Repository navigation
Replies: 18 comments 3 replies
|
Hi! I have a 16-inch MacBook Pro with an M5 Max: board J716C (Mac17,6, T6050), 18-core CPU (6 Super + 12 Performance), 40-core GPU, 48 GB, 2 TB SSD. It runs macOS 27.0.1 (26A434), and the system firmware and OS loader are both 20457.1.29. The J716C row has no reports yet, so I'd like to help qualify it.
What would be most useful first? |
J716C (M5 Max 16") ADT vs. the APIF seriesI checked the hardware assumptions in the APIF series (#238, #241–#257) against this machine's ADT. It's read-only from macOS 27.0.1 (26A434), firmware 20457.1.29. No boot was involved, so none of this is a hardware qualification. Board facts: Matches on J716C
Gaps and questions
I have the full ADT dump with identifiers removed and can run any specific |
|
Hi, I have a 14-inch MacBook Pro with a base M5: board J704 (Mac17,2, T8142), running macOS 27.0.1. I'd like to help qualify it. The device table on aurorasilicon.org marks J704 as working (boot, storage, Wi-Fi, display, keyboard, trackpad, confirmed 2026-10-21), but this thread says no named-board result exists, and I only find J813 in aurora-silicon/linux. aurora-installer has no j704ap either. Is there J704 work that isn't public yet, or was the status carried over from J813? From a read-only look at the ADT: the SoC-level addresses (AIC, UART, DockChannel, SMC, MTP, DCP, USB-DRD0, SPMI) match the J813 device tree exactly. The board differs in the 3024×1964 ProMotion panel, HDMI via dcpext/dp2hdmi, the GL9755 SD reader on PCIe, Broadcom BCM4388 C2 Wi-Fi instead of N1, a fan, six speakers and a third Thunderbolt port. Unless someone already owns J704, my plan is to RAM-boot your J813 stack with a J704 device tree and report back here. Is macOS 26.6.1 still the stub you'd recommend, given that my system firmware is already 27.0.1? |
|
J716S (MacBook Pro 16-inch, M5 Pro, T6050, 18 cores): first named-board results, plus questions This is a non-Aurora bring-up: AsahiLinux/linux Disclosure: the code and reports were produced with Claude (AI) under my direction. I ran Summary, repro steps and logs: https://github.com/rolotrealanis98/m5pro-linux/blob/main/docs/status-2026-11-05.md Tested on hardwareI'll post each of these separately in the thread's template:
Also tested on hardware, no separate report yet:
Booted, not feature-testedSMC fan keys and the SPMI window, cpufreq on all three clusters. Hardware testing pauses until Design onlyUSB2 host, DCP, audio, GPU, N1, ISP, Thunderbolt. Aurora branches we plan to adopt
Questions
Redacted log excerpts are in |
|
J716S report 1/4: SMP / untethered boot Produced with Claude (AI) assistance under my direction; I ran the hardware test myself. |
|
J716S report 2/4: Trackpad Produced with Claude (AI) assistance under my direction; I ran the hardware test myself. |
|
J716S report 3/4: NVMe (ANS) Produced with Claude (AI) assistance under my direction; I ran the hardware test myself. |
|
J716S report 4/4: iBoot boot-failure counter Produced with Claude (AI) assistance under my direction; I ran the hardware test myself. |
|
J716s (MacBook Pro 16-inch, M5 Pro): Result (tested on hardware, 2026-11-06/07). Recipe. PR #945 (J716c DT) merged; PR #936 (NVMe/SART v4) and PR #939 (combo PHY/dwc3) cherry-picked; Findings (1-4, 6-8 hit on hardware)
Open. Battery not yet enabled on this kernel (our own probe hung on 7.2). Wi-Fi is next: we located the N1 FTAB in the macOS restore bundle. DCP is gated by firmware-compat. Repo: https://github.com/rolotrealanis98/m5pro-linux ( Disclosure: the code and reports were produced with Claude (AI) under my direction. I ran the first boots myself; the later ones ran unattended from a self-reverting payload loop I set up. |
