builtin-report.c

/*
 * builtin-report.c
 *
 * Builtin report command: Analyze the perf.data input file,
 * look up and read DSOs and symbol information and display
 * a histogram of results, along various sorting keys.
 */
#include "builtin.h"

#include "util/util.h"

#include "util/color.h"
#include "util/list.h"
#include "util/cache.h"
#include "util/rbtree.h"
#include "util/symbol.h"
#include "util/string.h"
#include "util/callchain.h"
#include "util/strlist.h"

#include "perf.h"
#include "util/header.h"

#include "util/parse-options.h"
#include "util/parse-events.h"

#define SHOW_KERNEL	1
#define SHOW_USER	2
#define SHOW_HV		4

static char		const *input_name = "perf.data";
static char		*vmlinux = NULL;

static char		default_sort_order[] = "comm,dso";
static char		*sort_order = default_sort_order;
static char		*dso_list_str, *comm_list_str;
static struct strlist	*dso_list, *comm_list;

static int		input;
static int		show_mask = SHOW_KERNEL | SHOW_USER | SHOW_HV;

static int		dump_trace = 0;
#define dprintf(x...)	do { if (dump_trace) printf(x); } while (0)
#define cdprintf(x...)	do { if (dump_trace) color_fprintf(stdout, color, x); } while (0)

static int		verbose;
#define eprintf(x...)	do { if (verbose) fprintf(stderr, x); } while (0)

static int		full_paths;

static unsigned long	page_size;
static unsigned long	mmap_window = 32;

static char		default_parent_pattern[] = "^sys_|^do_page_fault";
static char		*parent_pattern = default_parent_pattern;
static regex_t		parent_regex;

static int		exclude_other = 1;
static int		callchain;

static u64		sample_type;

struct ip_event {
	struct perf_event_header header;
	u64 ip;
	u32 pid, tid;
	unsigned char __more_data[];
};

struct mmap_event {
	struct perf_event_header header;
	u32 pid, tid;
	u64 start;
	u64 len;
	u64 pgoff;
	char filename[PATH_MAX];
};

struct comm_event {
	struct perf_event_header header;
	u32 pid, tid;
	char comm[16];
};

struct fork_event {
	struct perf_event_header header;
	u32 pid, ppid;
};

struct period_event {
	struct perf_event_header header;
	u64 time;
	u64 id;
	u64 sample_period;
};

struct lost_event {
	struct perf_event_header header;
	u64 id;
	u64 lost;
};

struct read_event {
	struct perf_event_header header;
	u32 pid,tid;
	u64 value;
	u64 format[3];
};

typedef union event_union {
	struct perf_event_header	header;
	struct ip_event			ip;
	struct mmap_event		mmap;
	struct comm_event		comm;
	struct fork_event		fork;
	struct period_event		period;
	struct lost_event		lost;
	struct read_event		read;
} event_t;

static LIST_HEAD(dsos);
static struct dso *kernel_dso;
static struct dso *vdso;

static void dsos__add(struct dso *dso)
{
	list_add_tail(&dso->node, &dsos);
}

static struct dso *dsos__find(const char *name)
{
	struct dso *pos;

	list_for_each_entry(pos, &dsos, node)
		if (strcmp(pos->name, name) == 0)
			return pos;
	return NULL;
}

static struct dso *dsos__findnew(const char *name)
{
	struct dso *dso = dsos__find(name);
	int nr;

	if (dso)
		return dso;

	dso = dso__new(name, 0);
	if (!dso)
		goto out_delete_dso;

	nr = dso__load(dso, NULL, verbose);
	if (nr < 0) {
		eprintf("Failed to open: %s\n", name);
		goto out_delete_dso;
	}
	if (!nr)
		eprintf("No symbols found in: %s, maybe install a debug package?\n", name);

	dsos__add(dso);

	return dso;

out_delete_dso:
	dso__delete(dso);
	return NULL;
}

static void dsos__fprintf(FILE *fp)
{
	struct dso *pos;

	list_for_each_entry(pos, &dsos, node)
		dso__fprintf(pos, fp);
}

static struct symbol *vdso__find_symbol(struct dso *dso, u64 ip)
{
	return dso__find_symbol(kernel_dso, ip);
}

