Intel MacBook Pro (2016-2020) on Omarchy

Verdict

Bronze overall, and the year matters more than anything else.

A 2018 to 2020 MacBook Pro with a T2 chip is the machine Omarchy actually supports. The installer detects the T2 by PCI ID, swaps in the patched linux-t2 kernel, installs Apple’s Broadcom firmware and the t2fanrd fan daemon, and sets a kernel command line tuned for suspend. Keyboard, trackpad, audio, Wi-Fi, fans and the Touch Bar are all covered by that path. Call it silver on its own.

A 2016 or 2017 MacBook Pro with a Touch Bar is a different machine. Those carry the T1 chip, not the T2, and nothing in the partitioner detects it: the T1 is a USB device, so the lspci check that picks linux-t2 cannot see it, and the DMI match in the Apple quirk scripts only runs after the disk has been written. The manual is blunt about the result: on T1 models the Touch Bar is non-functional and sound is not functioning. Worse, a full-disk install destroys Apple boot data those machines need. That corner of the lineup is closer to broken than bronze.

Checked against v4.0.4 (2026-09-15) and the v4.0.4 source tree. T2 support first landed in v3.0.0.

What works

On T2 models (MacBookPro15,x and MacBookPro16,x), the internal keyboard and trackpad come up through the t2bce_vhci virtual USB bridge, speaker audio works via apple-t2-audio-config, the Touch Bar comes up on hid_appletb_kbd, which is in-tree, and fan control works through t2fanrd. Wi-Fi and Bluetooth use Apple’s Broadcom firmware from apple-bcm-firmware. Keyboard backlight control was fixed in v3.5.0.

Across the Intel-graphics-only models the internal panel and the iGPU work, and the SPI keyboard on 2016-2017 models reaches the LUKS prompt once fix-spi-keyboard.sh puts applespi in the initramfs. The models with an AMD dGPU are the exception, below. The manual reports a 36% performance gain over macOS on a 2019 MacBook Pro, which matches the general reason people do this.

What breaks

The T1 firmware wipe is the serious one. Issue #8323 and the source audit in #8271 show that the full-disk installer recreates the EFI system partition and nothing anywhere reads, copies or preserves an existing EFI/APPLE directory. On a T1 Mac that payload is what initializes the iBridge, so after the install the T1 shows up as 05ac:1281 in recovery mode instead of 05ac:8600, and the Touch Bar, camera and Touch ID stay dead. The audit’s own correction is worth repeating: T2 machines are not affected, because their Touch Bar is a plain USB HID device. That narrows it to four model IDs: MacBookPro13,2, 13,3, 14,2 and 14,3.

It is not a one-way door. Reporters on the same thread recovered a wiped MacBookPro14,2 from Linux with the community t1-revive tool, which re-requests the machine’s own firmware from Apple’s signing server and writes it back to the new ESP, and a side-by-side install that leaves Apple’s EFI partition alone avoids the problem entirely. Both are third-party work with a short tested list, so treat them as a fallback rather than a plan.

The other half of #8323 is smaller but real: fix-spi-keyboard.sh installs macbook12-spi-driver-dkms without pulling linux-headers, so the DKMS build never runs and apple_ibridge, apple_ib_tb and apple_ib_als never exist. The built-in keyboard does not depend on it, because applespi is in-tree, but the Touch Bar drivers that package carries do.

Kernel regression on the 15-inch T1. Issue #12119 reports that linux-omarchy 7.2.5-3, which 4.0.4 pushed to everyone, hangs during amdgpu init on a MacBookPro13,3 and then cannot find the NVMe root. The stock Arch kernel boots fine on the same box. Adding intel_iommu=off fixes NVMe, the Apple SPI keyboard and AMD graphics in one go. A second reporter hit the same CONFIG_INTEL_IOMMU_DEFAULT_ON default on unrelated hardware. Still open.

Audio on 2016-2017. The Cirrus Logic CS8409 codec in those models is not configured correctly by the in-tree driver, so speakers stay silent and headphone switching is broken. Issue #11619 proposes the out-of-tree snd_hda_macbookpro driver, but the attached PR was closed without merging. Nothing ships for this.

Wi-Fi needs a regulatory domain, not just the Broadcom quirk. On BCM43602 machines the shipped feature_disable=0x82000 option is not enough. In #9019 a MacBookPro14,3 sat in the default world domain for about 47 minutes before it would associate, because the hardware gives cfg80211 no regulatory hint to start from. Setting options cfg80211 ieee80211_regdom=GB fixed it from cold boot. Issue #11745 is the same thing on a MacBookPro14,1, where Wi-Fi looked connected but dropped every packet until iw reg set. In the other direction, #9802 shows the same quirk disabling SAE on a MacBookPro16,1, so WPA3-only networks fail with ssid-not-found.

