x86, 32-bit: improve lazy TLB handling code
Impact: micro-optimize the 32-bit TLB flush code
Use the faster x86_{read,write}_percpu() accessors here.
Signed-off-by: Jan Beulich <jbeulich@novell.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
diff --git a/arch/x86/kernel/tlb_32.c b/arch/x86/kernel/tlb_32.c
index f4049f3..4290d91 100644
--- a/arch/x86/kernel/tlb_32.c
+++ b/arch/x86/kernel/tlb_32.c
@@ -34,9 +34,8 @@
*/
void leave_mm(int cpu)
{
- if (per_cpu(cpu_tlbstate, cpu).state == TLBSTATE_OK)
- BUG();
- cpu_clear(cpu, per_cpu(cpu_tlbstate, cpu).active_mm->cpu_vm_mask);
+ BUG_ON(x86_read_percpu(cpu_tlbstate.state) == TLBSTATE_OK);
+ cpu_clear(cpu, x86_read_percpu(cpu_tlbstate.active_mm)->cpu_vm_mask);
load_cr3(swapper_pg_dir);
}
EXPORT_SYMBOL_GPL(leave_mm);
@@ -104,8 +103,8 @@
* BUG();
*/
- if (flush_mm == per_cpu(cpu_tlbstate, cpu).active_mm) {
- if (per_cpu(cpu_tlbstate, cpu).state == TLBSTATE_OK) {
+ if (flush_mm == x86_read_percpu(cpu_tlbstate.active_mm)) {
+ if (x86_read_percpu(cpu_tlbstate.state) == TLBSTATE_OK) {
if (flush_va == TLB_FLUSH_ALL)
local_flush_tlb();
else
@@ -238,7 +237,7 @@
unsigned long cpu = smp_processor_id();
__flush_tlb_all();
- if (per_cpu(cpu_tlbstate, cpu).state == TLBSTATE_LAZY)
+ if (x86_read_percpu(cpu_tlbstate.state) == TLBSTATE_LAZY)
leave_mm(cpu);
}