While we've long done away with CONFIG_SMP, we still allow CONFIG_NR_CPUS to be set to 1. Hence at least randconfig builds may fail, and the first of the two issues addressed was actually observed in the RISC-V bring-up work. I didn't check whether Arm would also have issues like these.
1: sched: fix build when NR_CPUS == 1 2: x86: fix build when NR_CPUS == 1 Jan