Skip to content

arm64: dts: apple: t8122-j615: Show the software camera indicator (slot 3) - #2717

Merged
RyanTheTide merged 1 commit into
integration/aurorafrom
airs-camera/j615-camera-indicator
Nov 22, 2026
Merged

RyanTheTide merged 1 commit into
integration/aurorafrom
airs-camera/j615-camera-indicator

Conversation

@RyanTheTide

@RyanTheTide RyanTheTide commented Nov 21, 2026 •

Copy link
Copy Markdown
Contributor

Problem. The 15-inch M3 MacBook Air (J615) has no camera indicator light. #2561 added a best-effort software indicator that the DCP draws in a reserved swap slot, but J615 doesn't opt in.

Change. DT only, J615 only: apple,camera-indicator-surface = <3> on the DCP and apple,camera-indicator = <&dcp> on the ISP. The J615 DCP has no apple,iomfb-surfaces, so on 27.2 firmware DRM owns slots 0 and 1, and the indicator needs slot 2 or 3. Slot 2 does not show under the desktop (the same result as on the J700, #2651). Slot 3 does.

Testing. J615 on its production kernel 7.3.0-aurora-fleet78 (carries #2561 and the J615 DCP driver), stock modules, default command line. Each trial was a one-shot boot whose only change was these two properties in the J615 DTB. The rows read the DCP blend CRC with the camera off, streaming, and off again, under GNOME, with the panel backlight at 0.

  • Slot 2: on == off in round 1. In rounds 2 and 3 the CRC moved with desktop redraws but did not return at stop: no indicator.
  • Slot 3: the CRC changes while the camera streams in 3/3 rounds, and the on CRC is identical in rounds 2 and 3. At stop it returns exactly to the off CRC (rounds 1 and 3; in round 2 the desktop redrew once mid-round, and that new off CRC is round 3's start and end).
  • DPMS off/on with the camera running, the camera started while the display is off, and s2idle with the camera running all keep the indicator, and stop clears it (arm64: dts: apple: t8122-j615: Show the software camera indicator (slot 3) #2717 (comment)). While streaming, the indicator is a 48x48 plane at the panel's top-right corner in DRM plane state.
  • Both trial boots log best-effort software camera indicator via the DCP at ISP probe, with 0 errors, 0 failed units, and SELinux enforcing.
  • This DTB builds. dtbs_check reports 0 warnings, the same as the base. checkpatch --strict is clean. Built from the J615's own production DTB source (fleet78 + arm64: dts: apple: t8122-j613-j615: describe the Touch ID data-ready line #2678) plus this commit, the DTB differs from the one it boots only by these two properties and the DCP phandle they need.

Still to run.

  • A photo of the panel. The rows above are blend-CRC only.
Notes for review
  • Slot 3 sits above both slots DRM uses here. The binding requires a slot above the highest enabled one, and dcp_indicator_probe() enforces that.
  • The 13-inch J613 boots 26.6.2 firmware, where DRM uses only slot 0. It is not opted in here because it has not been tested.
  • The J713 (T8132) has the same DCP generation profile (not H17G), so the indicator can run there too. Its DT opt-in is a separate change once it has a row.

Assisted-by: LLM (Claude:Opus 5.5)

…ot 3)

The 15-inch M3 MacBook Air has no camera indicator light. Opt it into
the best-effort software indicator: apple,camera-indicator-surface on
the DCP and apple,camera-indicator on the ISP.

On 27.2 firmware the J615 DCP has no apple,iomfb-surfaces, so DRM owns
swap slots 0 and 1 and the indicator needs slot 2 or 3. With slot 2 the
indicator does not change the composited frame under the desktop; with
slot 3 it does, and the frame returns to the camera-off one at stop.

Signed-off-by: Ryan Murray <ryan@aurorasilicon.org>
@RyanTheTide

Copy link
Copy Markdown
Contributor Author

Rows from "Still to run": J615, 7.3.0-aurora-fleet78, default Stage 2 with only this PR's two properties added to its DTB (one-shot boot), stock modules, GNOME session, panel backlight at 0.

  • Plane state: while the camera streams, a second plane joins the CRTC (plane_mask 1 -> 5): a 48x48 AR24 framebuffer at +2820+12, the top-right corner of the 2880-wide panel.
  • Blend CRC, camera off / on / off: on differs 3/3. The on CRC is identical in rounds 2 and 3, and off returns exactly in 2/3 (in round 1 the desktop redrew).
  • DPMS off/on with the camera running, 3/3: the indicator is in the frame after DPMS on and gone after stop. The desktop's own CRC changes across a DPMS cycle, so I compare against the stop.
  • Camera started while the display is off, 3/3: shown after DPMS on. It clears at stop (exact in 2/3; in row 2 the desktop redrew).
  • s2idle for 20 s (PMU RTC) with the camera running: the same on CRC before and after resume, and exactly the off CRC after stop.
  • 0 swap failures, dropped presents, oopses or warnings in the kernel log.

