perf syscalltbl: Remove struct syscalltbl

The syscalltbl held entries of system call name and number pairs,
generated from a native syscalltbl at start up. As there are gaps in
the system call number there is a notion of index into the
table. Going forward we want the system call table to be identifiable
by a machine type, for example, i386 vs x86-64. Change the interface
to the syscalltbl so (1) a (currently unused machine type of EM_HOST)
is passed (2) the index to syscall number and system call name mapping
is computed at build time.

Two tables are used for this, an array of system call number to name,
an array of system call numbers sorted by the system call name. The
sorted array doesn't store strings in part to save memory and
relocations. The index notion is carried forward and is an index into
the sorted array of system call numbers, the data structures are
opaque (held only in syscalltbl.c), and so the number of indices for a
machine type is exposed as a new API.

The arrays are computed in the syscalltbl.sh script and so no start-up
time computation and storage is necessary.

Signed-off-by: Ian Rogers <irogers@google.com>
Reviewed-by: Howard Chu <howardchu95@gmail.com>
Reviewed-by: Charlie Jenkins <charlie@rivosinc.com>
Reviewed-by: Namhyung Kim <namhyung@kernel.org>
Acked-by: Arnaldo Carvalho de Melo <acme@kernel.org>
Link: https://lore.kernel.org/r/20250319050741.269828-6-irogers@google.com
Signed-off-by: Namhyung Kim <namhyung@kernel.org>

Author: captain5050

tools/perf/builtin-trace.c

@@ -149,7 +149,6 @@ enum summary_mode {
 
 struct trace {
 	struct perf_tool	tool;
-	struct syscalltbl	*sctbl;
 	struct {
 		/** Sorted sycall numbers used by the trace. */
 		struct syscall  *table;
@@ -188,6 +187,14 @@ struct trace {
 		pid_t		*entries;
 		struct bpf_map  *map;
 	}			filter_pids;
+	/*
+	 * TODO: The map is from an ID (aka system call number) to struct
+	 * syscall_stats. If there is >1 e_machine, such as i386 and x86-64
+	 * processes, then the stats here will gather wrong the statistics for
+	 * the non EM_HOST system calls. A fix would be to add the e_machine
+	 * into the key, but this would make the code inconsistent with the
+	 * per-thread version.
+	 */
 	struct hashmap		*syscall_stats;
 	double			duration_filter;
 	double			runtime_ms;
@@ -2141,7 +2148,7 @@ static int syscall__read_info(struct syscall *sc, struct trace *trace)
 		return 0;
 	}
 
-	name = syscalltbl__name(trace->sctbl, sc->id);
+	name = syscalltbl__name(sc->e_machine, sc->id);
 	if (name == NULL) {
 		sc->nonexistent = true;
 		return -EEXIST;
@@ -2241,10 +2248,14 @@ static int trace__validate_ev_qualifier(struct trace *trace)
 
 	strlist__for_each_entry(pos, trace->ev_qualifier) {
 		const char *sc = pos->s;
-		int id = syscalltbl__id(trace->sctbl, sc), match_next = -1;
+		/*
+		 * TODO: Assume more than the validation/warnings are all for
+		 * the same binary type as perf.
+		 */
+		int id = syscalltbl__id(EM_HOST, sc), match_next = -1;
 
 		if (id < 0) {
-			id = syscalltbl__strglobmatch_first(trace->sctbl, sc, &match_next);
+			id = syscalltbl__strglobmatch_first(EM_HOST, sc, &match_next);
 			if (id >= 0)
 				goto matches;
 
@@ -2264,7 +2275,7 @@ static int trace__validate_ev_qualifier(struct trace *trace)
 			continue;
 
 		while (1) {
-			id = syscalltbl__strglobmatch_next(trace->sctbl, sc, &match_next);
+			id = syscalltbl__strglobmatch_next(EM_HOST, sc, &match_next);
 			if (id < 0)
 				break;
 			if (nr_allocated == nr_used) {
@@ -2722,6 +2733,7 @@ static int trace__sys_enter(struct trace *trace, struct evsel *evsel,
 	int id = perf_evsel__sc_tp_uint(evsel, id, sample), err = -1;
 	int augmented_args_size = 0;
 	void *augmented_args = NULL;
+	/* TODO: get e_machine from thread. */
 	struct syscall *sc = trace__syscall_info(trace, evsel, EM_HOST, id);
 	struct thread_trace *ttrace;
 