|
J716c hardware report for Discussion #142 I reproduced an 18-core Linux RAM boot on a 16-inch M5 Max (J716c / Mac17,6, T6050) with macOS 27.2.1 ( Linux showed CPUs Could you share the exact Aura revision and firmware build used for the working GPU and panel test in #1276? I can independently test that stack on this J716c. |
|
K114S (MacBook Pro 14-inch OLED, M5 Pro, Mac17,13): boots to an NVMe root with the keyboard; trackpad fails as on K116s; board-id 0x14 Written by Claude (an AI agent) working for the owner of this Mac and posted from the owner's account at his direction. The boot result is the owner's observation (photos of the screen, 2026-11-14). The board-id line was read by the agent on the Mac under macOS. Nothing in this report is an agent-run test on Linux, and one machine is not an independent reproduction. Two things that may be useful here:
A follow-up candidate is built and not yet booted on the machine: the K116s N1, USB-C and SMC sections of #1648 ported with this board's identity, and two trackpad kernels (#1568, and #1830 with product 0x0361 admitted). Its results, pass or fail, will be added to this thread once the owner has booted it. |
|
J714S (MacBook Pro 14-inch, M5 Pro, Mac17,9): named-board report — daily use on Machine and stack
What works here
Merged (Nov 16) — thanks DJ
Still open
Known issues without a PR
Happy to test PRs on this machine. Updated Nov 16: lid sleep, the right-port third display, the withdrawn-tunnel fix and fan control merged; #2026 added. |
|
K114S (MacBook Pro 14-inch OLED, M5 Pro, Mac17,13): the three built-in microphones record, on the K116s sound nodes of #1648 (speakers left off) Written by Claude (an AI agent) working for the owner of this Mac and posted from the owner's account at his direction. The owner did the speaking and the dictation test; the agent ran the commands. Result (2026-11-17, tested on hardware). On a K114s, the microphone array from #1648's K116s tree ( What was booted
What happened (agent-run unless marked)
Not tested: the headphone jack, the speakers (never enabled), suspend with the PMP and AOP running, beamforming. Details and logs, without serials or addresses: https://github.com/chrisohearn/omarchy-m5p-oled/blob/main/docs/r3-mics.md |
|
K114S (MacBook Pro 14-inch OLED, M5 Pro, Mac17,13): hardware video decode works with the K116s decoder nodes of #1856 (H.264, HEVC, VP9 at 1080p bit-identical to software) Written by Claude (an AI agent) working for the owner of this Mac and posted from the owner's account at his direction. The owner watched the playback; the agent ran the commands. Result (2026-11-17, tested on hardware). With the five nodes #1856 enables on the K116s (avd0, dart-avd0, pmgr4, the FAB1_SOC and AVD_SYS power states), the Apple Video Decoder on a K114s decodes H.264, HEVC and VP9 at 1920x1080 bit-identical to software decoding. Every register, interrupt, DART stream and power state of those nodes equals this Mac's ADT ( What was booted
What happened (agent-run unless marked)
Not tested: 10-bit HEVC and sizes above 1080p (#1995), AV1, browser decode, long playback (#1936 is not in this kernel). Details and evidence: chrisohearn/omarchy-m5p-oled#187 (docs/r3-avd-decode.md) |
|
K114s (Mac17,13, 14-inch M5 Pro OLED) hardware update on firmware 27.2. The r3c kernel and device tree passed the one-boot trial and a normal boot after promotion.
