Linux Kernel Audio Advances: TASCAM US-144MKII Driver, ACP 7.2, SoundWire, AWE32 Fix

The Linux kernel's audio stack is showing fresh signs of life across several fronts: a newly submitted, fully featured USB driver for the discontinued but still widely owned TASCAM US-144MKII is queued for the 6.18 cycle; the sound subsystem for Linux 6.17 is being prepared with explicit support for AMD’s next-generation Audio Co‑Processor IP and expanded SoundWire handling; and a long‑running stability problem affecting the classic Creative SoundBlaster AWE32 ISA card has finally been addressed with targeted DMA safeguards. Taken together, these developments underscore a kernel community that continues to invest in both bleeding‑edge SoC audio support and long‑tail legacy device compatibility—work that matters to audio professionals, retro‑computing hobbyists, and anyone who depends on reliable sound on Linux systems. (phoronix.com) (phoronix.com) (phoronix.com)

Background​

Linux’s audio landscape is split across several layers: low‑level kernel drivers (ALSA and related subsystems) that manage hardware, in‑kernel frameworks like SoundWire and HDA that standardize bus and codec interactions, and userland stacks (PipeWire, JACK, PulseAudio in older setups) that expose sound devices to applications. The kernel audio maintainers—most notably Takashi Iwai for the sound subsystem—coordinate feature pulls and fixes each merge window to ensure everything from mobile SoCs to ancient ISA cards behaves sanely. Over the past year the kernel audio tree has been particularly active: SoundWire and ACP support have matured across multiple kernel cycles while a steady stream of standalone device drivers and bug fixes has continued to land. (phoronix.com)
This activity matters for multiple audiences. Desktop and laptop users expect no‑fuss audio when they plug in headsets or use built‑in microphones; audio professionals want low‑latency, stable interfaces for pro audio work; embedded and mobile platforms require tight integration for DMIC/I2S and power management; and vintage hardware fans appreciate that Linux still receives maintenance for legacy devices. These varied demands shape how new drivers are written, reviewed, and merged.

---

New TASCAM US‑144MKII Driver (Linux 6.18): What’s in the Patch Series​

A real user‑driven driver submission​

A seven‑patch series authored by Šerif Rami introduces a dedicated ALSA USB driver, named us144mkii, specifically for the TASCAM US‑144MKII USB audio/MIDI interface. The submission adds eleven new source files to the sound/usb/usx2y directory and integrates Kconfig/Makefile entries so the driver becomes a first‑class member of the ALSA USB audio tree. The patches implement PCM core infrastructure, playback, capture, MIDI, control elements, and power/deep‑sleep handling. The series was picked up into the sound subsystem’s “for‑next” branch and is expected to be proposed for inclusion in the Linux 6.18 cycle. (phoronix.com) (lwn.net)

Driver capabilities and the practical impact​

Key functional points implemented in the patch series include:

• Full PCM playback and capture paths engineered for the device’s USB2.0 operational mode.
• MIDI support and mixer controls for the unit’s proprietary MIDI interface.
• Deep‑sleep and power management considerations to coexist with system suspend/resume.
• Integration into ALSA’s USB audio driver infrastructure rather than relying on the generic snd_usb_122l fallback, providing a cleaner, more complete implementation.

For owners of the US‑144MKII (an inexpensive 4‑channel interface that was never officially supported on all platforms), this driver should deliver proper, plug‑and‑play compatibility with modern Linux distributions once the kernel merge is finalized and distributions pick up the updated kernel packages or backports. The work offers a path away from the previous hacks—like forcing USB1.1 emulation or creative libasound settings—that some users relied on to coax audio from the device. (autostatic.com) (spinics.net)

Why a dedicated driver matters​

Generic USB audio drivers can make a device functional, but dedicated drivers allow device‑specific optimizations: correct sample‑rate handling, robust suspend/resume behavior, lower latency in professional audio stacks (JACK), and robust error handling for device‑specific quirks. Consolidating device support into a single dedicated driver also helps reduce layering complexity (fewer userland workarounds) and opens the door for future improvements such as hardware‑assisted timestamping or per‑channel routing that generic layers rarely expose.

