Hi, By default, the userspace perf tool opens per-cpu task-bound events when sampling, so for N logical events requested by the user, the tool will open N * NR_CPUS events.
In the kernel, we mux events with a hrtimer, periodically rotating the flexible group list and trying to schedule each group in turn. We skip groups whose cpu filter doesn't match. So when we get unlucky, we can walk N * (NR_CPUS - 1) groups pointlessly for each hrtimer invocation. This has been observed to result in significant overhead when running the STREAM benchmark on 272 core Xeon Phi systems. One way to avoid this is to place our events into an rb tree sorted by CPU, so that our hrtimer can skip to the current CPU's list and ignore everything else. This patch set moves event groups into rb trees and implements skipping to the current CPU's list on hrtimer interrupt. The patch set was tested on Xeon Phi using perf_fuzzer and tests from here: https://github.com/deater/perf_event_tests Patches in the set are expected to be applied one after another in the mentioned order and they are logically split here into three parts to simplify the review process. Thanks, Alexey --- Alexey Budankov (3): perf/core: use rb trees for pinned/flexible groups perf/core: use context tstamp_data for skipped events on mux interrupt perf/core: add mux switch to skip to the current CPU's events list on mux interrupt include/linux/perf_event.h | 54 +++-- kernel/events/core.c | 584 +++++++++++++++++++++++++++++++++------------ 2 files changed, 473 insertions(+), 165 deletions(-)