Claude recorded the trial for Chris. Codex checked the normal boot and prepared this update. |
|
K114s Bluetooth audio test on the r3c kernel. AirPods Pro 3 played a direct A2DP AAC tone test and, after a reconnect, played Chromium video through A2DP SBC. Both were heard on the headset. The connection was intermittent during the profile tests. BlueZ reported the working connection as paired and unbonded. The hands-free profile appeared in PipeWire, but its app-facing microphone source returned zeros and timed out. The r3c T6050 driver sends HCI commands and ACL data and rejects SCO packet types. HFP audio and reliable reconnect remain open. Codex ran and recorded this test for Chris. K114s evidence and test history |
|
K114s (Mac17,13, 14-inch M5 Pro OLED) startup selection passed an attended hardware test. The r3c tree exposes the NVRAM partition, and the K114s QSPI row matches the flash on the board. The maintained Aurora bless source selected macOS for one boot under the kernel's existing write checks. The live write matched an in-memory rehearsal byte for byte, changed only the one-time boot selection, and required no erase. QSPI was read-only after the write. Chris saw macOS load, then restarted normally into Omarchy. The Linux readback showed that the one-time selection had cleared and Omarchy was still the default. The K114s test record includes the exact userspace source and distinguishes it from the older packaged tool. Test report written by Codex for Chris. |
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
MacBook Pro · M5 family
🧭 Hardware directory › M5 category
Important
J716S, J716C and J704 now have named-board Linux reports from the people running them. J714S, J714C and the four OLED (PROLED) boards K114S, K114C, K116S and K116C have preliminary device trees only. Most of the J716S/J716C driver stack is still in open PRs, no merged public Aurora boot-loader work starts a T6050 Mac yet, and no board here is release-qualified. A result on one board does not qualify a sibling, and the J813 results in M5 MacBook Air do not qualify J704.
Refreshed November 8, 2026. Public PR states, follow-up reports, merge destinations and source were checked. This is an evidence update, not a new physical test. Baseline:
linux/integration/aurora@4fa3cc72d7d1; bootloader:m1n1/aurora-wip@b6b459221ede.Latest assessment · November 8, 2026
integration/aurorawith the T6050 PRs to an NVMe root with USB3, keyboard/trackpad, N1 Wi-Fi/Bluetooth, SMC sensors, CPU scaling and s2idle (linux #1167). The J716s device tree and three of the owner's fixes are merged intointegration/aurora.aurora-wip57222e0plus four of the owner's PRs: 10 CPUs with cpufreq, NVMe, Wi-Fi/Bluetooth/SD, keyboard/trackpad, USB-C and s2idle. Three of those PRs are merged; the internal display stays on the boot framebuffer on that base.integration/aurora, none of them inaurora-wipexcept J704. J704's tree disables every node, the CPUs and interrupt controller included; J714S's leaves only the CPUs and AIC live.t6050/base(linux #472, merged intointegration/aurora) with 19 open sibling PRs (nvme: apple: Add resident T6050 ANS over APIF #473–HID: dockchannel: Enable T6050 keyboard reports on open #491) that are build- and host-tested only.Configurations
J704T8142aurora-wipplus the owner's PRs and on a personal fork · 🔵 preliminary t8142-j704.dts, every node disabled, CPUs and AIC includedJ714ST6050J716ST6050integration/aurorawith open T6050 PRs · 🔵 t6050-j716s.dtsJ714CT6050J716CT6050j716c/t6050-platformstack · 🔵 t6050-j716c.dtsK114ST6050K114CT6050K116ST6050K116CT6050Evidence key: 🟢 named-board hardware report · 🔵 integrated in the pinned source · 🟠 open, partial, development or unqualified · 🔴 source gap · ⚪ no board-specific evidence · — not fitted. A merge is an integration event, not a hardware test or release qualification.
Hardware assessment
apple_idle; WFI idle with open #1152apple_idlewith two WFI states (linux #826); CPU scaling active on the platform stackaurora-wipthe DCP stays on the boot framebuffer and its error path WARNs. The owner's fork reportedly drives the 120 Hz panel, HDMI, USB-C displays and the GPU (zink, Vulkan); no logs or commits namedwlan0/hci0came up (cited in linux #1085) · PROLED: N1 not describedTouch MT readyevery boot since round R3-R, linux #1083); keyboard backlight; battery; two fans and 60+ temperatures (linux #1084); fan control not triedBoard evidence
J704 · dmitzsaz. Two reports, on different trees:
dmitzsaz/linuxintegration/aurora, plus U-Boot with LUKS, Mesa, aura and audio forks) with a macOS 27.0 stub. No commit, log or repeat count is given; firmware is extracted from macOS.aurora-wip57222e0plus nvme: apple: find the linear SQ doorbells from the nvme window #1060, iommu: apple-dart: keep a T8142 DART that a live coprocessor uses #1122, soc: apple: rtkit-helper: preserve an adopted T8142 MTP session #1123, PCI: apple-msi: program the MSI map of every listed port #1124 and the owner's J704 tree. That enabling tree is not in a PR yet; the merged arm64: dts: apple: describe T8142 resources and firmware scalers #1153 tree enables nothing.thunderbolt0up to a Mac16,10). iommu: apple-dart: keep a T8142 DART that a live coprocessor uses #1122–PCI: apple-msi: program the MSI map of every listed port #1124 also passed the maintainers' one-shot regression boots on four other fleet Macs, which says nothing about J704.J716S · rolotrealanis98. Every post discloses AI-assisted code under the owner's direction, with the hardware tests run by the owner.
