perf lock contention: Add --lock-cgroup option

The --lock-cgroup option shows lock contention stats break down by
cgroups.

Add LOCK_AGGR_CGROUP mode and use it instead of use_cgroup field.

  $ sudo ./perf lock con -ab --lock-cgroup sleep 1
   contended   total wait     max wait     avg wait   cgroup

           8     15.70 us      6.34 us      1.96 us   /
           2      1.48 us       747 ns       738 ns   /user.slice/.../app.slice/app-gnome-google\x2dchrome-6442.scope
           1       848 ns       848 ns       848 ns   /user.slice/.../session.slice/org.gnome.Shell@x11.service
           1       220 ns       220 ns       220 ns   /user.slice/.../session.slice/pipewire-pulse.service

For now, the cgroup mode only works with BPF (-b).

Committer notes:

Remove -g as it is used in the other tools with a clear meaning of
collect/show callchains. As agreed with Namhyung off list.
Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Hao Luo <haoluo@google.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Song Liu <song@kernel.org>
Cc: bpf@vger.kernel.org
Link: https://lore.kernel.org/r/20230906174903.346486-4-namhyung@kernel.orgSigned-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

tools/perf/Documentation/perf-lock.txt

@@ -208,6 +208,9 @@ CONTENTION OPTIONS
 	Show results using a CSV-style output to make it easy to import directly
 	into spreadsheets. Columns are separated by the string specified in SEP.
 
+--lock-cgroup::
+	Show lock contention stat by cgroup.  Requires --use-bpf.
+
 
 SEE ALSO
 --------

tools/perf/builtin-lock.c

@@ -60,6 +60,7 @@ static bool combine_locks;
 static bool show_thread_stats;
 static bool show_lock_addrs;
 static bool show_lock_owner;
+static bool show_lock_cgroups;
 static bool use_bpf;
 static unsigned long bpf_map_entries = MAX_ENTRIES;
 static int max_stack_depth = CONTENTION_STACK_DEPTH;
@@ -619,6 +620,7 @@ static int get_key_by_aggr_mode_simple(u64 *key, u64 addr, u32 tid)
 		*key = tid;
 		break;
 	case LOCK_AGGR_CALLER:
+	case LOCK_AGGR_CGROUP:
 	default:
 		pr_err("Invalid aggregation mode: %d\n", aggr_mode);
 		return -EINVAL;
@@ -1103,6 +1105,7 @@ static int report_lock_contention_begin_event(struct evsel *evsel,
 			if (lock_contention_caller(evsel, sample, buf, sizeof(buf)) < 0)
 				name = "Unknown";
 			break;
+		case LOCK_AGGR_CGROUP:
 		case LOCK_AGGR_TASK:
 		default:
 			break;
@@ -1653,6 +1656,9 @@ static void print_header_stdio(void)
 	case LOCK_AGGR_ADDR:
 		fprintf(lock_output, "  %16s   %s\n\n", "address", "symbol");
 		break;
+	case LOCK_AGGR_CGROUP:
+		fprintf(lock_output, "  %s\n\n", "cgroup");
+		break;
 	default:
 		break;
 	}
@@ -1680,6 +1686,9 @@ static void print_header_csv(const char *sep)
 	case LOCK_AGGR_ADDR:
 		fprintf(lock_output, "%s%s %s%s %s\n", "address", sep, "symbol", sep, "type");
 		break;
+	case LOCK_AGGR_CGROUP:
+		fprintf(lock_output, "%s\n", "cgroup");
+		break;
 	default:
 		break;
 	}
@@ -1720,6 +1729,9 @@ static void print_lock_stat_stdio(struct lock_contention *con, struct lock_stat
 		fprintf(lock_output, "  %016llx   %s (%s)\n", (unsigned long long)st->addr,
 			st->name, get_type_name(st->flags));
 		break;
+	case LOCK_AGGR_CGROUP:
+		fprintf(lock_output, "  %s\n", st->name);
+		break;
 	default:
 		break;
 	}
@@ -1770,6 +1782,9 @@ static void print_lock_stat_csv(struct lock_contention *con, struct lock_stat *s
 		fprintf(lock_output, "%llx%s %s%s %s\n", (unsigned long long)st->addr, sep,
 			st->name, sep, get_type_name(st->flags));
 		break;
+	case LOCK_AGGR_CGROUP:
+		fprintf(lock_output, "%s\n",st->name);
+		break;
 	default:
 		break;
 	}
@@ -1999,6 +2014,27 @@ static int check_lock_contention_options(const struct option *options,
 		return -1;
 	}
 
