Linus Torvalds released Linux kernel 7.3-rc1 on August 30, 2026, and buried inside a diff that pushes the kernel past roughly 40.98 million lines of code sits a change that Apple Silicon Linux users have wanted for years: initial USB4 and Thunderbolt support for the M1, M2, and M3 chips, according to LinuxCompatible’s coverage of the release. The work arrives as a 19-patch series posted for review on the Linux kernel mailing list on August 30-31, 2026, and it marks the first time USB4/Thunderbolt code for Apple Silicon has moved out of the downstream Asahi Linux tree and into the path toward mainline acceptance, Phoronix reported. The same submission was independently corroborated by outlets including Appleismo, SoftZone, and LinuxEncaja, and by September 2026 PauseHardware was describing the change simply as USB4/Thunderbolt support landing for three Apple Silicon generations, M1 through M3.
For a platform that has spent five years running Linux almost entirely on donated engineering time from a small reverse-engineering community, this is a meaningful checkpoint. It does not mean Apple laptops suddenly gain full external-GPU or 8K-display docking support under Linux. It means the plumbing that makes USB4 and Thunderbolt ports usable at all is finally being reviewed for inclusion in the kernel that ships with every major distribution, rather than living exclusively in a specialized fork most users will never install.
Don't miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
What the USB4/Thunderbolt Patch Series Actually Does
The 19-patch series was authored by Sven Peter, a developer closely associated with the Asahi Linux project, and it targets the Apple M1, M2, and M3 system-on-chips, per Phoronix. Until this submission, USB4 and Thunderbolt enablement for Apple Silicon existed only in Asahi Linux’s own downstream kernel tree, meaning anyone who wanted the feature had to run Asahi’s build rather than a stock distribution kernel. The new series is the formal attempt to bring that functionality upstream so it eventually ships as part of the standard Linux kernel that Fedora, Debian, Arch, and every other major distribution pulls from.
According to the LinuxCompatible writeup of 7.3-rc1, the series adds several discrete pieces of infrastructure rather than a single monolithic driver. The breakdown, as reported, looks like this:
Components added by the 19-patch USB4/Thunderbolt series (Linux 7.3-rc1):
- Thunderbolt switch driver
- Apple USB4/Thunderbolt NHI (host interface) driver
- Ring interrupt configuration
- DROM reading from device tree
- New Apple PMGR driver (power management)Each piece maps to a real hardware requirement. The NHI, or native host interface, is the controller-level driver that lets the kernel talk to the USB4/Thunderbolt silicon directly. The switch driver handles the routing logic for how devices chain together over a Thunderbolt connection. DROM, short for Device ROM, is where descriptor information about a connected device normally lives, and on Apple hardware that data has to be read from the device tree rather than queried from the controller the way it would be on a PC.
Who’s Behind the Push: Sven Peter and the Asahi Linux Connection
Sven Peter has been one of the more visible engineers inside Asahi Linux, the volunteer-driven project that has spent years reverse-engineering Apple Silicon Macs to make them bootable under a standard Linux kernel instead of macOS. Asahi’s contributors have already shipped working GPU drivers, audio support, and basic Thunderbolt functionality inside their own tree; what changes now is that this specific USB4/Thunderbolt work is being resubmitted for review against the mainline kernel maintained by Torvalds and the broader kernel community, rather than staying isolated in Asahi’s fork, Phoronix noted.
That distinction matters more than it might sound. Code that lives only in a downstream tree has to be manually rebased every time the upstream kernel changes, a maintenance burden that falls entirely on a small volunteer team. Code that gets merged into mainline is maintained collectively, tested against the kernel’s regression infrastructure, and ships automatically to any distribution that tracks a recent kernel release. Upstreaming is slower and involves more scrutiny, but it is the difference between a permanent side project and a permanent feature.
