Skip to content

irqchip/apple-aic: Use the SIQ_CFG routing on every AICv2+ controller - #2222

Merged
RyanTheTide merged 2 commits into
aurora-silicon:integration/aurorafrom
jgoyvaerts:m4pro/aic-t6041-affinity
Nov 22, 2026
Merged

RyanTheTide merged 2 commits into
aurora-silicon:integration/aurorafrom
jgoyvaerts:m4pro/aic-t6041-affinity

Conversation

@jgoyvaerts

@jgoyvaerts jgoyvaerts commented Nov 19, 2026 •

Copy link
Copy Markdown
Contributor

#447 made AICv2+ IRQs report the CPUs that SIQ_CFG says accept them, but only for T8140, through the per-SoC siq_cfg_routing flag. On the M4 Pro (T6040), whose device tree lists apple,t6041-aic3 and then apple,t8122-aic3, the controller matched the generic aic3_info. Every IRQ had an empty effective affinity, and genirq had no CPU to bind the irq threads to.

The hard IRQ is always taken on CPU0. An irq thread stays on whichever CPU it last ran on, so when that CPU is idle, CPU0 has to wake it first. For the GPU's mailbox IRQ (irq/70), that wakeup is on the path of every GPU completion.

SIQ_CFG is part of AICv2 and later, not of one SoC. m1n1 and aura write it on every CPU of the M1, M2 and M3 families (value 2, "Enable IRQs (at least necessary on t600x)"), the sleep driver saves and restores it on the M5 parts, and #447 and this PR read it on T8140 and T6040. Two commits:

  1. Keep the routed IRQ affinity across CPU hotplug. irqchip/apple-aic: Report the effective affinity of AICv2+ IRQs #447 read SIQ_CFG only when a CPU started, and an IRQ kept the accepting set it had when its affinity was set. Its comment said this was enough only while one CPU accepts IRQs. Under m1n1 and aura every M1–M3 CPU accepts, and aurora's spin-table can take CPUs offline and back across a sleep. So SIQ_CFG is now read in a CPUHP_AP_ONLINE_DYN callback, the CPU is dropped from the set when it goes offline, and every requested AIC IRQ gets its affinity set again when the set changes (aic_irq_cpus_update()).
  2. Use the routing on every AICv2+ controller. The siq_cfg_routing flag is gone. AICv2 and AIC3 at EL2 use aic2_routed_chip (aic_irq_domain_map()), and the EL2 check stays. No aic3_t6041_info entry is added. aic3_t8140_info stays because it still carries fw_suspend. At EL1 nothing changes.

Tested on the J616s (M4 Pro, 48 GiB, 25G83)

These numbers come from the first revision, which had a T6041-only entry. The behaviour on T6040 is the same, because only CPU0 accepts IRQs there and the new hotplug path does nothing. The kernel runs at EL2 (CTXR window).

  • Every AIC IRQ that has been activated reports effective_affinity_list = 0.
  • The irq/70 thread's Cpus_allowed_list is 0, and all of its interrupts land on CPU0.
  • AIC lines that were never enabled (idle mailboxes' send/receive IRQs) still read empty. That is because genirq sets the effective mask at activation.

GPU IRQ to irq-thread wakeup, gpubench compute.light at depth 1 (trimmed mean):

Submitter Before After
Pinned to P-cluster 0 4.45 us 0.95 us
Pinned to P-cluster 1 6.36 us 3.73 us
Unpinned 5.08 us 1.80 us

Depth-1 latency went from 75.1 to 68.7 us (P0), 91.9 to 86.1 us (P1) and 86.7 to 79.2 us (unpinned). In the full gpubench record, idle.wake is 8-14% faster and render.light clear at depth 1 is 14% faster. The before and after kernels were separate boots with our pending G17 changes (including #2217). The after kernel also had a drm_sched placement change that we later dropped. It does not touch IRQ routing or the irq thread.

This revision: drivers/irqchip/irq-apple-aic.o builds for arm64 with W=1 and no warnings after each commit; checkpatch gives 0 errors and 0 warnings on both.

Hardware

RyanTheTide posted rows on the same code (7d5104b9 -> c2e95d38, commit messages only) for all three AICv2+ classes: M2 (T8112, AIC2), A18 Pro (T8140, AIC3) and M4 Pro (J616s, T6040, AIC3), with s2idle, CPU hotplug where available, 0 failed units and 0 new kernel errors. See the thread for the numbers.

