Generate flame graphs from the 'perf' profile.

This is less helpful than expected on both oil.ovm-dbg and oil.ovm-opt.

It works at a function level, and the interpreter loop is in a single
function with a big switch statement.

benchmarks/perf.sh

@@ -62,6 +62,7 @@ debug-symbols() {
#
# Wow.
# Maybe I need counters in optimized mode?
# Yeah what I really want is per opcode total!

_record() {

@@ -70,7 +71,15 @@ _record() {
  # work?
  #perf record -o perf.data -- _bin/oil.ovm-dbg osh -n benchmarks/testdata/abuild | wc -l

  perf record -o perf.data -- _bin/oil.ovm-opt osh --ast-format none -n benchmarks/testdata/abuild
  # call graph recording.  This helps it be less "flat" in opt mode.  Otherwise
  # everything is PyEval_EvalFrameEx.
  local flag='-g'
  local bin=_bin/oil.ovm-opt 

  # This shows more details
  #local bin=_bin/oil.ovm-dbg

  perf record $flag -o perf.data -- $bin osh --ast-format none -n benchmarks/testdata/abuild
  #perf record -o perf.data -- _bin/osh --ast-format none -n benchmarks/testdata/abuild
}
record() { sudo $0 _record; }
@@ -82,6 +91,10 @@ record() { sudo $0 _record; }
# After recording, run perf-data, then 'perf report'.  It automatically shows
# perf.data.

report() {
  perf report -n --stdio "$@"
}

_perf-data() {
  # This gets run as root
  chmod 644 perf.data
@@ -99,4 +112,19 @@ _stat() {
}
stat() { sudo $0 _stat; }

# NOTE: I used this before with python-flamegraph too.
flamegraph() {
  ~/git/other/FlameGraph/flamegraph.pl "$@"
}

stackcollapse-perf() {
  ~/git/other/FlameGraph/stackcollapse-perf.pl "$@"
}

# http://www.brendangregg.com/FlameGraphs/cpuflamegraphs.html
make-graph() {
  perf script | stackcollapse-perf > out.perf-folded
  flamegraph out.perf-folded > perf-kernel.svg
}

"$@"