Which Apple Silicon Chips Are Covered
The patch series explicitly targets M1, M2, and M3 hardware, but the rollout inside the series itself is staggered. The final three patches in the set add device-tree nodes for the M1 and M2 generations specifically. M3 device-tree nodes are described as coming later, once more of the surrounding stack has landed upstream, even though Phoronix reported that the underlying Thunderbolt/USB4 code has already been tested successfully on M3 hardware. In practice, that means M3 support is functionally close but administratively behind M1 and M2 in this particular submission, and Asahi Linux’s own Linux 7.2 progress report, published in August 2026, already describes USB 3.0 and Thunderbolt as working across all M3 devices, suggesting the device-tree bookkeeping in the mainline series is the main thing still catching up to reality on the ground.
This lands alongside separate, related progress on M3 variants: Phoronix has also reported that initial mainline boot support for the M3 Pro, M3 Max, and M3 Ultra variants is being added for the Linux 7.3 cycle, following on from Linux 7.2, which first allowed base M3 chips to boot the mainline kernel, though not yet at a stage useful for daily driving. The USB4/Thunderbolt patches and the M3 Pro/Max/Ultra boot patches are separate efforts converging on the same kernel cycle, which is part of why 7.3 is shaping up as a notably active release for Apple Silicon users on Linux.
| Apple SoC | USB4/Thunderbolt patch coverage | Mainline boot status | Source |
|---|---|---|---|
| M1 (base, Pro, Max, Ultra) | Device-tree nodes included in the 19-patch series | Supported for several kernel cycles | Phoronix |
| M2 (base, Pro, Max, Ultra) | Device-tree nodes included in the 19-patch series | Supported for several kernel cycles | Phoronix |
| M3 (base) | Code tested successfully; device-tree nodes to follow once stack lands | Mainline boot enabled starting Linux 7.2 | Phoronix |
| M3 Pro / Max / Ultra | Not yet in this series | Initial boot support targeted for Linux 7.3 | Phoronix |
| M4 | Not covered; separate NVMe-only patch set exists for the M4 (t8132) | Ongoing, separate patch track | Phoronix |
What Works Today, and What Still Doesn’t
It’s worth being precise about what “USB4 and Thunderbolt support” means at this stage, because the phrase covers a wide range of capability. Per Phoronix, the current implementation supports XDomain connections, which is the protocol Thunderbolt uses for direct computer-to-computer links, and USB3 tunneling carried over a USB4 connection. That’s a meaningful baseline. It is not, however, the full Thunderbolt feature set that macOS users take for granted.
Two capabilities are explicitly missing from this patch set, according to the reporting: PCIe tunneling and DisplayPort tunneling. Both require further reverse-engineering work that has not been completed yet. PCIe tunneling is the mechanism that lets external GPU enclosures and other PCIe expansion devices work over Thunderbolt, so its absence means the classic eGPU-over-Thunderbolt use case is not yet functional on mainline Linux for Apple Silicon. DisplayPort tunneling is what allows a single Thunderbolt cable to carry both data and an external display signal, so multi-monitor setups that rely on that mechanism won’t work yet either.
Suspend is the other notable gap. Phoronix reported that the system currently has to actively prevent suspending whenever an active USB4/Thunderbolt connection is present, because letting the machine suspend under those conditions causes it to resume into an SError, a hardware-level exception. Until that’s resolved, laptop users who rely on Thunderbolt docks or displays will need to disconnect before closing the lid or otherwise triggering sleep.
The Power Management Piece: Apple’s New PMGR Driver
Alongside the Thunderbolt-specific drivers, the series introduces a new Apple PMGR driver that handles miscellaneous power controls on the SoC. LinuxCompatible reported that this driver saves roughly 1 watt of power during s2idle, the low-power suspend state, on M1 Pro, M1 Max, and M1 Ultra systems. A single watt sounds modest, but on a laptop battery, idle power draw compounds over hours, and s2idle is the suspend mode most Linux laptops actually use day to day rather than deeper suspend states that aren’t well supported on ARM hardware generally.