Not done

  • Not checked whether T8152 (M6, generic AIC3 entry) implements SIQ_CFG. Nothing in the tree reads it there, and its boards are disabled inventories.
  • No KUnit test: the code reads Apple system registers that qemu does not have.

Assisted-by: LLM (Claude:Opus 5.5)

@RyanTheTide

Copy link
Copy Markdown
Contributor

Thanks for tracking this down; the IRQ-thread wakeup numbers are a clear win.

We select behaviour by a hardware property rather than by the list of SoCs it was tested on, and this adds a second per-SoC entry for the same mechanism: aic3_t6041_info with .siq_cfg_routing = true (drivers/irqchip/irq-apple-aic.c:364-370, matched at :392-395), next to the existing T8140 one.

SIQ_CFG is the AICv2-and-later system register that says which CPUs accept IRQs, so the effective-affinity read is a property of the controller version (and of running at EL2, which the code already checks), not of T6040/T6041. Could you set siq_cfg_routing in the shared AIC3 info (and the AICv2 info if the register is there too) and drop the per-SoC entry? If some generation cannot use it, a comment naming the hardware reason on that one entry would be the way to keep it off.

If you make it the default, we can run rows on our M1 Pro, M2, M4 Pro and A18 Pro.

With SIQ_CFG routing, each AICv2+ IRQ reports the CPUs that accept IRQs
as its effective affinity. Each CPU's SIQ_CFG is read when that CPU
starts, and an IRQ copies the set only when its affinity is set. An IRQ
keeps the set it was given. So a CPU that starts later and accepts IRQs
is missing from every IRQ requested before it came up, and a CPU that
goes offline stays in every mask. That is harmless while only the boot
CPU accepts IRQs, but not when several CPUs do.

Read SIQ_CFG in a CPU's online callback, drop the CPU from the set in
its offline callback, and set the affinity of every requested AIC IRQ
again whenever the set changes. genirq then updates the IRQ's effective
affinity and its thread's CPU mask. CPUs that do not change the set cost
nothing beyond the register read.

Signed-off-by: Jonathan Goyvaerts <jonathan.goyvaerts@gmail.com>
SIQ_CFG, the per-CPU register that says whether a CPU accepts AIC IRQs,
is part of AICv2 and later. m1n1 and aura write it on every CPU of the
M1, M2 and M3 families, and the sleep driver saves and restores it on
the M5 parts. Reporting the accepting CPUs as the effective affinity of
AICv2+ IRQs is therefore a property of the controller version, plus the
EL2 check that is already there. It was enabled only for T8140, through
a per-SoC flag.

T6040 and T6041 take every AIC IRQ on CPU 0, like T8140: after 3 hours
of uptime /proc/interrupts counts none of the 72 AIC IRQs on any other
CPU. Their device tree lists "apple,t6041-aic3" and "apple,t8122-aic3",
which matches the generic AIC3 entry, so their IRQs reported no
effective affinity and genirq let the threaded handlers run anywhere. A
SCHED_FIFO IRQ thread stays on whichever CPU it last ran on, and when
that CPU is idle in WFI the hard IRQ on CPU 0 must wake it with an IPI
before the handler can run.

For the G17 GPU completion IRQ on T6040 (gpubench compute.light at
depth 1, ftrace, about 10,000 IRQs per run, five boots), the time from
the hard IRQ to the thread running is:

  thread's CPU idle (99% of IRQs on an idle desktop)  4.4-9.6 us
  thread's CPU running another task                   1.3-2.4 us
  thread on CPU 0                                     1.3-1.5 us

The one boot that measured 2 us had a compiler busy on every CPU during
the run, so the thread's CPU was almost never idle.

Drop the flag and use the routed irqchip on every AICv2+ controller at
EL2. At EL1 nothing changes. Where the boot loader lets every CPU accept
IRQs (m1n1 and aura on M1 to M3), each IRQ now reports all of them,
which the previous commit keeps up to date across CPU hotplug.

