From nobody Tue Dec 16 05:55:46 2025 Received: from smtp.kernel.org (aws-us-west-2-korg-mail-1.web.codeaurora.org [10.30.226.201]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 649B8214A68; Wed, 5 Feb 2025 20:54:45 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=10.30.226.201 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1738788886; cv=none; b=pwQDOWoTRZvPRUbxmR8BZc1n3Rlgr7TpWNKybntYpAYQjtLnWrJ3y+QZpnzcoTOUGOxWDEjF7zdYOlKsPNjH/wFqVe1DRkOVNGs+gZnROaqzPtxuijTlrxg+DP86BbzAn3Z/DD33RPSumyabMWBGvvgm6Lm2vE1LhRMntLbPVK8= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1738788886; c=relaxed/simple; bh=/WTAgk9sQpEsxfpN55ZnoqFPeaI1ULSneciI0bQqFs8=; h=From:To:Cc:Subject:Date:Message-ID:In-Reply-To:References: MIME-Version; b=pbyHyaU+KYGvBjRAORBCohSfoUo5rb6oyy629g/yGrDkXfsV/Qg4zlE82iJs28fOKKXGcHLX2Fk8uqVmxGK+FcxG7YrCdy0GOLr60B8jz+45vlgC+7rx/TPwAxcyDAHi2pWyJJ+qYp9acOa7Sa1c63hUUD4vanYchap9R4T9E4Q= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=kernel.org header.i=@kernel.org header.b=CrpcE3qm; arc=none smtp.client-ip=10.30.226.201 Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=kernel.org header.i=@kernel.org header.b="CrpcE3qm" Received: by smtp.kernel.org (Postfix) with ESMTPSA id 83B56C4CEE2; Wed, 5 Feb 2025 20:54:45 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=kernel.org; s=k20201202; t=1738788885; bh=/WTAgk9sQpEsxfpN55ZnoqFPeaI1ULSneciI0bQqFs8=; h=From:To:Cc:Subject:Date:In-Reply-To:References:From; b=CrpcE3qmgBL2kUjv+pgn/w2QGlwq4FJEr+oNRUFiNS3ZCSRtTJ5xjsELyxqbzs4UH cne4yoP02b8jFsl6U3HDYnYPegeDNWiATtIsBvZZUZBip/FLfe0wRIb1Gq/d8o2A/a SShksRmVdxvc+8Vzw2MBW8gwPj34FhCUAN7ojYaMqeLLEAso3VruOAz4AiHoFRxYbu Q2Q7pDMTZNg1XAxdQnxBqZI1S37kjh6CSQRyLQF+qQ7BqZADbukzAP29ii91sqgR3j 9FFB3R/mP6uA2AnmHXzMyVnAyR9rjvTMZMyt29lybyKtdyC3csbSqyYbfsZpEVZQho gQDk0vxje/lxg== From: Namhyung Kim To: Arnaldo Carvalho de Melo , Ian Rogers , Kan Liang Cc: Jiri Olsa , Adrian Hunter , Peter Zijlstra , Ingo Molnar , LKML , linux-perf-users@vger.kernel.org, Howard Chu Subject: [PATCH v3 2/4] perf trace: Convert syscall_stats to hashmap Date: Wed, 5 Feb 2025 12:54:41 -0800 Message-ID: <20250205205443.1986408-3-namhyung@kernel.org> X-Mailer: git-send-email 2.48.1.502.g6dc24dfdaf-goog In-Reply-To: <20250205205443.1986408-1-namhyung@kernel.org> References: <20250205205443.1986408-1-namhyung@kernel.org> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" It was using a RBtree-based int-list as a hash and a custom resort logic for that. As we have hashmap, let's convert to it and add a custom sort function for the hashmap entries using an array. It should be faster and more light-weighted. It's also to prepare supporting system-wide syscall stats. No functional changes intended. Acked-by: Howard Chu Signed-off-by: Namhyung Kim --- tools/perf/builtin-trace.c | 117 ++++++++++++++++++++++++++++--------- 1 file changed, 88 insertions(+), 29 deletions(-) diff --git a/tools/perf/builtin-trace.c b/tools/perf/builtin-trace.c index 7e0324a2e9182088..5e37f05737b75a14 100644 --- a/tools/perf/builtin-trace.c +++ b/tools/perf/builtin-trace.c @@ -39,6 +39,7 @@ #include "util/synthetic-events.h" #include "util/evlist.h" #include "util/evswitch.h" +#include "util/hashmap.h" #include "util/mmap.h" #include #include @@ -63,7 +64,6 @@ #include "print_binary.h" #include "string2.h" #include "syscalltbl.h" -#include "rb_resort.h" #include "../perf.h" #include "trace_augment.h" =20 @@ -1519,17 +1519,50 @@ struct thread_trace { struct file *table; } files; =20 - struct intlist *syscall_stats; + struct hashmap *syscall_stats; }; =20 +static size_t syscall_id_hash(long key, void *ctx __maybe_unused) +{ + return key; +} + +static bool syscall_id_equal(long key1, long key2, void *ctx __maybe_unuse= d) +{ + return key1 =3D=3D key2; +} + +static struct hashmap *alloc_syscall_stats(void) +{ + return hashmap__new(syscall_id_hash, syscall_id_equal, NULL); +} + +static void delete_syscall_stats(struct hashmap *syscall_stats) +{ + struct hashmap_entry *pos; + size_t bkt; + + if (syscall_stats =3D=3D NULL) + return; + + hashmap__for_each_entry(syscall_stats, pos, bkt) + free(pos->pvalue); + hashmap__free(syscall_stats); +} + static struct thread_trace *thread_trace__new(struct trace *trace) { struct thread_trace *ttrace =3D zalloc(sizeof(struct thread_trace)); =20 if (ttrace) { ttrace->files.max =3D -1; - if (trace->summary) - ttrace->syscall_stats =3D intlist__new(NULL); + if (trace->summary) { + ttrace->syscall_stats =3D alloc_syscall_stats(); + if (IS_ERR(ttrace->syscall_stats)) { + free(ttrace); + ttrace =3D NULL; + } + } } =20 return ttrace; @@ -1544,7 +1577,7 @@ static void thread_trace__delete(void *pttrace) if (!ttrace) return; =20 - intlist__delete(ttrace->syscall_stats); + delete_syscall_stats(ttrace->syscall_stats); ttrace->syscall_stats =3D NULL; thread_trace__free_files(ttrace); zfree(&ttrace->entry_str); @@ -2463,22 +2496,19 @@ struct syscall_stats { static void thread__update_stats(struct thread *thread, struct thread_trac= e *ttrace, int id, struct perf_sample *sample, long err, bool errno_summary) { - struct int_node *inode; - struct syscall_stats *stats; + struct syscall_stats *stats =3D NULL; u64 duration =3D 0; =20 - inode =3D intlist__findnew(ttrace->syscall_stats, id); - if (inode =3D=3D NULL) - return; - - stats =3D inode->priv; - if (stats =3D=3D NULL) { + if (!hashmap__find(ttrace->syscall_stats, id, &stats)) { stats =3D zalloc(sizeof(*stats)); if (stats =3D=3D NULL) return; =20 init_stats(&stats->stats); - inode->priv =3D stats; + if (hashmap__add(ttrace->syscall_stats, id, stats) < 0) { + free(stats); + return; + } } =20 if (ttrace->entry_time && sample->time > ttrace->entry_time) @@ -4617,18 +4647,45 @@ static size_t trace__fprintf_threads_header(FILE *f= p) return printed; } =20 -DEFINE_RESORT_RB(syscall_stats, a->msecs > b->msecs, +struct syscall_entry { struct syscall_stats *stats; double msecs; int syscall; -) +}; + +static int entry_cmp(const void *e1, const void *e2) +{ + const struct syscall_entry *entry1 =3D e1; + const struct syscall_entry *entry2 =3D e2; + + return entry1->msecs > entry2->msecs ? -1 : 1; +} + +static struct syscall_entry *thread__sort_stats(struct thread_trace *ttrac= e) { - struct int_node *source =3D rb_entry(nd, struct int_node, rb_node); - struct syscall_stats *stats =3D source->priv; + struct syscall_entry *entry; + struct hashmap_entry *pos; + unsigned bkt, i, nr; + + nr =3D ttrace->syscall_stats->sz; + entry =3D malloc(nr * sizeof(*entry)); + if (entry =3D=3D NULL) + return NULL; + + i =3D 0; + hashmap__for_each_entry(ttrace->syscall_stats, pos, bkt) { + struct syscall_stats *ss =3D pos->pvalue; + struct stats *st =3D &ss->stats; =20 - entry->syscall =3D source->i; - entry->stats =3D stats; - entry->msecs =3D stats ? (u64)stats->stats.n * (avg_stats(&stats->stats= ) / NSEC_PER_MSEC) : 0; + entry[i].stats =3D ss; + entry[i].msecs =3D (u64)st->n * (avg_stats(st) / NSEC_PER_MSEC); + entry[i].syscall =3D pos->key; + i++; + } + assert(i =3D=3D nr); + + qsort(entry, nr, sizeof(*entry), entry_cmp); + return entry; } =20 static size_t thread__dump_stats(struct thread_trace *ttrace, @@ -4636,10 +4693,10 @@ static size_t thread__dump_stats(struct thread_trac= e *ttrace, { size_t printed =3D 0; struct syscall *sc; - struct rb_node *nd; - DECLARE_RESORT_RB_INTLIST(syscall_stats, ttrace->syscall_stats); + struct syscall_entry *entries; =20 - if (syscall_stats =3D=3D NULL) + entries =3D thread__sort_stats(ttrace); + if (entries =3D=3D NULL) return 0; =20 printed +=3D fprintf(fp, "\n"); @@ -4648,8 +4705,10 @@ static size_t thread__dump_stats(struct thread_trace= *ttrace, printed +=3D fprintf(fp, " (msec) = (msec) (msec) (msec) (%%)\n"); printed +=3D fprintf(fp, " --------------- -------- ------ -------- --= ------- --------- --------- ------\n"); =20 - resort_rb__for_each_entry(nd, syscall_stats) { - struct syscall_stats *stats =3D syscall_stats_entry->stats; + for (size_t i =3D 0; i < ttrace->syscall_stats->sz; i++) { + struct syscall_entry *entry =3D &entries[i]; + struct syscall_stats *stats =3D entry->stats; + if (stats) { double min =3D (double)(stats->stats.min) / NSEC_PER_MSEC; double max =3D (double)(stats->stats.max) / NSEC_PER_MSEC; @@ -4660,10 +4719,10 @@ static size_t thread__dump_stats(struct thread_trac= e *ttrace, pct =3D avg ? 100.0 * stddev_stats(&stats->stats) / avg : 0.0; avg /=3D NSEC_PER_MSEC; =20 - sc =3D &trace->syscalls.table[syscall_stats_entry->syscall]; + sc =3D &trace->syscalls.table[entry->syscall]; printed +=3D fprintf(fp, " %-15s", sc->name); printed +=3D fprintf(fp, " %8" PRIu64 " %6" PRIu64 " %9.3f %9.3f %9.3f", - n, stats->nr_failures, syscall_stats_entry->msecs, min, avg); + n, stats->nr_failures, entry->msecs, min, avg); printed +=3D fprintf(fp, " %9.3f %9.2f%%\n", max, pct); =20 if (trace->errno_summary && stats->nr_failures) { @@ -4677,7 +4736,7 @@ static size_t thread__dump_stats(struct thread_trace = *ttrace, } } =20 - resort_rb__delete(syscall_stats); + free(entries); printed +=3D fprintf(fp, "\n\n"); =20 return printed; --=20 2.48.1.502.g6dc24dfdaf-goog