intel-gpu-top: Support for client statscontext-stats

Adds support for per-client engine busyness stats i915 exports in sysfs
and produces output like the below:

==========================================================================
intel-gpu-top -  935/ 935 MHz;    0% RC6; 14.73 Watts;     1097 irqs/s

      IMC reads:     1401 MiB/s
     IMC writes:        4 MiB/s

          ENGINE      BUSY                                 MI_SEMA MI_WAIT
     Render/3D/0   63.73% |███████████████████           |      3%      0%
       Blitter/0    9.53% |██▊                           |      6%      0%
         Video/0   39.32% |███████████▊                  |     16%      0%
         Video/1   15.62% |████▋                         |      0%      0%
  VideoEnhance/0    0.00% |                              |      0%      0%

  PID            NAME     RCS          BCS          VCS         VECS
 4084        gem_wsim |█████▌     ||█          ||           ||           |
 4086        gem_wsim |█▌         ||           ||███        ||           |
==========================================================================

Apart from the existing physical engine utilization it now also shows
utilization per client and per engine class.

v2:
 * Version to match removal of global enable_stats toggle.
 * Plus various fixes.

v3:
 * Support brief backward jumps in client stats.

v4:
 * Support device selection.

v5:
 * Rebase for class aggregation.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>

1 files changed, 523 insertions, 11 deletions

diff --git a/tools/intel_gpu_top.c b/tools/intel_gpu_top.c
index 72ad7cbe..b2fd7371 100644
--- a/tools/intel_gpu_top.c
+++ b/tools/intel_gpu_top.c
@@ -424,6 +424,51 @@ static struct engines *discover_engines(char *device)
 	return engines;
 }
 
+static int
+filename_to_buf(const char *filename, char *buf, unsigned int bufsize)
+{
+	int fd, err;
+	ssize_t ret;
+
+	fd = open(filename, O_RDONLY);
+	if (fd < 0)
+		return -1;
+
+	ret = read(fd, buf, bufsize - 1);
+	err = errno;
+	close(fd);
+	if (ret < 1) {
+		errno = ret < 0 ? err : ENOMSG;
+
+		return -1;
+	}
+
+	if (ret > 1 && buf[ret - 1] == '\n')
+		buf[ret - 1] = '\0';
+	else
+		buf[ret] = '\0';
+
+	return 0;
+}
+
+static uint64_t filename_to_u64(const char *filename, int base)
+{
+	char buf[64], *b;
+
+	if (filename_to_buf(filename, buf, sizeof(buf)))
+		return 0;
+
+	/*
+	 * Handle both single integer and key=value formats by skipping
+	 * leading non-digits.
+	 */
+	b = buf;
+	while (*b && !isdigit(*b))
+		b++;
+
+	return strtoull(b, NULL, base);
+}
+
 #define _open_pmu(type, cnt, pmu, fd) \
 ({ \
 	int fd__; \
@@ -625,23 +670,290 @@ static void pmu_sample(struct engines *engines)
 	}
 }
 
