[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <2abe618d-a2c4-3b22-ac9d-37bc91d05d41@linux.intel.com>
Date: Thu, 18 May 2023 15:47:25 -0400
From: "Liang, Kan" <kan.liang@...ux.intel.com>
To: Ian Rogers <irogers@...gle.com>,
Peter Zijlstra <peterz@...radead.org>,
Ingo Molnar <mingo@...hat.com>,
Arnaldo Carvalho de Melo <acme@...nel.org>,
Mark Rutland <mark.rutland@....com>,
Alexander Shishkin <alexander.shishkin@...ux.intel.com>,
Jiri Olsa <jolsa@...nel.org>,
Namhyung Kim <namhyung@...nel.org>,
Adrian Hunter <adrian.hunter@...el.com>,
James Clark <james.clark@....com>,
Andrii Nakryiko <andrii@...nel.org>,
Eduard Zingerman <eddyz87@...il.com>,
linux-perf-users@...r.kernel.org, linux-kernel@...r.kernel.org,
Ahmad Yasin <ahmad.yasin@...el.com>,
Stephane Eranian <eranian@...gle.com>,
Andi Kleen <ak@...ux.intel.com>,
Perry Taylor <perry.taylor@...el.com>,
Samantha Alt <samantha.alt@...el.com>,
Caleb Biggers <caleb.biggers@...el.com>,
Weilin Wang <weilin.wang@...el.com>,
Edward Baker <edward.baker@...el.com>
Subject: Re: [PATCH v1 1/2] perf expr: Make the evaluation of & and | logical
and lazy
On 2023-05-04 3:58 p.m., Ian Rogers wrote:
> Currently the & and | operators are only used in metric thresholds
> like (from the tma_retiring metric):
> tma_retiring > 0.7 | tma_heavy_operations > 0.1
>
> Thresholds are always computed when present, but a lack events may
> mean the threshold can't be computed. This happens with the option
> --metric-no-threshold for say the metric tma_retiring on Tigerlake
> model CPUs. To fully compute the threshold tma_heavy_operations is
> needed and it needs the extra events of IDQ.MS_UOPS,
> UOPS_DECODED.DEC0, cpu/UOPS_DECODED.DEC0,cmask=1/ and
> IDQ.MITE_UOPS. So --metric-no-threshold is a useful option to reduce
> the number of events needed and potentially multiplexing of events.
>
> Rather than just fail threshold computations like this, we may know a
> result from just the left or right-hand side. So, for tma_retiring if
> its value is "> 0.7" we know it is over the threshold. This allows the
> metric to have the threshold coloring, when possible, without all the
> counters being programmed.
>
> Signed-off-by: Ian Rogers <irogers@...gle.com>
> ---
The patch works well on my machine.
Tested-by: Kan Liang <kan.liang@...ux.intel.com>
Thanks,
Kan
> tools/perf/tests/expr.c | 40 +++++++++++++++++++
> tools/perf/util/expr.y | 86 +++++++++++++++++++++++++++++++++--------
> 2 files changed, 109 insertions(+), 17 deletions(-)
>
> diff --git a/tools/perf/tests/expr.c b/tools/perf/tests/expr.c
> index cbf0e0c74906..45c7fedb797a 100644
> --- a/tools/perf/tests/expr.c
> +++ b/tools/perf/tests/expr.c
> @@ -184,6 +184,46 @@ static int test__expr(struct test_suite *t __maybe_unused, int subtest __maybe_u
> NULL, ctx) == 0);
> TEST_ASSERT_VAL("find ids", hashmap__size(ctx->ids) == 0);
>
> + /* The expression is a constant 0.0 without needing to evaluate EVENT1. */
> + expr__ctx_clear(ctx);
> + TEST_ASSERT_VAL("find ids",
> + expr__find_ids("0 & EVENT1 > 0", NULL, ctx) == 0);
> + TEST_ASSERT_VAL("find ids", hashmap__size(ctx->ids) == 0);
> + expr__ctx_clear(ctx);
> + TEST_ASSERT_VAL("find ids",
> + expr__find_ids("EVENT1 > 0 & 0", NULL, ctx) == 0);
> + TEST_ASSERT_VAL("find ids", hashmap__size(ctx->ids) == 0);
> + expr__ctx_clear(ctx);
> + TEST_ASSERT_VAL("find ids",
> + expr__find_ids("1 & EVENT1 > 0", NULL, ctx) == 0);
> + TEST_ASSERT_VAL("find ids", hashmap__size(ctx->ids) == 1);
> + TEST_ASSERT_VAL("find ids", hashmap__find(ctx->ids, "EVENT1", &val_ptr));
> + expr__ctx_clear(ctx);
> + TEST_ASSERT_VAL("find ids",
> + expr__find_ids("EVENT1 > 0 & 1", NULL, ctx) == 0);
> + TEST_ASSERT_VAL("find ids", hashmap__size(ctx->ids) == 1);
> + TEST_ASSERT_VAL("find ids", hashmap__find(ctx->ids, "EVENT1", &val_ptr));
> +
> + /* The expression is a constant 1.0 without needing to evaluate EVENT1. */
> + expr__ctx_clear(ctx);
> + TEST_ASSERT_VAL("find ids",
> + expr__find_ids("1 | EVENT1 > 0", NULL, ctx) == 0);
> + TEST_ASSERT_VAL("find ids", hashmap__size(ctx->ids) == 0);
> + expr__ctx_clear(ctx);
> + TEST_ASSERT_VAL("find ids",
> + expr__find_ids("EVENT1 > 0 | 1", NULL, ctx) == 0);
> + TEST_ASSERT_VAL("find ids", hashmap__size(ctx->ids) == 0);
> + expr__ctx_clear(ctx);
> + TEST_ASSERT_VAL("find ids",
> + expr__find_ids("0 | EVENT1 > 0", NULL, ctx) == 0);
> + TEST_ASSERT_VAL("find ids", hashmap__size(ctx->ids) == 1);
> + TEST_ASSERT_VAL("find ids", hashmap__find(ctx->ids, "EVENT1", &val_ptr));
> + expr__ctx_clear(ctx);
> + TEST_ASSERT_VAL("find ids",
> + expr__find_ids("EVENT1 > 0 | 0", NULL, ctx) == 0);
> + TEST_ASSERT_VAL("find ids", hashmap__size(ctx->ids) == 1);
> + TEST_ASSERT_VAL("find ids", hashmap__find(ctx->ids, "EVENT1", &val_ptr));
> +
> /* Test toplogy constants appear well ordered. */
> expr__ctx_clear(ctx);
> TEST_ASSERT_VAL("#num_cpus", expr__parse(&num_cpus, ctx, "#num_cpus") == 0);
> diff --git a/tools/perf/util/expr.y b/tools/perf/util/expr.y
> index 250e444bf032..6b110f9f95c9 100644
> --- a/tools/perf/util/expr.y
> +++ b/tools/perf/util/expr.y
> @@ -123,20 +123,6 @@ static struct ids handle_id(struct expr_parse_ctx *ctx, char *id,
> * constant value using OP. Its invariant that there are no ids. If computing
> * ids for non-constants union the set of IDs that must be computed.
