83c001d3c2
Reduce the cost of TLB invalidation on x86 by using per-CPU completion flags Reduce contention during TLB invalidation operations by using a per-CPU completion flag, rather than a single atomically-updated variable. On a Westmere system (2 sockets x 4 cores x 1 threads), dtrace measurements show that smp_tlb_shootdown is about 50% faster with this patch; observations with VTune show that the percentage of time spent in invlrng_single_page on an interrupt (actually doing invalidation, rather than synchronization) increases from 31% with the old mechanism to 71% with the new one. (Running a basic file server workload.) Submitted by: Anton Rang <rang at acm.org> Reviewed by: cem (earlier version) Sponsored by: Dell EMC Isilon Differential Revision: https://reviews.freebsd.org/D8041 |
||
|---|---|---|
| .. | ||
| xen | ||
| _align.h | ||
| _inttypes.h | ||
| _limits.h | ||
| _stdint.h | ||
| _types.h | ||
| acpica_machdep.h | ||
| apicreg.h | ||
| apicvar.h | ||
| apm_bios.h | ||
| bus.h | ||
| busdma_impl.h | ||
| cputypes.h | ||
| dump.h | ||
| elf.h | ||
| endian.h | ||
| fdt.h | ||
| float.h | ||
| fpu.h | ||
| frame.h | ||
| init.h | ||
| legacyvar.h | ||
| mca.h | ||
| metadata.h | ||
| mptable.h | ||
| ofw_machdep.h | ||
| pci_cfgreg.h | ||
| psl.h | ||
| ptrace.h | ||
| pvclock.h | ||
| reg.h | ||
| segments.h | ||
| setjmp.h | ||
| sigframe.h | ||
| signal.h | ||
| specialreg.h | ||
| stack.h | ||
| stdarg.h | ||
| sysarch.h | ||
| trap.h | ||
| ucontext.h | ||
| vdso.h | ||
| vmware.h | ||
| x86_smp.h | ||
| x86_var.h | ||