+enum client_status {
+	FREE = 0, /* mbz */
+	ALIVE,
+	PROBE
+};
+
+struct clients;
+
+struct client {
+	struct clients *clients;
+
+	enum client_status status;
+	unsigned int id;
+	unsigned int pid;
+	char name[128];
+	unsigned int samples;
+	unsigned long total;
+	struct engines *engines;
+	unsigned long *val;
+	uint64_t *last;
+};
+
+struct clients {
+	unsigned int num_clients;
+
+	unsigned int num_classes;
+	struct engine_class *class;
+
+	char sysfs_root[64];
+
+	struct client *client;
+};
+
+#define for_each_client(clients, c, tmp) \
+	for ((tmp) = (clients)->num_clients, c = (clients)->client; \
+	     (tmp > 0); (tmp)--, (c)++)
+
+static struct clients *init_clients(const char *drm_card)
+{
+	struct clients *clients = malloc(sizeof(*clients));
+	const char *slash;
+	ssize_t ret;
+
+	memset(clients, 0, sizeof(*clients));
+
+	if (drm_card) {
+		slash = rindex(drm_card, '/');
+		assert(slash);
+	} else {
+		slash = "card0";
+	}
+
+	ret = snprintf(clients->sysfs_root, sizeof(clients->sysfs_root),
+		       "/sys/class/drm/%s/clients/", slash);
+	assert(ret > 0 && ret < sizeof(clients->sysfs_root));
+
+	return clients;
+}
+
+static uint64_t
+read_client_busy(const struct client *client, unsigned int class)
+{
+	char buf[256];
+	ssize_t ret;
+
+	ret = snprintf(buf, sizeof(buf), "%s/%u/busy/%u",
+		       client->clients->sysfs_root, client->id, class);
+	assert(ret > 0 && ret < sizeof(buf));
+	if (ret <= 0 || ret == sizeof(buf))
+		return 0;
+
+	return filename_to_u64(buf, 10);
+}
+
+static struct client *
+find_client(struct clients *clients, enum client_status status, unsigned int id)
+{
+	struct client *c;
+	int tmp;
+
+	for_each_client(clients, c, tmp) {
+		if ((status == FREE && c->status == FREE) ||
+		    (status == c->status && c->id == id))
+			return c;
+	}
+
+	return NULL;
+}
+
+static void update_client(struct client *c, unsigned int pid, char *name)
+{
+	uint64_t val[c->clients->num_classes];
+	unsigned int i;
+
+	if (c->pid != pid)
+		c->pid = pid;
+
+	if (strcmp(c->name, name))
+		strncpy(c->name, name, sizeof(c->name) - 1);
+
+	for (i = 0; i < c->clients->num_classes; i++)
+		val[i] = read_client_busy(c, c->clients->class[i].class);
+
+	c->total = 0;
+
+	for (i = 0; i < c->clients->num_classes; i++) {
+		if (val[i] < c->last[i])
+			continue; /* It will catch up soon. */
+
+		c->val[i] = val[i] - c->last[i];
+		c->total += c->val[i];
+		c->last[i] = val[i];
+	}
+
+	c->samples++;
+	c->status = ALIVE;
+}
+
+static void
+add_client(struct clients *clients, unsigned int id, unsigned int pid,
+	   char *name)
+{
+	struct client *c;
+
+	if (find_client(clients, ALIVE, id))
+		return;
+
+	c = find_client(clients, FREE, 0);
+	if (!c) {
+		unsigned int idx = clients->num_clients;
+
+		clients->num_clients += (clients->num_clients + 2) / 2;
+		clients->client = realloc(clients->client,
+					  clients->num_clients * sizeof(*c));
+		assert(clients->client);
+
+		c = &clients->client[idx];
+		memset(c, 0, (clients->num_clients - idx) * sizeof(*c));
+	}
+
+	c->id = id;
+	c->clients = clients;
+	c->val = calloc(clients->num_classes, sizeof(c->val));
+	c->last = calloc(clients->num_classes, sizeof(c->last));
+	assert(c->val && c->last);
+
+	update_client(c, pid, name);
+}
+
+static void free_client(struct client *c)
+{
+	free(c->val);
+	free(c->last);
+	memset(c, 0, sizeof(*c));
+}
+
+static char *
+read_client_sysfs(const char *sysfs_root, unsigned int id, const char *field)
+{
+	char buf[256];
+	ssize_t ret;
+
+	ret = snprintf(buf, sizeof(buf), "%s/%u/%s", sysfs_root, id, field);
+	assert(ret > 0 && ret < sizeof(buf));
+	if (ret <= 0 || ret == sizeof(buf))
+		return NULL;
+
+	ret = filename_to_buf(buf, buf, sizeof(buf));
+	if (ret)
+		return NULL; /* Client exited. */
+
+	return strdup(buf);
+}
+
+static void scan_clients(struct clients *clients)
+{
+	struct dirent *dent;
+	struct client *c;
+	char *pid, *name;
+	unsigned int id;
+	int tmp;
+	DIR *d;
+
+	if (!clients)
+		return;
+
+	for_each_client(clients, c, tmp) {
+		if (c->status == ALIVE)
+			c->status = PROBE;
+	}
+
+	d = opendir(clients->sysfs_root);
+	if (!d)
+		return;
+
+	while ((dent = readdir(d)) != NULL) {
+		if (dent->d_type != DT_DIR)
+			continue;
+		if (!isdigit(dent->d_name[0]))
+			continue;
+
+		id = atoi(dent->d_name);
+
+		c = find_client(clients, PROBE, id);
+
+		name = read_client_sysfs(clients->sysfs_root, id, "name");
+		pid = read_client_sysfs(clients->sysfs_root, id, "pid");
+
+		if (name && pid) {
+			if (!c)
+				add_client(clients, id, atoi(pid), name);
+			else
+				update_client(c, atoi(pid), name);
+		} else if (c) {
+			c->status = PROBE; /* Will be deleted below. */
+		}
+
+		if (name)
+			free(name);
+		if (pid)
+			free(pid);
+	}
+
+	closedir(d);
+
+	for_each_client(clients, c, tmp) {
+		if (c->status == PROBE)
+			free_client(c);
+	}
+}
+
+static int cmp(const void *_a, const void *_b)
+{
+	const struct client *a = _a;
+	const struct client *b = _b;
+	long tot_a = a->total;
+	long tot_b = b->total;
+
+	tot_a *= a->status == ALIVE && a->samples > 1;
+	tot_b *= b->status == ALIVE && b->samples > 1;
+
+	tot_b -= tot_a;
+
+	if (!tot_b)
+		return (int)b->id - a->id;
+
+	while (tot_b > INT_MAX || tot_b < INT_MIN)
+		tot_b /= 2;
+
+	return tot_b;
+}
+
 static const char *bars[] = { " ", "▏", "▎", "▍", "▌", "▋", "▊", "▉", "█" };
 
