Files
UniXNU/docs/devices/aston.md
2026-11-15 05:54:19 +00:00

106 KiB

OnePlus Ace 3 PJE110

MoyuZJ912/UniXNU-aston

OnePlus Ace 3 PJE110 (aston), SM8550, iPadOS 27.0.1 (24A446). CPU and persistent Data fixes are source changes awaiting device verification. Inherited sheng observations are reference-board evidence only.

Source record: porting.toml, b8832d72e731349c948e82f68c8fbdfc9501ae1d.

This page is generated from the device record.

Area Status
Configuration implemented
Firmware and boot planned
CPU and interrupts implemented
Memory and DMA planned
Storage and root implemented
Display planned
Input planned
GPU implemented
bluetooth planned
usb planned
Network planned
Audio planned
Power implemented
Image assembly implemented
Diagnostics implemented
arm64e-os implemented

Configuration

implemented

Hardware and dependencies

  • UniXNU 5dc00d2014, SM8550 97665e28578483082b674e8375a8fa65f4cf49cf, XNU ac9718fb1af618d5ce8678d0dc6e8a58f252216f.

Configuration

  • board.name=aston, kit=sheng, os_profile=ios-27.0.1-24A446, kernel.defconfig=aston_ios27_defconfig (branch ios27).
  • The first archive stages the pinned RamBlock publication patch before either source-storage or RamBlock compilation.

Procedure

  • Run the package workflow with the original installation seed, then inspect effective source pins and patch order.

Development findings

  • ArmCpu 93a5030 defaults armcpu-edge to zero, skipping ml_cpu_init_completed. Aston now selects armcpu-edge=1 after fixing the board Edge matrix.
  • The RamBlock fix is applied to an isolated build-tree copy; revision and normalized source SHA-256 are checked and source.json records the patched input. Shared checkouts and source pins are unchanged.

Completion criteria

  • All source pins resolve; every selected component has a record; status shows the intended overrides and patch order.

Current limits

  • No local.toml or unixnu.lock was present during this change. Full macOS kernel compilation and device results remain pending.

Next action

  • Build the branch on the pinned macOS runner using the existing Data seed.

Evidence

Firmware and boot

planned

Components: ShengPreflight.

Hardware and dependencies

  • uefi-sheng firmware, stock firmware DTB, GOP handoff, PSCI SMC interface and vendor firmware inventory.

Configuration

  • firmware.dsc = Platforms/Xiaomi/shengPkg/sheng.dsc; firmware.config = sheng.
  • proprietary-files.txt records input paths and provenance.
  • The firmware DSC and FDF embed UniXNU, UniXNUDxe, DarwinBoot, UniXNURamDiskDxe, DiskIoDxe and FAT support.

Procedure

  • Run ./unixnu build firmware -D ../UniXNU-sheng --uefi ../uefi-sheng.
  • Record the firmware commit, output FD and handoff memory map before a boot attempt.
  • Build the payload, then run ./unixnu bootimg with the firmware DTB. Inspect boot.json ramdisk_bytes and the required firmware modules.

Development findings

  • The firmware DTB is an input to boot-image assembly. XNU consumes the separately generated ADT.
  • The sheng preflight component belongs with board integration and its firmware requirements.
  • The shared UniXNURamDiskDxe reads the Mu-Silicium entry-argument HOB and FDT initrd range to publish the payload block device. ShengBootMemoryDxe owns memory expansion.
  • Configured menu entries require a payload FAT filesystem containing EFI/UniXNU/unixnu.toml. Check the UNIXNURAMDISK initrd block-device status when entries are absent.
  • fetch-sheng-dts.py writes the firmware device tree at build/sheng/sheng.dtb and the copy the GPU generators bind to at build/adreno-firmware/sheng.dtb.
  • firmware/novatek/novatek_nt36532_n81a_fw_csot.bin is the pinned touch firmware of proprietary-files.txt; the build reads the extracting machine cache first and this copy on a clean checkout.
  • board.toml [firmware.image] pins the official fastboot archive (url, SHA-256, MD5, bytes); the firmware preflight downloads it, streams out only the porting inputs, deletes the archive and generates Binaries/sheng in the firmware checkout from them. No extracted binaries live in any repository.
  • A completed extraction skips both scripts: porting-inputs.json records the archive as deleted with the pinned source url and SHA-256, so a rerun rebuilds Binaries/sheng without downloading again.
  • The extraction skip sits after the board.toml [firmware.image] load: it compares against the pinned table, which the first lines of the script have not read yet.
  • import-sheng-uefi.py regenerates Binaries/sheng only when its manifest differs from the cache: a matching tree is left in place.
  • Wi-Fi firmware files are pinned by firmware/wcn7850/manifest.json and packaged under the shared Data directory; VFS loading waits for addon installation.
  • Bluetooth firmware and NVM files are pinned by firmware/bluetooth/manifest.json and staged by scripts/stage-bluetooth-firmware.py under firmware/qca in the Data addon.

Completion criteria

  • Firmware image builds, enters UniXNU, publishes the framebuffer and preserves the documented memory reservations.

Current limits

  • Inherited sheng implementation notes below are reference material, not an aston hardware qualification.
  • Boot-image assembly requires the selected firmware DTB and proprietary input files.

Next action

  • Capture firmware revision and memory reservations with the next image verification.

Evidence

CPU and interrupts

implemented

Components: ArmCpu, ArmDTPlatform.

Hardware and dependencies

  • Eight SM8550 CPUs: cpu0-2 A510, cpu3-6 A715/A710, cpu7 X3. EPSS frequency domains contain 3, 4 and 1 CPUs.

Configuration

  • dts/aston.dts exports E/P cluster types. armcpu-edge=1 enables cross-cluster scheduler initialization. sheng-cpufreq=4 uses the governor.

Procedure

  • Apply the complete selected patch series and run scripts/check-aston-cpu-storage.py against the patched XNU and pinned RamBlock.
  • On the phone, record per-CPU startup, idle counts, busy share, frequency votes, frame latency and battery current.

Development findings

  • The former policy forbids P-to-E migration for high QoS but allows it for lower QoS, violating the XNU non-decreasing-strictness invariant used by early-exit searches.
  • The replacement preserves P preference for foreground work, permits weighted P-to-E spill at all QoS levels, and since iOS boot 7 the mirrored weighted E-to-P spill. The XNU matrix assertion is restored.
  • Missing or offline processors return UINT64_MAX. The governor rebases those samples and regressing counters instead of treating a parked CPU as 100 percent busy.
  • iOS boot 7 (diag, 390 s): with E-to-P closed, utility and background work (their buckets prefer the boot cluster) kept the three A510 cores at 100 % and 2016 MHz from 140 s to 350 s while the A715/A710 and X3 domains stayed under 20 % busy. No performance controller recommends thread groups to P here. E-to-P is now weighted_spill; scripts/check-aston-cpu-storage.py runs XNU's monotonicity and transitivity validators over the E, P, P psets.
  • iOS boot 8 (default entry, 370 s): the A715/A710 domain averaged 2043 MHz with its busiest CPU 42 % busy at 300 s, the E domain stayed at 2016 MHz from 108 to 220 s, the L3 at its top vote throughout. The governor now has no P-domain floor outside a touch boost (sheng: half the top level), jumps to 60 % of the top on a 90 % busy window instead of to the top (as Android's uag hispeed), and votes the L3 at the highest domain share of its maximum (sheng-cpufreq-pfloor-pct, -hispeed-pct, -l3 restore sheng's behavior).
  • Idle cores only executed WFI. The kernel runs under Gunyah, which traps WFI and blocks the vCPU, so a physical core and its cluster stay at the shallowest level unless the guest asks PSCI for a power-down state; Android does (kalama.dtsi: CPU rail-pc 0x40000004, cluster l3-off 0x41000044, APSS llcc-off 0x4100c344).
  • unixnu-patches/ArmCpu/0001-psci-cpu-suspend-idle.patch registers a processor_idle hook: XNU calls it before its WFI and again from cpu_idle_exit. With armcpu-idle=2 (all but the boot core) or 3 (all, plus the cluster state for long idle) it asks CPU_SUSPEND with the kernel reset vector as entry when the virtual timer is at least 10 ms away, with the timer armcpu-idle-early-us (2 ms) early. A powered-down core returns through resume_idle_cpu and arm_init_idle_cpu (XNU sets every booted core's reset handler to it); the hook then restores the GIC CPU interface and the timer's compare value. Three wake-ups more than 50 ms late, or a refused CPU state, turn it off. The default (armcpu-idle=1) only reads PSCI_VERSION and PSCI_FEATURES(CPU_SUSPEND). BootLog [idle].
  • iOS boot 9 (armcpu-idle=3, 318 s): PSCI_VERSION 1.1 (0x10001) and PSCI_FEATURES(CPU_SUSPEND) 3 (extended state IDs, OS-initiated mode supported), stop=0, but no core made a single request. The kernel's idle loop ran 0.7 to 1.7 million times per core (2,300 to 5,400 entries per second), the cores were 49 to 87 % busy and Preferences alone woke the CPU 7354 times per second for 6 s. Either XNU never called the hook or every wake-up was nearer than 10 ms. Android's idle states (live devicetree): silver-c4 entry 550 us, exit 750 us, min residency 6700 us; gold-c4 600/1300/8136; gold-plus-c4 500/1350/7480; cluster-d4 0x41000044 750/2350/9144; cluster-e3 0x4100c344 2800/4400/10150. The hook now counts its entries, entries with the timer off or masked, and entries whose deadline was under 1 ms, 1 to 4 ms and 4 ms to the threshold away, and the farthest deadline; BootLog prints them as n<entries>/<off>/<1ms>/<4ms>/<threshold> f<ms>.