This power-management addition is a useful signal about where the Asahi and upstream kernel effort is focused right now: not just making features work, but making the machine usable as a daily driver where battery life doesn’t fall apart the moment you leave macOS behind.
A Brief History: From DART to USB4STREAM to Mainline Thunderbolt
This week’s patch series doesn’t come out of nowhere. It’s the latest step in a multi-year effort to get Apple Silicon’s USB4 and Thunderbolt hardware working under mainline Linux, piece by piece.
Back when Linux 6.8 was in development, Phoronix reported on early Apple M1 USB4/Thunderbolt DART support being submitted for that cycle. DART is Apple’s implementation of an IOMMU, the Device Address Resolution Table, which peripherals need for direct memory access. The M1 SoC uses a second, distinct variant of DART specifically for its USB4 PCIe ports, one that supports 64 concurrent streams rather than the 16 supported elsewhere on the chip, which required its own MMIO layout changes to support in the kernel. That was foundational, low-level plumbing, not user-facing functionality.
Then came Linux 7.2, which introduced USB4STREAM, an Intel-developed capability for transferring data directly between two computers over a USB4 or Thunderbolt link, treating the connection less like a traditional host-to-device interface and more like a high-speed peer link; LinuxKernelNewbies later documented that USB4STREAM actually landed in the kernel around July 2026, arriving across multiple separate commits rather than a single patch. Linux 7.2 also enabled base M3 chips to boot the mainline kernel for the first time, though Phoronix was careful to note that support wasn’t yet at a practically useful stage for end users at that point.
Now, with 7.3-rc1, the pieces converge: the IOMMU groundwork from 6.8, the general USB4 infrastructure matured through 7.2, and the M3 boot enablement all feed into a kernel cycle that finally ships initial Thunderbolt/USB4 device support for M1, M2, and (partially) M3 hardware.
| Kernel cycle | Key Apple Silicon milestone | Source |
|---|---|---|
| Linux 6.8 development cycle | Apple M1 USB4/Thunderbolt DART (IOMMU) support submitted | Phoronix |
| Linux 7.2 | USB4STREAM introduced; base M3 gains mainline boot capability (not yet practical) | Phoronix, Linux Journal |
| Linux 7.3-rc1 (released August 30, 2026) | 19-patch series adds initial USB4/Thunderbolt device support for M1, M2, M3; M3 Pro/Max/Ultra boot support added; kernel reaches ~40.98 million lines | Phoronix, LinuxCompatible |
That timeline also echoes a much older episode. Back in 2019, Intel’s own open-source engineers submitted the original USB4 support patches for the Linux kernel, a set of roughly 22 patches adding under 4,000 lines of new code, as TechPowerUp reported at the time. That work laid the generic USB4 subsystem foundation that Apple Silicon support is now being built on top of, seven years later, adapted for Apple’s non-standard controller implementation.
Why Upstreaming Matters More Than It Sounds
It would be easy to read this as a niche update relevant only to the small population of developers running Linux on a MacBook. The more useful way to read it is as a test of whether a community-driven reverse-engineering project can graduate its work into permanent kernel infrastructure. Asahi Linux has already proven it can make Apple Silicon boot Linux and run a usable desktop. What it hasn’t fully proven yet is that its lower-level hardware-enablement work can survive the mainline kernel’s review process, which is deliberately conservative, code-quality focused, and slow.
Getting USB4/Thunderbolt code reviewed on the kernel mailing list is a step toward that validation. If the series lands cleanly, it sets a precedent for other Asahi-originated drivers, including anything still pending around GPU features, camera support, or additional power management, to make the same journey from downstream fork to permanent upstream citizen.
How This Compares to macOS and Windows Thunderbolt Support
Apple’s own macOS has supported Thunderbolt on its Silicon Macs since the M1 launched in late 2020, with full PCIe tunneling, DisplayPort tunneling, and eGPU support (where Apple allows it) built in from day one, because Apple designed both the silicon and the OS together. Windows PCs with USB4 controllers from Intel, AMD partners, or third-party silicon vendors also generally ship with vendor-validated driver stacks that support the complete tunneling feature set out of the box, since USB4 support has been a mainstream PC platform requirement for several kernel and driver generations at this point.