+static void n_spaces(const unsigned int n)
+{
+	unsigned int i;
+
+	for (i = 0; i < n; i++)
+		putchar(' ');
+}
+
 static void
 print_percentage_bar(double percent, int max_len)
 {
-	int bar_len = percent * (8 * (max_len - 2)) / 100.0;
-	int i;
+	int bar_len, i, len = max_len - 2;
+	const int w = 8;
+
+	assert(max_len > 0);
+
+	bar_len = percent * len / 100.0;
+	if (bar_len > len)
+		bar_len = len;
+	bar_len *= w;
 
 	putchar('|');
 
-	for (i = bar_len; i >= 8; i -= 8)
-		printf("%s", bars[8]);
+	for (i = bar_len; i >= w; i -= w)
+		printf("%s", bars[w]);
 	if (i)
 		printf("%s", bars[i]);
 
-	for (i = 0; i < (max_len - 2 - (bar_len + 7) / 8); i++)
-		putchar(' ');
+	len -= (bar_len + (w - 1)) / w;
+	n_spaces(len);
 
 	putchar('|');
 }
@@ -744,6 +1056,18 @@ json_close_struct(void)
 		fflush(stdout);
 }
 
+static void
+__json_add_member(const char *key, const char *val)
+{
+	assert(json_indent_level < ARRAY_SIZE(json_indent));
+
+	fprintf(out, "%s%s\"%s\": \"%s\"",
+		json_struct_members ? ",\n" : "",
+		json_indent[json_indent_level], key, val);
+
+	json_struct_members++;
+}
+
 static unsigned int
 json_add_member(const struct cnt_group *parent, struct cnt_item *item,
 		unsigned int headers)
@@ -1046,8 +1370,6 @@ print_header(const struct igt_device_card *card,
 		memmove(&groups[0], &groups[1],
 			sizeof(groups) - sizeof(groups[0]));
 