Signed-off-by: Jonathan Goyvaerts <jonathan.goyvaerts@gmail.com>
@jgoyvaerts
jgoyvaerts force-pushed the m4pro/aic-t6041-affinity branch from 41fd5db to 7d5104b Compare November 19, 2026 14:51
@jgoyvaerts jgoyvaerts changed the title irqchip/apple-aic: Report the effective affinity of T6041 IRQs irqchip/apple-aic: Use the SIQ_CFG routing on every AICv2+ controller Nov 19, 2026
@jgoyvaerts

Copy link
Copy Markdown
Contributor Author

Done in 7d5104b, as two commits:

  • siq_cfg_routing is gone. AICv2 and AIC3 at EL2 use the routed chip (aic_irq_domain_map()), and the EL2 check stays. No T6041 entry. aic3_t8140_info stays only for fw_suspend.
  • SIQ_CFG exists on AICv2: m1n1 and aura write it (value 2) on every M1/M2/M3 CPU, and the sleep driver uses it on M5. That means every CPU accepts IRQs there, the case irqchip/apple-aic: Report the effective affinity of AICv2+ IRQs #447's comment left open. So the first commit reads SIQ_CFG in a CPUHP_AP_ONLINE_DYN callback, drops a CPU from the set when it goes offline, and sets every requested IRQ's affinity again when the set changes (aic_irq_cpus_update()).

Tested: arm64 build of irq-apple-aic.o with W=1 and no warnings after each commit; checkpatch clean. Not booted yet. On T6040 only CPU0 accepts, so the earlier numbers still hold there.

Yes please, rows on the M1 Pro, M2, M4 Pro and A18 Pro would help most. On M1 Pro and M2, effective_affinity_list should list every CPU, also after a secondary goes offline and comes back. That is the hotplug path, which we cannot exercise here.

@jgoyvaerts

Copy link
Copy Markdown
Contributor Author

One gap I couldn't close: I found nothing showing that T8152 (M6) implements SIQ_CFG. It takes the generic AIC3 entry, so if the register isn't there, the EL2 read in the online callback would fault at boot. No M6 board is enabled today. If you know it lacks the register, I'll give it its own entry with that reason in a comment.

@RyanTheTide

Copy link
Copy Markdown
Contributor

Thanks for 7d5104b, and for adding the hotplug tracking rather than leaving it as a comment. Since this now runs on every AICv2+ machine and has not booted yet, we will run rows on our M1 Pro, M2, M4 Pro and A18 Pro, including effective_affinity_list after a secondary goes offline and comes back. On M6 we have nothing either way; keeping the generic entry until a board is enabled seems right, and if bring-up shows the register is absent, an entry with that reason in a comment is the right shape.

@RyanTheTide

Copy link
Copy Markdown
Contributor

Row from our M2 (MacBook Air, T8112, AIC2), as offered: production kernel (fleet74) vs the same tree plus this PR's two commits, one-shot boot, same machine.

  • effective_affinity_list on the 77 AIC2 IRQs: production reports all 77 empty; with this PR 62/77 read 0-7, matching smp_affinity & online CPUs. The 15 still empty are the mailbox send/receive IRQs that never started (count 0).
  • s2idle x2 on the PMU RTC: 2/2, same boot, a 256 MiB direct NVMe read at 1.2-1.4 GB/s after each wake, affinity unchanged, no kernel errors.
  • Boot health: Enforcing, 0 failed units, 0 AVC, 0 kernel errors; warnings identical to production.
  • CPU hotplug: not testable here, the M2 uses spin-table enable with no cpuN/online.

Rows on our M4 Pro (T6040) and A18 Pro are queued; they also cover hotplug.

@RyanTheTide

Copy link
Copy Markdown
Contributor

Row from our A18 Pro (MacBook Neo, T8140, AIC2): production 7.3.0-40 vs the same tree plus this PR, one-shot boot, same machine.

  • effective_affinity_list on 86 AIC2 IRQs: 74 read CPU0 and 12 are empty (never started) on both kernels, so no change here, as expected on a part where only CPU0 takes device IRQs.
  • CPU hotplug of a P core (cpu5) and an E core (cpu1): offline/online clean, the effective CPU never points at an offline core, a 256 MiB direct NVMe read keeps working before, during and after, no kernel errors.
  • s2idle x2 on the PMU RTC: 2/2, same boot, affinity unchanged.
  • Boot health: Enforcing, 0 failed units, 0 AVC.

Together with the M2 row above, this covers both AIC2 cases we have. An M4 Pro row is still queued.

@RyanTheTide

Copy link
Copy Markdown
Contributor

Row from our M4 Pro (MacBook Pro 16-inch, J616s, T6040): production fleet kernel vs the same tree plus this PR, one-shot boot.

