[BUG] LG Gram 17Z90U (PTL, RT713 L3 + RT1320 L1): speakers silent while RT1320 is in PS0, unmuted and streaming #6007

Summary

The internal speakers on an LG Gram 17 (17Z90U-G.AU89U1) are completely silent. The machine support from #5750 / #5767 works on this unit: the card is created, sof-ptl-rt713-l3-rt1320-l1-2ch.tplg loads, and the headphone jack (RT713) works. During playback to the Speaker PCM, the RT1320 reports PS0, FU21 is unmuted, the DAPM path is On and the PCM hw_ptr advances, but nothing is audible. There are no errors in dmesg.

This looks like the RT1320 VC problem being debugged in #5925 (same amp: dev_id 0x6981, version_id 2), on a different OEM and with a single amp on link 1. I'm filing it separately because the platform is different; happy to move it to #5925 if you prefer.

Hardware

• LG Gram 17, 17Z90U-G.AU89U1, SKU EVO, BIOS P1ZK3080 (01/22/2026)
• Intel Core Ultra 7 355 (Panther Lake)
• SoundWire: link 1 RT1320 (sdw:0:1:025d:1320:01), link 3 RT713 (sdw:0:3:025d:0713:01)
• RT1320 driver reports version_id=2, dev_id=0x6981 (VC)
• Linux only (no Windows installed), so I can't do the Windows warm-restart comparison from [BUG] No Audio - Built-in speakers silent on MSI Stealth 16 AI+ (Intel Panther Lake / Dual RT1320 SoundWire) #5925.

Software

• Omarchy (Arch Linux), kernel linux-omarchy 7.2.8-9 (based on 7.2.8)
• sof-firmware 2026.09.2 (SOF FW 2.15.0.1, topology ABI 3:29:1)
• linux-firmware 20260916 (realtek/rt1320/rt1320-patch-code-vc.bin.zst present)
• alsa-ucm-conf 1.2.16.1, PipeWire 1.6.9, WirePlumber 0.5.18

Test

With PipeWire and WirePlumber stopped:

speaker-test -D plughw:0,2 -c 2 -r 48000 -t sine -f 440 -l 0

It runs normally for several minutes (Front Left / Front Right), but the speakers stay silent.

While it was playing:

• /proc/asound/card0/pcm2p/sub0/status: RUNNING, hw_ptr advances by about 48000 frames/s
• DAPM (rt1320 component): DP1RX, FU21, OT23 L/R, SPOL/SPOR, PDE 23 all On; bias_level On
• RT1320 registers read with cache_bypass=1:
  • PDE23 requested PS 0x41001988 = 0x00, actual PS 0x41081980 = 0x00
  • FU21 mute ch1/ch2 0x41000189 / 0x4100018a = 0x00
  • Function status 0x41080000 = 0x41
  • RT1320_KR0_INT_READY 0x1000f021 = 0x1f
  • DSP FW status 0x3fc2bfc4 = 0x04
• Mixer: rt1320 OT23 Playback Switch on/on, rt1320-1 FU21 Playback Volume 87/87, Speaker Switch on, Pre/Post Mixer Speaker Playback Volume 45/45, rt1320-1 DSP FW Update off
• dmesg during playback (rt1320 dyndbg enabled): only rt1320_sdw_hw_params rt1320-aif1 and rt1320_bus_config clk_base=1, with no warnings or errors

MCU patch and DSP FW (re the question in #5925)

• rt1320-patch-code-vc.bin.zst is installed, CONFIG_FW_LOADER_COMPRESS_ZSTD=y, and the boot log has no rt1320_load_mcu_patch: Failed to load or Direct firmware load message, so the MCU patch appears to load.
• No platform DSP FW file is available. The driver would look for realtek/rt1320/rt1320_LG_17Z90U-G.AU89U1_EVO.dat. Is that file needed for speaker output on this platform?

History

• 7.2.5 + sof-firmware 2025.12.2: silent, and dmesg showed rt1320_pde_transition_delay PDE to PS0 is NOT ready
• 7.2.8 + sof-firmware 2026.09.2: that warning is gone, still silent

The attached boot log was captured before I enabled rt1320 dyndbg, so it doesn't show func_status at first init. I can do a cold boot (AC unplugged, 30 s wait) with dyndbg enabled from boot, test a mainline or sof-dev kernel, or try a patch if that helps.

Attachments

lg-gram-17z90u-rt1320-logs.zip:

• alsa-info.txt
• dmesg-boot.txt (first 120 s of boot)
• dmesg-during-playback.txt
• dapm-during-playback.txt
• pcm2p-status-during-playback.txt
• rt1320-regmap-during-playback.txt, rt1320-regmap-mbq-during-playback.txt (cache_bypass=1, captured during playback; the main regmap was read in two passes because a full dump takes several minutes)

cc @jack-cy-yu @bardliao

lg-gram-17z90u-rt1320-logs.zip