Hi, I'm sitting on a few performance improvements that I'm hoping to try get polished up enough to merge, but it's taking a while, so I juts post them out for review because I think most are at the stage where they are good enough to start getting some reviews on.
After this series, light weight context switch (yield, threads, same CPU) improves about 10% on my POWER8 (2.1m -> 2.3m per second with powernv_defconfig and tick based time accounting). Ping-pong context switches improve similarly, particularly when you force them to go to nap. I'm still gathering up numbers. I haven't been able to get POWER9 numbers yet. Thanks, Nick Nicholas Piggin (14): powerpc/64s: optimize hypercall/syscall powerpc/64: syscall avoid restore_math call if possible powerpc/64s: idle move soft interrupt mask logic into C code powerpc/64s: process interrupts from system reset wakeup powerpc/64s: msgclr when handling doorbell exceptions powerpc/64s: branch to idle handler with virtual mode offset powerpc/64s: idle avoid SRR usage in idle sleep/wake paths powerpc/64s: idle set polling before enabling irqs powerpc/64s: idle read mostly for common globals powerpc/64: CTRL[RUN] run-latch setting optimisation powerpc/64s: idle no memory barrier after break from idle powerpc/64s: Leave IRQs hard enabled over context switch for radix powerpc/64: context switch can avoid reservation clear powerpc/64: context switch additional hwsync can be avoided arch/powerpc/include/asm/barrier.h | 4 + arch/powerpc/include/asm/dbell.h | 13 +++ arch/powerpc/include/asm/exception-64s.h | 17 ++- arch/powerpc/include/asm/hw_irq.h | 1 + arch/powerpc/include/asm/machdep.h | 1 + arch/powerpc/include/asm/ppc-opcode.h | 3 + arch/powerpc/include/asm/processor.h | 8 +- arch/powerpc/kernel/asm-offsets.c | 1 + arch/powerpc/kernel/entry_64.S | 94 +++++++++------ arch/powerpc/kernel/exceptions-64s.S | 191 ++++++++++++++++++++++++------- arch/powerpc/kernel/idle_book3s.S | 135 ++++++---------------- arch/powerpc/kernel/process.c | 30 +++-- arch/powerpc/kvm/book3s_hv_rmhandlers.S | 8 +- arch/powerpc/platforms/powernv/idle.c | 122 ++++++++++++++++++-- arch/powerpc/platforms/powernv/subcore.c | 3 +- drivers/cpuidle/cpuidle-powernv.c | 37 +++--- drivers/cpuidle/cpuidle-pseries.c | 22 ++-- kernel/sched/core.c | 9 ++ 18 files changed, 469 insertions(+), 230 deletions(-) -- 2.11.0