: 148c038ca663f09720ebdaa77e5c63d0e8da4573
> --
> 2.34.1
>
>
> --
> kernel-team mailing list
> kernel-team at lists.ubuntu.com
> https://lists.ubuntu.com/mailman/listinfo/kernel-team
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 873 bytes
Desc: not available
URL: <https://lists.ubuntu.com/archives/kernel-team/attachments/20260619/c9c1dc9d/attachment-0001.sig>