Asahi Linux 7.1 brings M3 audio and custom video-decode firmware
The Asahi team got sound out of M3 MacBook speakers without using a line of Apple's audio firmware, and wrote their own video decoder from scratch rather than lift Apple's proprietary blob.
On M3 machines, the speakers and headphone jack now play through I2S controllers unchanged since the M1, and CPU frequency switching with big.LITTLE scheduling finally places work on the right cores.
Rather than extract Apple's video-decode blob, the team wrote custom AVD firmware, giving a V4L2 driver that decodes 10-bit AVC up to 4K.
One developer warns the custom-firmware bet could collapse the day Apple decides to sign firmware blobs in hardware.
Sound from M3 speakers, without Apple's audio firmware
M3-series MacBooks now produce high-quality audio on Asahi Linux through the built-in speakers and the headphone jack, reusing the same I2S controllers and NCO clocking that have not changed since the M1. Alongside audio, the M3 gained CPU frequency switching and proper big.LITTLE scheduling, so tasks land on efficiency or performance cores as intended, on top of working PCIe, WiFi, Bluetooth, NVMe, keyboard, trackpad and SMC sensors.
The headline engineering feat is video decode. Instead of extracting Apple's proprietary AVD firmware blob, the team wrote their own, producing a V4L2 driver that decodes 10-bit AVC up to 4K and opens a path to VA-API and Vulkan Video.
Two macOS 27 regressions needed patches. A firmware flag that marks a volume bootable was previously ignored, so the installer now sets it automatically; and the SMC changed a battery reading from a 32-bit integer to a single byte, triggering emergency shutdowns until kernel 7.0.12 caught up. The m1n1 bootloader reached 1.6.0, now requiring Rust for its stage 2 build and adding M3 support (including the SPMI controller and PCIe init) plus groundwork for M4 and A18 Pro parts.
Why won't Apple just fund this?
The recurring thread question is why a cash-rich hardware company leaves this to volunteers. Jeffrey Paul, who blogs at sneak.berlin put the business case bluntly:
"It is baffling to me that Apple … doesn't throw 2 or 3 of their thousands of FTEs on this."
The report's technical prose drew a correction, too. It described the audio bus by calling I²S "an I²C-based bus," and a developer who notes SPI, not I²C, is I²S's close cousin flagged the mixup:
"I²S has nothing to do with I²C."
The most sober voice, a software engineer who blogs at zacbentley.com, named the load-bearing assumption behind writing custom firmware at all: it runs only because Apple's coprocessor does not verify what it loads.
"it doesn't seem unlikely that they will introduce hardware signing for firmware blobs or the data they supply at runtime … that's a reasonable security concern for Apple to address."
Every M3 gain in this release rides on that door staying open.
Get breakdowns like this every day
The top Hacker News stories, unpacked — by email, Telegram, or RSS.
Box3D: Erin Catto extends Box2D into an open source 3D engine
Erin Catto, whose Box2D powers the physics in Angry Birds and countless indie games, has released Box3D, a 3D rigid-body engine that began as a fork of Valve's in-house Rubikon.