For a metric like: EVENT1 if #smt_on else EVENT2 currently EVENT1 and EVENT2 will be measured and then when the metric is reported EVENT1 or EVENT2 will be printed depending on the value from smt_on() during the expr parsing. Computing both events is unnecessary and can lead to multiplexing as discussed in this thread: https://lore.kernel.org/lkml/20201110100346.2527031-1-irog...@google.com/
This change modifies expression parsing so that constants are considered when building the set of ids (events) and only events not contributing to a constant value are measured. v7. fixes the fix to be in the correct patch. v6. rebases and fixes issues raised by Namhyung Kim <namhy...@kernel.org>, a memory leak and a function comment. v5. uses macros to reduce boiler plate in patch 5/5 as suggested by Andi Kleen <a...@linux.intel.com>. v4. reduces references to BOTTOM/NAN in patch 5/5 by using utility functions. It improves comments and fixes an unnecessary union in a peephole optimization. v3. fixes an assignment in patch 2/5. In patch 5/5 additional comments are added and useless frees are replaced by asserts. A new peephole optimization is added for the case CONST IF expr ELSE CONST, where the the constants are identical, as we don't need to evaluate the IF condition. v2. is a rebase. Ian Rogers (5): perf metric: Restructure struct expr_parse_ctx. perf metric: Use NAN for missing event IDs. perf metric: Rename expr__find_other. perf metric: Add utilities to work on ids map. perf metric: Don't compute unused events. tools/perf/tests/expr.c | 159 +++++++++++----- tools/perf/tests/pmu-events.c | 42 +++-- tools/perf/util/expr.c | 137 ++++++++++++-- tools/perf/util/expr.h | 21 ++- tools/perf/util/expr.l | 9 - tools/perf/util/expr.y | 343 ++++++++++++++++++++++++++-------- tools/perf/util/metricgroup.c | 44 +++-- tools/perf/util/stat-shadow.c | 54 ++++-- 8 files changed, 591 insertions(+), 218 deletions(-) -- 2.30.0.284.gd98b1dd5eaa7-goog