Skip to content

drm/apple: add native HDMI-CEC and 2 ms IRQ_HPD pulse tracking - #2230

Merged
DjDeveloperr merged 2 commits into
aurora-silicon:integration/aurorafrom
astaff:t6041/j616c-hdmi-cec
Nov 21, 2026
Merged

DjDeveloperr merged 2 commits into
aurora-silicon:integration/aurorafrom
astaff:t6041/j616c-hdmi-cec

Conversation

@astaff

@astaff astaff commented Nov 19, 2026 •

Copy link
Copy Markdown
Contributor

This branch starts from integration/aurora and merges cleanly on top of #1954, #2216, and #2389 with zero conflicts.

Summary

This change adds HDMI-CEC adapter support (CONFIG_DRM_APPLE_CEC) for built-in HDMI ports behind the compact EPIC CEC service (dcpav-cec-interface-epic on RTKit endpoint 0x2b):

  • Add CONFIG_DRM_APPLE_CEC: Select CEC_CORE when enabled and register a standard Linux /dev/cecX adapter per fixed HDMI-A connector when endpoint 0x2b is active (CEC_CAP_DEFAULTS | CEC_CAP_CONNECTOR_INFO | CEC_CAP_NEEDS_HPD, 9500 ms transfer timeout). Endpoint 0x29 on 14.x firmware remains unsupported pending a capture of its report layout.
  • Preserve bare compact EPIC service-call reports: Keep the 64-byte struct epic_service_call header when the compact subheader is followed directly by a service-call header. Continue to skip the eight-byte descriptor on wrapped reports. Add afk_service_retire_channel() so retiring a terminated service does not disable a singleton slot reused in place for a later announcement.
  • Implement CEC calls and reports on endpoint 0x2b. Group 0 calls open (4), close (5), set the logical address mask (6), read EDID attributes (7), and transmit a frame (9, 7500 ms timeout). The mask includes the broadcast address while enabled and clears on disable. Group 2 reports carry termination (1), a received frame (2), HPD state (3), or EDID attributes (4).
  • Keep CEC lifecycle, HPD-low close, and TX completion safe: Read the physical address directly because HPD status can still read low on plug, poll physical address after fixed-connector hotplug EDID read, close the active CEC client before DPTX HPD low, drain all CEC works on system_dfl_wq before cec_unregister_adapter() outside dcp_typec_fabric_lock, keep cec->srv pinned across DRM component unbind/rebind, and guard TX completions with a per-transfer sequence and 8500 ms completion window.
  • Track <= 2 ms HDMI IRQ_HPD pulses without 500 ms masking: Enable dcp_hdmi_init() for every DCP with an hdmi-hpd GPIO and a fixed HDMI-A connector, suppressing further requests if firmware refuses DPTX_CMD_SINK_IRQ. Classify <= 2 ms HPD low pulses in hard IRQ context with a 2 ms hrtimer so the IRQF_ONESHOT threaded handler does not mask converter IRQ_HPD pulses, and wake the threaded HPD handler via irq_wake_thread() when HPD stays low past 2 ms.
  • Add KUnit and host selftests: Add tests in afk-parser-test, apple-dcp-cec, apple-dcp-cec-adap, desktop.py, and extract-hpd.py.

Testing

  • ./scripts/checkpatch.pl --strict -g HEAD: 0 errors, 0 warnings, 0 checks.
  • tools/testing/selftests/drivers/gpu/apple-dcp/run.py: 88/88 passed.
  • tools/testing/selftests/drivers/gpu/apple-afkv2/run-afkv2-tests.sh: all checks passed.
  • tools/asahi/run-host-tests.py --only dcp: 27/27 passed.
  • arm64 QEMU KUnit (CONFIG_PROVE_LOCKING=y, CONFIG_DRM_APPLE_KUNIT_TEST=y): 380 passed (10/10 in apple-dcp-cec and apple-dcp-cec-adap, 9/9 in apple-epic-parser), 0 failed, 0 lockdep or kernel warnings.
  • Hardware testing on J616c (T6041, MacBook Pro 16" M4 Max, firmware 27.2) with an HDMI-CEC sink (BenQ PJ) diagnosed and verified the two initial PA issues fixed here (bare compact epic_service_call report framing on endpoint 0x2b and HPD status reading low on plug).