@@ -2796,6 +2808,7 @@ static int trace__fprintf_sys_enter(struct trace *trace, struct evsel *evsel,
 	struct thread_trace *ttrace;
 	struct thread *thread;
 	int id = perf_evsel__sc_tp_uint(evsel, id, sample), err = -1;
+	/* TODO: get e_machine from thread. */
 	struct syscall *sc = trace__syscall_info(trace, evsel, EM_HOST, id);
 	char msg[1024];
 	void *args, *augmented_args = NULL;
@@ -2871,6 +2884,7 @@ static int trace__sys_exit(struct trace *trace, struct evsel *evsel,
 	struct thread *thread;
 	int id = perf_evsel__sc_tp_uint(evsel, id, sample), err = -1, callchain_ret = 0, printed = 0;
 	int alignment = trace->args_alignment;
+	/* TODO: get e_machine from thread. */
 	struct syscall *sc = trace__syscall_info(trace, evsel, EM_HOST, id);
 	struct thread_trace *ttrace;
 
@@ -3224,6 +3238,7 @@ static int trace__event_handler(struct trace *trace, struct evsel *evsel,
 
 	if (evsel == trace->syscalls.events.bpf_output) {
 		int id = perf_evsel__sc_tp_uint(evsel, id, sample);
+		/* TODO: get e_machine from thread. */
 		struct syscall *sc = trace__syscall_info(trace, evsel, EM_HOST, id);
 
 		if (sc) {
@@ -3731,9 +3746,9 @@ static struct bpf_program *trace__find_syscall_bpf_prog(struct trace *trace, str
 	return trace->skel->progs.syscall_unaugmented;
 }
 
-static void trace__init_syscall_bpf_progs(struct trace *trace, int id)
+static void trace__init_syscall_bpf_progs(struct trace *trace, int e_machine, int id)
 {
-	struct syscall *sc = trace__syscall_info(trace, NULL, EM_HOST, id);
+	struct syscall *sc = trace__syscall_info(trace, NULL, e_machine, id);
 
 	if (sc == NULL)
 		return;
@@ -3742,22 +3757,22 @@ static void trace__init_syscall_bpf_progs(struct trace *trace, int id)
 	sc->bpf_prog.sys_exit  = trace__find_syscall_bpf_prog(trace, sc, sc->fmt ? sc->fmt->bpf_prog_name.sys_exit  : NULL,  "exit");
 }
 
-static int trace__bpf_prog_sys_enter_fd(struct trace *trace, int id)
+static int trace__bpf_prog_sys_enter_fd(struct trace *trace, int e_machine, int id)
 {
-	struct syscall *sc = trace__syscall_info(trace, NULL, EM_HOST, id);
+	struct syscall *sc = trace__syscall_info(trace, NULL, e_machine, id);
 	return sc ? bpf_program__fd(sc->bpf_prog.sys_enter) : bpf_program__fd(trace->skel->progs.syscall_unaugmented);
 }
 
-static int trace__bpf_prog_sys_exit_fd(struct trace *trace, int id)
+static int trace__bpf_prog_sys_exit_fd(struct trace *trace, int e_machine, int id)
 {
-	struct syscall *sc = trace__syscall_info(trace, NULL, EM_HOST, id);
+	struct syscall *sc = trace__syscall_info(trace, NULL, e_machine, id);
 	return sc ? bpf_program__fd(sc->bpf_prog.sys_exit) : bpf_program__fd(trace->skel->progs.syscall_unaugmented);
 }
 