+	if (show_lock_cgroups && !use_bpf) {
+		pr_err("Cgroups are available only with BPF\n");
+		parse_options_usage(usage, options, "lock-cgroup", 0);
+		parse_options_usage(NULL, options, "use-bpf", 0);
+		return -1;
+	}
+
+	if (show_lock_cgroups && show_lock_addrs) {
+		pr_err("Cannot use cgroup and addr mode together\n");
+		parse_options_usage(usage, options, "lock-cgroup", 0);
+		parse_options_usage(NULL, options, "lock-addr", 0);
+		return -1;
+	}
+
+	if (show_lock_cgroups && show_thread_stats) {
+		pr_err("Cannot use cgroup and thread mode together\n");
+		parse_options_usage(usage, options, "lock-cgroup", 0);
+		parse_options_usage(NULL, options, "threads", 0);
+		return -1;
+	}
+
 	if (symbol_conf.field_sep) {
 		if (strstr(symbol_conf.field_sep, ":") || /* part of type flags */
 		    strstr(symbol_conf.field_sep, "+") || /* part of caller offset */
@@ -2060,7 +2096,8 @@ static int __cmd_contention(int argc, const char **argv)
 	con.machine = &session->machines.host;
 
 	con.aggr_mode = aggr_mode = show_thread_stats ? LOCK_AGGR_TASK :
-		show_lock_addrs ? LOCK_AGGR_ADDR : LOCK_AGGR_CALLER;
+		show_lock_addrs ? LOCK_AGGR_ADDR :
+		show_lock_cgroups ? LOCK_AGGR_CGROUP : LOCK_AGGR_CALLER;
 
 	if (con.aggr_mode == LOCK_AGGR_CALLER)
 		con.save_callstack = true;
@@ -2524,6 +2561,7 @@ int cmd_lock(int argc, const char **argv)
 	OPT_BOOLEAN('o', "lock-owner", &show_lock_owner, "show lock owners instead of waiters"),
 	OPT_STRING_NOEMPTY('x', "field-separator", &symbol_conf.field_sep, "separator",
 		   "print result in CSV format with custom separator"),
+	OPT_BOOLEAN(0, "lock-cgroup", &show_lock_cgroups, "show lock stats by cgroup"),
 	OPT_PARENT(lock_options)
 	};
 

tools/perf/util/bpf_lock_contention.c

@@ -152,7 +152,10 @@ int lock_contention_prepare(struct lock_contention *con)
 	skel->bss->needs_callstack = con->save_callstack;
 	skel->bss->lock_owner = con->owner;
 
-	if (con->use_cgroup) {
+	if (con->aggr_mode == LOCK_AGGR_CGROUP) {
+		if (cgroup_is_v2("perf_event"))
+			skel->bss->use_cgroup_v2 = 1;
+
 		read_all_cgroups(&con->cgroups);
 	}
 
@@ -214,12 +217,12 @@ static const char *lock_contention_get_name(struct lock_contention *con,
 			return "siglock";
 
 		/* global locks with symbols */
-		sym = machine__find_kernel_symbol(machine, key->lock_addr, &kmap);
+		sym = machine__find_kernel_symbol(machine, key->lock_addr_or_cgroup, &kmap);
 		if (sym)
 			return sym->name;
 