  • effective_affinity_list: production reports 137/137 empty; with this PR 115/137 read CPU0 (the only CPU taking device IRQs on this part) and the 22 still empty never started.
  • s2idle x2: 2/2, no resets; the only kernel errors are the dwc3-apple genpd_prepare returns -11 lines production also shows on every cycle.
  • CPU hotplug: not testable here (spin-table secondaries cannot go offline).

With the M2 and A18 Pro rows above, that is every AIC2 case in our lab: PASS on all three.

@jgoyvaerts
jgoyvaerts force-pushed the m4pro/aic-t6041-affinity branch from 7d5104b to c2e95d3 Compare November 21, 2026 06:14
@jgoyvaerts

Copy link
Copy Markdown
Contributor Author

Commit messages only, for the pre-PR checklist: the commits no longer carry Assisted-by (item 4), and the disclosure is now the PR body's last line. The tree is identical to 7d5104b956 (git diff 7d5104b956 c2e95d38b7 is empty). New head c2e95d38b7.

Assisted-by: LLM (Claude:Opus 5.5)

@RyanTheTide

Copy link
Copy Markdown
Contributor

Checked: the diff of c2e95d3 is line for line the one of 7d5104b, so only the commit messages changed. Our rows on 7d5104b (M2, M4 Pro and A18 Pro: effective affinity, P+E hotplug, s2idle) apply to this head. CI is green. No further points from us.

@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.

Review of head c2e95d38b7160796d4992956f4e2767ccc22fb92 against the pre-PR checklist.

Code, policy, CI & evidence

  • 0 blockers. CI (host-tests) is green.
  • §1–§3: Removes the per-SoC siq_cfg_routing gate in favour of controller version (version == 2 || version == 3) + is_kernel_in_hyp_mode(). Moving SIQ_CFG tracking to CPUHP_AP_ONLINE_DYN (aic_irq_cpu_online / aic_irq_cpu_offline) keeps aic_irq_cpus accurate across CPU bringup and hotplug, refreshes requested IRQs via irq_set_affinity() (safe in hotplug context — takes desc->lock only, no cpus_read_lock()), and gracefully falls back to cpu_possible_mask if cpuhp_setup_state() fails or aic_irq_cpus is empty.
  • §4 & §6: Both commits (d3d404c0e448, c2e95d38b716) carry human author/committer and human Signed-off-by: only, with PR-body disclosure present.
  • §5: Hardware rows from @RyanTheTide on the identical code diff (7d5104b9 -> c2e95d38, commit messages only) cover all three AICv2+ classes — M2 (T8112, AIC2: 62/77 active IRQs 0-7, s2idle x2 + NVMe, Enforcing, 0 failed units, 0 AVC, 0 kernel errors), A18 Pro (T8140, AIC3: 74/86 0, P+E hotplug + NVMe, s2idle x2, Enforcing, 0 failed units, 0 AVC), and M4 Pro (J616s, T6040, AIC3: 115/137 0, s2idle x2, 0 new kernel errors).

Non-blocking notes

  1. In aic_irq_cpus_update() (drivers/irqchip/irq-apple-aic.c:1123), cpumask_copy(&aff, irq_get_affinity_mask(virq)) assumes irq_get_affinity_mask(virq) is non-NULL right after irq_has_action(virq). Static DT AIC IRQs are rarely disposed at runtime, but checking for NULL (or guarding with rcu_read_lock()) would eliminate any TOCTOU window.
  2. On a multi-CPU-accepting part, if an IRQ's saved smp_affinity were ever pinned solely to a CPU that is currently offline, irq_set_affinity(virq, &aff) (force=false) would return -EINVAL without refreshing effective_affinity; not hit under normal masks, just worth keeping in mind for any follow-up.
  3. The PR body's "Not done" section still says "Not booted yet", which is now superseded by the M2, A18 Pro, and M4 Pro hardware rows in the thread.

Assisted-by: LLM (Gemini 4 Neon)

@RyanTheTide

Copy link
Copy Markdown
Contributor

Merge-train adversarial review (MERGETRAIN-LINUX2) at c2e95d3: NO BLOCKERS. With Dj's approval at this head, that makes the joint verdict MERGE. Before the merge we will boot it on our Macs, as one test kernel of integration 24dc422 plus #2222 and #2225: M1 Pro, M2, J616s, Neo and the J615. Results will follow here.