@RyanTheTide

Copy link
Copy Markdown
Contributor

Thanks for the CEC work and the thorough KUnit coverage.

One thing we would like changed: in this tree we select behaviour by a hardware property, not by the list of SoCs where it has been tried. dcp_hdmi_init() now admits HDMI by dcpext compatible (drivers/gpu/drm/apple/dcp-hdmi.c:330-333, used at :347: apple,t6030-dcpext || apple,t6041-dcpext). The same function already checks what actually describes a built-in HDMI port: dcp->hdmi_hpd (the HPD GPIO from the DT) and fixed_connector_type == DRM_MODE_CONNECTOR_HDMIA.

Could you drop the compatible list and rely on those two properties, so every DCP with a described HDMI port gets the IRQ_HPD pulse tracking and /dev/cec0? If there is a DCP generation where this must not run (different AV endpoint, firmware without the CEC service), please gate on that instead and say why in a comment. We can run rows on our M1 Pro, M2, M4 Pro and A18 Pro machines if you make it the default.

@RyanTheTide

Copy link
Copy Markdown
Contributor

Review of 1266e4a (read against integration 0abc2f3). No hardware row yet: the J616s here has nothing on its HDMI port, so I'm still building this commit for a registration/PA/TX-without-sink row.

Blocking

  1. A TX completion can land on the wrong message (cec.c dcp_cec_tx_work, cec-proto.h:52-53). xfer_timeout_ms is 8000, but tx_work can spend up to 1000 ms in Open plus 7500 ms in sendCEC. When the core times out (or the adapter is disabled mid-transmit), it clears transmit_in_progress, and the next adap_transmit overwrites tx_msg. The old work then calls cec_transmit_attempt_done() and completes the new message with the old status. Suggest a per-transmit sequence number, completing only if it is still current, and either opening outside the TX path or setting the timeout to open + send + margin.
  2. 0x29 DCPDPCECInterface uses the 0x2b report table (cec-proto.h:39-49, cec.c:802-858). The existing comment in av.c describes the 14.x reports on that interface as physical address, link training, and link start/stop. If report 2 there is a link event, it gets decoded as an RX CEC frame and handed to userspace followers. Could you share the capture each report table comes from, or keep 0x29 unsupported until it is captured?
  3. Terminate followed by a re-announce retires the fresh instance (cec.c:401-441). afk_compact_slot() reuses an enabled singleton slot in place, so after Close returns, afk_service_retire(term_srv) hits the newly announced service. Suggest saving the channel when taking term_srv and retiring only if it is unchanged, or leaving the retire to the compact teardown.
  4. Teardown order (cec.c:714-722). cec_unregister_adapter() runs before cancel_work_sync() of the tx/rx/lifecycle works, so the stopping checks leave a window where a work item touches a freed adapter. Set stopping, cancel all three, then unregister.

Should fix

  • There is no Close on HPD low. On 14.7 the firmware holds the HPD-low reply for up to 10 s while a client is open (see dpavservep.c).
  • dcp_unlink holds dcp_typec_fabric_lock across the CEC unregister. That is up to 1 s of AFK call plus up to 7.5 s of TX cancel, and it blocks hotplug on every port. dcp_comp_unbind already unregisters outside the lock.
  • dcp-hdmi.c 2 ms timer:
    • unplug is now handled from unplug_work;
    • pulses under 100 µs are dropped;
    • the service work moved from system_freezable_wq to system_unbound_wq, so it can reach DPTX during suspend;
    • a rising edge near 2 ms can be counted twice, by the timer and by the IRQ.
  • After a component rebind the adapter has no service until the next relink, because the pin is dropped at unbind.
  • .reusable = true is newly added to the audio/video singletons with no explanation.

