This bug occurs very rarely.
This bug is not specific to MP or IKS. I have seen this bug on the both the kernels when we were build IKS and MP separately.
The probability of bug occurrence is CPU hotplug/ and unplug in a loop. You can ref the bug description for more info. Where i was doing hotplug and unplug available CPUs in random order.
The reason we found this found while running cpu_hotplug_latency_with_load & cpu_hotplug_latency_without_load is that test case performs CPU hotplug/unplug continuously in a loop to measure the latency of CPU hotplug.
This is the one of the evidence to confirm it happended when "cpu_down" called in Kernel.
[ 1265.306986] [<c03cb7ef>] (_cpu_down+0xb7/0x1b4) from [<c03cb907>] (cpu_down+0x1b/0x30)
[ 1265.330348] [<c03cb907>] (cpu_down+0x1b/0x30) from [<c03cc04b>] (store_online+0x27/0x50)
Hi Basil,
This bug occurs very rarely.
This bug is not specific to MP or IKS. I have seen this bug on the both the kernels when we were build IKS and MP separately.
The probability of bug occurrence is CPU hotplug/ and unplug in a loop. You can ref the bug description for more info. Where i was doing hotplug and unplug available CPUs in random order.
The reason we found this found while running cpu_hotplug_ latency_ with_load & cpu_hotplug_ latency_ without_ load is that test case performs CPU hotplug/unplug continuously in a loop to measure the latency of CPU hotplug.
This is the one of the evidence to confirm it happended when "cpu_down" called in Kernel.
[ 1265.306986] [<c03cb7ef>] (_cpu_down+ 0xb7/0x1b4) from [<c03cb907>] (cpu_down+ 0x1b/0x30) 0x1b/0x30) from [<c03cc04b>] (store_ online+ 0x27/0x50)
[ 1265.330348] [<c03cb907>] (cpu_down+
-Naresh