diff options
Diffstat (limited to 'todo.org')
| -rw-r--r-- | todo.org | 25 |
1 files changed, 24 insertions, 1 deletions
@@ -46,11 +46,34 @@ below): input-side-spec.org (DRAFT, four decisions open). * Archsetup Open Work -** TODO [#B] Ship the Framework 13 AMD internal-mic workaround in the installer :bug:velox:audio:solo: +** DONE [#B] Ship the Framework 13 AMD internal-mic workaround in the installer :bug:velox:audio:solo: +CLOSED: [2026-10-06 Tue] :PROPERTIES: :CREATED: [2026-10-06 Tue] :LAST_REVIEWED: 2026-10-06 :END: +Shipped 2026-10-06 as archsetup commit =5189b7c=. =configure_framework_amd_internal_mic= runs in +=desktop_environment= right after the pipewire/wireplumber package loop, +gated by =is_framework_13_ryzen_ai_300= (DMI: vendor Framework, product_name +carrying "Laptop 13" and "AMD Ryzen AI 300"; ratio's "AI Max 300", the Intel +13, the 7040 13, and the 16 all miss). It installs +=configs/wireplumber/50-disable-ucm.conf= with =install -Dm644=. The shipped +file matches the PDM card by name pattern (=~alsa_card.pci-.*-platform-acp-pdm-mach=) +rather than the one PCI path home's hand-written version pinned, so a +renumbered bus on the same board is still caught; the live velox file was +replaced with the shipped one and wireplumber restarted, with the card set and +default source unchanged. The audio failure taxonomy gained the bug's row and +a Cluster 5 note so the doctor's use-ucm remedy treats the drop-in as the fix, +not an offending config. No wireplumber restart from the installer +(no user session bus during install), so it applies at the next wireplumber +start. Fixture test: =tests/installer-steps/test_configure_framework_amd_internal_mic.py=, +17 cases across the gate and the step; full =make test-unit= green, 57 +suites. Verified live: the gate fires on velox's real DMI, the step's output +matches the live file byte for byte, the default source is the analog-stereo +one, and a 4 s capture with the source briefly unmuted settled at rms about +0.05 (room noise at +30 dB capture gain) after a 0.6 s unmute transient, not +the 0.99999 rail. The source was muted at the PipeWire level before and +after; that state was restored. From home's 2026-10-06 handoff, verified live the same morning. On velox (Framework 13, Ryzen AI 9 HX 370, linux-lts 6.18.55, BIOS 04.02) the internal microphone was dead in every app. Root cause is a known upstream kernel bug in |