static int load_kernel(void)
{
	int err;

	kernel_dso = dso__new("[kernel]", 0);
	if (!kernel_dso)
		return -1;

	err = dso__load_kernel(kernel_dso, vmlinux, NULL, verbose);
	if (err) {
		dso__delete(kernel_dso);
		kernel_dso = NULL;
	} else
		dsos__add(kernel_dso);

	vdso = dso__new("[vdso]", 0);
	if (!vdso)
		return -1;

	vdso->find_symbol = vdso__find_symbol;

	dsos__add(vdso);

	return err;
}

static char __cwd[PATH_MAX];
static char *cwd = __cwd;
static int cwdlen;

static int strcommon(const char *pathname)
{
	int n = 0;

	while (pathname[n] == cwd[n] && n < cwdlen)
		++n;

	return n;
}

struct map {
	struct list_head node;
	u64	 start;
	u64	 end;
	u64	 pgoff;
	u64	 (*map_ip)(struct map *, u64);
	struct dso	 *dso;
};

static u64 map__map_ip(struct map *map, u64 ip)
{
	return ip - map->start + map->pgoff;
}

static u64 vdso__map_ip(struct map *map, u64 ip)
{
	return ip;
}

static inline int is_anon_memory(const char *filename)
{
	return strcmp(filename, "//anon") == 0;
}

static struct map *map__new(struct mmap_event *event)
{
	struct map *self = malloc(sizeof(*self));

	if (self != NULL) {
		const char *filename = event->filename;
		char newfilename[PATH_MAX];
		int anon;

		if (cwd) {
			int n = strcommon(filename);

			if (n == cwdlen) {
				snprintf(newfilename, sizeof(newfilename),
					 ".%s", filename + n);
				filename = newfilename;
			}
		}

		anon = is_anon_memory(filename);

		if (anon) {
			snprintf(newfilename, sizeof(newfilename), "/tmp/perf-%d.map", event->pid);
			filename = newfilename;
		}

		self->start = event->start;
		self->end   = event->start + event->len;
		self->pgoff = event->pgoff;

		self->dso = dsos__findnew(filename);
		if (self->dso == NULL)
			goto out_delete;

		if (self->dso == vdso || anon)
			self->map_ip = vdso__map_ip;
		else
			self->map_ip = map__map_ip;
	}
	return self;
out_delete:
	free(self);
	return NULL;
}

static struct map *map__clone(struct map *self)
{
	struct map *map = malloc(sizeof(*self));

	if (!map)
		return NULL;

	memcpy(map, self, sizeof(*self));

	return map;
}

static int map__overlap(struct map *l, struct map *r)
{
	if (l->start > r->start) {
		struct map *t = l;
		l = r;
		r = t;
	}

	if (l->end > r->start)
		return 1;

	return 0;
}

static size_t map__fprintf(struct map *self, FILE *fp)
{
	return fprintf(fp, " %Lx-%Lx %Lx %s\n",
		       self->start, self->end, self->pgoff, self->dso->name);
}


struct thread {
	struct rb_node	 rb_node;
	struct list_head maps;
	pid_t		 pid;
	char		 *comm;
};

static struct thread *thread__new(pid_t pid)
{
	struct thread *self = malloc(sizeof(*self));

	if (self != NULL) {
		self->pid = pid;
		self->comm = malloc(32);
		if (self->comm)
			snprintf(self->comm, 32, ":%d", self->pid);
		INIT_LIST_HEAD(&self->maps);
	}

	return self;
}

static int thread__set_comm(struct thread *self, const char *comm)
{
	if (self->comm)
		free(self->comm);
	self->comm = strdup(comm);
	return self->comm ? 0 : -ENOMEM;
}

static size_t thread__fprintf(struct thread *self, FILE *fp)
{
	struct map *pos;
	size_t ret = fprintf(fp, "Thread %d %s\n", self->pid, self->comm);

	list_for_each_entry(pos, &self->maps, node)
		ret += map__fprintf(pos, fp);

	return ret;
}


static struct rb_root threads;
static struct thread *last_match;

static struct thread *threads__findnew(pid_t pid)
{
	struct rb_node **p = &threads.rb_node;
	struct rb_node *parent = NULL;
	struct thread *th;