 		/* try semi-global locks collected separately */
-		if (!bpf_map_lookup_elem(lock_fd, &key->lock_addr, &flags)) {
+		if (!bpf_map_lookup_elem(lock_fd, &key->lock_addr_or_cgroup, &flags)) {
 			if (flags == LOCK_CLASS_RQLOCK)
 				return "rq_lock";
 		}
@@ -227,8 +230,8 @@ static const char *lock_contention_get_name(struct lock_contention *con,
 		return "";
 	}
 
-	if (con->use_cgroup) {
-		u64 cgrp_id = key->lock_addr;
+	if (con->aggr_mode == LOCK_AGGR_CGROUP) {
+		u64 cgrp_id = key->lock_addr_or_cgroup;
 		struct cgroup *cgrp = __cgroup__find(&con->cgroups, cgrp_id);
 
 		if (cgrp)
@@ -329,7 +332,8 @@ int lock_contention_read(struct lock_contention *con)
 			ls_key = key.pid;
 			break;
 		case LOCK_AGGR_ADDR:
-			ls_key = key.lock_addr;
+		case LOCK_AGGR_CGROUP:
+			ls_key = key.lock_addr_or_cgroup;
 			break;
 		default:
 			goto next;

tools/perf/util/bpf_skel/lock_contention.bpf.c

@@ -118,6 +118,9 @@ int needs_callstack;
 int stack_skip;
 int lock_owner;
 
+int use_cgroup_v2;
+int perf_subsys_id = -1;
+
 /* determine the key of lock stat */
 int aggr_mode;
 
@@ -130,6 +133,29 @@ int data_fail;
 int task_map_full;
 int data_map_full;
 
+static inline __u64 get_current_cgroup_id(void)
+{
+	struct task_struct *task;
+	struct cgroup *cgrp;
+
+	if (use_cgroup_v2)
+		return bpf_get_current_cgroup_id();
+
+	task = bpf_get_current_task_btf();
+
+	if (perf_subsys_id == -1) {
+#if __has_builtin(__builtin_preserve_enum_value)
+		perf_subsys_id = bpf_core_enum_value(enum cgroup_subsys_id,
+						     perf_event_cgrp_id);
+#else
+		perf_subsys_id = perf_event_cgrp_id;
+#endif
+	}
+
+	cgrp = BPF_CORE_READ(task, cgroups, subsys[perf_subsys_id], cgroup);
+	return BPF_CORE_READ(cgrp, kn, id);
+}
+
 static inline int can_record(u64 *ctx)
 {
 	if (has_cpu) {
@@ -364,10 +390,13 @@ int contention_end(u64 *ctx)
 			key.stack_id = pelem->stack_id;
 		break;
 	case LOCK_AGGR_ADDR:
-		key.lock_addr = pelem->lock;
+		key.lock_addr_or_cgroup = pelem->lock;
 		if (needs_callstack)
 			key.stack_id = pelem->stack_id;
 		break;
+	case LOCK_AGGR_CGROUP:
+		key.lock_addr_or_cgroup = get_current_cgroup_id();
+		break;
 	default:
 		/* should not happen */
 		return 0;

tools/perf/util/bpf_skel/lock_data.h

@@ -6,7 +6,7 @@
 struct contention_key {
 	u32 stack_id;
 	u32 pid;
-	u64 lock_addr;
+	u64 lock_addr_or_cgroup;
 };
 
 #define TASK_COMM_LEN  16
@@ -39,6 +39,7 @@ enum lock_aggr_mode {
 	LOCK_AGGR_ADDR = 0,
 	LOCK_AGGR_TASK,
 	LOCK_AGGR_CALLER,
+	LOCK_AGGR_CGROUP,
 };
 
 enum lock_class_sym {

tools/perf/util/lock-contention.h

@@ -144,7 +144,6 @@ struct lock_contention {
 	int owner;
 	int nr_filtered;
 	bool save_callstack;
-	bool use_cgroup;
 };
 
 #ifdef HAVE_BPF_SKEL