Completion criteria

  • The ordered patch series applies, native policy fixtures pass, all enabled CPUs start, and measured UI latency and power improve without scheduler warnings.

Current limits

  • The E-to-P spill has no device result yet. Boot 7 drew up to 2.7 A from the 2S pack with every domain at its top level; the mid domain was throttled in 1799 of 16113 busy windows (lowest 499 MHz). L3 remains at the inherited maximum vote; deep idle and thermal feedback are not implemented by this change.

Next action

  • Compare the same workload on this branch and the previous image; retain the CPU and battery logs.

Evidence

Memory and DMA

planned

Hardware and dependencies

  • Handoff DRAM at 0x80000000 with size 0x80000000; splash reservation at 0xb8000000 with size 0x2b00000.
  • Payload owned-memory request is 0x130000000. Device DMA requires controller stream IDs, address widths and ownership.

Configuration

  • device_tree.source = device:dts/sheng.dts; include_dirs = [soc:dts].
  • payload.adt.owned_memory_size and firmware reservations describe separate allocations.

Procedure

  • Compare firmware RAM ranges, reserved regions, DTS ranges and ADT memory-map entries.
  • Check DMA buffers against controller address width, cache policy and SMMU mapping before submitting work.

Development findings

  • The sheng WLAN node exports the wlan ADT name used by wifid matching. It adds service-discovery metadata to the shared DTS; DRAM reservations and DMA resource ownership continue to use the existing board and controller descriptions.
  • Physical register translation and CPU virtual mappings have different purposes. Preserve physical resource ranges in hardware descriptions.
  • GPU and UFS use different SMMU instances and stream assignments.
  • The DTS carries each CPU's cluster type for the ADT (see the cpu area).
  • The GPU memory host fixture uses the current trace and task-role interfaces. Image backing reuse and deferred physical release are bounded by the Metal runtime; foreground placement and room-worker allocations can still overlap.
  • The exported Bluetooth node carries transport encoding and baud as little-endian u32 properties; its runtime MAC address is derived from the chosen platform identity.

Completion criteria

  • All handoff regions are accounted for and each enabled DMA engine accesses its allocated range with the documented cache policy.

Current limits

  • Inherited sheng implementation notes below are reference material, not an aston hardware qualification.
  • The DTS declares the current handoff window and splash region; memory sizing requires the selected firmware map.

Next action

  • Record the firmware memory map and each DMA consumer when changing reserved memory.

Evidence

Storage and root

implemented

Components: ShengUfs, VirtioMmioBlock, RamBlock, VerifiedRoot.

Hardware and dependencies

  • The unixnu GPT partition spans LBA 115424000..123812607 at 4096 bytes per block. The Data seed is 8 GiB.
  • RamBlock 4be2d7eb05ef5a98126c91e0302c149140c0e3e1 and the existing installation seed.

Configuration

  • native-source-storage=1 native-ufs-root=1 sheng-data-persist=2. Android userdata is not the root partition.

Procedure

  • Preserve the existing Data seed. Stage the RamBlock patch before compiling either storage archive.
  • Record the data persistence row, the Data mount capacity checkpoint and Files crash reason; verify a file survives reboot.

Development findings

  • The pinned RamBlock publishes its seed before selectVolumeSeed attaches persistent pages. APFS can cache seed checkpoint metadata before the block backend changes.
  • Aston starts UFS before RamBlock, so the persistent window is available before RamBlock registerService. The patch attaches it before publication and removes late attachment from volume lookup.
  • A refused, mismatched or unavailable writable store is not published as an apparently writable seed-only disk. Reset remains explicit; no automatic reformat, seed replacement or GPT change is added.
  • This ordering defect is established from source. Its contribution to the reported zero-capacity UI and Files crashes needs an aston boot log.
  • Device 2026-11-09 (boot 5, root 158d94bc, diag): the Data mount reports status 0, total 8589934592, free 6893985792 bytes, and the store row D st=2 er=0, so the kernel capacity and persistence are sound; Settings' 0 KB is a userland symptom.
  • The Data store read back with scripts/pull-aston-data.py and opened with scripts/apfs-read.py has no /containers/Data/Application and no /containers/Shared/AppGroup. The seed carries com.apple.migration.plist (buildVersion 24A446), so datamigrator skipped the first-boot migration that registers the system apps. Files and DockFolderViewService crash in +[FILocalStorageNode sharedNodeWithDomain:] on a nil local-storage URL (10 and 78 reports) and dasd aborts in -[NSURL initFileURLWithPath:] (50 reports).
  • patches/xnu-12377/aston-migration-rerun.patch removes the marker before launchd starts when sheng-rerun-migration=1 (menu entry Rebuild app data), so that boot runs the migration once; the BootLog capacity row reports migration=<status>.
  • iOS boot 8: Settings > iPhone Storage read Zero KB of Zero KB. SpaceAttribution (SAVolumeSizer, 24A446) sizes a volume only when its block storage device has Protocol Characteristics with Physical Interconnect Location = Internal, and none of the root windows or the Data store had one. AstonStorageTraits.cpp gives every block storage device that publishes without it &#123;Physical Interconnect = Apple Fabric, Location = Internal&#125;, as on an iPhone. libMobileGestalt reads TotalDiskCapacity from disk0 (the OS window).
  • RootLayout.h gScratch (kDataMax, 128 KiB) is allocated by the first findRootLayout in the kernel instead of being static __DATA; host tests keep the static buffer. native_ufs_object_read reads any root image object for other kernel code (the Wi-Fi firmware object).
  • iOS boot 9: AstonStorageTraits marked 3 block storage devices (storage st=1 marked=3 had=0) and the kernel reports the Data mount at 15032385536 bytes total, 13830021120 free, yet Settings still read Zero KB of Zero KB. No Preferences crash report was written. The unified log (Persist, Special, Signpost tracev3 and uuidtext/dsc 1892D244) is in the Data store; its chunksets are LZBITMAP (ZBM) compressed.
  • Boot 9 root cause of Zero KB: /var/log/com.apple.xpc.launchd/launchd.log has 'Failed to bootstrap path: path = /System/Library/LaunchDaemons/com.apple.spaceattributiond.plist, error = 138: Service cannot be loaded on this hardware', and every lookup of com.apple.spaceattributiond (Preferences, fileproviderd, fileindexerd, managedassetsd, milod, mobileassetd) failed with error 3. The job carries _LimitLoadToDeviceTree [filesystems:e-apfs]. The APFS kext reads the same /filesystems e-apfs property in enhanced_apfs_enabled_helper (callers: -apfs_newvf_enabled, apfs_volume_fixup, apfs_meta_crypto_state_unwrap), so the board does not set it. unixnu-patches/0004 adds [payload.runtime] launchd_job_device_tree; board.toml names com.apple.spaceattributiond, and the merged launchd cache drops that one key (checked against the cache read from the phone: only that job changes). Of the cache's other device-tree gates only wifid (wlan) concerns this board; iomfb_bics_daemon (hilo), biometrickitd.pearl, threadradiod, chassisplatformhostd, securityresearchdeviceinit and CommCenter stay off.
  • The unified log can be read off the Data store: its tracev3 chunksets are LZBITMAP with version byte 0x0c (bit 3 always set, bit 2: one 4-bit token per 8 bytes without run-length escapes and token 15 a table entry, bit 0: 32 KiB blocks); inline string arguments are then readable without the uuidtext format strings.
  • Boot 10/11 (root 67282aed): launchd now loads spaceattributiond (the merged cache has no device-tree gate for it), but it exits at once, 38 times a boot (exit(0) after 20 ms to 1.5 s). Its executable (SpaceAttribution.framework/spaceattributiond, decmpfs type 4, read with tools/logtools/apfs-decmpfs.py) logs 'Space Attribution Framework is NOT supported' when +[SAHelper isSAFSupported] fails, which asks every path of a list (/private/var/mobile, /private/var) for VOL_CAP_FMT_CLONE_MAPPING ('full clone tracking', bsd/sys/attr.h) and logs "%s: Volume (%@) doesn't support clone mapping". The Data volume (the CI-made seed) has role 0, features 0x2, incompatible features 0x1 (case insensitive); the APFS kext keeps clone mapping, dir-stats for SAF and attribution tags with its Enhanced APFS mode (/filesystems e-apfs, '=== enhanced apfs is enabled ===') and refuses attribution tags for some volume roles.

Completion criteria

  • The first APFS probe reads the persisted checkpoint, Data reports nonzero capacity, Files opens and can save a file, and the same file survives reboot without persistence errors.

Current limits

  • Native fixtures use API stand-ins. They do not mount APFS or establish that existing on-device Data is undamaged. The reported storage symptoms are not yet confirmed resolved.

Next action

  • Find how the APFS kext turns clone mapping on for a volume (feature bit, role, Enhanced APFS at mount or format) before choosing between a new Data seed with that feature and role Data, and e-apfs with its mount-time fixups.

Evidence

Display

planned

Components: GopFramebuffer.

Hardware and dependencies

  • UEFI GOP framebuffer, stride, pixel format, splash reservation and board panel/backlight controls.

Configuration

  • Select GopFramebuffer; CONFIG_XNU_NATIVE_LEGACY_DISPLAY=y; artwork-display-gamut = sRGB.
  • sheng-vsync=120 requests board display timing.
  • sheng-dpu-scanout=2 lets the display pipes fetch the compositor surface through a stage 1 window of the display bank (TCR2 SEP 7); off by default. The flip waits for the GPU fence of the frame (always on).
  • sheng-backlight=1 starts the backlight driver; sheng-backlight-control=0 leaves out backlight-control and the AppleARMBacklight metaclass answer.

Procedure

  • Verify GOP address, dimensions, stride and pixel layout before enabling presentation.
  • Build the late display archive with the selected OS ABI records, then check display publication and frame presentation.

Development findings

  • iPadOS 27 reads the nits range from IODisplayParameters.BrightnessMilliNits and sets brightness-nits as signed 16.16. CoreBrightness needs backlight-marketing-table and backlight-calibration-parameters to create the display. BrightnessControl reads LmaxProduct as 4 bytes of 16.16 and LmidProduct and LminProduct as numbers; a zero LmaxProduct makes every slider position the maximum. iPadOS 17 uses brightness and FadePeriod; the service accepts both.
  • ADT vmm-present makes MobileGestalt answer IsVirtualDevice, and iPadOS 27 CoreBrightness then selects VMBLControl, which drops every brightness request. AMFI and many frameworks need the answer, so UniXNU scripts/appsync/BacklightMode.m changes only backboardd: initBLControl: gets mode 1 instead of 3.
  • The board archive records SmmuBypass.h and AppsSmmuLock.h from the resolved board package directory as Bluetooth build dependencies.
  • The 24A446 display adapter declares fixed original targets. XnuPrelink relocates BTI/B stubs after native segment placement; a cached native-slide value is no longer used. The 2026-11-02 device log showed the unrelocated constructor branch jumping to 0xfffffff00dbb4c74 instead of 0xfffffff00a288c74, an excess of 0x392c000.
  • The display source and ABI binding are selected independently from the physical panel configuration.
  • The board configuration declares sRGB for the current 8-bit framebuffer path.
  • scripts/build-gop-legacy-display.py also builds the board driver objects in the display archive, including the ADSP, IPC router, FAT reader and battery manager.
  • With sheng-dpu=2 DisplayTimingProbe reads the VIG 0 and 1 pipes, CTL 0 and the tear checks; sheng-dpu=3 also rewrites the pipe addresses and times the flush latch (5.4 ms, no start needed: the panel refreshes by itself).
  • Direct scanout: the compositor submits a swap when the command buffer is scheduled, so the flip waits until the GPU wrote the frame; the wait removes the flicker (content changes of the shown surface fell from 808 to 2 per run). Direct scanout needs the 8-bit BGRA framebuffer.
  • The GENI I2C core also moves 1 to 8 byte writes and 1 to 4 byte reads (the audio amplifiers use 32-bit registers); the backlight keeps its 8-bit transfers.
  • The legacy display builder now refuses arm64e/iOS 26 before reading original ABI addresses. Its existing use of kernel_target.TRIPLE does not qualify the iOS 17 class layout for 23F84. Source input transport compilation is independent of this builder.
  • The 24A446 adapter creates a BGRA IOSurface through the original IOSurfaceRoot factory, checks allocation and geometry, and transfers its reference to the family default-surface field. The void create_default_fb_surface slot is invoked by the original start method.
  • build-gop-legacy-display-27.py reads the device checkout from UNIXNU_DEVICE (set by the release) before the sibling directory name, so a checkout under any name works., and the HID source is the tree pin (refer/IOHIDFamily of the feed) instead of a hand-placed driver-sources copy.
  • The 24A446 board archive resolves QcaBluetooth from the assembled tree before the device fallback, matching the selected runtime component.
  • The late probe starts USB through the weak sheng_usb_probe, and the board package build compiles QcomDwc3.
  • package/AstonBootScreen.cpp draws a boot screen into the firmware framebuffer from the board probes (bsd_init) until iOS draws: title, a moving bar, elapsed time, the stage (Data mounted, data migration asked for) and the newest kernel message buffer lines (18x42 pixel cells, 70 columns). It stops without clearing once the legacy display accepts a swap or its sentinel pixel changes; aston-boot-screen=0 turns it off. Host run on a 1264x2780 buffer with the boot 5 message buffer, ASan and UBSan clean; not yet on the device.
  • Screen class (iOS boot 7): the panel is 1264 x 2780 pixels, 421 x 927 points at scale 3, while iPhone17,2 is 440 x 956 (artwork-device-subtype 2868); ProgressUI found no screen class for it (stackshot "unsupported screen class" during the first-boot migration). The largest iPhone size at 3x that fits the panel is 402 x 874 (1206 x 2622), the iPhone 16 Pro and the 16 Pro Max's zoomed size. aston-display=1 (menu entry "iOS 402x874", patches/xnu-12377/aston-display-window.patch) narrows PE_state.video to that window at (16, 81), the 16 KiB-aligned origin nearest the centre, with the panel's row stride; every framebuffer consumer maps from PE_current_console. package/AstonDisplayWindow.cpp paints the borders black and TouchProbe maps contacts into the window (TouchBoard.h tb_window_axis, border contacts clamp to the edge). scripts/check-aston-display-window.py checks the origin and the touch mapping. No device result yet.
  • The panel is the AA551 P 3 A0004 DSC command mode AMOLED on DSI0 (Android cmdline, OnePlus display-devicetree 4498f99): DCS 0x51 high byte first on page 0 (FF 08 38 00), levels 1..4094 with 1..3349 the normal range. sheng-backlight=2 runs sheng's iOS brightness service with that DCS write (AstonPanel.cpp, the TPG FIFO command path of UniXNU-asteroids 6370e66) in place of sheng's I2C backlight chips.
  • iOS boot 9: L st=4 req=0 and [aston] panel sent=0 although the unified log has 86 'setDisplayBrightness (permanent)' calls: CoreBrightness answers 'No container for handle Display(id: 1)' (com.apple.CoreBrightness.DisplayError 1, 17 times), so on iPhone17,2 it never builds a display container over the board's backlight service. CoreBrightness' code is not in the extracted subcaches; its own messages use static format strings, which need the uuidtext/dsc file to read.
  • Own display driver, stage 0: the panel stays lit by Qualcomm's boot chain (XBL continuous splash; Mu-Silicium's GOP and iOS only reuse its buffer at 0xb8000000), whose DPU sends that buffer by autorefresh. The phone's devicetree (qcom,mdss_mdp@ae00000, read 2026-11-14) gives the SDE catalog relative to MDSS 0xae00000: MDP 0x1000, CTL 0x16000 + 0x1000 n (CTL 0 preferred for the primary display), VIG pipes 0x5000/0x7000/0x9000/0xb000, DMA 0x25000..0x2f000, mixers 0x45000.. (0 and 1 primary), merge 3D 0x4f000.., DSC 0x81000 and 0x82000 (encoders +0x100/+0x200, controls +0xf00/+0xf80, DSC 1.2), INTF 0x35000.. (INTF_1 0x36000 = DSI0, its tear interrupt block 0x36800), interrupt SPI 83; DSI0 0xae94000, its PHY 0xae95000 and PLL 0xae95500; the display clock controller 0xaf00000 (dispcc-sm8550.c: MDSS GDSC 0x9000, INT2 0xb000, MDP branch 0x800c, pclk0 0x8004, byte0 0x8028, vsync 0x8024). The panel timing (1264x2780, porches 26/26/2 and 42/22/2, 1113.6 MHz, 10 bpc, DSC 632x20 slices at 8 bpp, two encoders, TE pin, 120/90/60 Hz) is in aston-android/panel.
  • UniXNU-Pagani's DpuProbe hung the OnePlus 13T reading display blocks off the boot loader's path and the apps SMMU; sheng's DisplayTimingProbe (sheng-dpu) reads CTL 0..3, all ten pipes and INTF_2, fine on sheng's dual-DSI panel but off this board's path. AstonDisplaySurvey.cpp (aston-dpu=1, menu entry "iOS display survey") reads once only what the running path uses: the clock controller, then, with the MDSS GDSC and the MDP, AHB, pclk0 and byte0 branches on, the MDSS and MDP tops and CTL 0; with CTL 0 driving INTF_1 the pipes, mixers, DSC and merge blocks CTL 0 marks active, INTF_1 and its tear block, DSI0 and its PHY, and INTF_1's frame and tear counts over a second. Nothing is written. BootLog [dpu] and [dpu-dump] rows.
  • iOS boot 10 ("iOS display survey", all reads at once): the boot log stopped after its first write at 26.9 s, before any board probe ran, with no panic and a black screen: the survey stopped the phone, as UniXNU-Pagani's probe did the 13T. The survey now waits 15 s and takes paced steps 3 s apart (stage 100 + n in [dpu] st=): the two GDSCR words first, then the five branches, the rest of the clock controller, the MDSS top, the MDP top, CTL 0, INTF_1, the pipes, mixers, DSC and merge blocks CTL 0 names, DSI0, its PHY and the refresh.
  • Android's view (2026-11-14, root, debugfs mounted at /data/local/tmp/dbg and unmounted after, aston-android/display-regs): the SDE resource manager gives the primary display one CTL, two mixers, two DSPPs, two pingpongs, two DSC encoders and one interface, no merge 3D; 1264x2780 at 120 Hz, DSC 10 bpc 8 bpp. The Iris7P is in analog bypass (abyp mode 1, display mode SLEEP-ABYPASS): DSI0 reaches the panel through the bypass switches and the bridge sleeps. The SDE debug window for MDSS (sde_off/sde_reg) takes offsets below 0x100 only (MDSS_HW_VERSION 0x90000001); DSI0's controller and PHY windows take 0x1000, and both were read (dsi-android.txt; DSI_CTRL 0x1f5, as iOS sees it).