> */
> -#define BINARY_LONG_OP(RESULT, OP, LHS, RHS) \
> - if (!compute_ids || (is_const(LHS.val) && is_const(RHS.val))) { \
> - assert(LHS.ids == NULL); \
> - assert(RHS.ids == NULL); \
> - if (isnan(LHS.val) || isnan(RHS.val)) { \
> - RESULT.val = NAN; \
> - } else { \
> - RESULT.val = (long)LHS.val OP (long)RHS.val; \
> - } \
> - RESULT.ids = NULL; \
> - } else { \
> - RESULT = union_expr(LHS, RHS); \
> - }
> -
> #define BINARY_OP(RESULT, OP, LHS, RHS) \
> if (!compute_ids || (is_const(LHS.val) && is_const(RHS.val))) { \
> assert(LHS.ids == NULL); \
> @@ -213,9 +199,75 @@ expr: NUMBER
> }
> | ID { $$ = handle_id(ctx, $1, compute_ids, /*source_count=*/false); }
> | SOURCE_COUNT '(' ID ')' { $$ = handle_id(ctx, $3, compute_ids, /*source_count=*/true); }
> -| expr '|' expr { BINARY_LONG_OP($$, |, $1, $3); }
> -| expr '&' expr { BINARY_LONG_OP($$, &, $1, $3); }
> -| expr '^' expr { BINARY_LONG_OP($$, ^, $1, $3); }
> +| expr '|' expr
> +{
> + if (is_const($1.val) && is_const($3.val)) {
> + assert($1.ids == NULL);
> + assert($3.ids == NULL);
> + $$.ids = NULL;
> + $$.val = (fpclassify($1.val) == FP_ZERO && fpclassify($3.val) == FP_ZERO) ? 0 : 1;
> + } else if (is_const($1.val)) {
> + assert($1.ids == NULL);
> + if (fpclassify($1.val) == FP_ZERO) {
> + $$ = $3;
> + } else {
> + $$.val = 1;
> + $$.ids = NULL;
> + ids__free($3.ids);
> + }
> + } else if (is_const($3.val)) {
> + assert($3.ids == NULL);
> + if (fpclassify($3.val) == FP_ZERO) {
> + $$ = $1;
> + } else {
> + $$.val = 1;
> + $$.ids = NULL;
> + ids__free($1.ids);
> + }
> + } else {
> + $$ = union_expr($1, $3);
> + }
> +}
> +| expr '&' expr
> +{
> + if (is_const($1.val) && is_const($3.val)) {
> + assert($1.ids == NULL);
> + assert($3.ids == NULL);
> + $$.val = (fpclassify($1.val) != FP_ZERO && fpclassify($3.val) != FP_ZERO) ? 1 : 0;
> + $$.ids = NULL;
> + } else if (is_const($1.val)) {
> + assert($1.ids == NULL);
> + if (fpclassify($1.val) != FP_ZERO) {
> + $$ = $3;
> + } else {
> + $$.val = 0;
> + $$.ids = NULL;
> + ids__free($3.ids);
> + }
> + } else if (is_const($3.val)) {
> + assert($3.ids == NULL);
> + if (fpclassify($3.val) != FP_ZERO) {
> + $$ = $1;
> + } else {
> + $$.val = 0;
> + $$.ids = NULL;
> + ids__free($1.ids);
> + }
> + } else {
> + $$ = union_expr($1, $3);
> + }
> +}
> +| expr '^' expr
> +{
> + if (is_const($1.val) && is_const($3.val)) {
> + assert($1.ids == NULL);
> + assert($3.ids == NULL);
> + $$.val = (fpclassify($1.val) == FP_ZERO) != (fpclassify($3.val) == FP_ZERO) ? 1 : 0;
> + $$.ids = NULL;
> + } else {
> + $$ = union_expr($1, $3);
> + }
> +}
> | expr '<' expr { BINARY_OP($$, <, $1, $3); }
> | expr '>' expr { BINARY_OP($$, >, $1, $3); }
> | expr '+' expr { BINARY_OP($$, +, $1, $3); }
Powered by blists - more mailing lists