Full review

#2222 irqchip/apple-aic: Use the SIQ_CFG routing on every AICv2+ controller @c2e95d38

VERDICT: NO BLOCKERS

Commits: d3d404c (hotplug tracking), c2e95d3 (drop siq_cfg_routing). Line numbers are in drivers/irqchip/irq-apple-aic.c at c2e95d3.

What I checked and found to hold:

  • Gate rules: the per-SoC siq_cfg_routing flag is removed. Selection is now controller version plus is_kernel_in_hyp_mode() (:823, :1470-1474). There is no new compatible, no board or firmware gate, and no param, env knob or delay. aic3_t8140_info keeps only fw_suspend.
  • Early-boot cpuhp_setup_state(CPUHP_AP_ONLINE_DYN, ...) at :1472 runs from init_IRQ, before cpuhp_threads_init() (kernel/smp.c:1121). cpuhp_invoke_ap_callback() (kernel/cpu.c:1132) then calls the callback directly on the boot CPU, because st->thread is still NULL. So aic_irq_cpu_online() reads CPU0's own SIQ_CFG, as the comment at :1129 claims. Later CPUs run it on their own cpuhp thread. This is correct.
  • Offline ordering: the ONLINE_DYN teardown runs before __cpu_disable()/irq_migrate_all_off_this_cpu(). When irq_do_set_affinity() (kernel/irq/manage.c:233) runs, the dying CPU is still in cpu_online_mask, so tmp_mask is non-empty and the routed chip writes the already-shrunk aic_irq_cpus. migrate_one_irq() then finds the dying CPU absent from the effective mask. No IRQ is left pointing at an offline CPU, which matches the Neo hotplug row.
  • Removing the SIQ_CFG read from aic_init_cpu() loses nothing on resume. apple_sleep_return() restores the boot CPU's SIQ_CFG itself (drivers/soc/apple/sleep.c:5635), and the secondaries pass through the new online callback when they thaw.
  • On multi-accepting parts (M1 Pro/M2/M3 under m1n1/aura) the routed chip now reports every accepting CPU. IRQ threads get bound to effectively all CPUs, so they behave as before, and the M2 row confirms 0-7.

Blockers: none.

Nits:

  1. :1123-1124 cpumask_copy(&aff, irq_get_affinity_mask(virq)); irq_set_affinity(virq, &aff); copies desc->irq_common_data.affinity without the desc lock. Suppose a /proc/irq/N/smp_affinity write (irqbalance) lands between the copy and irq_set_affinity(). The stale copy is then written back over the user's mask, and the user's update is lost. Passing irq_get_affinity_mask(virq) straight to irq_set_affinity() closes this, because the mask is then read under the lock. Dj's NULL/TOCTOU note covers the same line.
  2. irq_set_affinity_locked() sets IRQD_AFFINITY_SET on every requested AIC IRQ at each set change (manage.c, irqd_set(data, IRQD_AFFINITY_SET)). The requested mask is unchanged, so this is harmless. It does make every IRQ look user-pinned to genirq on its next startup.
  3. The irq_has_action() filter at :1121 skips chained parents. pinctrl-apple-gpio's group IRQs are chained on the AIC (drivers/pinctrl/pinctrl-apple-gpio.c:502). Their effective affinity is not refreshed on online, and on offline it is refreshed only through migrate_one_irq(). They are not threaded, so the only effect is a stale value in /proc.
  4. A CPU-pinned smp_affinity whose CPU is offline makes the online-time irq_set_affinity() return -EINVAL, which is ignored. Effective affinity then misses newly online accepting CPUs. This is cosmetic and is Dj's note 2.
  5. If every accepting CPU is offline (CPU0 on T6040/T8140, if aurora's spin-table lets cpu0 go offline), aic_irq_cpus becomes empty. :633-634 then falls back to cpu_possible_mask, which includes offline CPUs. The fallback predates this PR, and the hardware still routes to the parked CPU, so this PR does not change that. It may be worth a comment.
  6. The PR body's "Not done" no longer says "Not booted yet", which is fine. The A18 Pro row in the thread calls the T8140 controller "AIC2", but it is AIC3 (the chip name string is "AIC2" for both).