-	pops->open_struct(NULL);
-
 	*consumed = print_groups(groups);
 
 	if (output_mode == INTERACTIVE) {
@@ -1204,7 +1526,7 @@ print_engine(struct engines *engines, unsigned int i, double t,
 			      engine->display_name, engine_items[0].buf);
 
 		val = pmu_calc(&engine->busy.val, 1e9, t, 100);
-		print_percentage_bar(val, max_w - len);
+		print_percentage_bar(val, max_w > len ? max_w - len : 0);
 
 		printf("%s\n", buf);
 
@@ -1219,7 +1541,6 @@ print_engines_footer(struct engines *engines, double t,
 		     int lines, int con_w, int con_h)
 {
 	pops->close_struct();
-	pops->close_struct();
 
 	if (output_mode == INTERACTIVE) {
 		if (lines++ < con_h)
@@ -1243,6 +1564,9 @@ static void init_engine_classes(struct engines *engines)
 	unsigned int i, num;
 	int max = -1;
 
+	if (engines->num_classes)
+		return;
+
 	for (i = 0; i < engines->num_engines; i++) {
 		struct engine *engine = engine_ptr(engines, i);
 
@@ -1404,6 +1728,148 @@ print_engines(struct engines *engines, double t, int lines, int w, int h)
 	return lines;
 }
 
+static int
+print_clients_header(struct clients *clients, int lines,
+		     int con_w, int con_h, int *class_w)
+{
+	if (output_mode == INTERACTIVE) {
+		const char *pidname = "   PID              NAME ";
+		unsigned int num_active = 0;
+		int len = strlen(pidname);
+
+		if (lines++ >= con_h)
+			return lines;
+
+		printf("\033[7m");
+		printf("%s", pidname);
+
+		if (lines++ >= con_h || len >= con_w)
+			return lines;
+
+		if (clients->num_classes) {
+			unsigned int i;
+			int width;
+
+			for (i = 0; i < clients->num_classes; i++) {
+				if (clients->class[i].num_engines)
+					num_active++;
+			}
+
+			*class_w = width = (con_w - len) / num_active;
+
+			for (i = 0; i < clients->num_classes; i++) {
+				const char *name = clients->class[i].name;
+				int name_len = strlen(name);
+				int pad = (width - name_len) / 2;
+				int spaces = width - pad - name_len;
+
+				if (!clients->class[i].num_engines)
+					continue; /* Assert in the ideal world. */
+
+				if (pad < 0 || spaces < 0)
+					continue;
+
+				n_spaces(pad);
+				printf("%s", name);
+				n_spaces(spaces);
+				len += pad + name_len + spaces;
+			}
+		}
+
+		n_spaces(con_w - len);
+		printf("\033[0m\n");
+	} else {
+		if (clients->num_classes)
+			pops->open_struct("clients");
+	}
+
+	return lines;
+}
+
+static int
+print_client(struct client *c, struct engines *engines, double t, int lines,
+	     int con_w, int con_h, unsigned int period_us, int *class_w)
+{
+	struct clients *clients = c->clients;
+	unsigned int i;
+
+	if (output_mode == INTERACTIVE) {
+		printf("%6u %17s ", c->pid, c->name);
+
+		for (i = 0; i < clients->num_classes; i++) {
+			double pct;
+
+			if (!clients->class[i].num_engines)
+				continue; /* Assert in the ideal world. */
+
+			pct = (double)c->val[i] / period_us / 1e3 * 100 /
+			      clients->class[i].num_engines;
+
+			/*
+			 * Guard against possible time-drift between sampling
+			 * client data and time we obtained our time-delta from
+			 * PMU.
+			 */
+			if (pct > 100.0)
+				pct = 100.0;
+
+			print_percentage_bar(pct, *class_w);
+
+			lines++;
+		}
+
+		putchar('\n');
+	} else if (output_mode == JSON) {
+		char buf[64];
+
+		snprintf(buf, sizeof(buf), "%u", c->id);
+		pops->open_struct(buf);
+
+		__json_add_member("name", c->name);
+
+		snprintf(buf, sizeof(buf), "%u", c->pid);
+		__json_add_member("pid", buf);
+
+		pops->open_struct("engine-classes");
+
+		for (i = 0; i < clients->num_classes; i++) {
+			double pct;
+
+			snprintf(buf, sizeof(buf), "%s",
+				 clients->class[i].name);
+			pops->open_struct(buf);
+
+			pct = (double)c->val[i] / period_us / 1e3 * 100;
+			snprintf(buf, sizeof(buf), "%f", pct);
+			__json_add_member("busy", buf);
+
+			__json_add_member("unit", "%");
+
+			pops->close_struct();
+		}
+
+		pops->close_struct();
+		pops->close_struct();
+	}
+
+	return lines;
+}
+
+static int
+print_clients_footer(struct clients *clients, double t,
+		     int lines, int con_w, int con_h)
+{
+	if (output_mode == INTERACTIVE) {
+		if (lines++ < con_h)
+			printf("\n");
+	} else {
+		if (clients->num_classes)
+			pops->close_struct();
+	}
+
+	return lines;
+}
+
 static bool stop_top;
 
