Nadav noticed that the cpumask allocations in native_send_call_func_ipi() are noticeable in microbenchmarks.
Use the new cpumask_or_equal() function to simplify the decision whether the supplied target CPU mask is either equal to cpu_online_mask or equal to cpu_online_mask except for the CPU on which the function is invoked. cpumask_or_equal() or's the target mask and the cpumask of the current CPU together and compares it to cpu_online_mask. If the result is false, use the mask based IPI function, otherwise check whether the current CPU is set in the target mask and invoke either the send_IPI_all() or the send_IPI_allbutselt() APIC callback. Make the shorthand decision also depend on the static key which enables shorthand mode. That allows to remove the extra cpumask comparison with cpu_callout_mask. Reported-by: Nadav Amit <na...@vmware.com> Signed-off-by: Thomas Gleixner <t...@linutronix.de> --- V2: New patch --- arch/x86/kernel/apic/ipi.c | 24 +++++++++++------------- 1 file changed, 11 insertions(+), 13 deletions(-) --- a/arch/x86/kernel/apic/ipi.c +++ b/arch/x86/kernel/apic/ipi.c @@ -83,23 +83,21 @@ void native_send_call_func_single_ipi(in void native_send_call_func_ipi(const struct cpumask *mask) { - cpumask_var_t allbutself; + if (static_branch_likely(&apic_use_ipi_shorthand)) { + unsigned int cpu = smp_processor_id(); - if (!alloc_cpumask_var(&allbutself, GFP_ATOMIC)) { - apic->send_IPI_mask(mask, CALL_FUNCTION_VECTOR); + if (!cpumask_or_equal(mask, cpumask_of(cpu), cpu_online_mask)) + goto sendmask; + + if (cpumask_test_cpu(cpu, mask)) + apic->send_IPI_all(CALL_FUNCTION_VECTOR); + else if (num_online_cpus() > 1) + apic->send_IPI_allbutself(CALL_FUNCTION_VECTOR); return; } - cpumask_copy(allbutself, cpu_online_mask); - cpumask_clear_cpu(smp_processor_id(), allbutself); - - if (cpumask_equal(mask, allbutself) && - cpumask_equal(cpu_online_mask, cpu_callout_mask)) - apic->send_IPI_allbutself(CALL_FUNCTION_VECTOR); - else - apic->send_IPI_mask(mask, CALL_FUNCTION_VECTOR); - - free_cpumask_var(allbutself); +sendmask: + apic->send_IPI_mask(mask, CALL_FUNCTION_VECTOR); } #endif /* CONFIG_SMP */