-static int trace__bpf_sys_enter_beauty_map(struct trace *trace, int key, unsigned int *beauty_array)
+static int trace__bpf_sys_enter_beauty_map(struct trace *trace, int e_machine, int key, unsigned int *beauty_array)
 {
 	struct tep_format_field *field;
-	struct syscall *sc = trace__syscall_info(trace, NULL, EM_HOST, key);
+	struct syscall *sc = trace__syscall_info(trace, NULL, e_machine, key);
 	const struct btf_type *bt;
 	char *struct_offset, *tmp, name[32];
 	bool can_augment = false;
@@ -3854,8 +3869,8 @@ static struct bpf_program *trace__find_usable_bpf_prog_entry(struct trace *trace
 	return NULL;
 
 try_to_find_pair:
-	for (int i = 0; i < trace->sctbl->syscalls.nr_entries; ++i) {
-		int id = syscalltbl__id_at_idx(trace->sctbl, i);
+	for (int i = 0, num_idx = syscalltbl__num_idx(sc.e_machine); i < num_idx; ++i) {
+		int id = syscalltbl__id_at_idx(sc.e_machine, i);
 		/* calling trace__syscall_info() may invalidate '_sc' */
 		struct syscall *pair = trace__syscall_info(trace, NULL, sc.e_machine, id);
 		struct bpf_program *pair_prog;
@@ -3941,35 +3956,35 @@ static struct bpf_program *trace__find_usable_bpf_prog_entry(struct trace *trace
 	return NULL;
 }
 
-static int trace__init_syscalls_bpf_prog_array_maps(struct trace *trace)
+static int trace__init_syscalls_bpf_prog_array_maps(struct trace *trace, int e_machine)
 {
 	int map_enter_fd = bpf_map__fd(trace->skel->maps.syscalls_sys_enter);
 	int map_exit_fd  = bpf_map__fd(trace->skel->maps.syscalls_sys_exit);
 	int beauty_map_fd = bpf_map__fd(trace->skel->maps.beauty_map_enter);
 	int err = 0;
 	unsigned int beauty_array[6];
 
-	for (int i = 0; i < trace->sctbl->syscalls.nr_entries; ++i) {
-		int prog_fd, key = syscalltbl__id_at_idx(trace->sctbl, i);
+	for (int i = 0, num_idx = syscalltbl__num_idx(e_machine); i < num_idx; ++i) {
+		int prog_fd, key = syscalltbl__id_at_idx(e_machine, i);
 
 		if (!trace__syscall_enabled(trace, key))
 			continue;
 
-		trace__init_syscall_bpf_progs(trace, key);
+		trace__init_syscall_bpf_progs(trace, e_machine, key);
 
 		// It'll get at least the "!raw_syscalls:unaugmented"
-		prog_fd = trace__bpf_prog_sys_enter_fd(trace, key);
+		prog_fd = trace__bpf_prog_sys_enter_fd(trace, e_machine, key);
 		err = bpf_map_update_elem(map_enter_fd, &key, &prog_fd, BPF_ANY);
 		if (err)
 			break;
-		prog_fd = trace__bpf_prog_sys_exit_fd(trace, key);
+		prog_fd = trace__bpf_prog_sys_exit_fd(trace, e_machine, key);
 		err = bpf_map_update_elem(map_exit_fd, &key, &prog_fd, BPF_ANY);
 		if (err)
 			break;
 
 		/* use beauty_map to tell BPF how many bytes to collect, set beauty_map's value here */
 		memset(beauty_array, 0, sizeof(beauty_array));
-		err = trace__bpf_sys_enter_beauty_map(trace, key, (unsigned int *)beauty_array);
+		err = trace__bpf_sys_enter_beauty_map(trace, e_machine, key, (unsigned int *)beauty_array);
 		if (err)
 			continue;
 		err = bpf_map_update_elem(beauty_map_fd, &key, beauty_array, BPF_ANY);
@@ -4005,9 +4020,9 @@ static int trace__init_syscalls_bpf_prog_array_maps(struct trace *trace)
 	 * first and second arg (this one on the raw_syscalls:sys_exit prog
 	 * array tail call, then that one will be used.
 	 */
-	for (int i = 0; i < trace->sctbl->syscalls.nr_entries; ++i) {
-		int key = syscalltbl__id_at_idx(trace->sctbl, i);
-		struct syscall *sc = trace__syscall_info(trace, NULL, EM_HOST, key);
+	for (int i = 0, num_idx = syscalltbl__num_idx(e_machine); i < num_idx; ++i) {
+		int key = syscalltbl__id_at_idx(e_machine, i);
+		struct syscall *sc = trace__syscall_info(trace, NULL, e_machine, key);
 		struct bpf_program *pair_prog;
 		int prog_fd;
 
