On 01/11/17 15:59, James Morse wrote: > dpm_suspend() calls the freeze/thaw callbacks for hibernate before > disable_non_bootcpus() takes down secondaries. > > This leads to a fun race where the freeze/thaw callbacks reset the > SDEI interface (as we may be restoring a kernel with a different > layout due to KASLR), then the cpu-hotplug callbacks come in to > save the current state, which has already been reset. > > We solve this with a 'frozen' flag that stops the hotplug callback > from overwriting the saved values. > > This patch moves the flag under the 'events' spinlock we take > in the hotplug callbacks, to avoid depending on cpu-hotplug's > mechanics to ensure the callback sees the correct value. Scratch this. All this has really done is moved the race around. Will had a much better suggestion that makes it look like all the CPUs are down. That makes hotplug and power-management behave the same. Thanks, James _______________________________________________ kvmarm mailing list kvmarm@xxxxxxxxxxxxxxxxxxxxx https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