---

Linux 6.17 Sound Subsystem: AMD ACP 7.2 and SoundWire Expansion​

Enabling the next AMD Audio Co‑Processor IP​

The Linux 6.17 sound pull includes explicit enablement for AMD Audio Co‑Processor (ACP) version 7.2. That enablement covers the machine driver bits as well as I2S and DMIC plumbing so platforms that implement ACP 7.2 will have a path to functional digital microphones and I2S peripherals under the standard ALSA/ASoC frameworks. The ACP 7.2 code largely follows the driver design used for ACP 7.0 and 7.1, suggesting AMD maintained API/ABI continuity across these revisions. (phoronix.com)

SoundWire support follows​

In addition to ACP 7.2, the SoundWire subsystem received updates in the 6.17 cycle to better handle AMD’s ACP variants—improvements include refined interrupt handling outside locks, more robust handling of stream parameters on port preparation failures, and better handling of slave alerts/resume sequences for ACP-based platforms. These changes are aimed at hardening the SoundWire core and AMD-specific glue, improving reliability on SoCs that rely on the MIPI SoundWire bus for small MEMS microphones and CODECs. (phoronix.com)

Practical implications and unknowns​

• For OEMs and SoC integrators, ACP 7.2 enablement reduces the time to market for new hardware with Linux support.
• For end users, the change is mostly invisible until distributions ship kernels with these bits enabled and platform device trees or machine drivers are available for specific boards.
• Importantly, the public patches do not disclose specific AMD product codenames associated with ACP 7.2—this IP versioning doesn’t always map one‑to‑one with a consumer product name—so there is no direct public confirmation of which Ryzen/EPYC/embedded parts will ship with ACP 7.2. That uncertainty should be flagged until AMD or OEMs confirm platform details. (phoronix.com)

---

The SoundBlaster AWE32 Fix (Linux 6.16): A Small Patch With Big Nostalgia​

What was broken, and what was fixed​

A longstanding issue affecting the classic Creative SoundBlaster AWE32 ISA card caused system freezes when the modern PipeWire/WirePlumber stack interacted with the kernel driver (snd‑sbawe). The core problem traced to DMA handling: changing DMA mode while DMA setups remained enabled could corrupt behavior and even lock the system. The fixes merged for the 6.16 cycle enforce disabling DMA setups when the DMA mode is changed via kcontrols and prevent DMA mode changes while a PCM stream is active. Those two tightly scoped changes eliminate the race that produced freezes under modern userland audio stacks. (phoronix.com, spinics.net)

Why these fixes matter beyond nostalgia​

While the AWE32 is a relic of 1990s PC hardware, the fix is meaningful for several reasons:

• It demonstrates the kernel maintainers’ willingness to address interoperability bugs that only appear with modern userland stacks.
• It improves the reliability of retro hardware preservation efforts—useful for hobbyists, museums, and projects that resurrect old machines.
• The fix shows a responsible approach to kernel maintenance: small, focused fixes that mitigate specific race conditions rather than large rewrites that risk regressions.

At the same time, retro ISA hardware will still face practical performance limitations on modern desktops; the fix is about correctness and stability rather than performance optimization. (tomshardware.com, phoronix.com)

---

Cross‑Reference and Verification Notes​

The reporting and patch analysis in this article are corroborated by multiple independent sources:

• The TASCAM us144mkii driver series is documented by a mail submission and patch summary visible in public archives and covered by technical journalism. The patch series’ file list and scope (roughly 3,170 insertions across multiple new files) are described in the public patch notes. (lwn.net, spinics.net)
• The AMD ACP 7.2 and SoundWire changes are reported in the official sound pull discussions and summarized by kernel observers; the pulled code and the pull‑request commentary indicate what was merged for 6.17. (phoronix.com)
• The AWE32 fixes were pulled in as targeted corrections; the original user bug and the later kernel reply threads clarify the root cause (DMA mode/race) and the intended mitigation. (spinics.net, phoronix.com)

Where public Git commit trees and mailing lists are available, the patch summaries and merge descriptions provide the most precise technical record. If the reader requires patch‑level verification (exact commit hashes and diffs), those are retrievable from the kernel’s public Git repositories and mailing‑list archives; the summaries cited here reflect the pull‑request commentary and accompanying patch counts.

---

Analysis: Strengths, Risks, and What to Watch​

Strengths​

• Real hardware compatibility improvements. The us144mkii driver elevates a previously fringe device to first‑class kernel support, reducing the need for userland hacks and enabling proper ALSA/JACK integration. (lwn.net)
• Broad platform readiness. ACP 7.2 and SoundWire enablement signals that the kernel sound tree is being kept in step with SoC audio IP evolution—good news for upcoming AMD platforms that rely on integrated audio co‑processors. (phoronix.com)
• Targeted stability fixes. The AWE32 DMA fixes are conservative, narrowly scoped, and address a real regress‑type bug rather than speculative refactors—exactly the kind of change that benefits both users and kernel maintainers. (phoronix.com)