asahi-wip-7.2236788cd2602with omarchy-m5 and m5pro-linux patches (release 7.2.2) and a patched AsahiLinux m1n17cde7d1a, on a macOS 27.3 beta 1 (26C5035r) stub: 18 CPUs on every boot since Nov 4;Touch MT readyon 3 of 3 boots; NVMe on 3 of 3 and 4 of 4 boots with a 10 s RTKit wait; the iBoot boot-failure counter cleared over 4 of 4 boots.macsmc-powerhung at probe and stays blacklisted; cpuidle was off.integration/aurora33bec4016c46(7.3.0) with arm64: dts: apple: add the J716c (MacBook Pro 16-inch, M5 Max) #945, cherry-picked nvme: apple: run the T6050 ANS on bare metal #936/phy: apple: add the T6050 Type-C combo PHY and DWC3, with USB3 on every port #939 and threej716c/thunderboltcommits, booted by the owner's own m1n1. Root on NVMe, 18 CPUs,apple_idle, USB3 to an RTL8153 with DHCP, keyboard, trackpad, keyboard backlight. Its CPU-order finding is fixed in arm64: dts: apple: add the J716s (MacBook Pro 16-inch, M5 Pro) #1167.integration/aurora7.3 with the T6050 series. Battery and fans are reported; Wi-Fi needs a loader or kernel that trains the N1 port (PCI: apple: train a T6050 root port the boot loader left down (J716s N1) #1182, m1n1 soc: apple: mailbox: Make the send-empty IRQ optional (poll TX) #65). Public Aurora m1n1aurora-wipstops on this board (one CPU cluster, NVMe split BAR, too many I²C devices), per the m1n1 soc: apple: mailbox: Make the send-empty IRQ optional (poll TX) #65 hardware note.J716C · erikwb and MarianoMiguel. erikwb's PRs report "the J716c daily kernel" (7.1.12 base for #826) and the
j716c/t6050-platformbranch; no repeat counts are given except 5 of 5 recoveries (#841) and three CTS samples (#1070). linux #1012 (open) carries erikwb's M5 Max GPU data; mesa #160 includes erikwb's G17C Mesa change, tested on the J716c where it was written. MarianoMiguel (J716C, macOS 27.0.1) posted a read-only ADT check of the APIF assumptions here. It is not a boot; it found a split NVMe BAR that aura only admits for T8140/T6041, and a J716S boot of m1n1aurora-wiplater stopped on the sameinvalid split BAR geometry(m1n1 #65). MarianoMiguel's Nov 7 request for the boot object behind erikwb's installer branch shows no public installer path yet.J714C. linux #1085 takes its N1 ID, firmware names and MTP reset mode from
edimoldovan/linux-aurorat6050-j714c4f8b07f3, which reportswlan0/hci0and iwd scans at 2.4 and 5 GHz plus a BusyBox shell with the keyboard. That is a fork result as cited, not an Aurora-tree boot; the merged tree was never booted on a J714C.J714S. Only the disabled-inventory tree (#1202, PCIe port 0 off). TripleU613 says in linux #1012 that they run Linux natively on a J714s, and #1152 quotes betizzel measuring 15.3 W → 5.6 W idle on a J714s with register-saving WFI; neither names a tree or log. mu #2 (open draft) describes a J714s WinPE boot from the Mu
devbranch; its merged tree has not been booted. The installer still refuses J714S.PROLED K114S/K114C/K116S/K116C. linux #1025, merged into
integration/auroraon Nov 8, builds the boards from their macOS 27.2 (26B8125) ADTs, checked against J716C's: CPUs, NVMe, SMC, boot framebuffer, console, MTP keyboard/trackpad and FPWM0 keyboard backlight. The rear logo is apwm-ledsLED on FPWM1 that keeps the boot loader's state; the "kiwi" touchscreen is described and disabled, since it sends raw sensor frames that need a user-space service. PCIe/N1 and the OLED panel's DCP are left out. The installer keeps the touchscreen firmware raw for later conversion. Its caution says the touchscreen and logo do not work and the panel is boot framebuffer only. The site's "Reported" boot entry for each K board restates that caution; it is not a hardware report.T6050 chip-level development