	/*
	 * Font-end cache - PID lookups come in blocks,
	 * so most of the time we dont have to look up
	 * the full rbtree:
	 */
	if (last_match && last_match->pid == pid)
		return last_match;

	while (*p != NULL) {
		parent = *p;
		th = rb_entry(parent, struct thread, rb_node);

		if (th->pid == pid) {
			last_match = th;
			return th;
		}

		if (pid < th->pid)
			p = &(*p)->rb_left;
		else
			p = &(*p)->rb_right;
	}

	th = thread__new(pid);
	if (th != NULL) {
		rb_link_node(&th->rb_node, parent, p);
		rb_insert_color(&th->rb_node, &threads);
		last_match = th;
	}

	return th;
}

static void thread__insert_map(struct thread *self, struct map *map)
{
	struct map *pos, *tmp;

	list_for_each_entry_safe(pos, tmp, &self->maps, node) {
		if (map__overlap(pos, map)) {
			if (verbose >= 2) {
				printf("overlapping maps:\n");
				map__fprintf(map, stdout);
				map__fprintf(pos, stdout);
			}

			if (map->start <= pos->start && map->end > pos->start)
				pos->start = map->end;

			if (map->end >= pos->end && map->start < pos->end)
				pos->end = map->start;

			if (verbose >= 2) {
				printf("after collision:\n");
				map__fprintf(pos, stdout);
			}

			if (pos->start >= pos->end) {
				list_del_init(&pos->node);
				free(pos);
			}
		}
	}

	list_add_tail(&map->node, &self->maps);
}

static int thread__fork(struct thread *self, struct thread *parent)
{
	struct map *map;

	if (self->comm)
		free(self->comm);
	self->comm = strdup(parent->comm);
	if (!self->comm)
		return -ENOMEM;

	list_for_each_entry(map, &parent->maps, node) {
		struct map *new = map__clone(map);
		if (!new)
			return -ENOMEM;
		thread__insert_map(self, new);
	}

	return 0;
}

static struct map *thread__find_map(struct thread *self, u64 ip)
{
	struct map *pos;

	if (self == NULL)
		return NULL;

	list_for_each_entry(pos, &self->maps, node)
		if (ip >= pos->start && ip <= pos->end)
			return pos;

	return NULL;
}

static size_t threads__fprintf(FILE *fp)
{
	size_t ret = 0;
	struct rb_node *nd;

	for (nd = rb_first(&threads); nd; nd = rb_next(nd)) {
		struct thread *pos = rb_entry(nd, struct thread, rb_node);

		ret += thread__fprintf(pos, fp);
	}

	return ret;
}

/*
 * histogram, sorted on item, collects counts
 */

static struct rb_root hist;

struct hist_entry {
	struct rb_node		rb_node;

	struct thread		*thread;
	struct map		*map;
	struct dso		*dso;
	struct symbol		*sym;
	struct symbol		*parent;
	u64			ip;
	char			level;
	struct callchain_node	callchain;
	struct rb_root		sorted_chain;

	u64			count;
};

/*
 * configurable sorting bits
 */

struct sort_entry {
	struct list_head list;

	char *header;

	int64_t (*cmp)(struct hist_entry *, struct hist_entry *);
	int64_t (*collapse)(struct hist_entry *, struct hist_entry *);
	size_t	(*print)(FILE *fp, struct hist_entry *);
};

static int64_t cmp_null(void *l, void *r)
{
	if (!l && !r)
		return 0;
	else if (!l)
		return -1;
	else
		return 1;
}

/* --sort pid */

static int64_t
sort__thread_cmp(struct hist_entry *left, struct hist_entry *right)
{
	return right->thread->pid - left->thread->pid;
}

static size_t
sort__thread_print(FILE *fp, struct hist_entry *self)
{
	return fprintf(fp, "%16s:%5d", self->thread->comm ?: "", self->thread->pid);
}

static struct sort_entry sort_thread = {
	.header = "         Command:  Pid",
	.cmp	= sort__thread_cmp,
	.print	= sort__thread_print,
};