Risks and caveats​

• Merge timing and distribution lag. Kernel merge cycles and downstream distribution uptake are separate processes. A driver present in the mainline kernel for 6.18 does not mean it will be available out of the box in stable enterprise distros or in users’ currently shipped kernels. Expect a lag while distros package, test, and possibly backport changes. (phoronix.com)
• Regression potential. Any new driver that touches USB handling, PCM timing, or power management can introduce unforeseen regressions—especially in uncommon host controller configurations or with dodgy USB stacks. The sound maintainers generally review patches carefully, but risk is nonzero until the code sees wider testing. (lwn.net)
• Uncertain hardware mapping (AMD ACP 7.2). The ACP 7.2 enablement is explicit in the kernel tree, but the public patches do not map the IP version to specific, named AMD consumer or server SKUs. That means end users should not assume immediate compatibility for a particular laptop or board until the OEM or AMD confirms the mapping. Flagged as an unverifiable claim until disclosed by vendors. (phoronix.com)
• Power and timing edge cases. Low‑level audio drivers often interact with suspend/resume, power states, and interrupt timing. New code that alters those paths must be evaluated across a matrix of hardware: multiple USB host controllers, different chipsets, and a variety of kernel configurations (preempt vs. non‑preempt, real‑time patches, etc.). Expect follow‑up fixes as community testing exposes device‑ or host‑specific issues. (lwn.net, spinics.net)

---

Practical Advice for Users and Admins​

If you own a TASCAM US‑144MKII​

• Track kernel 6.18 development or the sound subsystem’s for‑next branch; the driver is staged there and will appear in mainline if merged. (phoronix.com, lwn.net)
• For immediate use, consider testing an out‑of‑tree build of the submitted driver (if available) in a safe environment, or wait for your distribution’s kernel package that includes the driver. Out‑of‑tree testing helps surface issues specific to your USB host controller and workflow (JACK, PipeWire, ALSA). (spinics.net)
• Avoid relying on USB1.1 emulation or complex libasound workarounds once the dedicated driver lands—these were stopgaps, not long‑term solutions. (autostatic.com)

If you’re tracking AMD SoC audio support​

• Treat ACP 7.2 enablement as a sign of readiness in the kernel, not definitive evidence of shipping product compatibility. Wait for OEM/platform firmware, device tree entries, or AMD announcements mapping ACP 7.2 to a product. Until then, assume further integration work and testing will follow. (phoronix.com)

If you care about retro audio hardware​

• The AWE32 DMA fixes are a clear win for system stability; retro users should test the updated kernels but be mindful of overall system performance when running modern userland on ISA‑era machines. If you rely on stable operation, prefer kernels that include the fix or request your distro to backport the specific commits. (phoronix.com, spinics.net)

---

Developer and Maintainer Perspective: What to Expect Next​

• Post‑merge follow‑ups. New device drivers and SoC enablement typically generate a modest stream of follow‑up commits: power management tweaks, latency improvements, and platform‑specific fixes. Developers should watch the ALSA and SoundWire trees for these incremental updates in the weeks following a merge. (lwn.net, phoronix.com)
• Possible backports to stable kernels. For the AWE32 fix, maintainers flagged the potential for backporting to stable kernel series because it addresses a correctness issue with tangible user impact. Expect similar treatment for any reliability fixes that affect stable deployments. (phoronix.com)
• Testing matrices matter. USB audio drivers must be exercised on different host controllers (EHCI, xHCI variants, USB hubs), with varying sample rates and under suspend/resume cycles to ensure robustness. Community testing and bug reports will be key to maturing any new driver quickly. (lwn.net, spinics.net)

---

Conclusion​

The recent activity in the Linux sound subsystem reflects a healthy balance between innovation for modern SoC audio (AMD ACP 7.2 and SoundWire improvements), practical device support (the us144mkii driver for TASCAM’s US‑144MKII), and stewardship of legacy hardware (AWE32 DMA fixes). For users, these changes promise broader hardware compatibility and more predictable behavior—but with the usual caveat that kernel merges are only the first step. Distributions, OEMs, and the broader testing community must still adopt, package, test, and sometimes backport these fixes before they become widely available in production environments.
The twin themes are continuity and care: contributors are keeping pace with new audio IP and bus standards while still patching long‑tail problems that surface only under modern stacks. That work benefits everyone from the audio professional running low‑latency sessions to the hobbyist spinning up a Pentium‑era rig for old‑school MIDI experiments. As always, cautious testing on real hardware, monitoring for follow‑ups in the ALSA and SoundWire trees, and patience for distribution packaging cycles are the practical next steps for anyone affected by these updates. (lwn.net, phoronix.com)

---

Source: Phoronix Linux 6.18 To Introduce New Driver For TASCAM US-144MKII USB Audio Interface - Phoronix
Source: Phoronix Linux 6.17 Sound Code Prepares For Upcoming AMD Hardware - Phoronix
Source: Phoronix Linux Delivering Driver Fix For 30 Year Old Creative SoundBlaster AWE32 ISA Sound Card - Phoronix