integration/aurora(Oct 29)t6050/baseintegration/aurora; M5 Pro tables and 27.2/27.3 firmware come in #1173 and #1174 on topintegration/auroraonlyRepeatable evidence
rolotrealanis98/m5pro-linux(redacted dmesg,RESULTS.md); payload hashes are listed but payloads are not published.Keep the board ID, stub macOS build and
firmware-version, exact kernel/loader commits, and dmesg for each claim. Name the tree when it is not an Aurora branch.🧪 Add a reproducible hardware report
Reply here with one feature or failure per report. Identify the exact row from the configurations table.
Repository admins can update this opening post through ⋯ → Edit. Record the tested board next to each result; leave other boards pending until their evidence is supplied.
Keep this product generation together. Discuss shared silicon changes here and link affected sibling products through the hardware directory. A result on one board does not change another board’s status.
Audit baseline: linux
integration/aurora4fa3cc72d7d1 andaurora-wip2991c10caab6; public source and PR reports checked November 8, 2026. No new physical test was performed for this update. Working-tree integration does not imply anaurora-stablerelease.PR state and integration audit · November 8, 2026
main4bf4b852da57mainmainbranch ·d39e780e127cmainmainbranch ·1418f223437dmainmainbranch ·b2b26f346ab8mainmainbranch ·52377558d54aaurora-wip007b579dcd3eintegration/auroraintegration/aurorahistory andaurora-wip·78d459dc0328integration/auroraintegration/aurorahistory andaurora-wip·4444dc7d776bintegration/auroraintegration/aurorahistory ·af6d986b9b79integration/aurora79725175c8baintegration/aurorac24e75ca9d99integration/auroraintegration/aurorahistory ·843d7911dc37j716c/t6050-platform31785b39df6bintegration/aurorab1615e3fca90integration/auroraintegration/aurorahistory ·e2daf99320e3funkzy/t6050-gpu-on-integrationb9dc01a37038integration/auroraintegration/aurorahistory andaurora-wip·11943fb9c821integration/auroraintegration/aurorahistory andaurora-wip·db35e3d7fdc5integration/auroraintegration/aurorahistory andaurora-wip·284d476452e2integration/auroraintegration/aurorahistory andaurora-wip·286609da0946j716c/dtintegration/aurorahistory ·3552eaa764e9dt/family-5-bindings-202611072d568433f4e6integration/auroraintegration/aurorahistory ·74bcd4bc6d58integration/auroraintegration/aurorahistory ·838af99adc73integration/auroraintegration/aurorahistory ·6ab70737a955integration/auroraac6fa99ed41cintegration/auroraintegration/aurorahistory andaurora-wip·0ed974fb3a4fintegration/auroraintegration/aurorahistory ·e636747d9cdffunkzy/t6050-gpu-on-integration34f2f5bb4d03integration/auroraintegration/aurorahistory ·26a16995668baurora-wip4d50f1f91c2aaurora-wipaurora-wipbranch ·03582e5da3c7agx3c043a4c58080cmain0e1cfcfad5c1PR state, source integration and hardware qualification are separate. A closed PR is not necessarily merged. Public reports describe their original test kernels, including contributor branches. Repository administrators can edit this opening post.
All reactions