[Bug 2169490] Re: [Regression] snd_hda_intel fails to initialize on Intel 5-series (Ibex Peak) HDA: "CORB reset timeout#2, CORBRP = 65535" / "no codecs initialized" — works on 6.8.0-31, broken on 6.8.0-139

** Tags added: kernel-daily-bug -- You received this bug notification because you are subscribed to linux in Ubuntu. Matching subscriptions: Bgg, Bmail, Nb https://bugs.launchpad.net/bugs/2169490 Title: [Regression] snd_hda_intel fails to initialize on Intel 5-series (Ibex Peak) HDA: "CORB reset timeout#2, CORBRP = 65535" / "no codecs initialized" — works on 6.8.0-31, broken on 6.8.0-139 Status in linux package in Ubuntu: New Bug description: ## Description After a kernel update within the 6.8.x series, the Intel onboard HDA audio controller stopped working: the snd_hda_intel driver fails to communicate with the codec at boot, so no sound card (other than the GPU HDMI audio) is exposed to ALSA, and PipeWire falls back to a "Dummy Output". This is a kernel regression, bisected upstream to commit: 4d4c10f763d7 "PCI: Explicitly put devices into D0 when initializing" which is tracked as an upstream regression on lists.linux.dev for the same hardware (Intel 5-series PCH "Ibex Peak" HDA controller 8086:3b56 + Realtek ALC269VB). See regressions@lists.linux.dev thread "[REGRESSION][BISECTED] HDA controller MMIO...". On these old chipsets, after this commit the HDA controller MMIO registers read 0xFFFF, causing CORB timeouts and codec enumeration failure. ## Hardware - Machine: [Acer model, e.g. Acer Aspire — DMI: `sudo dmidecode -s system-product-name`] - HDA controller: 00:1b.0 [8086:3b56] rev 05 (Intel 5 Series/3400 Series, Ibex Peak) - Codec: Realtek ALC269VB (Subsystem Id: 0x10250359) - GPU: AMD Radeon HD 5000 series (HDMI audio on card 1 works fine) - OS: Ubuntu 24.04 LTS, GNOME, PipeWire 1.0.5 ## Working / broken kernel versions - WORKS: 6.8.0-31-generic (original 24.04 kernel) - BROKEN: 6.8.0-139-generic (+several previous versions) - Also verified working on the same machine with Linux Mint / kernel 6.1.0-49-amd64 - Neither a cold boot (full shutdown, battery removed, several minutes) nor module options (probe_only=1, probe_mask=1, single_cmd=1, power_save=0) work around the issue. ## Steps to reproduce 1. Boot Ubuntu 24.04 with kernel >= 6.8.0-139 2. Run: aplay -l -> Only "card 1: HDMI [HDA ATI HDMI]" is listed; the analog HDA card is missing. 3. Run: wpctl status -> Only sink available: "Dummy Output". ## Relevant dmesg output [ 7.236527] snd_hda_intel 0000:00:1b.0: enabling device (0000 -> 0002) [ 7.238138] snd_hda_intel 0000:00:1b.0: number of I/O streams is 30, forcing separate stream tags [ 7.342564] snd_hda_intel 0000:00:1b.0: CORB reset timeout#2, CORBRP = 65535 [ 7.342733] hdaudio hdaudioC0D0: no AFG or MFG node found [ 7.342751] hdaudio hdaudioC0D1: no AFG or MFG node found [ 7.342763] hdaudio hdaudioC0D2: no AFG or MFG node found [ 7.342778] hdaudio hdaudioC0D3: no AFG or MFG node found [ 7.342781] snd_hda_intel 0000:00:1b.0: no codecs initialized ## Expected behavior The ALC269VB codec is enumerated and the analog sound card appears as "card 0: MID [HDA Intel MID], device 0: ALC269VB Analog", as it does with kernel 6.8.0-31 and with kernel 6.1.0-49-amd64 on the same machine. ## Additional context The upstream regression was bisected to commit 4d4c10f763d7 ("PCI: Explicitly put devices into D0 when initializing"), first appearing in v6.16 upstream and still present as of recent mainline rc kernels. If this commit (or its backport) was pulled into the Ubuntu 6.8.x stable updates, that would explain the breakage starting with 6.8.0-[xx]. A revert or a quirk for the 5-series HDA controller (8086:3b56) would restore audio for this hardware family. ## Workaround Staying on linux-image-unsigned-6.8.0-31-generic restores audio. To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2169490/+subscriptions