Structure

  • dcp_hdmi_compatible() is a compatible list (t6030/t6041 dcpext). The hardware property is already in the DT: hdmi-hpd-gpios plus the fixed HDMI-A connector. Keying on that covers M1/M2 Pro, the minis, the Studios and the Mac Pro too.
  • The ops table is picked from a firmware/profile list (cec.c:1019-1021). Picking it by which interface the firmware actually announces would be simpler.
  • The adapter is registered for every fixed HDMI-A connector, but on paths where the firmware never announces a CEC interface (14.7 non-native, and so far every M1/M2 machine) userspace gets a /dev/cecN that never works. Registering on the first CEC announce avoids that.
  • The CEC ops are declared in three places (av.c, cec.c, dcp-v27-sleep.inc), and there is a second AV endpoint starter next to avep_init.
  • The names DCPAVCECInterfaceEPClient, setLogicalAddressMask, getEDIDAttributes and getHPDStatus are not the names the firmware announces (dcpav-cec-interface-epic, DCPDPCECInterface). Where do the command numbers, body offsets (+0x10, +0x20, +0x30), DPCD 0x300e and the 0xe0118000/0xe0118001 codes come from? A note on the capture would help reviewers.

These check out: Kconfig is default y, there are no module parameters, adap_log_addr adds the claimed LA to the mask, RX length is bounded at 16, report sizes are checked, and RX is delivered from a work item.

@RyanTheTide

Copy link
Copy Markdown
Contributor

Thanks for this, it fills the CEC gap on the HDMI machines. Review of 1266e4a against integration/aurora 0abc2f3 (it applies on the tip as is):

  1. Compact (ep 0x2b) report layout. dcp_cec_parse_report() requires a 0x40-byte EPIC call header with the ipcx magic and group 2 at the start of data. On the compact path, afk_recv_compact_message() passes the payload after the sub-header and, for command-style reports, after struct epic_compact_desc. Compact calls carry the same call header after the descriptor (afk_recv_handle_compact_call() checks the magic), so this is probably right. The KUnit cases only build the header-first layout, though, and the hardware result in the description covers adapter registration only. A dyndbg hex dump of one HPD or EDID-attributes report on 0x2b would settle it, e.g. echo 'file cec.c +p; file afk.c +p' > /sys/kernel/debug/dynamic_debug/control plus a replug. If the header is missing there, every 0x2b report gets -EINVAL and the physical address never arrives through the report path.
  2. Reuse the shared header definitions. cec-proto.h redefines the EPIC magic (DCP_CEC_EPIC_MAGIC == EPIC_SERVICE_CALL_MAGIC in afk.h) and parses group/command/length by raw offsets. struct epic_service_call from afk.h (as av_report_terminate() uses) would keep the two from drifting apart.
  3. Default behaviour on the Mac mini (J616s, apple,t6041-dcpext, 26.x desktop). Two things change with nothing configured. AV endpoint 0x2b now starts on HDMI external DCPs without hdmi_audio (before, the endpoint stayed down unless hdmi_audio was set). And dcp_hdmi_init() now arms the 2 ms IRQ_HPD hrtimer on that compatible. Both are reasonable, but neither has run on a J616s yet, and the J616c is a 27.2 machine. We will run it there and post the numbers: /dev/cec0 and physical address versus EDID, TX status with no CEC follower, hotplug x5, s2idle x3 with HDMI attached, DCP/RTKit/DART fault count.
  4. Minor: dcp_cec_init_avep() adds a devm cleanup action on every successful start. dcp_start() and dcp_external_audio_work() are each one-shot today, so this is fine as long as the endpoint is never started twice per bind. A short comment saying so would help the next reader.

The adapter is selected by a hardware property (fixed HDMI-A connector from the DT, plus the RTKit endpoint being present), so every HDMI machine whose DCP publishes the interface gets it without a board list.

@RyanTheTide

Copy link
Copy Markdown
Contributor

J616s (M4 Pro 16", T6040, DCP 26.x desktop external path), this commit on a fleet kernel based on 48cac48. Nothing is plugged into the HDMI port.

  • /dev/cec0 registers at boot on dcpext0 (3a2e00000.dcp):
    • caps 0x35e, including Needs HPD and Connector Info, which points at card1 HDMI-A-1;
    • 4 logical addresses, CEC 2.0.
  • The firmware's dcpav-cec-interface-epic announcement on ep 0x2b is matched at 3.7 s.
  • PA is f.f.f.f. cec-ctl --playback is accepted, and the LA stays unallocated until HPD.
  • CEC_TRANSMIT of Image View On and Standby to the TV returns ENONET, because the core refuses without HPD. This row therefore says nothing about sendCEC, ACK/NACK or RX. That needs a CEC sink.
  • One s2idle cycle (RTC wake) is clean, and /dev/cec0 is still there afterwards.
  • SELinux enforcing with 0 AVCs, and 0 failed units.
  • One new kernel warning at boot: workqueue: work func dcp_cec_lifecycle_work [appledrm] enqueued on deprecated workqueue. Use system_{percpu|dfl}_wq instead. All six queue_work(system_unbound_wq, ...) calls in cec.c should move to system_dfl_wq.

@astaff

astaff commented Nov 19, 2026 •

Copy link
Copy Markdown
Contributor Author

Thanks for the thorough review and the J616s boot/s2idle test report, @RyanTheTide!

We also tested 1266e4a7d273 on J616c (T6041, firmware 27.2) with a live HDMI-CEC sink (BenQ PJ projector), which exposed two firmware behaviors that directly confirmed your compact report-framing question:

  1. Compact 0x2b report framing (your Comment 3 mux: apple-display-crossbar: replace the T602x selector fields instead of OR-ing them #1): On t604xdcp, AFKEPClient::sendReport (flags=2, non-blocking EPIC_SUBTYPE_STD_SERVICE report) puts struct epic_service_call (magic = EPIC_SERVICE_CALL_MAGIC at +0x0c) directly after struct epic_compact_sub_hdr with no 8-byte struct epic_compact_desc. afk_recv_compact_message() was stripping 8 bytes whenever !afk_compact_report_dictionary(), which cut the start off struct epic_service_call and caused dcp_cec_parse_report() (and Terminate) to drop every async EDID_ATTRS, HPD, and RX_FRAME report. afk_recv_compact_message() now preserves bare struct epic_service_call reports alongside dictionaries.
  2. getHPDStatus on plug + DDC EDID timing: t604xdcp runs DCPDPService::setHPDStatus(1) before publishCECInterfaceGated() allocates DCPDPCECInterface, leaving getHPDStatus (cmd 10) at 0 on initial plug, and announces dispext0:dcpav-cec-interface-ep ~180 ms before DDC EDID completes. dcp_cec_lifecycle_open() now queries getEDIDAttributes directly (dropping getHPDStatus), accepts async EDID_ATTRS reports, and polls getEDIDAttributes after fixed-connector hotplug EDID read.

Force-pushed b821247e62b9 resolving all review items and hardware findings in one commit (merging cleanly with #2389):

  • Converter / DT gating (dcp-hdmi.c): Dropped dcp_hdmi_compatible(); dcp_hdmi_init() now keys on dcp->hdmi_hpd && fixed_connector_type == DRM_MODE_CONNECTOR_HDMIA and marks hdmi->unsupported if firmware refuses DPTX_CMD_SINK_IRQ.
  • TX timeout and stale completion (cec.c, cec-proto.h): Removed Open from tx_work, set sendCEC timeout to 7500 ms, completion guard window to 8500 ms, and adap->xfer_timeout_ms to 9500 ms, and gated completion on a per-transfer tx_seq incremented on adap_transmit, adap_enable(false), HPD-low close, Terminate, and unregister.
  • 14.x endpoint 0x29 (av.c, dcp.c): Left DCPDPCECInterface on 0x29 and dcp_external_audio_work() untouched until 0x29 frames are captured on hardware.
  • Singleton reuse vs. Terminate (afk.h, cec.c): Added afk_service_retire_channel(srv, chan) and saved service->channel when Terminate arrives so an in-place singleton reuse by afk_compact_slot() keeps its new channel enabled.
  • Unregister ordering, fabric lock, and rebind (cec.c, dcp-fabric.c): dcp_cec_unregister() drains tx_work, lifecycle_work, and rx_work before cec_unregister_adapter(), runs outside dcp_typec_fabric_lock in dcp_unlink(), keeps cec->srv pinned across component unbind/rebind, starts compact 0x12 (afk_start_service) on ep->wq before publishing cec->srv, and avoids calling dcp_cec_register() from ep->wq.
  • Close before DPTX HPD low (dptxep.c, cec.c): Added dcp_cec_close_port() in dptxport_release_display() and dptxport_set_hpd_timeout(!hpd) (plus RPT_HPD false), with close_gen cancelling any concurrent Open.
  • 2 ms hrtimer and IRQ_HPD edges (dcp-hdmi.c, dcp-hdmi.h): Removed unplug_work in favor of irq_wake_thread(dcp->hdmi_hpd_irq, dcp) (with synchronize_irq() in dcp_hdmi_disable()), stopped dropping < 100 us pulses, restored system_freezable_wq in dcp-hdmi.c, moved all cec.c workqueues to system_dfl_wq, and added time-bounded claimed_ns so pulses near 2 ms are counted once without swallowing later fast pulses.
  • Deduplication and .reusable (cec-proto.h, cec.h, av.c, dcp-v27-sleep.inc): Reused struct epic_service_call / EPIC_SERVICE_CALL_MAGIC from afk.h, shared DCP_CEC_COMPACT_SERVICE_OPS across av.c, cec.c, and dcp-v27-sleep.inc, dropped .reusable = true from compact singletons, gated /dev/cecN registration on dcp_cec_has_endpoint(), and documented the t604xdcp method/DPCD provenance and avep_cleanup_registered guard.

@astaff

astaff commented Nov 19, 2026

Copy link
Copy Markdown
Contributor Author

Hardware-verified 85639ff63ce4 on J616c (T6041, MacBook Pro 16" M4 Max, firmware 27.2) with the BenQ PJ HDMI-CEC sink connected on 3a2e00000.dcp (card2-HDMI-A-1):

  • Physical Address resolves automatically to 1.0.0.0.
  • cec-ctl --playback --osd-name MacBookPro claims Logical Address: 4 (Playback Device 1) (Logical Address Mask: 0x0010).
  • cec-ctl -S completes bidirectional CEC TX/RX over the 2 ms hrtimer IRQ_HPD path and bare compact 0x2b RX_FRAME reports, discovering device 0 (TV) at 1.0.0.0 (CEC Version: 1.3a, Power Status: On).
  • 0 kernel, workqueue, DCP, RTKit, or DART warnings.

@astaff

astaff commented Nov 19, 2026

Copy link
Copy Markdown
Contributor Author

Also ran the Linux V4L2/CEC adapter compliance suite (cec-compliance -d /dev/cec0 -A) live on J616c (3a2e00000.dcp) against the BenQ PJ HDMI-CEC sink:

Find remote devices:
	Polling: OK

CEC API:
	CEC_ADAP_G_CAPS: OK
	Invalid ioctls: OK
	CEC_DQEVENT: OK
	CEC_ADAP_G/S_PHYS_ADDR: OK
	CEC_ADAP_G/S_LOG_ADDRS: OK
	CEC_TRANSMIT: OK
	CEC_RECEIVE: OK
	CEC_TRANSMIT/RECEIVE (non-blocking): OK (Presumed)
	CEC_G/S_MODE: OK
	CEC_EVENT_LOST_MSGS: OK

Total for apple-dcp device /dev/cec0: 11, Succeeded: 11, Failed: 0, Warnings: 1

(Warnings: 1 is the projector rate-limiting replies during CEC_EVENT_LOST_MSGS's 58-message back-to-back GET_CEC_VERSION flood; 0 kernel warnings in dmesg.)

@RyanTheTide

Copy link
Copy Markdown
Contributor

Thanks for 85639ff and the J616c results with a real CEC sink, including the compliance run. Selecting by the HDMI-A connector and HPD GPIO, and the fixes for the TX sequence, the 0x29 interface, the retire channel, the teardown order and the workqueue, cover what we raised. It still needs to land after #1954.

@RyanTheTide

Copy link
Copy Markdown
Contributor

Thanks again for the CEC work; #1954 has landed, so this is close.

One small request before it goes in: a few comments describe the firmware's internal classes, methods and dispatch tables rather than the interface on the wire, at cec-proto.h:7-8, cec.c:414, afk.c:1211 and afk-parser-test.c:152, and the same names appear in the commit message and the PR description. Could you describe these by what the driver observes instead (message layout, report order, timing)? The announced service names the driver matches on are fine as they are.

Register a CEC adapter for fixed HDMI-A connectors served by the
compact EPIC CEC endpoint (0x2b). Accept both bare and
descriptor-wrapped service-call reports. Use group 0 calls to open
and close the service, set logical addresses, read the physical
address and transmit frames. Handle group 2 reports for received
frames, HPD, EDID attributes and service termination.

HPD status can still read low on plug. Query the physical address
on open and refresh it on an EDID attributes report or after the
hotplug EDID read. Close the client before forwarding HPD low and
drain pending work before unregistering the adapter. Track each
transmit by sequence so a late reply cannot complete a later frame.

Classify short HPD pulses in hard IRQ context with a 2 ms timer.
Wake the threaded handler for a sustained low. Add parser, adapter
and HPD test coverage.

Signed-off-by: Artyom Astafurov <artyom.astafurov@gmail.com>
@astaff
astaff force-pushed the t6041/j616c-hdmi-cec branch from b821247 to 7577337 Compare November 20, 2026 03:21
@RyanTheTide

Copy link
Copy Markdown
Contributor

Thanks for 7577337. The comments in cec-proto.h, afk.c, cec.c and the tests, the commit message and the description now describe the message layout, report order and timing the driver sees, which is what we asked for. With #1954 in, CI green and the J616c CEC sink row, this is ready for the next step.

@DjDeveloperr

Copy link
Copy Markdown
Contributor

Cross-board review of 7577337 against integration/aurora e0ee46a. It merges cleanly. Integration has moved under dcp-fabric.c, dcp.c and dcp-v27-sleep.inc since this PR's CI run, so the combination is being built on today's base in #2532 (CI only). Ryan's review covers the CEC and IRQ_HPD design; this comment only covers what the series changes outside the HDMI-CEC boards.

  1. Our machines (J700 T8140, J714s T6050, J773g T8132). None of their DCPs has apple,connector-type = "HDMI-A" with hdmi-hpd-gpios. So dcp_hdmi_init(), dcp_cec_init_avep(), dcp_cec_register() and dcp_cec_service_init() all return early. On firmware-27 external DCPs, swapping the CEC entry in dcp_v27_ext_av_services changes nothing for non-HDMI ports: the old init was a dev_dbg, the teardown is still afk_service_disable(), and the AFK core still starts the service.
  2. The afk.c framing change is not CEC-only. On every compact AV role in av_desktop_audio_role() (26.x desktop external, the J700's H17P USB-C DCP, firmware-27 external DCPs such as the M5's Type-C ports), a bare service-call report now reaches av_audiosrv_report() whole. That makes its Terminate check (av_report_terminate()) live there for the first time. Before, stripping 8 bytes put the magic at the wrong offset, so it never matched. This is probably what that handler was written for, but it changes how the audio interface is closed on unplug or a sink change on the USB-C displays. We'll run M5 external-display rows after the merge: hotplug, DP audio, and deep sleep with displays attached.
  3. IRQ_HPD tracking now runs on every HDMI board. On the base, dcp_hdmi_init() admitted only apple,t6030-dcpext. It now runs on every board with a described HDMI port: T600x J314/J316/J375, T602x J414/J416/J474/J475/J180d, T603x J514/J516, T6040 J614s/J616s, T6041, T6050 J775d, T8103 J274 and T8112 J473. On all of them, a falling HPD edge with the link up is held by the 2 ms hrtimer before the presence thread runs, and on unplug dcp_dp2hdmi_record_edge() now starts 2 ms later. This is what was asked for. @RyanTheTide, your M1 Pro / M2 / M4 Pro rows (hotplug ×5, s2idle with HDMI attached, an HBR3 mode where the sink has one) would be the post-merge check for it.
  4. Minor: afk_compact_av_capable() (afk.c) has the same condition as av_desktop_audio_role() (av.c:304). A single helper would keep the two from drifting apart. This doesn't block anything.

On our side, this goes in after a J700 boot/panel/deep-sleep row on a kit with the series. CONFIG_DRM_APPLE_CEC defaults to y, so appledrm now pulls in CEC core on every machine, and that row checks the J700 boot path with it.

@RyanTheTide

Copy link
Copy Markdown
Contributor

On the HDMI side of the cross-board review: we will run the post-merge rows on our M1 Pro, M2 and M4 Pro with HDMI attached: hotplug five times, s2idle with the sink connected, and an HBR3 mode where the sink offers one. We will post them here.

@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.

Re-review of 7577337. Integration had moved 332 commits since this PR's CI run, including dcp.c, iomfb.c, dcp-internal.h and the drm/apple Makefile, so I refreshed the branch with gh pr update-branch. That added merge commit 27c69b4 on integration/aurora 024cec8 and left your commit unchanged. CI is green on 27c69b4: host-tests, source, kunit, rust-build and DWC3/ATC.

Since then integration has taken 31 more commits. One of them touches a file in this PR: af33408 (#2602) reworks dcp_v27_sleep_settle() in dcp-v27-sleep.inc. That's a different function from your two hunks there (the CEC check in dcp_v27_ext_busy() and DCP_CEC_COMPACT_SERVICE_OPS in dcp_v27_ext_av_services[]), and the two share no symbols. The merge is clean, and a CEC report that arrives during the settle is now taken and the checks rerun, which is what we want.

Nothing has changed in the series since the 06:42 cross-board review, and nothing there needs a change from you:

  • J700, J714s, J773g: no DCP on these has an HDMI-A connector with hdmi-hpd-gpios, so the CEC and IRQ_HPD paths all return early.
  • Framing: with the bare service-call framing in afk.c, av_report_terminate() now actually runs on the compact AV roles, as it was written to.
  • Combination build: together with #1951 and #2492 it built and passed CI in #2532.
  • Nit: afk_compact_av_capable() and av_desktop_audio_role() (av.c) have the same condition, so a shared helper would keep them in step. That can be a follow-up.

Merging. Post-merge verification:

  • @RyanTheTide's HDMI rows on the M1 Pro, M2 and M4 Pro: hotplug five times, s2idle with the sink connected, and an HBR3 mode where the sink offers one.
  • Our rows:
    • J700 boot, panel and deep sleep with CEC core built in;
    • M5 external displays (hotplug, DP audio, deep sleep with displays attached), for the Terminate path on the compact AV roles.

A revert is ready if any of these turns something up.

Thanks for the CEC work and the J616c sink and compliance runs.

@DjDeveloperr
DjDeveloperr merged commit 9f93c19 into aurora-silicon:integration/aurora Nov 21, 2026
7 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.

3 participants