Skip to content

perf-core-2020-06-01

tagged this 01 Jun 15:31
  - Add AMD Fam17h RAPL support
  - Introduce CAP_PERFMON to kernel and user space
  - Add Zhaoxin CPU support
  - Misc fixes and cleanups

Tooling changes:

  perf record:

    - Introduce --switch-output-event to use arbitrary events to be setup
      and read from a side band thread and, when they take place a signal
      be sent to the main 'perf record' thread, reusing the --switch-output
      code to take perf.data snapshots from the --overwrite ring buffer, e.g.:

	# perf record --overwrite -e sched:* \
		      --switch-output-event syscalls:*connect* \
		      workload

      will take perf.data.YYYYMMDDHHMMSS snapshots up to around the
      connect syscalls.

    - Add --num-synthesize-threads option to control degree of parallelism of the
      synthesize_mmap() code which is scanning /proc/PID/task/PID/maps and can be
      time consuming. This mimics pre-existing behaviour in 'perf top'.

  perf bench:

    - Add a multi-threaded synthesize benchmark.
    - Add kallsyms parsing benchmark.

  Intel PT support:

    - Stitch LBR records from multiple samples to get deeper backtraces,
      there are caveats, see the csets for details.
    - Allow using Intel PT to synthesize callchains for regular events.
    - Add support for synthesizing branch stacks for regular events (cycles,
      instructions, etc) from Intel PT data.

  Misc changes:

    - Updated perf vendor events for power9 and Coresight.
    - Add flamegraph.py script via 'perf flamegraph'
    - Misc other changes, fixes and cleanups - see the Git log for details.

Signed-off-by: Ingo Molnar <mingo@kernel.org>
Assets 2
Loading