Completion criteria

  • A known frame appears with correct geometry and colors, then userland presentation and backlight controls work.

Current limits

  • Inherited sheng implementation notes below are reference material, not an aston hardware qualification.
  • The late display builder needs qualified prelink candidates and ABI records. A timing request requires a measured presentation result. The 24A446 default-surface adapter still requires a hardware boot result.
  • On iPadOS 27 backlight off on sleep is not verified. The range of 2, 140 and 500 nits is not measured and only sets the slider's scale.

Next action

  • Record frame timing, pixel format and presentation evidence for each image used to qualify display.
  • On iPadOS 27, verify that sleep turns the backlight off.

Evidence

Input

planned

Components: ShengHidInput, ShengNovatekTouch, QcomGeniSpi.

Hardware and dependencies

  • Novatek NT36532 sensor, GENI SPI controller, pin configuration, reset/interrupt lines, firmware and touch coordinate geometry.

Configuration

  • Select ShengHidInput, ShengNovatekTouch and QcomGeniSpi; sheng-touch=4 and sheng-touch-hid=1 select board integration.

Procedure

  • Read TouchBoard.h and verify the SPI clock and mode against the controller setup.
  • Validate protocol reports, contact tracking and coordinate orientation before publishing HID events.

Development findings

  • backboardd caches the event service's usage pair when it first sees the service, and drops every touch if that pair is not 0x0d/4. The DriverKit driver writes 1/6 and registers the service, so HidInputProbe checks every 2 ms until the service has been registered for 10 s.
  • The shared SMMU fixture adds UART14 stream 0x423 beside touch without modifying its route and accepts a 64-bit DMA buffer address.
  • The arm64e HID adapter uses the same fixed-target stub contract as the display adapter. Host verification checks all 51 IOHIDFamily stub destinations after fileset placement.
  • Controller transfer logic, sensor protocol and board sensor geometry are separate inputs.
  • Track contact IDs across movement and lift events; a register read alone covers only transport access.
  • sheng_buttons_press() holds a button mask for a given time through the HID sink, for unattended tests (the audio lock sound test uses it).
  • iPadOS 26 source-interface build: IOHIDDevice reports and independent touch/button, GENI SPI, Novatek and firmware archives compile. Four host model suites pass ASan/UBSan. The source path does not patch a foreign vtable.
  • On arm64e the event driver's vtable is copied and every slot is re-signed with the 220 discriminators read from the 24A446 kernelcache; each slot is checked against the original table first, and AppleUserHIDEventService, this device's event driver, takes the copy as well. A search over 16-bit values can accept a second discriminator with the same 24-bit signature and stops touch events.
  • With latency capture enabled, the HID input path records the time of each touch report handed to the event driver; these times bound the analysed touch windows. Contact IDs and coordinates are not recorded.

Completion criteria

  • Touches map to the display, simultaneous contacts keep stable IDs, release events arrive, and buttons produce expected HID reports.

Current limits

  • Inherited sheng implementation notes below are reference material, not an aston hardware qualification.
  • Sensor firmware and board dimensions remain sheng inputs.
  • 23F84 HID linking/startup and edge gestures remain unqualified. Apple IOHIDFamily 2238.120.5 omits event-type/struct headers and the complete device build needs private reporting/analytics dependencies.

Next action

  • Attach contact and orientation observations when firmware or geometry changes.

Evidence

GPU

implemented

Components: Adreno740, AdrenoMetal, AirNir, Sm8550Platform, AdrenoPM4, AdrenoFirmware, ArmSmmu4K.

Hardware and dependencies

  • A740 GPU/GMU registers, GPU SMMU streams 0/1 and GMU stream 5, signed ZAP firmware and board power votes.
  • Turnip, metal2vulkan, AIR translation and the selected OS GPU/Metal ABI.

Configuration

  • Select Adreno740, AdrenoMetal and AirNir; shared SM8550 operations arrive through component dependencies.
  • Board scripts combine shared GMU code with package/ShengGmuVotes.c.
  • sheng-gpu-preempt defaults to 2: requested high priorities use ring 0, normal/low work uses ring 1. Mode 1 sends only low/background to ring 1; mode 0 selects one ring.

Procedure

  • Qualify power, firmware, SMMU and GMU startup before command submission.
  • Check memory mapping, command completion and a known rendered image before staging the Metal plugin.