T2 odds and ends, all open: Bluetooth audio drops out on the 16-inch 2019 because hci_uart_bcm wins the bind race against hci_bcm4377 (#10119); USB hot-plug does nothing on the same model unless the device was present at boot (#11926); the trackpad hangs off the T2 virtual USB bridge and is therefore classified as external, so disable-while-typing never activates (#10815); and the MacBookPro16,1 gmux can leave the internal panel on the AMD dGPU and hang the machine at session start (#9609).

Suspend is model-specific. On MacBookPro14,1 the default deep sleep leaves the Thunderbolt controller dead and resume takes 75 to 107 seconds; if anything is plugged into a Thunderbolt port at lid close, only a cold boot brings the port back (#10593). Wake-from-suspend problems on T2 models are still reported in #3883, which is open from 2025.

What Omarchy does for this model

Four scripts in install/hardware/apple/ run at install time, plus a migration.

fix-t2.sh gates on lspci finding 106b:1801 or 106b:1802. It installs linux-t2, apple-t2-audio-config, apple-bcm-firmware and t2fanrd, force-loads t2bce_vhci and hci_bcm4377, adds the T2 bridge and HID modules to the initramfs, and appends intel_iommu=on iommu=pt pm_async=off mem_sleep_default=deep to the Limine command line.

fix-spi-keyboard.sh and fix-suspend-nvme.sh both match DMI product_name against MacBookPro13,[123] and MacBookPro14,[123] plus the small MacBooks, and the two lists differ by one entry: only the keyboard script also matches MacBook12,1. The first installs macbook12-spi-driver-dkms and puts applespi in the initramfs; the second clears d3cold_allowed on the NVMe device so it wakes.

fix-brcmfmac-supplicant.sh writes the feature_disable=0x82000 module option. Since v4.0.0 it applies to non-T2 Macs too, gated on the chip IDs brcmfmac actually binds rather than on the T2 bridge.

Migration 1785944594.sh rewrote the T2 suspend defaults for the t2bce kernel, added the missing [Fan2] section that was crashing t2fanrd on dual-fan models, and removes tiny-dfr if present. That came from #6559 and shipped in 4.0.0.

Variants

Prefer the 13-inch T2 models (MacBookPro15,2, 15,4, 16,2, 16,3). Intel graphics only, so no gmux or dGPU problems, and they get the full T2 stack.

Be careful with the 15-inch and 16-inch T2 models. They work, but the AMD discrete GPU adds the gmux hang and dual-GPU power problems on top.

Avoid the 2016-2017 Touch Bar models unless you accept losing the Touch Bar, camera and speakers. The 2016 15-inch (MacBookPro13,3) currently also needs a manual intel_iommu=off to boot 4.0.4.

The 2012-2015 models are a separate story: they use the BCM4360 or BCM4331 with the out-of-tree wl driver, handled by fix-bcm43xx.sh, and none of the T1 or T2 machinery applies.

Before you install

• Back up the whole disk, not just files. On a T1 Mac the Apple EFI/APPLE payload is not recoverable from a file backup.
• Disable Secure Boot and allow external boot from macOS Recovery, per the Mac support chapter.
• Know your model identifier before you commit. The T1 machines are the 2016-2017 Touch Bar models, MacBookPro13,2, 13,3, 14,2 and 14,3. The non-Touch-Bar 13,1 and 14,1 have no T1 at all, and 15,x and 16,x are T2.
• Have a USB keyboard ready for the LUKS prompt. The SPI keyboard fix for these machines landed in v3.0.0, after #1666, but on the 2016 15-inch the 4.0.4 kernel still floods applespi with -12 errors until you add intel_iommu=off.
• Plan to set a regulatory domain by hand if Wi-Fi associates but passes no traffic.
• On a 15-inch 2016, be ready to edit the Limine entry and add intel_iommu=off at first boot.

Related

• /hardware/t2-mac/ for the T2 stack itself
• /hardware/apple-macbook-air-intel/ for the Air, which shares the T2 path
• /hardware/apple-silicon-macs/ for M-series machines
• /fix/suspend-wont-resume-s2idle/ and /hardware/suspend-sleep/
• /fix/bluetooth-stops-after-resume/
• /fix/no-sound-from-laptop-speakers/
• /fix/kernel-panic-after-update-limine/
• /switch/should-you-dual-boot/