@@ -4032,7 +4047,7 @@ static int trace__init_syscalls_bpf_prog_array_maps(struct trace *trace)
 		 * Get syscall info again as find usable entry above might
 		 * modify the syscall table and shuffle it.
 		 */
-		sc = trace__syscall_info(trace, NULL, EM_HOST, key);
+		sc = trace__syscall_info(trace, NULL, e_machine, key);
 		sc->bpf_prog.sys_enter = pair_prog;
 
 		/*
@@ -4457,8 +4472,13 @@ static int trace__run(struct trace *trace, int argc, const char **argv)
 		goto out_error_mem;
 
 #ifdef HAVE_BPF_SKEL
-	if (trace->skel && trace->skel->progs.sys_enter)
-		trace__init_syscalls_bpf_prog_array_maps(trace);
+	if (trace->skel && trace->skel->progs.sys_enter) {
+		/*
+		 * TODO: Initialize for all host binary machine types, not just
+		 * those matching the perf binary.
+		 */
+		trace__init_syscalls_bpf_prog_array_maps(trace, EM_HOST);
+	}
 #endif
 
 	if (trace->ev_qualifier_ids.nr > 0) {
@@ -4483,7 +4503,8 @@ static int trace__run(struct trace *trace, int argc, const char **argv)
 	 *  So just disable this beautifier (SCA_FD, SCA_FDAT) when 'close' is
 	 *  not in use.
 	 */
-	trace->fd_path_disabled = !trace__syscall_enabled(trace, syscalltbl__id(trace->sctbl, "close"));
+	/* TODO: support for more than just perf binary machine type close. */
+	trace->fd_path_disabled = !trace__syscall_enabled(trace, syscalltbl__id(EM_HOST, "close"));
 
 	err = trace__expand_filters(trace, &evsel);
 	if (err)
@@ -4796,7 +4817,7 @@ static struct syscall_entry *syscall__sort_stats(struct hashmap *syscall_stats)
 	return entry;
 }
 
-static size_t syscall__dump_stats(struct trace *trace, FILE *fp,
+static size_t syscall__dump_stats(struct trace *trace, int e_machine, FILE *fp,
 				  struct hashmap *syscall_stats)
 {
 	size_t printed = 0;
@@ -4827,7 +4848,7 @@ static size_t syscall__dump_stats(struct trace *trace, FILE *fp,
 			pct = avg ? 100.0 * stddev_stats(&stats->stats) / avg : 0.0;
 			avg /= NSEC_PER_MSEC;
 
-			sc = trace__syscall_info(trace, /*evsel=*/NULL, EM_HOST, entry->syscall);
+			sc = trace__syscall_info(trace, /*evsel=*/NULL, e_machine, entry->syscall);
 			if (!sc)
 				continue;
 
@@ -4854,14 +4875,14 @@ static size_t syscall__dump_stats(struct trace *trace, FILE *fp,
 }
 
 static size_t thread__dump_stats(struct thread_trace *ttrace,
-				 struct trace *trace, FILE *fp)
+				 struct trace *trace, int e_machine, FILE *fp)
 {
-	return syscall__dump_stats(trace, fp, ttrace->syscall_stats);
+	return syscall__dump_stats(trace, e_machine, fp, ttrace->syscall_stats);
 }
 
-static size_t system__dump_stats(struct trace *trace, FILE *fp)
+static size_t system__dump_stats(struct trace *trace, int e_machine, FILE *fp)
 {
-	return syscall__dump_stats(trace, fp, trace->syscall_stats);
+	return syscall__dump_stats(trace, e_machine, fp, trace->syscall_stats);
 }
 
 static size_t trace__fprintf_thread(FILE *fp, struct thread *thread, struct trace *trace)
@@ -4887,7 +4908,8 @@ static size_t trace__fprintf_thread(FILE *fp, struct thread *thread, struct trac
 	else if (fputc('\n', fp) != EOF)
 		++printed;
 