Development findings

  • Each ring restores its executed page-table snapshot, saves GMEM and runs the SRAM postamble. GMU keepalive covers restore-register writes through the incoming WPTR update; board writes retry AHB-fence drops.
  • The shared process queue keeps FIFO across priority changes. Current live Metal requests determine priority, with task GPU role selecting foreground or background.
  • All AdrenoMetal clients use one sample by default; NATIVE_METAL_SINGLE_SAMPLE=0 preserves a process's requested sample counts. The frontend implements heap aliasing, swizzles, mip/slice copies, function constants and callbacks, specialized link names, backing-allocation statistics, queue capacity and full viewport/scissor arrays.
  • Writable same-user caches cover frontend, runtime and Turnip IR3 artifacts; restricted clients use private caches. Normal builds enable Mesa disk caching and apply the recorded vendor patches.
  • The optional sheng-wallpaper-freeze=1 mode retains a Poster renderer's first completed large IOSurface draw and parks later runtime commits. PosterBoard and compositor clients keep running. Event 60 identifies the completion boundary and parked commands.
  • The current sheng iPadOS 27 image boots and accepts touch interaction, and shader priming covers 280 of 280 selected libraries. Foreground and room-worker 64 MiB allocations still overlap. Focused latency capture retains full-width runtime memory phases, command and pass timings, paired callbacks and allocation roles. GPU pixel collection and per-frame content scans are disabled.
  • The kernel GPU ring keeps submission, retirement, display arrival, fence, latch, build identity and the render server's render, encode and completion thread identities; client processes forward nothing else. The decoders read only these records and write touch-window subsets.
  • check-gpu-trace.py checks that Turnip's GMEM swap trace (passes, transfers, attachments) leaves only the render server, and the kernel store keeps events 36, 37 and 44.
  • AdrenoMetal carries 25 patches over metal2vulkan 963f0e64. The lowering thread, commit submission and pipeline warm pass are off by default and sheng-metal-flow=1 enables them for testing. They showed no measured frame rate gain, and one of four boots with them stopped presenting at the logo. pipeline_waits stays Exact because the upstream default drops draws of a pipeline not made in time.
  • iOS 27 (iPhone17,2), stage B: the GPU arguments of iPadOS b5bb234 under the names their readers use since UniXNU 47ec6a5 (sheng-gpu-power=9, sheng-gpu-metal=1, sheng-gpu-dcvs=1 and sheng-fb-gpu=1 for this board; fb-gpu=1 for GopFramebuffer, fb-bgra=1 for the Metal plugin, gpu-addon=1 for the addon mount) and the Metal addon built ([payload.runtime] gpu_addon). The GMU, SQE and ZAP images (151 KiB) are in the native __TEXT: the iPhone17,2 collection has 125 KiB left for __DATA_CONST and 432 KiB for __TEXT.
  • The Metal plugin writes QuartzCore options at fixed addresses of the iPad 24A446 shared cache (CE33B2BA). The iPhone17,2 cache (1892D244) has the same QuartzCore code at the same addresses, and its __DATA_DIRTY, __AUTH and __AUTH_CONST 0x4000 higher. Every ADRP reference of QuartzCore's __text into those segments (7978 and 6450) differs by exactly 0x4000 at the same instruction, those into __DATA (7027) by 0. unixnu-patches/AdrenoMetal keys the addresses by the cache UUID read from the mapped cache header, after UniXNU-asteroids 5650eda. An unknown cache gets only the environment (marker 0x6EF0). The caches were read from the IPSWs' sealed SystemOS images with scripts/apfs-read.py, which now follows sealed trees (hashed nodes, child ids relative to the tree root, fext extents).
  • iOS boot 8: the GPU spent 72-88 % of each 10 s at 680 MHz while 27-32 % busy (a touch held the top level 1 s); Android idles the A740 at 220 MHz (kgsl min_pwrlevel 7). The DCVS floor default is now level 1 (220 MHz) and the touch boost 300 ms.

Completion criteria

  • GPU submissions complete, mapping/fence handling is correct, a known image renders, and the selected Metal client presents it.

Current limits

  • Inherited sheng implementation notes below are reference material, not an aston hardware qualification.
  • A740 coverage depends on the selected firmware and OS ABI; board qualification requires its own memory and power evidence.

Next action

  • Qualify concurrent Metal clients, repeated preemption, rendered pixels and frame pacing on sheng with iPadOS 27.
  • Run third-party Metal applications with cryptid 0 on the metal2vulkan 963f0e64 build. Find why backboardd stopped presenting with sheng-metal-flow=1 before enabling any of its three paths.

Evidence

bluetooth

planned

Components: QcaBluetooth.

Hardware and dependencies

Configuration

  • UART14 is QUP2 SE6 at 0x898000, GIC INTID 493, wrapper 0x8c0000 and apps-SMMU stream 0x423; CTS/RTS/TX/RX use GPIO 76/77/78/79 and BT_EN uses GPIO 81.
  • The board's 3.2 Mbaud operating rate uses a 51.2 MHz serial source and 16x oversampling.
  • package/BluetoothFirmware.h and stage-bluetooth-firmware.py share firmware/qca under the Data addon.
  • sheng-bluetooth=1 starts the bound UART14 worker after Data addon installation; log-bluetooth=1 enables initialization messages. The drivers/QcaBluetooth submodule or an explicit --local-module checkout supplies kernel and runtime sources.

Procedure

  • Stage firmware with scripts/stage-bluetooth-firmware.py OUTPUT --source FIRMWARE_QCA_DIRECTORY.
  • Run scripts/check-wcn-power.py and scripts/check-bluetooth-board.py ASSEMBLED_TREE after changing shared power.
  • Build the protocol, GENI DMA and Skywalk sources with the selected arm64e kernel command, then package BlueTool and the native controller adapter.
  • Boot the integrated image with log-bluetooth=1, then check pairing, reconnection, audio, sleep and recovery.

Development findings

  • On iPadOS 27.0.1 (24A446) the device pairs AirPods Pro, reconnects after reboot without pairing again, toggles power, plays A2DP at normal volume and quality, and shows noise control and battery. Music over Bluetooth during 2.4 and 5 GHz speed tests stays normal.
  • WcnPower records independent WLAN and Bluetooth ownership of the six shared supplies; the last owner releases them. Bluetooth-only startup omits the PCIe NoC vote. Wi-Fi startup uses this ownership interface.
  • The firmware VFS reader waits for Data addon installation, accepts controller firmware names, bounds file size and rejects short reads.
  • The protocol core uses the byte-identical Linux btqca.c initialization through namespaced host services. H4 backpressure retains input, and IBS control bytes enter only between TX packets.
  • The GENI engine provides bounded DMA buffers, deferred IRQ work, wire drain checks, quiesce/resume and error reporting. UART.SK and BlueTool sources compile for the 27 target.
  • BluetoothProbe binds UART14 clocks, 32 KiB of DMA buffers through the firmware-managed bypass bank, IRQ 493 and GPIO79 wake. The 64-bit physical DMA addresses use the same shared mapping planner as touch and audio; Bluetooth uses no private page tables. Shared configuration locks preserve touch and audio resources; GIC routes dispatch directly to registered handlers. Hardware faults abort command waiters immediately.
  • The version-locked 27 adapter selects type zero, clears vendor hooks, preserves generic callbacks and skips the empty H4 pre-stop hook. Real HCI reports supply eleven capabilities, including coded and 2M PHY; an asynchronous 0x203b query supplies up to six native advertising sets. Model, cache, lower flags and broadcaster mode agree.
  • The board publishes a bluetooth IOService before userland starts, with H4 encoding, 3.2 Mbaud and a private stable MAC. Public addresses use network order and convert to Linux bdaddr_t at initialization. The runtime signs three arm64e binaries, preserves bluetoothd entitlements and uses read-only system-path overlays; BlueTool runs as root with transactions enabled.
  • The kernel archive and runtime builder accept the same resolved QcaBluetooth checkout. UART clock-restore failures end the transport, preserve the hardware error and stop command waits and wake retries. The board startup entry uses C linkage to match the late platform probe. The archive record hashes the Bluetooth board, core, shim and SoC inputs before kernel linkage.
  • Controller shutdown and faults clear the UART.SK queues, wake both rings and return the controller error from sync callbacks before data movement. In-flight notifications retain both rings until disconnect can release them.
  • BlueTool starts the controller by reading kern.qbt_power_on because its sandbox denies sysctl writes; connecting to UART.SK also starts a stopped controller. Closing state belongs to the nexus that owns the session, so a channel torn down after its control socket closed cannot block the next connection.
  • Extended scanning uses 0x2041 and 0x2042, because the controller rejects legacy scan commands once extended advertising commands have been used. Apple's scan extension stays off because it has no type zero backend.
  • A random address (0x2005) is queued before a scan can be enabled, and starting an advertising set always sends its address (0x2035). When no resolvable address can be generated, the refresh uses a non-resolvable address and completes, so the daemon does not restart on the busy flag.
  • Pairing keys need the keychain, which needs AppleKeyStoreState software key wrapping. The board returns /chosen/unique-chip-id through native_aks_device_id.

Completion criteria

  • The integrated kernel and payload boot with independent Bluetooth power and live UART DMA/IRQ.
  • The native stack scans, pairs, reconnects, stores keys, supports BLE and plays Bluetooth audio.
  • Idle power, audio continuity, recovery and Wi-Fi coexistence have device measurements tied to the image revision.

Current limits

  • Inherited sheng implementation notes below are reference material, not an aston hardware qualification.
  • Calls and microphone (HFP/SCO), sleep and wake, idle power and recovery are not tested.
  • The kern.qbt_* sysctl names compile but have not run on the device; the device results use the earlier names. A build that resolves the driver from the submodule has not been run.
  • Features that need Apple vendor commands are unavailable. Non-connectable advertising uses a non-resolvable address because the secondary IRK is zero.

Next action

  • Boot an image built from the submodule to confirm the kern.qbt_* sysctls, then test calls and microphone, sleep and wake, idle power and recovery.

Evidence

usb

planned

Components: QcomDwc3.

Hardware and dependencies

  • AppleUSBXDCI, AppleUSBXHCI, IOUSBDeviceFamily, IOUSBHostFamily and AppleUSBDeviceMux from the 24A446 kernelcache.
  • Linux reference https://github.com/ianchb/sm8550-mainline for the SoC sequences and sm8550-xiaomi-sheng.dts for the board.

