The editorial argues that two of the three headline items carry engineering risk that 'boring changelog' framing obscures. Speaker support is a thermal-protection problem where failure means physically destroying $200-$500 voice coils, and Thunderbolt bring-up is the first non-Intel Thunderbolt host controller running Linux ever.
The team's own progress report frames 7.1 as an incremental drop covering USB-C DisplayPort alt-mode on M1/M2, MacBook Pro speaker drivers, early Thunderbolt host controller bring-up, graphics improvements, and mainline kernel merges targeting Linux 6.10. They present the work as steady, methodical progress across the stack.
By submitting the progress report and driving 537 points of discussion, the submitter signals the community sees 7.1 as a notable, positive milestone worth surfacing to the broader HN audience. The strong upvote count reflects agreement that the release is meaningful progress.
The editorial notes that 7.1 is the first substantive release since Hector Martin stepped back from project leadership last year. The 537-point HN thread implicitly confirms that Asahi Lina, Sven Peter, Alyssa Rosenzweig, Janne Grunau and the roughly dozen remaining maintainers can execute on frontier hardware bring-up without Marcan sitting at the center of every decision.
Asahi Linux published progress report 7.1 last week, five months after 7.0. The headline items: USB-C DisplayPort alternate-mode support across M1 and M2 machines, speaker drivers for the 14" and 16" MacBook Pro (Pro and Max SKUs), and early Thunderbolt host controller bring-up. The report also documents graphics stack improvements from Alyssa Rosenzweig's team, mainline kernel merges targeting Linux 6.10, and a handful of firmware loader fixes.
The team frames 7.1 as an incremental drop. The team is being modest. Two of those three headline items carry a level of engineering risk that "boring changelog" tends to obscure. Speaker support on the MacBook Pro is not an audio problem; it's a thermal-protection problem where the failure mode is physical destruction of the hardware. Thunderbolt bring-up is the first non-Intel Thunderbolt host controller running Linux, ever.
Hector Martin (marcan) stepped back from project leadership last year, and 7.1 is the first substantive release since that transition. The 537-point Hacker News thread mostly confirms what the release notes state without saying it out loud: the remaining maintainer team — Asahi Lina, Sven Peter, Alyssa Rosenzweig, Janne Grunau, and roughly a dozen others — can ship deep hardware work without Marcan sitting at the center of every decision.
Modern Apple laptops use active thermal protection in the speaker driver path. The MacBook Pro's woofers are driven by a Cirrus Logic CS35L41-class amplifier that runs a real-time voltage/temperature model of the voice coil. When the model estimates the coil is nearing thermal limits, the DSP attenuates the amplifier output. That model is what stops you destroying $200-$500 of speaker hardware every time a Zoom call peaks or someone drops the volume slider onto a bass-heavy track.
Ship a naive Linux driver that ignores the DSP protocol and drives the amplifier directly, and the voice coils cook. Not "eventually degrade." Cook, in minutes, permanently. macOS ships Apple's reference implementation of the protection loop, tuned against per-unit factory-calibrated coefficients stored in NVRAM. Linux had to reverse-engineer the entire calibration chain: the coefficients' storage layout, the DSP's runtime state machine, the safe boundaries around the voltage limiter, and the loading protocol for the protection firmware itself.
Asahi's speaker driver is a rare piece of Linux code whose primary safety property is "does not destroy the peripheral it controls." Multiple release cycles were spent solely on the calibration and protection loop. Development runs on the team's test hardware bricked speakers. The 7.1 release is the first time the driver is signed off as safe for production use on M1/M2 Pro and Max machines — a signoff that carries the implicit warranty that early adopters will not be reporting fried tweeters on the mailing list.
Thunderbolt is the quieter engineering feat. Apple's M-series Thunderbolt implementation uses an on-die controller unrelated to the Intel Titan Ridge / Barlow Ridge lineage that has shipped in every prior Thunderbolt host. The USB4/TB3 protocol layer is public; the vendor-specific glue that lights up the PCIe lanes, negotiates power delivery, and coordinates DisplayPort tunneling through the same physical port is not. Every commercial Thunderbolt host controller Linux has ever spoken to was Intel silicon — Asahi's bring-up is the first non-Intel Thunderbolt host to enumerate under Linux. The bring-up is "early" in 7.1's language, meaning the initial handshake works but PCIe tunneling and DisplayPort-over-TB are still on the roadmap.
Hacker News comment sentiment split between "this is faster than any Linux-on-ARM port in history" and "Apple ships new silicon annually and Asahi is always a year behind." Both are true. Nouveau took a decade to reach parity with Nvidia's binary blob; Broadcom's B43 wifi work took roughly as long. Asahi hit basic desktop usability on M1 in 18 months and Pro/Max feature parity in five years. The reverse-engineering-per-generation cost curve is genuinely bending, mostly because the team learned to invest heavily in tooling before writing drivers.
If you have been waiting for Asahi to cross a "daily driver" line, 7.1 is that line for M1/M2 Pro and Max hardware. USB-C DP alt-mode means external displays work at native resolution without dongles or lidded workarounds. Speaker support on the Pro chassis means you can move off macOS on the machine you actually own, not just the base M1 Air that most "I switched to Linux" posts quietly assume. Fedora Asahi Remix remains the recommended distribution — the team ships kernel and firmware bits there first, with Arch and NixOS ports lagging by weeks.
Thunderbolt is not yet production-ready. "Early" in the changelog means enumeration works but TB-attached PCIe devices, external GPU pass-through, and daisy-chained displays are development targets. If your workflow depends on a TB dock or an eGPU, wait for 7.2 or 7.3. Realistically, plan on late 2026 for the full stack.
The broader lesson for anyone building drivers for adversarial hardware — modern GPUs, custom NPUs, integrated modems — is that the "destroy the peripheral" failure class is now a mainstream driver category. It used to be a QA problem confined to overclocking utilities and vendor-specific firmware update tools. Now it's a routine reverse-engineering problem, because manufacturers keep pushing physical-safety logic into firmware and DSPs to squeeze thermal headroom out of small chassis. Any team reverse-engineering a modern SoC — Qualcomm's X Elite, MediaTek Dimensity, whatever ships in the next Snapdragon-powered Copilot+ PC — inherits this risk profile. If your driver test plan does not include "can this brick the device," you are not testing seriously.
Apple's M4 shipped in the MacBook Air in Q1 2026 and the Pro line in Q2. Asahi's stated policy is to finish M1/M2 before starting M3, and M3 before M4 — so a two-generation lag is now the working assumption, not a temporary state. The interesting question over the next twelve months is whether upstreaming (a growing fraction of Asahi's stack now lives in mainline Linux 6.9 and 6.10) reduces the marginal cost of each subsequent generation enough to close that gap, or whether Apple's annual silicon cadence keeps the treadmill running indefinitely. 7.1 is the empirical argument that a small, well-organized team can do this work without the original founder in the driver's seat. Whether they can do it faster than Apple ships new hardware is the next test.
Top 10 dev stories every morning at 8am UTC. AI-curated. Retro terminal HTML email.