Confirming the fix on the 24.04 HWE track (linux-hwe-7.0 7.0.0-38.38~24.04.4),
Meteor Lake-P, with before/after data from the same machine.
Hardware
- Lenovo ThinkPad P1 Gen 7 (21KV001CUS), BIOS N48ET25W (1.12), 2025-01-02
- iGPU: 00:02.0 Meteor Lake-P [Intel Arc Graphics] [8086:7d55] rev 08, driver
i915
- dGPU: 01:00.0 NVIDIA AD106M RTX 4070 Max-Q [10de:2820], 3D controller, no
display outputs; nvidia 595.99.02 open modules, prime on-demand. Every
output is wired to the iGPU (eDP-1, Type-C DP, and a TV on DP-1).
- Firmware: DMC i915/mtl_dmc.bin v2.21, GuC 70.36.0, HuC 8.5.4
- Ubuntu 24.04.5, GNOME 46 on Wayland, s2idle only (no S3)
BAD — 7.0.0-28-generic (7.0.0-28.28~24.04.1), July 2026
- Two boots, 2026-07-21 to 07-27: 111 and 58 `i915 *ERROR*` lines.
- Same signature as the reports above: `Failed to bring PHY A to idle`,
`Timeout waiting for DDI BUF A to get active`, `Timed out waiting for DP
idle patterns`, `[CRTC:150:pipe A] flip_done timed out`,
`[CONNECTOR:507:eDP-1] commit wait timed out`, plus `PHY A/B failed to
request refclk`.
- Triggers seen: eDP wedged after a Type-C monitor unplug, and after
screen-lock DPMS (2026-07-23 12:58). Suspend/resume cleared a wedge
without reboot.
- Workaround since 2026-07-27: pinned GRUB to 6.17.0-41-generic.
GOOD — 7.0.0-38-generic (7.0.0-38.38~24.04.4), since 2026-10-09
- Verified the backport is in the shipped module, not just the changelog:
`objdump -dr` on /lib/modules/7.0.0-38-generic/.../i915.ko shows
`intel_pps_on` called from `intel_ddi_pre_pll_enable` and no longer from
`mtl_ddi_pre_enable_dp` — exactly the code motion in 062499cc4813. On
6.17.0-41's i915.ko it is the other way round.
- s2idle dwells, all clean, resume about 1 s each:
2026-10-09 08:42 -> 08:46 4 min 24 s
2026-10-09 18:20 -> 10-10 09:05 14 h 45 min
2026-10-10 09:09 -> 10-12 08:34 47 h 25 min
- Also clean: screen lock -> DPMS off -> wake; Type-C monitor unplug x2;
TV hot-plug/unplug on DP-1 (an external port, i.e. the PHY B path the commit
message says it does not directly address — no errors here either).
- Across 3 days of uptime: zero `i915 *ERROR*`, zero `timed out`, zero
`mismatch`, zero `DDI BUF` lines. Only the normal GuC/HuC reload messages on
each resume.
So 24.04 users on the HWE track can drop the 6.17 pin at 7.0.0-38.38~24.04.4.
Note for anyone on the Ubuntu-built NVIDIA modules: the 7.0.0-38 module is
only built against 595.99.02, so moving to it removes the 595.84 modules for
the 6.17 kernels.
The fix reached both tracks via the "upstream stable patchset 2026-08-26"
(LP: #2165189) rather than a targeted SRU for this bug, so nothing will close
this report automatically. With #23 (Lenovo lab), #24 (resolute) and this
(noble HWE) all confirming 7.0.0-38.38 — and bug 2167549 (same signature,
same P1 Gen 7 / Meteor Lake-P hardware) already closed Fix Released on the
same patch — I'm setting the linux (Ubuntu) task to Fix Released. Please
reopen if you still see the signature on 7.0.0-38 or later.
** Changed in: linux (Ubuntu)
Status: Confirmed => Fix Released
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/2150605
Title:
`i915 Arrow Lake-S: PHY A / C10 DPLL state mismatch on resume from
long s2idle dwell — slow wake (5-10s) with retry storm`
To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2150605/+subscriptions
--
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
