[qcom-6.18.y] [rb3gen2] Hotplug test failed

[sgaud-quic]

Failure : Hotplug test failed
Kernel Version : v6.18.30
Branch : qcom-6.18.y
Tag : qcom-6.18.y-20260527
Failed job : https://lava.infra.foundries.io/scheduler/job/236168

[FAIL] 1970-01-01 00:00:46 - Failed to offline cpu5
[INFO] 1970-01-01 00:00:46 - Onlining cpu5
[PASS] 1970-01-01 00:00:46 - /sys/devices/system/cpu/cpu5/online is online as expected
[   41.954572] gcc-sc7280 100000.clock-controller: sync_state() pending due to 3d6a000.gmu
[   41.970259] gpu_cc-sc7280 3d90000.clock-controller: sync_state() pending due to 3d6a000.gmu
[INFO] 1970-01-01 00:00:47 - Offlining cpu6
/home/lava-236168/0/tests/0_hotplug/Runner/suites/Kernel/Baseport/hotplug/run.sh: line 46: echo: write error: Device or resource busy
[FAIL] 1970-01-01 00:00:48 - Failed to offline cpu6
[INFO] 1970-01-01 00:00:48 - Onlining cpu6
[PASS] 1970-01-01 00:00:48 - /sys/devices/system/cpu/cpu6/online is online as expected
[INFO] 1970-01-01 00:00:49 - Offlining cpu7
/home/lava-236168/0/tests/0_hotplug/Runner/suites/Kernel/Baseport/hotplug/run.sh: line 46: echo: write error: Device or resource busy
[   44.604250] xhci-pci-renesas 0001:04:00.0: can't setup: -110
[   44.610087] xhci-pci-renesas 0001:04:00.0: USB bus 1 deregistered
[   44.616375] xhci-pci-renesas 0001:04:00.0: init 0001:04:00.0 fail, -110
[   44.623188] xhci-pci-renesas 0001:04:00.0: probe with driver xhci-pci-renesas failed with error -110
[FAIL] 1970-01-01 00:00:50 - Failed to offline cpu7
[INFO] 1970-01-01 00:00:50 - Onlining cpu7
[PASS] 1970-01-01 00:00:50 - /sys/devices/system/cpu/cpu7/online is online as expected
[INFO] 1970-01-01 00:00:51 - Final CPU status:
tee: '': No such file or directory
1
1
1
1
1
1
1
1
[FAIL] 1970-01-01 00:00:51 - hotplug : Test Failed
+ true
+ /home/lava-236168/0/tests/0_hotplug/Runner/utils/send-to-lava.sh /home/lava-236168/0/tests/0_hotplug/Runner/suites/Kernel/Baseport/hotplug/hotplug.res
<<<LAVA_SIGNAL_TESTCASE TEST_CASE_ID=hotplug RESULT=FAIL>>>

[smuppand]

If an offline request returns EBUSY, it as runtime busy in sanity mode and retry before marking it as skip/warn. In strict mode, treat EBUSY as failure.

This avoids false failures while still validating real hotplug functional behavior, including CPU0 when supported by the platform. Fixed with qualcomm-linux/qcom-linux-testkit#463