Configuration

  • drivers/QcomDwc3 is a submodule of https://github.com/UniXNU/QcomDwc3.
  • sheng-usb=1 runs the driver; 2 adds register sampling; 3 also takes the controller off the bus and back once.
  • sheng-usb-host=<s> forces host mode after a delay; sheng-usb-back=<s> then returns to the cable's role.
  • sheng-aks-sls=3 reports no passcode.
  • board.toml adds the USB device and host stacks to the prelink roots for iPadOS 27 and leaves out AppleOnboardSerial and AppleSMC.
  • chip_id 0x8142 comes from board.toml; board-id 0xc and security-domain 1 come from the OS profile.
  • The provider's configuration-string is standardMuxOnly.

Procedure

  • Generate the class layouts again after the kernelcache changes.
  • Boot and connect to a Mac: the device lists as 05ac:12ab and lockdownd answers QueryType.
  • Unlock the tablet and tap Trust: the Mac reports the device as paired and starts a lockdown session.

Development findings

  • This CPU/storage change does not enable USB on aston. Existing sheng USB observations below are retained only as reference.
  • payload.runtime.appsync and appsync=1 in board.toml do not change the USB stacks.
  • AppleUSBXDCI programs the DWC3 itself; a subclass supplies the register window at base + 0xc000, one interrupter, power and PHYs. AppleUSBXDCI has no personality, so it needs an explicit kext load before the subclass registers.
  • The driver does no cache maintenance. Stream 0x40 needs a stage 1 identity context with write-back inner shareable blocks; through the shared bypass bank the CPU reads stale event words.
  • The eUSB2 reference is 38.4 MHz (xo_board 76.8 MHz, RPMH_CXO_CLK divides by 2). With the 19.2 MHz table chirp passes and high-speed data fails.
  • Slot 179 of the class keeps the family's configuration list when true, which leaves bNumConfigurations 0; it must stay false without an IOPort transport object.
  • lockdownd starts on the IOResources property AppleUSBCableType=USBHost, one of its launch events; configd's USBDeviceConfiguration plug-in then sets the description and the family goes on the bus.
  • UCSI reports the cable partner and selects the role. Until it answers, the charger's online value stands in for a host.
  • In host mode the family's desired power is 0 once ports are idle, so the xHCI subclass answers on. The device stack answers 0 while the host role owns the core.
  • A Mac looks up chip-id, board-id and security-domain in MobileDevice's device map. A pair that is missing makes it ask for a software update.
  • SpringBoard accepts a CFUserNotification from a process on the system partition: it compares statfs f_fsid of the executable with /System. The SystemOS union layer has its own number, so native-union-statfs-fsid reports the root volume's number for that layer.
  • lockdownd completes Trust directly only in lock state 3. Any other state asks for a passcode.

Completion criteria

  • A Mac enumerates the device, usbmuxd lists it and lockdownd answers, at 10 Gbps on a USB 3 cable and 480 Mbps on a USB 2 cable.
  • A physical reconnect re-enumerates.
  • Trust appears on the panel, the Mac pairs and starts a session, and lockdownd starts installation_proxy, afc, atc and diagnostics_relay.

Current limits

  • Inherited sheng implementation notes below are reference material, not an aston hardware qualification.
  • Sustained transfers and app installation have not run.
  • One cable orientation has run on USB 3.
  • Host mode runs Apple's xHCI stack and completes commands; enumerating a peripheral is not verified.
  • MobileDeviceMinimumVersion has no answer; a Mac treats that as no update needed.
  • fairplayd.A2 and adid stay disabled: with both enabled, SpringBoard does not finish starting.

Next action

  • Run a sustained transfer and an app installation.
  • Test the reverse cable orientation on USB 3.
  • Verify peripheral enumeration in host mode.

Evidence

Network

planned

Components: Ath12kWifi.

Hardware and dependencies

  • WCN7850 PCIe endpoint, MSI routing, DMA callbacks and hash-pinned amss.bin, m3.bin and board-2.bin.

Configuration

  • sheng-wifi=5 selects the mainline cfg80211/mac80211/ath12k runtime; wpa_supplicant 2.12 owns userspace authentication.
  • WifiProbe supplies power/PCIe/MSI and the 128 MiB DMA window; WifiNet exposes en0/Apple80211. The DTS exports /arm-io/wlan for wifid.
  • The signed service runs at /usr/libexec/unixnu-wpa_supplicant. package/WifiFirmware.h defines the shared loader/packer directory /private/var/ngpu/firmware/ath12k/WCN7850/hw2.0/.

Procedure

  • Run host checks affected by source changes, build the kernel/payload/boot image and verify packed inputs.
  • Check scanning, association, DHCP, traffic, cancellation and signal display on the device; measure throughput, jitter and heat.

Development findings

  • Device verification covers Wi-Fi toggling, SSID scanning, WPA2 association, DHCP, traffic and join-result delivery.
  • Accepted joins complete after port authorization, cancel once and time out after 30 seconds. Association and cached peer RSSI supply signal updates.
  • TX/RX use direct copies, shared DMA storage, bounded reuse and batching; mac80211 supplies aggregation, reordering, FQ and AQL.
  • Data firmware reads wait for addon installation. Packaging validates file hashes, service/cache signatures, stock jobs, trust modules and the preserved Data seed.
  • The shared runtime cache retains the native Wi-Fi jobs and adds the supplicant alongside the Bluetooth service configuration; the original Data seed and existing trust modules are preserved.
  • Aston WLAN facts (live devicetree, OnePlus cnss2 qmi.c): WCN7850 17cb:1107 on PCIe0, WLAN_EN 80, BT_EN 81, SW_CTRL 82, PERST# 94, WAKE# 96, as on the reference board; firmware WLAN.HMT.1.0-03639 (kiwi/amss.bin), board_id 0xff, board data /odm/etc/wifi/kiwi/bdwlan.elf (cnss_get_oplus_bdf_file_name: bdwlan.elf for a CN unit). The Wi-Fi stack is not in the iOS kernel: the iPadOS build with it needed 1920 KiB of __DATA, the iPhone17,2 collection has 1872 (32 KiB spare without it), and its code would fill the 11872 KiB __TEXT_EXEC room.
  • iOS build with Wi-Fi (ASTON_WIFI=1, sheng-wifi=5): ShengUfs' 128 KiB scratch buffer (RootLayout.h gScratch) is allocated at run time, so the kernel's __DATA has room for the stack. XnuPrelink's Layout run offline against the iPhone17,2 collection with CI 43018454672's segments plus the stack (893 KiB code, 75 KiB data, about 45 KiB constants) and minus the buffer leaves 192 KiB of code, 77 KiB of constants and 80 KiB of __DATA.
  • Firmware: linux-firmware 20261019 ath12k/WCN7850/hw2.0 amss.bin, m3.bin and board-2.bin are the reference board's files (same SHA-256). The Android kiwi files for hw2.0 are amss20.bin and phy_ucode20.elf, cnss builds; they are not used. ath12k looks a WCN7850 up in board-2.bin by bus, PCI IDs, subsystem IDs, chip ID 2 and board ID 255 and only then takes board.bin; it requires a regulatory database (board-2.bin element 'bus=pci', or regdb.bin) before the board data.
  • The kernel reads the firmware from the root image object "Wifi" (scripts/make-aston-wifi-object.py: amss.bin, m3.bin, a board-2.bin holding only linux-firmware's regulatory element, and this phone's /odm/etc/wifi/kiwi/bdwlan.elf as board.bin), so neither the firmware nor the board data is in the public build. NativeUfs exports native_ufs_object_read; WifiFirmware.c checks each file against the object's digests and falls back to the Data addon only without a valid object. BootLog: [aston] wifi object st= files= served= bad= err=.
  • WCN7850 PMU rails (phone devicetree chip_cfg@1, OnePlus kalama-cnss.dtsi): the reference board's six rails S5G, L15B, S2G, S4E, S4G, S6G, set to Android's 1.0, 1.8, 0.98, 0.95, 1.35 and 1.9 V. OPLUS' antenna switch (salami_overlay_common.dtsi oplus,wlan-wsa, GPIO 98, default state output high) is driven high with WLAN_EN. The sleep clock is PMK8550 gpio3 as on the reference board (the overlay adds only the Iris clock on gpio5).
  • iOS boot 9 (root 75bf9f05, deep idle 3 entry, 318 s): the Wifi object was valid (st=1, 4 files, 3 served). The card powered up, linked, took amss.bin over BHI and BHIe and reached mission mode; QMI answered (chip_id 2, board_id 0xff, soc_id 0x40170200, firmware 0x1103006c of 2026-03-06). The regulatory element downloaded, then the board data (bdwlan.elf, 88768 bytes) started at 783 ms; at 974 ms MHI reported a system error and the download timed out (qmi failed to load board data file: -110). ath12k_core_reset refused recovery before firmware ready (dev flags 0x12).
  • Android reports the card as 17cb:1107 with subsystem 17cb:1107 (/sys/bus/pci). linux-firmware's board-2.bin has a board element under exactly this card's name (bus=pci,vendor=17cb,device=1107,subsystem-vendor=17cb,subsystem-device=1107,qmi-chip-id=2,qmi-board-id=255, 88760 bytes, sha256 d4f92e9f...). The phone's bdwlan.elf has the same layout (one 87040-byte segment at 0x1000) but is built for OnePlus' cnss firmware. make-aston-wifi-object.py now keeps that board element in board-2.bin, so ath12k sends the board data linux-firmware's amss.bin was released with; bdwlan.elf stays as board.bin, used only without a match.
  • The RAM dump reader in WifiProbe.cpp (rddmScan: firmware error texts, crash record, export to the boot log's export area) was keyed to the MHI_CB_EE_RDDM status line, which ath12k prints only at debug level; with ath12k_debug_mask 0 it never ran in boot 9. It now also starts on ath12k's 'ignore reset dev flags' and 'firmware crashed' warnings.
  • iOS boot 11 (root 67282aed, iOS diag): with linux-firmware's board element for this card the board data downloaded, m3.bin followed and the firmware came up; ath12k then failed in core start: 'dma window full: want 69632 free 439424 largest 61440', 'failed to setup reo_dst_ring', 'failed to start core: -12'. The firmware's target memory took about 22 MiB in seven segments and the arena peaked at 49.9 MB of ShengUfs' 48 MiB window. The window is now 32 blocks of 2 MiB (64 MiB, kExtraBlocks in drivers/ShengUfs/NativeUfs.cpp).