@RyanTheTide

Copy link
Copy Markdown
Contributor Author

Merge-train adversarial review (MERGETRAIN-LINUX2) at 75bd323: NO BLOCKERS.

Code read only, nothing built here. Before the merge it still needs Dj's verdict at this head; the missing rows are listed below.

Full review

#2717 arm64: dts: apple: t8122-j615: Show the software camera indicator (slot 3) @75bd3233

VERDICT: NO BLOCKERS

The change is DT only and J615 only: arch/arm64/boot/dts/apple/t8122-j615.dts:21-25 and :37-39. &isp (t8122.dtsi:678) and &dcp resolve. The J615 DCP has no apple,iomfb-surfaces, so with 27.x firmware the driver enables slots 0 and 1 (drivers/gpu/drm/apple/dcp.c:7375-7376), and with 14.7/26.6 firmware only slot 0 (dcp.c:7372-7373). Slot 3 passes the surf <= top_surf check in dcp_indicator_probe() (dcp-indicator.c:61) on both firmware lines. The binding allows a maximum of 3 (apple,dcp.yaml:195-198). There is no gate, param or provenance problem.

Nits:

  • apple,t8122-isp is not in any binding (no Documentation/ hit), so dtbs_check does not validate apple,camera-indicator on this node. "0 warnings" means no schema ran here, not that the node passed one. This predates the PR but is worth a follow-up.
  • dcp_indicator_probe() returns -EINVAL from DCP probe (dcp.c:7405-7408) if a future loader or firmware enables slot 3 in apple,iomfb-surfaces. A best-effort indicator would then take down the internal display. This is the driver's behaviour from drm/apple, isp: add a best-effort camera indicator (unbooted) #2561, not this PR's, but this PR is the first non-J700 opt-in that depends on it.
  • The DT comment says slot 2 "does not show under the desktop" without a cause. If that is a DCP z-order or blend rule, slot 3 could also hide when DRM starts using slot 1 as an overlay (fullscreen video, direct scanout). No such row exists.

Missing rows: a panel photo, as the PR itself lists. A row with DRM actively using slot 1 (an overlay plane in use) while the camera streams. J613 is not opted in, so no row is needed there.

@DjDeveloperr DjDeveloperr left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Adversarial review at 75bd32335ca4 against pre-pr-checklist.md.

Checklist verification

  • §1–§3 (No model gating / production readiness / DT correctness): Clean. DT-only opt-in on arch/arm64/boot/dts/apple/t8122-j615.dts (apple,camera-indicator-surface = <3> on &dcp, apple,camera-indicator = <&dcp> on &isp), matching the DT-driven design from #2561 and #2651 with no kernel gates, cmdline parameters, or hacks. Verified against drivers/gpu/drm/apple/dcp.c:7372-7376, dcp-indicator.c:55-69, dcp-internal.h:48 (DCP_MAX_PLANES = 4), and Documentation/devicetree/bindings/display/apple,dcp.yaml:195-204: without apple,iomfb-surfaces, DRM owns slots 0–1 on 27.x firmware (top_surf = 1) and slot 0 on 14.7/26.6 (top_surf = 0), so slot 3 passes surf < SWAP_SURFACES && surf > top_surf on both firmware lines.
  • §4 & §6 (Attribution / clean-room / naming): Clean. Human author/committer (Ryan Murray <ryan@aurorasilicon.org>) and human Signed-off-by: only; zero AI trailers in the commit; Assisted-by: present in the PR body.
  • §5 (CI & hardware evidence): Complete on the live head. dt-checks and host-tests are green. J615 hardware runs on 7.3.0-aurora-fleet78 (stock modules, default cmdline, DTB verified to differ only by these two properties + phandle) cover ISP probe, DRM plane state (48×48 AR24 at +2820+12, plane_mask 1→5), blend-CRC off/on/off 3/3 (differentiating slot 3 from slot 2), DPMS off/on 3/3, camera start while display off 3/3, 20 s s2idle with camera streaming, and full regression (0 errors, 0 swap failures, 0 dropped presents, 0 failed units, SELinux enforcing).

Non-blocking follow-ups

  • A physical panel photo and a test with DRM actively using overlay slot 1 (direct scanout / fullscreen video) while streaming would be nice follow-ups to confirm slot 3 stays above active slot 1.
  • Pre-existing: apple,t8122-isp has no DT schema yet, so dtbs_check only validates &dcp; also, if a future loader ever enables slot 3 in apple,iomfb-surfaces, dcp_indicator_probe() returning -EINVAL fails DCP probe (#2561 behavior, worth softening in a separate driver follow-up).

Assisted-by: LLM (Gemini 4 Neon)

@RyanTheTide
RyanTheTide merged commit 527ae57 into integration/aurora Nov 22, 2026
2 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants