Skip to content

usb: dwc3: apple: take a device role off the bus for suspend-to-idle - #2728

Open
RyanTheTide wants to merge 2 commits into
integration/aurorafrom
airs-wifibt2/dwc3-apple-s2idle-gadget
Open

RyanTheTide wants to merge 2 commits into
integration/aurorafrom
airs-wifibt2/dwc3-apple-s2idle-gadget

Conversation

@RyanTheTide

Copy link
Copy Markdown
Contributor

Problem

dwc3-apple keeps a started controller running through suspend-to-idle in either role, and it has no system suspend callback. A device role (the Mac as a USB device, with a gadget function bound) therefore stays connected to its host while the system is frozen. Nothing services its endpoints meanwhile, so the host's OUT transfers pile up, and the link does not always recover. On an M4 MacBook Air (J713) with an NCM gadget bound, IPv6 ping over the link passed after the 3rd and 4th suspend-to-idle. After the 5th, the host's transmissions never completed again (cdc_ncm ... NETDEV WATCHDOG: transmit queue 0 timed out on the host), while device-to-host traffic still flowed.

Change

dwc3's own system suspend soft-disconnects a gadget for the sleep (dwc3_gadget_suspend()). dwc3-apple now does the same from its sleep preparation. For a suspend-to-idle with the controller started in device role and a gadget driver bound, it disconnects the gadget (usb_gadget_disconnect()) before the system freezes. It connects it again (usb_gadget_connect()) from the work that runs after the wake. The host sees a disconnect and enumerates the device again. Deep sleeps stop the controller anyway and are unchanged, as are host roles. Role changes are already held off while a sleep is being prepared, so the gadget stays bound across the two calls. No parameters, nothing keyed on a board or SoC.

Testing

  • J713, stock command line, lab kernel = the J713 default kernel + phy: apple: atc: admit suspend-to-idle with DWC3 running in device role #2700 (needed for s2idle with a host on the port) + this commit; NCM gadget on port0, the other computer enumerating it as cdc_ncm:
    • Before any sleep: ping 10/10, iperf3 host-to-device 280 Mbit/s.
    • 6/6 suspend-to-idle (20 s, PMU RTC) with ping 10/10 and iperf3 280 Mbit/s after each one. The host logs a clean USB disconnect at each sleep and a new enumeration after each wake, with 0 NETDEV WATCHDOG. 0 kernel errors on the J713.
    • The same row without this commit stalled host-to-device traffic after the 5th sleep (above).
  • W=1: no new warnings in drivers/usb/dwc3/. Host tests --only dwc3, --only apple-dcp, --only atc pass.

Still to run

  • A device-role suspend-to-idle row on another Mac that runs dwc3-apple in device role (M1/M2 generation). There, s2idle in device role was already admitted, so the same stall applies and this change should help.
Details for reviewers
  • Shared path: dwc3-apple drives every Apple Silicon USB-C controller. The new code only acts when !sleep_deep && state == DWC3_APPLE_DEVICE && core_active && dwc.gadget_driver and the gadget is connected. Host roles, host-only controllers (T6040/T6050, host_only) and deep sleeps never reach it.
  • usb_gadget_disconnect()/usb_gadget_connect() are called outside the glue lock: dwc3's pullup path has no dwc3-apple glue op (only pre_host_resume is set), so there is no re-entry.
  • The reconnect sits at the end of dwc3_apple_sleep_start_work(). If a role change after the wake already took the gadget down, the flag is just cleared.
  • Evidence (our own captures): host-side dmesg at each sleep and wake, iperf3 and ping per cycle, and the w/o-fix stall counters (host tx_packets frozen at 89, device rx_packets frozen at 89, device-to-host still incrementing).

Assisted-by: LLM (Claude:Opus 5.5)

dwc3-apple keeps a started controller running through suspend-to-idle
in either role and has no system suspend callback, so a device role
stays connected to its host while the system is frozen. Nothing
services its endpoints meanwhile: the host's OUT transfers pile up and
the link need not recover. On a J713 with an NCM gadget bound, IPv6
ping over the link passed after the 3rd and 4th suspend-to-idle, and
after the 5th the host's transmissions never completed again
('cdc_ncm ... NETDEV WATCHDOG: transmit queue 0 timed out') while
device-to-host traffic still flowed.

dwc3's own system suspend soft-disconnects a gadget for the sleep
(dwc3_gadget_suspend()). Do the same from the sleep preparation: for a
suspend-to-idle with the controller started in device role and a
gadget driver bound, disconnect the gadget before the system freezes,
and connect it again from the work that runs after the wake. The host
sees a disconnect and enumerates the device again. Deep sleeps stop
the controller anyway and are unchanged; host roles are unchanged.

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

Copy link
Copy Markdown
Contributor Author

Merge-train adversarial review (MERGETRAIN-LINUX2) at ee83542: BLOCKERS.

Not merging until the blockers below are fixed or answered; the details have file:line and a failure case.

Full review

#2728 usb: dwc3: apple: take a device role off the bus for suspend-to-idle @ee83542f

VERDICT: BLOCKERS

  1. Use-after-free of the gadget in dwc3_apple_sleep_gadget_on()
    (drivers/usb/dwc3/dwc3-apple.c:1797-1814, called at :1924).

    • dwc3_apple_sleep_start_work() sets sleep_phase = DWC3_APPLE_SLEEP_IDLE (:1916-1919) and drops
      the glue lock. Only then does it call dwc3_apple_sleep_gadget_on().
    • gadget_on() reads appledwc->dwc.gadget under the lock (:1810), drops the lock and calls
      usb_gadget_connect(gadget) (:1812).
    • Once the phase is IDLE, role changes are no longer deferred. The deferral at :1478 only applies
      while sleep_phase != IDLE.
    • So a role switch from the port controller can run dwc3_usb_role_switch_set() → dwc3_apple_exit()
      (:1431-1433) → dwc3_gadget_exit() in the window between :1811 and the end of
      usb_gadget_connect(). dwc3_gadget_exit() runs usb_del_gadget() + usb_put_gadget()
      (drivers/usb/dwc3/gadget.c:4982-4984), which frees the gadget and its udc.
    • usb_gadget_connect() then uses a freed gadget/gadget->udc: it takes udc->connect_lock
      and calls dwc3_gadget_pullup() on torn-down endpoints.
    • Scenario: the cable is unplugged, or the host does a role swap, during a 20 s s2idle. The port
      controller's interrupt after the wake delivers USB_ROLE_NONE just as the start work finishes.
      That is the very moment this patch reconnects.
    • The PR body says "Role changes are already held off while a sleep is being prepared, so the
      gadget stays bound". That is true for gadget_off (phase PREPARING) but false for gadget_on.
    • Fix: reconnect while the phase is still FINISHING. For example, compute the gadget under the lock
      before IDLE is published, drop the lock, connect, then retake the lock to publish IDLE. Or keep
      IDLE unpublished until the connect returns.
  2. A back-to-back suspend can stay connected through the next s2idle, from the same ordering
    (:1916-1924, :1965-1980, :1790).

    • After IDLE is published (:1918), dwc3_apple_pm_notify(PM_SUSPEND_PREPARE) can admit a new epoch
      and queue sleep_stop_work on system_dfl_wq, which is unbound and runs concurrently.
    • That work's gadget_off() sees !gadget->connected (the old epoch has not reconnected yet) and
      returns at :1790 without setting sleep_gadget_off.
    • The old start_work then runs gadget_on(), clears the flag and connects. Result: the gadget is
      connected through the second s2idle, which is the stall this PR is meant to fix. Worse, the
      connect can overlap with dpm_prepare/dpm_suspend.
    • The window is narrow (the start work runs with 0 delay for a kept device role, :1950-1952), but it
      is real for rapid suspend loops (rtcwake loops, a lid bouncing). Fixing blocker 1 by reconnecting
      before IDLE also closes this one.

Nits:

  • gadget->connected is read at :1790 without udc->connect_lock. usb_gadget_disconnect()
    already returns 0 when the gadget is not connected, so you could call it unconditionally and
    record the flag only when the pullup was really on. Today the flag is the only state, so a racing
    sysfs soft_connect write goes unseen.
  • After a deferred DEVICE→DEVICE re-init in start_work (:1898-1909, sleep_stop + sleep_start),
    gadget_on() reconnects a brand-new gadget whose function driver may be rebinding and may itself
    call usb_gadget_connect(). That is harmless, but the comment at :1807 ("A role change after the
    sleep took the gadget down") does not describe this case.
  • Behaviour change for lab tethers: every s2idle now drops gadget links such as a CDC-ACM/NCM debug
    console, so console output across suspend/resume is lost until the host enumerates the device
    again. This matches upstream dwc3, but please name it in the commit message.

Missing rows:

  • A device-role s2idle row on an M1/M2-generation Mac (the PR lists this as still to run). The code
    path is shared by every dwc3-apple controller that is not host_only, so M1, M2, M1/M2 Pro, M3 and
    Neo are all affected and unproven.
  • A negative row: unplug the cable or swap roles during s2idle in device role, with KASAN, to exercise
    blocker 1.
  • Back-to-back s2idle with no idle gap in device role (blocker 2).
  • An aborted-suspend row in device role: pm_test=devices, or a wakeup during prepare, so the
    disconnect is followed by an immediate reconnect from PM_POST_SUSPEND.

@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 pre-PR checklist review at ee83542f3a26adf500a2b21a322eef3c432d9af2 (REQUEST_CHANGES).

Checklist Summary

  • §1 (No model gating): Pass — gated on !sleep_deep, state == DWC3_APPLE_DEVICE, core_active, and dwc.gadget_driver; no board/SoC/firmware gate.
  • §2 (No hacks / default cmdline): Pass.
  • §3 (Correctness / locking / s2idle): Blockers found (see below).
  • §4 (Clean-room & attribution): Pass — human author/committer (Ryan Murray <ryan@aurorasilicon.org>) with Signed-off-by: only in commit; PR body includes Assisted-by:.
  • §5 (Evidence): Incomplete — missing non-J713 device-role s2idle row and negative/race rows (see below).
  • §6 (Naming): Pass.
  • CI: 7/7 checks PASS (host-tests, source, rust-build, kunit, DWC3/ATC builtin-no-atc, DWC3/ATC modular-no-atc, DWC3/ATC builtin-modular-atc).

Blockers

  1. Use-after-free of gadget / gadget->udc in dwc3_apple_sleep_gadget_on() (drivers/usb/dwc3/dwc3-apple.c:1797-1814, called from :1924):

    • dwc3_apple_sleep_start_work() publishes appledwc->sleep_phase = DWC3_APPLE_SLEEP_IDLE at :1916-1919 and releases appledwc->lock at :1921 before calling dwc3_apple_sleep_gadget_on() at :1924.
    • dwc3_apple_sleep_gadget_on() copies gadget = appledwc->dwc.gadget under appledwc->lock (:1810), drops the lock at :1811, and then calls usb_gadget_connect(gadget) at :1812.
    • Once DWC3_APPLE_SLEEP_IDLE is visible, dwc3_usb_role_switch_set() (:1478) no longer defers role changes. A cable unplug or role swap delivered by the port controller right after wake can enter dwc3_apple_exit() (:1431-1433) -> dwc3_gadget_exit() (drivers/usb/dwc3/gadget.c:4982-4984) -> usb_del_gadget() + usb_put_gadget(), freeing gadget and gadget->udc in the window between :1811 and the completion of usb_gadget_connect(). usb_gadget_connect() then locks freed gadget->udc->connect_lock and invokes dwc3_gadget_pullup() on a torn-down gadget.
    • Fix: Reconnect while sleep_phase is still DWC3_APPLE_SLEEP_FINISHING (before publishing DWC3_APPLE_SLEEP_IDLE), so role switches remain deferred until usb_gadget_connect() returns.
  2. Back-to-back suspend race can leave the gadget connected through the next s2idle (drivers/usb/dwc3/dwc3-apple.c:1916-1924, :1965-1982, :1790):

    • Because DWC3_APPLE_SLEEP_IDLE is published before dwc3_apple_sleep_gadget_on() runs, dwc3_apple_pm_notify(PM_SUSPEND_PREPARE) can immediately admit a new suspend epoch and queue sleep_stop_work concurrently on system_dfl_wq.
    • The new epoch's dwc3_apple_sleep_gadget_off() can observe !gadget->connected (or race on sleep_gadget_off before gadget_on() clears it at :1806) and return without recording sleep_gadget_off = true, after which the previous epoch's dwc3_apple_sleep_gadget_on() calls usb_gadget_connect(), reconnecting the gadget into the second s2idle (or concurrently with dpm_prepare/dpm_suspend).
    • Fixing Blocker 1 by completing dwc3_apple_sleep_gadget_on() before publishing DWC3_APPLE_SLEEP_IDLE also closes this race.

Secondary Issues / Nits

  • Unlocked gadget->connected read (dwc3-apple.c:1790): PM_SUSPEND_PREPARE runs before tasks are frozen, so userspace can still toggle soft_connect or unbind configfs UDC concurrently. gadget->connected is protected by gadget->udc->connect_lock.
  • Call placement in dwc3_apple_sleep_prepare_work() (dwc3-apple.c:1823): dwc3_apple_sleep_gadget_off() runs before the appledwc->shutting_down check (:1826) and before dwc3_apple_sleep_start() (:1833); moving it after those checks (for !sleep_deep) avoids acting during teardown or missing a device role just started by dwc3_apple_sleep_start().
  • Comment / state hygiene: Clearing sleep_gadget_off in dwc3_apple_exit() keeps the flag consistent across deferred DEVICE -> DEVICE re-init (:1905-1908), and the commit message should note that s2idle now drops and re-enumerates USB gadget links (e.g., CDC-ACM/NCM debug consoles).

Required §5 Rows After Fix

  • Device-role s2idle row on a second generation (M1/M2-class Mac running dwc3-apple in device role).
  • Negative/race rows: cable unplug / role swap during s2idle (ideally with KASAN), back-to-back s2idle loop, and aborted suspend (pm_test=devices or prepare-phase wakeup).
  • Explicit SELinux Enforcing / 0 AVC denials / 0 failed units confirmation on the regression run.

Assisted-by: LLM (Gemini 4 Neon)

The start work after a suspend-to-idle published the IDLE phase, dropped
the glue lock and only then connected the gadget it had taken off its
host. Once IDLE is visible, two things can run under that connect:

- Role changes are no longer deferred. A cable pull or role swap that
  the port controller reports right after the wake goes through
  dwc3_apple_exit() and dwc3_gadget_exit(), which frees the gadget and
  its UDC while usb_gadget_connect() is still using them.
- A new sleep can be admitted. Its preparation sees a gadget that is not
  connected yet and leaves it alone, then the old epoch connects it, so
  the gadget stays on the bus through the next suspend-to-idle.

Connect the gadget while the epoch is still FINISHING and publish IDLE
afterwards. A role change that arrives during the connect is deferred
as before and applied under the glue lock before IDLE. Split the
deferred-role and IDLE steps of the start work into helpers so both
paths share them.

Also:
- Take the gadget off its host after the shutdown check and the
  controller start in the preparation, for suspend-to-idle only.
- Read gadget->connected and disconnect under the gadget's device lock,
  which soft_connect writes and function driver binding also hold, so a
  racing soft_connect write is not missed. The glue lock is not held
  there, since dwc3_gadget_exit() takes the device lock under it.
- Forget a pending reconnect in dwc3_apple_exit(): a gadget that is set
  up again connects when its function driver binds.

With this, every suspend-to-idle in device role drops the gadget link
and the host enumerates the device again after the wake. A CDC-ACM or
NCM console on the gadget loses whatever is sent while it is
disconnected. dwc3's own system suspend does the same.

A new host test, dwc3-sleep-gadget.py, builds the sleep works, the role
switch and dwc3_apple_exit() against mocks. It checks that the connect
runs while the epoch still defers role changes and admits no sleep,
across back-to-back sleeps, role changes before and during the
reconnect, and teardown. Negative controls publish IDLE before the
connect, skip the role re-check, keep the flag across an exit and
disconnect a gadget that is not connected. The harnesses that stub
struct dwc3_apple carry the new field.

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

Copy link
Copy Markdown
Contributor Author

Both blockers and the nits are fixed in 1a91e65, a new commit on top of ee83542.

  • Use-after-free (blocker 1): the start work now connects the gadget again while the epoch is still FINISHING, and only then publishes IDLE (drivers/usb/dwc3/dwc3-apple.c:1965-1977). A role change that arrives during the connect is deferred as before. It is applied under the glue lock before IDLE, in the new dwc3_apple_sleep_role() (:1892), and dwc3_apple_exit()/dwc3_gadget_exit() can't run while usb_gadget_connect() is using the gadget. The deferred-role and IDLE steps are now helpers (dwc3_apple_sleep_role(), dwc3_apple_sleep_idle() at :1913), so both paths share them.
  • Back-to-back suspend (blocker 2): the same reorder closes it. dwc3_apple_pm_notify() refuses a new epoch until IDLE, and IDLE now comes after the gadget is back on the bus, so the next preparation always sees it connected and disconnects it.
  • Unlocked gadget->connected read: dwc3_apple_sleep_gadget_off() (:1798) reads the flag and disconnects under the gadget's device lock (:1803). soft_connect writes and function driver bind/unbind hold that same lock, so a racing write is no longer missed. The glue lock is not held there, because dwc3_gadget_exit() takes the device lock under it.
  • Placement in the preparation: the disconnect now runs after the shutting_down check and dwc3_apple_sleep_start(), for suspend-to-idle only (:1848).
  • sleep_gadget_off is cleared in dwc3_apple_exit() (:1433), so a gadget that is set up again connects itself when its function driver binds. The misleading comment is gone with the old gadget_on().
  • The commit message now says that every device-role suspend-to-idle drops the gadget link and the host enumerates it again, so a CDC-ACM/NCM console loses output while it is disconnected.

Tests: a new host test, tools/testing/selftests/drivers/apple/dwc3-sleep-gadget.py (registered in run-host-tests and the selftest Makefile), builds the real sleep works, the gadget helpers, dwc3_usb_role_switch_set() and dwc3_apple_exit() against mocks. Its 4/4 cases cover:

  • three back-to-back sleeps, with the connect asserting that the epoch is still FINISHING and the glue lock is free;
  • a role change during the connect (deferred, the gadget stays alive, applied before IDLE) and one before the wake (no reconnect);
  • deep sleep, host, unbound and soft-disconnected gadgets (left alone);
  • teardown before and after the preparation.

Four negative controls fail as they should: IDLE before the connect, no role re-check after it, the flag kept across an exit, and disconnecting a gadget that is not connected. The other harnesses that stub struct dwc3_apple carry the new field. Host tests on this head: --only dwc3 2/2, --only atc 6/6, --only neo-stability 11/11, apple-dcp 1/1. The same suites also pass on integration/aurora with this PR and #2700 (da78109) merged together; both PRs merge cleanly alone and together. drivers/usb/dwc3/dwc3-apple.o builds with W=1 (clang, fleet81 config, DWC3 dual-role) with no compiler warnings, and checkpatch --strict shows 0 errors (its only warning is the generic MAINTAINERS note for a new file; tools/testing/selftests/drivers/apple/ is already covered). The dwc3-apple KUnit suite only covers dwc3-apple-state.h, which this does not change.

Hardware rows still to run:

  • J713, device role with NCM bound, stock command line: back-to-back rtcwake -m freeze -s 20 with no idle gap, x5; then an aborted suspend (echo devices > /sys/power/pm_test; echo freeze > /sys/power/state) showing the disconnect and an immediate re-enumeration; ping and iperf3 after each. State SELinux enforcing, 0 AVC (ausearch --input-logs -m avc), 0 failed units and no new kernel errors.
  • A second generation running dwc3-apple in device role (an M1/M2-class Mac such as an M2 Air or M2 mini) with a host on the port and NCM bound: suspend-to-idle x3 with ping/iperf3 after each.
  • A cable pull or role swap during suspend-to-idle in device role, ideally on a KASAN build. This needs someone at the lab.

Assisted-by: LLM (Claude:Opus 5.5)

@RyanTheTide

Copy link
Copy Markdown
Contributor Author

Review at 1a91e65: no blockers.

Both blockers from the review at ee83542 and its three nits are fixed in 1a91e65. I found no new locking, lifetime or error-path bug in the reworked sleep works. The hardware rows asked for in that review (a second-generation device-role s2idle row, back-to-back and aborted-suspend rows, cable pull or role swap during s2idle, SELinux/AVC/failed-unit confirmation) are still to run. They are evidence still owed, not code findings.

One non-blocking nit: dwc3_apple_sleep_gadget_back() reconnects whenever this sleep disconnected the gadget. If userspace writes soft_connect = disconnect between the sleep's disconnect and the freezer, that write is a no-op (the gadget is already off), and after the wake the gadget connects again anyway. This could be fixed by recording the disconnect in a way that soft_connect can clear, or by leaving it as a documented limit. It is not worth holding the merge for.

Full review

Head 1a91e65 (2 commits on a84637d: ee83542, 1a91e65).

Findings from the ee83542 review

  1. Use-after-free in gadget_on() (fixed). dwc3_apple_sleep_start_work() now calls usb_gadget_connect() while sleep_phase is still FINISHING (dwc3-apple.c:1962-1977). dwc3_usb_role_switch_set() defers every role change while the phase is not IDLE (:1480), so dwc3_apple_exit()/dwc3_gadget_exit() can't free the gadget under the connect. A role change that arrives during the connect is applied by the second dwc3_apple_sleep_role() call under the glue lock, before dwc3_apple_sleep_idle() publishes IDLE. I checked every other caller of dwc3_apple_exit():
    • hub_check (:1411), the tunnel work (:1725) and the J714s ss check only act in host role and IDLE.
    • sleep_stop() runs only from the sleep works.
    • remove and shutdown run dwc3_apple_pm_quiesce() first. It sets shutting_down and cancel_*_syncs both sleep works before exit, so a connect in flight finishes before the gadget is freed.
  2. Back-to-back race (fixed). dwc3_apple_pm_notify() admits a new epoch only from IDLE (:2020). IDLE is now published after the connect returns, so the next prepare always sees connected. During the connect, the notifier's trylock/phase check refuses the next sleep with -EBUSY. That refusal behaviour already existed for FINISHING.
  3. Nit: unlocked connected read (fixed). The read and usb_gadget_disconnect() now run under device_lock(&gadget->dev) (:1803). That is the lock soft_connect_store() and gadget bind/unbind take (udc/core.c:1809). The lock order is device lock -> connect_lock, the same as soft_connect_store(). The glue lock is not held there, which avoids the inversion with dwc3_gadget_exit() (glue lock -> gadget device lock). The gadget pointer taken under the glue lock stays valid until the device lock is taken, because PREPARING defers role changes and teardown cancels sleep_stop_work before exit. dwc3-apple has no pre_run_stop glue op, so pullup never comes back into the glue lock.
  4. Nit: placement (fixed). The disconnect now runs after the shutting_down check and dwc3_apple_sleep_start(), and only for s2idle (:1847-1862). Every path through the scoped_guard either returns or sets s2idle.
  5. Nit: flag hygiene (fixed). sleep_gadget_off is cleared in dwc3_apple_exit()'s DEVICE case (:1433). gadget_back() also clears it and re-checks shutting_down, state, core_active and gadget_driver.
  6. The commit message now says that s2idle drops the gadget link and the host enumerates the device again (CDC-ACM/NCM).

Other paths checked

  • Aborted suspend: PM_POST_SUSPEND queues the start work, which flush_work()s the stop work first, so the flag is final before gadget_back() reads it.
  • Hibernation and deep sleep: not admitted, or they take the old stop path. Unchanged.
  • usb_gadget_connect() racing with a configfs UDC unbind after thaw: serialised by connect_lock, and it only sets connected when allow_connect/started is false. Safe.
  • finish is host-only (sleep in place), so calling dwc3_apple_sleep_idle() on the gadget path can't double-queue the PHY finish.

Other Macs, parameters, delays, trailers, merge

  • Scope: there is no chip, board or firmware gate. The notifier is online for every dwc3-apple, so M1/M2 device roles get the same s2idle disconnect. That matches dwc3's own dwc3_gadget_suspend().
  • No new module parameter or cmdline option, and no new delay.
  • Trailers: Signed-off-by: Ryan Murray only, in both commits. The PR body ends with Assisted-by:.
  • Merge: git merge-tree onto integration/aurora 95a13cf is clean. Since the base, no commits touch drivers/usb/dwc3 or the UDC core.

Tests at the head (buildbox)

  • dwc3-sleep-gadget.py: 4/4 passed, and its 4 negative controls were rejected as expected.
  • run-host-tests --only dwc3 2/2, atc 6/6, neo-stability 11/11, apple-dcp 1/1, all passed.

Still to run (hardware)

  • J713 NCM: back-to-back rtcwake -m freeze x5, aborted suspend via pm_test=devices, and Enforcing with 0 AVC.
  • A second-generation device-role row (M2 or M1 class).
  • Cable pull or role swap during s2idle, ideally on a KASAN build.

Assisted-by: LLM (Claude:Opus 5.5)

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