Completion criteria

  • Scanning, association, address configuration and sustained bidirectional traffic pass with the selected firmware.

Current limits

  • Inherited sheng implementation notes below are reference material, not an aston hardware qualification.
  • Current-image device checks pending: cancellation/timeout completion, runtime signal display, Data firmware startup, throughput, jitter and heat.

Next action

  • Boot the image with the 64 MiB window and check the wifi-log rows past core start: WMI ready, mac80211 registration, scanning, association, DHCP and traffic.

Evidence

Audio

planned

Hardware and dependencies

  • ADSP firmware from modem_a, SMEM, SMP2P, GLINK, the GPR channel adsp_apps.
  • Speakers: six CS35L43 amplifiers on QUP1 SE5 and QUP2 SE0 (I2C 0x40 to 0x42), shared reset gpio90, secondary MI2S (gpio121 to 124).
  • Microphones: WCD938x codec on the TX SoundWire master (0x6d30000), reset gpio108, supply ldob15; TX and VA macros.

Configuration

  • sheng-audio takes bits (sheng-adsp=2 is required): any nonzero value serves the audio service, 2 plays a 1.5 s 750 Hz test tone at boot, 4 brings up the microphone path and reads two seconds of capture, 8 runs the lock sound test.
  • sheng-audio-pa moves the DSP period buffer (default 0xd0000000); sheng-audio-tone-ms and sheng-audio-atten set the tone.
  • The payload addon carries hal/ShengAudio.driver under audio/; patches/xnu/zzyd-sheng-audio-addon-mount.patch mounts it over /System/Library/Audio/Plug-Ins/HAL. Its CDHash joins the owner trust cache.

Procedure

  • Read the [audio], [amps], [audio-glink], [audio-buffer], [mic], [mic regs] and [mic capture] log lines with scripts/read-sheng-log.py.
  • Read [audio-props], [audio-service], [audio-io], [audio-thread], [audio-clock], [audio-requests] and [audio-sync] for the iOS path: the plug-in's volume and peak, late and silent periods, write-ahead and read-behind, IO cycle intervals, clock skew, the audio server's IO thread scheduling and host clock, and the plug-in's last start and stop requests.
  • Run scripts/check-audioreach-host.py, scripts/check-cs35l43-host.py and scripts/check-geni-i2c-host.py after changing the codecs.

Development findings

  • The ADSP image loads in place from the split adsp and adsp_dtb files in /image of the modem_a FAT volume.
  • The DSP offers 24 intents of 128 bytes on adsp_apps and grants a 4 KiB one on request; the GLINK remote intent table keeps it.
  • The amplifiers play ASP RX1 directly (DSP bypassed, no amplifier firmware): top three slot 0, bottom three slot 1, gain 18 dB, 16-bit I2S at 48 kHz from the SoC as clock master.
  • The DSP takes buffers as a 32-bit address with the stream index (0x1001 & 0xf) in the upper word. The period buffer is low DRAM outside the kernel's window, mapped write-combined; the audio streams stay in SMMU pass-through.
  • Programming a translating SMMU context for the audio streams resets the board; the hypervisor keeps those context banks.
  • PRM commands go from the PRM port (2) and use PRM clock ids: the TX core clock is 0x30c. The macro and digital codec votes make the LPASS registers accessible.
  • The VA macro gates the TX SoundWire master's clock on SM8550; with its gate and the TX macro's on, the link comes up and the codec enumerates as device 1.
  • The primary microphone is AMIC3 on micbias 3 into ADC2, codec data port 1 channel 2, master port 2, TX decimator 0 from SWR_MIC1 and codec DMA TX3, mono 48 kHz.
  • iOS sees the plug-in device only with the Thunderbolt transport type: VirtualAudio makes it a driver device with one output and one input port. Built-in devices need Apple port descriptions, and a USB device needs a location the audio server does not take from a plug-in.
  • The plug-in and the service share one timeline anchored at boot on the 19.2 MHz host timer (400 ticks per frame, computed in 128-bit). Playback and capture rings hold 65536 frames; the service takes 480-frame periods and keeps four in flight.
  • The DSP answers a playback buffer once it has copied it, so the steady lead is measured per run (about 1600 to 2000 frames) and the published lead is at least 300 ms.
  • Capture answers carry a byte count and an offset; only those bytes are taken, and a short period is normal.
  • CoreAudio takes the host clock to be 24 MHz (one constant behind AudioGetHostClockFrequency). The plug-in finds that constant and the service sets it to 19.2 MHz on its shared page; without it the audio server's IO cycles ran 25 percent long and playback and recording stuttered.
  • The zero time stamps start at the frame IO starts on, every 65536 frames, with the raw clock algorithm. A period under the safety offset stops the audio server's IO.
  • The audio server's IO thread runs with a time constraint (priority 97, period 409600 ticks, computation 204800).
  • The audio server is arm64e on iPadOS 27: dlsym signs the pointer it returns, so the plug-in strips the signature to read the two instructions behind AudioGetHostClockFrequency and calls through the signed pointer. The VirtualAudio image-added rebind is built for arm64 only, because the slot pointers there are authenticated with their slot address and a plain rebind faults.
  • The audio overlay hides the system HAL plug-in directory from audiomxd. board.toml audio_system_plugins lists BTAudioHALPlugin.driver, and the runtime builder copies its unchanged bundle from the OS image. Other system plug-ins stay hidden.
  • Audio takes AppsSmmuLock and uses sm8550_gcc_update32 for its GCC clock vote when present, so the vote and shared SMMU setup do not race Bluetooth and touch.
  • The microphone rail request holds AppsSmmuLock because TCS 2 is shared with the USB rails. ADSP messages reach 560 bytes to fit UCSI.

Completion criteria

  • Each declared route carries the selected audio format through playback or capture and handles start/stop correctly.

Current limits

  • Inherited sheng implementation notes below are reference material, not an aston hardware qualification.
  • Playback sounds harsh in places; no speaker equalization or protection runs.

Next action

  • Qualify iOS playback and capture across apps and routes, then add speaker equalization and protection.

Evidence

Power

implemented

Components: AppleKeyStoreState.

Hardware and dependencies

  • The existing EPSS governor and bq28z610 read-only battery telemetry.

Configuration

  • sheng-cpufreq-boost-ms=80; sheng-cpufreq-frame-ms=0; sheng-battery=2.

Procedure

  • Compare idle, touch scrolling and sustained work with battery current and per-domain frequency logs.