-	printed += thread__dump_stats(ttrace, trace, fp);
+	/* TODO: get e_machine from thread. */
+	printed += thread__dump_stats(ttrace, trace, EM_HOST, fp);
 
 	return printed;
 }
@@ -4948,7 +4970,8 @@ static size_t trace__fprintf_total_summary(struct trace *trace, FILE *fp)
 	else if (fputc('\n', fp) != EOF)
 		++printed;
 
-	printed += system__dump_stats(trace, fp);
+	/* TODO: get all system e_machines. */
+	printed += system__dump_stats(trace, EM_HOST, fp);
 
 	return printed;
 }
@@ -5140,8 +5163,9 @@ static int trace__parse_events_option(const struct option *opt, const char *str,
 			*sep = '\0';
 
 		list = 0;
-		if (syscalltbl__id(trace->sctbl, s) >= 0 ||
-		    syscalltbl__strglobmatch_first(trace->sctbl, s, &idx) >= 0) {
+		/* TODO: support for more than just perf binary machine type syscalls. */
+		if (syscalltbl__id(EM_HOST, s) >= 0 ||
+		    syscalltbl__strglobmatch_first(EM_HOST, s, &idx) >= 0) {
 			list = 1;
 			goto do_concat;
 		}
@@ -5294,7 +5318,6 @@ static void trace__exit(struct trace *trace)
 			syscall__exit(&trace->syscalls.table[i]);
 		zfree(&trace->syscalls.table);
 	}
-	syscalltbl__delete(trace->sctbl);
 	zfree(&trace->perfconfig_events);
 }
 
@@ -5443,9 +5466,8 @@ int cmd_trace(int argc, const char **argv)
 	sigaction(SIGCHLD, &sigchld_act, NULL);
 
 	trace.evlist = evlist__new();
-	trace.sctbl = syscalltbl__new();
 
-	if (trace.evlist == NULL || trace.sctbl == NULL) {
+	if (trace.evlist == NULL) {
 		pr_err("Not enough memory to run!\n");
 		err = -ENOMEM;
 		goto out;

tools/perf/scripts/syscalltbl.sh

@@ -50,37 +50,27 @@ fi
 infile="$1"
 outfile="$2"
 
-nxt=0
-
-syscall_macro() {
-    nr="$1"
-    name="$2"
-
-    echo "	[$nr] = \"$name\","
-}
-
-emit() {
-    nr="$1"
-    entry="$2"
-
-    syscall_macro "$nr" "$entry"
-}
-
-echo "static const char *const syscalltbl[] = {" > $outfile
-
 sorted_table=$(mktemp /tmp/syscalltbl.XXXXXX)
 grep -E "^[0-9]+[[:space:]]+$abis" "$infile" | sort -n > $sorted_table
 
-max_nr=0
+echo "static const char *const syscall_num_to_name[] = {" > $outfile
 # the params are: nr abi name entry compat
 # use _ for intentionally unused variables according to SC2034
 while read nr _ name _ _; do
-    emit "$nr" "$name" >> $outfile
-    max_nr=$nr
+	echo "	[$nr] = \"$name\"," >> $outfile
 done < $sorted_table
+echo "};" >> $outfile
 
-rm -f $sorted_table
+echo "static const uint16_t syscall_sorted_names[] = {" >> $outfile
 
+# When sorting by name, add a suffix of 0s upto 20 characters so that system
+# calls that differ with a numerical suffix don't sort before those
+# without. This default behavior of sort differs from that of strcmp used at
+# runtime. Use sed to strip the trailing 0s suffix afterwards.
+grep -E "^[0-9]+[[:space:]]+$abis" "$infile" | awk '{printf $3; for (i = length($3); i < 20; i++) { printf "0"; }; print " " $1}'| sort | sed 's/\([a-zA-Z1-9]\+\)0\+ \([0-9]\+\)/\1 \2/' > $sorted_table
+while read name nr; do
+	echo "	$nr,	/* $name */" >> $outfile
+done < $sorted_table
 echo "};" >> $outfile
 
-echo "#define SYSCALLTBL_MAX_ID ${max_nr}" >> $outfile
+rm -f $sorted_table