Missing rows:

  • The point of commit d3d404c (refreshing the set when one of several accepting CPUs goes offline or comes online) has no hardware proof. The only hotplug row is Neo (T8140), where only CPU0 accepts, so the update returns early at :1114. The M2 and J616s rows had no cpuN/online. This code would be exercised by an M1 Pro, M2 Pro/Max or M3 that can hotplug, or by an M3/M5 firmware suspend (freeze_secondary_cpus) with several accepting CPUs.
  • Every AICv2+ Mac now reads SIQ_CFG at boot, at EL2, in the init-time callback. Undefined on any of them means a panic at boot. Unbooted are the M1 Pro/Max (T600x AIC2; a row was offered, not posted), M3 J615 (T8122, generic AIC3), M4 Air J713 (T8132, aic3_t8132_info), M5 T8142/T6050, and M6 T8152 (no board enabled, as the author notes). For M5 the sleep driver's own SIQ_CFG read (sleep.c:5309) is indirect evidence. For J713 the evidence is the same-generation T6040 row.

Commit hygiene: both commits are by Jonathan Goyvaerts (author and committer) with a human Signed-off-by, and there are no agent trailers in the commits. The disclosure is on the PR body's last line. Provenance: SIQ_CFG is the public m1n1 register name already used in the tree, and there are no kext names.

@RyanTheTide

Copy link
Copy Markdown
Contributor

Hardware rows (MERGETRAIN-LINUX2), test kernel 7.3.0-aurora-mt3 = integration/aurora 24dc422 + #2222 c2e95d3 + #2225 6643e60, on the default fleet cmdline. Each Mac got 3 one-shot warm boots and went back to its default:

Mac boots failed/avc/oops/fwcrash s2idle dEQP-VK smoke AIC effective affinity
M1 Pro (T6000) 3/3 0/0/0/0 4/4 4986 P (= fleet79) 142 IRQs on CPUs 0-9 (fleet78: all 142 empty)
M2 (T8112) 3/3 0/0/0/0 4/4 4986 P (= fleet79) 62/77 at boot, 77/77 after the hotplug refresh (fleet78: all 77 empty)
Neo (T8140) 3/3 0/0/0/0 4/4 8951 P (= fleet79) unchanged: 74 on CPU0, 12 empty
M3 Air J615 (T8122) 3/3 0/0/0/0 3/3 (no GPU on this kernel) 37/50 on CPUs 0-7 (default boot: all 53 empty)
  • CPU1 hotplug: Neo offline/online OK, with NVMe I/O while CPU1 was offline. M1 Pro, M2 and J615 refuse the offline (rc=1, EOPNOTSUPP), and production fleet78 refuses it the same way, so the refresh path only ran on the Neo.
  • No new kernel errors on any Mac. The J615 set is the same as on the tip kernel (AOP and ISP pieces from open PRs).
  • mm: populate a subpage window's uffd-wp markers with the faulting page #2225: its path needs 4K userspace (user4k=on), which the default cmdline doesn't enable, so on these Macs this is a no-regression row only.
  • The J616s (M4 Pro) row follows when the Mac is free.

Evidence: traces/kernel/20261122-mt3-fleetboot (lab store).

@RyanTheTide

Copy link
Copy Markdown
Contributor

J616s (M4 Pro, T6040 AIC) row: PASS. This is the row the merge train was still waiting for.

Kernel 7.3.0-aurora-mt4: integration/aurora 71abea7 (batch-3 tip) plus this PR at c2e95d3, on the one fleet cmdline. I ran three one-shot warm boots, and the Mac went back to its fleet78 default after each one.

boot failed units AVC kernel err oops fw crash s2idle
1 0 0 0 0 0 2/2
2 0 0 0 0 0 1/1
3 0 0 0 0 0 1/1
  • AIC: 139 IRQs on AIC2, all 14 CPUs online. Effective affinity is spread over 0-13 (118 IRQs), and no IRQ has an offline CPU in its effective mask.
  • dEQP-VK smoke (J=4): 16541 cases, 8951 Pass, 7590 NotSupported, 0 Fail, 0 Crash.
  • One G17 render timeout (recovery state 0) appeared during the smoke. The same message is already on the fleet78 default on this Mac (fleet78 row dmesg), so it is not from this PR.

The same kernel passed earlier on the M1 Pro, M2, Neo and J615 (posted above). With Dj's APPROVE and our review at this head, the train now merges it.

@RyanTheTide
RyanTheTide merged commit cc5c09b into aurora-silicon:integration/aurora Nov 22, 2026
1 check 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