 static void sigint_handler(int  sig)
@@ -1492,6 +1958,7 @@ static void process_stdin(unsigned int timeout_us)
 int main(int argc, char **argv)
 {
 	unsigned int period_us = DEFAULT_PERIOD_MS * 1000;
+	struct clients *clients = NULL;
 	int con_w = -1, con_h = -1;
 	char *output_path = NULL;
 	struct engines *engines;
@@ -1625,13 +2092,20 @@ int main(int argc, char **argv)
 
 	ret = EXIT_SUCCESS;
 
+	clients = init_clients(card.pci_slot_name[0] ? card.card : NULL);
+	init_engine_classes(engines);
+	clients->num_classes = engines->num_classes;
+	clients->class = engines->class;
+
 	pmu_sample(engines);
+	scan_clients(clients);
 	codename = igt_device_get_pretty_name(&card, false);
 
 	while (!stop_top) {
 		bool consumed = false;
-		int lines = 0;
+		int j, lines = 0;
 		struct winsize ws;
+		struct client *c;
 		double t;
 
 		/* Update terminal size. */
@@ -1650,10 +2124,18 @@ int main(int argc, char **argv)
 		pmu_sample(engines);
 		t = (double)(engines->ts.cur - engines->ts.prev) / 1e9;
 
+		scan_clients(clients);
+		if (clients) {
+			qsort(clients->client, clients->num_clients,
+			      sizeof(*clients->client), cmp);
+		}
+
 		if (stop_top)
 			break;
 
 		while (!consumed) {
+			pops->open_struct(NULL);
+
 			lines = print_header(&card, codename, engines,
 					     t, lines, con_w, con_h,
 					     &consumed);
@@ -1661,6 +2143,36 @@ int main(int argc, char **argv)
 			lines = print_imc(engines, t, lines, con_w, con_h);
 
 			lines = print_engines(engines, t, lines, con_w, con_h);
+
+			if (clients) {
+				int class_w;
+
+				lines = print_clients_header(clients, lines,
+							     con_w, con_h,
+							     &class_w);
+
+				for_each_client(clients, c, j) {
+					if (lines++ > con_h)
+						break;
+
+					assert(c->status != PROBE);
+					if (c->status != ALIVE)
+						break;
+
+					if (c->samples < 2)
+						continue;
+
+					lines = print_client(c, engines, t,
+							     lines, con_w,
+							     con_h, period_us,
+							     &class_w);
+				}
+
+				lines = print_clients_footer(clients, t, lines,
+							     con_w, con_h);
+			}
+
+			pops->close_struct();
 		}
 
 		if (stop_top)