Development findings

  • Aston previously inherited a 1000 ms maximum-frequency touch hold and a 300 ms GPU frame floor. The phone now requests an 80 ms hold and disables the frame floor.
  • These are board policy choices, not measured energy savings. The load governor and its existing downscale hysteresis remain active.
  • Device 2026-11-09 (boot 5, diag): the pack current stayed negative for the whole run and every [charger] and [usb] row was zero. Charging on this SoC is the battery manager in the audio DSP, which was never started.
  • sheng-adsp=2 now starts the DSP with the firmware of modem_a (read in place). The board's adspslpi region is 0x9ea00000 + 0x4880000 (devicetree and adsp.mdt span); sheng's 0x4680000 leaves one of the 57 load segments outside, so package/AdspProbe.cpp uses the aston size.
  • No driver of the fileset sets PE_halt_restart, so restart and power off spun until PEHaltRestart's 30 s timer panicked "Halt/Restart Timed Out" (boots 6 and 7; shutdown notifications finished in under 1 s). package/AstonHaltRestart.c issues PSCI SYSTEM_RESET or SYSTEM_OFF (the firmware reports PSCI 1.1 with both); a panic's own restart still stops on the panic screen. Charging (sheng-adsp=2) has no [adsp] row yet: that row is in the audio log group, now in the diag menu entry.
  • iOS showed 20 %, Android then 33 %: OPLUS restores the level it last showed from PMK8550 register 0x8250 (SDAM 0x82 + 0x50; observed under Android to follow the shown level) when it is within 20 of the gauge (oplus_chg_comm.c shutdown_soc). The iOS level was the gauge's own (bq28z610 RelativeStateOfCharge, the register Android reads as chip_soc). AstonSocBackup.c writes each change of the iOS level to that byte when the SPMI arbiter gives the apps EE write ownership and the byte already reads 0..100.

Completion criteria

  • Measured idle current and temperature stabilize, with no new frame stalls under the same workload.

Current limits

  • No thermal sensor feedback or deep idle implementation is added. L3 still uses the inherited maximum vote.

Next action

  • Collect temperature, current and frequency measurements before further tuning.

Evidence

Image assembly

implemented

Components: DarwinBoot, XnuPrelink, SoftwareAES.

Hardware and dependencies

  • Prelinked kernel, hardware DTB, generated ADT, trust cache, Data seed, platform identity and firmware DTB.

Configuration

  • Use aston_ipados27_defconfig and the package workflow; Ace3-UniXNU owns firmware and boot-image assembly.

Procedure

  • Build kernel and package, then payload with --unique-chip-id and the selected trust-cache/Data inputs.
  • Run ./unixnu bootimg -D ../UniXNU-sheng --uefi ../uefi-sheng --dtb <firmware.dtb> and inspect the assembly report.

Development findings

  • The build stages a hash-checked RamBlock copy before all consumers. source.json records its source revision and patch digest.
  • Existing Data seed identity is preserved; a normal rebuild must not request init_data or choose an erase action.
  • XnuPrelink keeps retained kexts at their collection addresses, so each native segment must fit before the first retained kext segment of its root. The iPhone17,2 24A446 collection, with this board's roots, leaves 11872 KiB for __TEXT_EXEC and __KLD, 1165 KiB for __DATA_CONST and __LASTDATA_CONST (to AppleARMPlatform), and 1872 KiB for __DATA (to Libm, which IOAVFamily needs). The iPad17,3 collection leaves 11968, 1571 and 1952 KiB.
  • The iPadOS build b5bb234 needs 11888, 1392 and 1920 KiB there. The iOS build leaves out the WiFi stack (893 KiB code, 75 KiB data), the embedded GPU firmware (151 KiB of constants, ASTON_GPU_FIRMWARE=1 restores it) and the sparse IOUserServer dispatch table (160 KiB of constants). An offline run of XnuPrelink's Layout with those sizes leaves 864, 93 and 16 KiB.
  • CI 42832522940's native kernel needed 1904 KiB of __DATA: BootLog's weak WiFi rings (native_wifi_text, _note, _pmic, _dump) still got zero defaults of their declared sizes (72 KiB) from the source storage build. package/WifiAbsent.c defines them with one element; a ld64.lld check shows the strong definitions drop the weak ones. Recomputed with section alignments, __DATA is 1840 KiB of 1872.
  • Prelink of that kernel against the iPhone17,2 kernelcache, run offline with emit_prelink.py: without reviewed pins 206 kext calls bind to the unmapped-call stub and 30 kext pointers stay unmapped. With UniXNU-ishtar's ported iPad17,3 pins, 23 and 25 remain. scripts/derive-fileset-pins.py adds 11 function and 9 data pins (e.g. task_is_privileged, csm_resolve_os_entitlements_from_proc, IOService::Start_Invoke, OSObject::metaClass, gIOParentMatchKey). It reads each unmapped target's kext slots at the same place in the iPad17,3 collection and takes the iPad pairing. With scripts/fileset-24A446-d94ap-symbols.json, 10 stubbed calls and 2 unmapped pointers remain. Each of those is unpaired on iPad17,3 too, where the b5bb234 build left 19 and 2 and booted.
  • iOS boot 7: Settings blocked at launch in SCNetworkInterfaceCopyAll, waiting in SCDynamicStoreNotifyWait for Plugin:InterfaceNamer. The kernel has no IONetworkingFamily, so configd's InterfaceNamer published only at its 300 s WaitStackTimeout (configd started at 55 s). [payload.runtime] overlay_tool (unixnu-patches/0003) writes the plugin's Info.plist with 5 s and 10 s timeouts into the GPU addon, and aston-interface-namer-overlay.patch mounts it over InterfaceNamer.bundle; BootLog's A row reports ifnamer=stage/status.
  • fairplayd.H2, the iPhone17,2 name of fairplayd.A2, relaunched every 10 s and blocked Freeform on com.apple.fairplayd.versioned; sheng-lite now disables it next to the A2 label.

Completion criteria

  • Kernel and prelink succeed, payload contains all declared inputs, and the boot image matches the bootloader format and size limit.

Current limits

  • Windows host checks do not compile the arm64e kernel. The pinned macOS/Xcode build and phone boot are pending.

Next action

  • Build the kernel and payload on macOS, then pack with the matching aston firmware.

Evidence

Diagnostics

implemented

Hardware and dependencies

  • Serial or framebuffer diagnostics, panic region, boot checkpoints and a recovery path.

Configuration

  • boot-progress=2 enables checkpoints; source and image revisions accompany every log.
  • On the iPadOS 27 native source path, sheng-log=1 plus boot-ui-trace=1 starts a separate process-capture worker. boot-ui-focus3=chronod adds the display-query client to backboardd and SpringBoard. Arm64e return-address PAC is stripped for diagnostic stacks.

Procedure

  • Run the existing check for the changed area and keep its command, output and input revision.
  • Record the last completed checkpoint and the first failing operation; link the captured output.

Development findings

  • The boot log records APFS Data capacity at mount separately from persistence state, with a status code for missing attributes or invalid geometry. It never substitutes invented capacity.
  • Files, fileproviderd and diskspaced exit reasons join the existing exit diagnostics. A successful block read does not prove that the storage UI or application containers work.
  • Windows native fixtures cover publication order, failure gates, CPU samples and XNU QoS rules; no WSL or QEMU is used.
  • Every restart panicked in vnode_rele_ext (usecount -ve, v_type VBAD) during vfs_unmountall's system pass on the sealed OS volume under the SystemOS union (dounmount, vflush, vclean, vnode_update_identity); sheng records the same panic on low-battery shutdown. patches/xnu-12377/aston-reboot-readonly-unmount.patch leaves read-only mounts mounted in that last pass; Data is unmounted and synced in the first pass as before.
  • scripts/pull-aston-data.py copies the changed Data pages from TWRP read only (hashes compared on the phone), scripts/build-aston-data-image.py lays them over the seed and scripts/apfs-read.py lists and extracts files (unencrypted APFS, stdlib), so iPadOS's own crash reports can be read after a test.

Completion criteria

  • Each claimed milestone has a reproducible command, a matching source/image revision and an accessible result.

Current limits

  • A Data mount capacity row is a boot-time snapshot, not live free-space accounting. No new device log is available.

Next action

  • Read [data mount capacity], [data] and sheng-exit records after testing the new image.

Evidence

arm64e-os

implemented

Hardware and dependencies

  • iPadOS 26.5.2 build 23F84 IPSW for iPad17,1 to iPad17,4; target identity iPad17,3 J820AP.
  • Public XNU source xnu-12377.121.6; arm64e profile requirements in https://unixnu.com/porting/arm64e.

Configuration

  • iPadOS 27.0.1 (24A446) on XNU 12377.121.6 with the selected ordered compatibility patches.

Procedure

  • Check FEAT_PAuth, EL1 key register writes and PAC instruction behavior on all eight CPUs.
  • Build the arm64e kernel from xnu-12377.121.6 with the ported 4 KiB patches, then prelink the 23F84 kexts.
  • Follow the bring-up order: arm_init, bsd_init, root mount, launchd, SpringBoard.

Development findings

  • The CPU and Data diagnostic patches apply after the existing device series. They do not change the XNU source pin or claim a new OS ABI qualification.

Completion criteria

  • The arm64e kernel reaches launchd on sheng from the 23F84 root volume, then SpringBoard.

Current limits

  • Full kernel link and device ABI qualification remain pending for this branch.

Next action

  • Build the current OS profile on the pinned Xcode runner.

Evidence