Mainline Linux on Apple Silicon is, by contrast, still assembling that same feature set piece by piece, years after the hardware shipped, entirely through reverse engineering rather than vendor documentation or cooperation. That gap is the whole story of Asahi Linux: a volunteer team rebuilding driver support for hardware whose vendor has not published the register-level documentation that would make the job straightforward.
| Capability | macOS on Apple Silicon | Windows on USB4 PCs | Mainline Linux on Apple Silicon (post 7.3-rc1) |
|---|---|---|---|
| XDomain / peer-to-peer links | Supported | Supported | Supported |
| USB3 tunneling over USB4 | Supported | Supported | Supported |
| PCIe tunneling (eGPU, expansion) | Supported | Supported | Not yet; pending further reverse engineering |
| DisplayPort tunneling | Supported | Supported | Not yet; pending further reverse engineering |
| Suspend with active connection | Supported | Supported | Not yet; suspend actively blocked during active connections |
Market and Developer Impact
The direct market impact here is narrow but real. Apple Silicon Macs running Linux remain a small slice of the installed base, used mostly by developers, security researchers, and hobbyists who want a Unix-like environment on hardware praised for its battery life and build quality. For that audience, working Thunderbolt docks, multi-display setups, and eventually eGPU support would remove one of the last major reasons to keep a second machine around, or to dual-boot instead of running Linux natively.
There’s a broader signal too. Apple has never officially supported Linux on its Silicon Macs and has not published register-level hardware documentation for the community to work from. Every driver in Asahi’s tree, and now in this upstream submission, exists because volunteer engineers reverse-engineered it from scratch. Continued progress on hard problems like Thunderbolt controller behavior, without vendor cooperation, is itself evidence of how far community reverse-engineering projects can go when enough sustained engineering time is put behind them, something enterprise Linux vendors and distribution maintainers watch closely when deciding how much confidence to place in Apple Silicon as a supported Linux platform going forward.
For distribution maintainers at Fedora, Debian, and Arch, each of which already ships experimental Asahi-adjacent kernel builds or repositories, mainline acceptance of this code means less patch-carrying overhead and a smoother path to declaring Apple Silicon Thunderbolt support “stable” rather than “experimental” in release notes down the line.
What’s Still Missing, and What Comes Next
Three gaps stand between this patch series and Thunderbolt parity with macOS: PCIe tunneling, DisplayPort tunneling, and suspend compatibility. All three, per Phoronix’s reporting, require additional reverse-engineering work that hasn’t happened yet. None of the reporting available gives a firm timeline for when that work will land, which is typical for a volunteer-driven reverse-engineering effort where progress depends on available contributor time rather than a corporate roadmap.
M3 device-tree nodes are also still pending in this specific series, even though the underlying code has reportedly already been tested successfully on M3 hardware. That’s a bookkeeping gap as much as a technical one, and it would be reasonable to expect M3 nodes to follow in a subsequent revision of the series or a follow-up patch set once the core infrastructure clears review.
Predictions: Where Apple Silicon Linux Support Goes From Here
Based on the pattern established across the 6.8, 7.2, and 7.3 kernel cycles, a few reasonable predictions follow for the next several release cycles:
- Expect the 19-patch series to go through multiple review rounds before final merge; kernel maintainers routinely request revisions on driver submissions of this size, so a clean first-pass merge into a stable release is unlikely.
- M3 device-tree nodes will likely land within one or two kernel cycles after the core series merges, given that the underlying code has already been validated on M3 hardware according to Phoronix.
- USB3-over-USB4 tunneling and XDomain support will become the practical baseline most Asahi and mainline Linux users rely on for the next year, while PCIe and DisplayPort tunneling remain aspirational until further reverse engineering closes those gaps.
- M3 Pro, Max, and Ultra boot support, targeted for 7.3, will likely follow the same staged pattern the base M3 went through in 7.2: bootable first, broadly usable later.
- Distribution-level “official support” announcements for Apple Silicon, similar to what SteamOS and other projects have done for other hardware platforms, are more likely once suspend handling and PCIe tunneling are resolved, since those are the gaps most likely to affect everyday laptop usability.
The Bigger Picture for Apple Silicon and Open Source
What makes this patch series notable isn’t any single feature, it’s the direction of travel. Every prior milestone in Apple Silicon Linux support, from basic boot capability to GPU acceleration to now Thunderbolt, has followed the same arc: ship first in Asahi’s downstream tree, prove it works, then submit for mainline review. USB4 and Thunderbolt are simply the latest subsystem to take that path, and the fact that it’s happening at all, on hardware Apple has never officially opened up to Linux, remains the more interesting story than the exact list of supported tunnel types in any given kernel release candidate.
Anyone running or considering Linux on an M1, M2, or M3 Mac should watch the 7.3 stable release, expected roughly two months after the -rc1 tag per the kernel’s usual release cadence, as the point where this specific code either merges cleanly or gets pushed to a later cycle after further review.
Frequently Asked Questions
Does this mean Thunderbolt docks and displays now work on Linux on a MacBook?
Partially. The patches enable USB3 tunneling over USB4 and XDomain peer-to-peer connections, per Phoronix, but DisplayPort tunneling isn’t supported yet, so a Thunderbolt dock’s display output won’t work through this code alone. USB peripherals connected through a Thunderbolt/USB4 dock should function once the series is merged and available in a distribution kernel.
Which Apple Silicon chips does the patch series cover?
M1 and M2 device-tree nodes are included directly in the 19-patch series. M3 support has reportedly been tested successfully but its device-tree nodes are expected in a follow-up once more of the stack is upstream, according to Phoronix.
Who wrote the patches?
Sven Peter, a developer associated with the Asahi Linux project, authored the series, per Phoronix’s reporting — SoftZone independently confirmed Peter posted the first USB4/Thunderbolt driver for Apple M1 through M3 on August 30, 2026.
Is this the same as Asahi Linux?
No. Asahi Linux is a separate downstream project that has carried USB4/Thunderbolt support in its own kernel tree for some time. This patch series is the effort to bring that functionality into the mainline Linux kernel that all major distributions use, rather than requiring users to run Asahi’s specific build.
Will external GPUs work over Thunderbolt on Apple Silicon Linux now?
Not yet. PCIe tunneling, which is what external GPU enclosures rely on, is explicitly not supported in this patch series and requires further reverse-engineering work, according to Phoronix.
Can I suspend my laptop with a Thunderbolt device connected?
Not reliably. The current implementation actively blocks suspend while an active USB4/Thunderbolt connection is present, because suspending under those conditions can cause the system to resume into an SError, per Phoronix’s reporting.
When will this land in a stable Linux release?
The patches are currently out for review as part of the Linux 7.3-rc1 development cycle, released August 30, 2026. Kernel release candidates typically take around seven to eight weeks to reach a stable release, though code can still be revised or pulled during that review window.
Does this patch series cover the Apple M4 or M5?
No. The series targets M1, M2, and M3 only. Separate patches exist for M4 NVMe storage support, and as of Borncity’s coverage of the submission, USB4/Thunderbolt work for the M4 and M5 generations is described only as “in development,” with no patches yet posted as part of this effort.
Related Coverage
- Tim Cook Steps Down as Apple CEO, Leaves $4T Legacy [2026]
- How to Fix NAT Type: Gaming Router Setup in 10 Steps [2026]
- PC Won't Turn On: 12 Steps, 45 Min Fix [2026]
- OpenAI Macs Squeeze Apple Supply, Speed Up M7 [2026]
- Apple Skips M6 Pro to Rush AI-First M7, Baltra Chip [2026]