/* --sort comm */

static int64_t
sort__comm_cmp(struct hist_entry *left, struct hist_entry *right)
{
	return right->thread->pid - left->thread->pid;
}

static int64_t
sort__comm_collapse(struct hist_entry *left, struct hist_entry *right)
{
	char *comm_l = left->thread->comm;
	char *comm_r = right->thread->comm;

	if (!comm_l || !comm_r)
		return cmp_null(comm_l, comm_r);

	return strcmp(comm_l, comm_r);
}

static size_t
sort__comm_print(FILE *fp, struct hist_entry *self)
{
	return fprintf(fp, "%16s", self->thread->comm);
}

static struct sort_entry sort_comm = {
	.header		= "         Command",
	.cmp		= sort__comm_cmp,
	.collapse	= sort__comm_collapse,
	.print		= sort__comm_print,
};

/* --sort dso */

static int64_t
sort__dso_cmp(struct hist_entry *left, struct hist_entry *right)
{
	struct dso *dso_l = left->dso;
	struct dso *dso_r = right->dso;

	if (!dso_l || !dso_r)
		return cmp_null(dso_l, dso_r);

	return strcmp(dso_l->name, dso_r->name);
}

static size_t
sort__dso_print(FILE *fp, struct hist_entry *self)
{
	if (self->dso)
		return fprintf(fp, "%-25s", self->dso->name);

	return fprintf(fp, "%016llx         ", (u64)self->ip);
}

static struct sort_entry sort_dso = {
	.header = "Shared Object            ",
	.cmp	= sort__dso_cmp,
	.print	= sort__dso_print,
};

/* --sort symbol */

static int64_t
sort__sym_cmp(struct hist_entry *left, struct hist_entry *right)
{
	u64 ip_l, ip_r;

	if (left->sym == right->sym)
		return 0;

	ip_l = left->sym ? left->sym->start : left->ip;
	ip_r = right->sym ? right->sym->start : right->ip;

	return (int64_t)(ip_r - ip_l);
}

static size_t
sort__sym_print(FILE *fp, struct hist_entry *self)
{
	size_t ret = 0;

	if (verbose)
		ret += fprintf(fp, "%#018llx  ", (u64)self->ip);

	if (self->sym) {
		ret += fprintf(fp, "[%c] %s",
			self->dso == kernel_dso ? 'k' : '.', self->sym->name);
	} else {
		ret += fprintf(fp, "%#016llx", (u64)self->ip);
	}

	return ret;
}

static struct sort_entry sort_sym = {
	.header = "Symbol",
	.cmp	= sort__sym_cmp,
	.print	= sort__sym_print,
};

/* --sort parent */

static int64_t
sort__parent_cmp(struct hist_entry *left, struct hist_entry *right)
{
	struct symbol *sym_l = left->parent;
	struct symbol *sym_r = right->parent;

	if (!sym_l || !sym_r)
		return cmp_null(sym_l, sym_r);

	return strcmp(sym_l->name, sym_r->name);
}

static size_t
sort__parent_print(FILE *fp, struct hist_entry *self)
{
	size_t ret = 0;

	ret += fprintf(fp, "%-20s", self->parent ? self->parent->name : "[other]");

	return ret;
}

static struct sort_entry sort_parent = {
	.header = "Parent symbol       ",
	.cmp	= sort__parent_cmp,
	.print	= sort__parent_print,
};

static int sort__need_collapse = 0;
static int sort__has_parent = 0;

struct sort_dimension {
	char			*name;
	struct sort_entry	*entry;
	int			taken;
};

static struct sort_dimension sort_dimensions[] = {
	{ .name = "pid",	.entry = &sort_thread,	},
	{ .name = "comm",	.entry = &sort_comm,	},
	{ .name = "dso",	.entry = &sort_dso,	},
	{ .name = "symbol",	.entry = &sort_sym,	},
	{ .name = "parent",	.entry = &sort_parent,	},
};

static LIST_HEAD(hist_entry__sort_list);

static int sort_dimension__add(char *tok)
{
	int i;

	for (i = 0; i < ARRAY_SIZE(sort_dimensions); i++) {
		struct sort_dimension *sd = &sort_dimensions[i];

		if (sd->taken)
			continue;

		if (strncasecmp(tok, sd->name, strlen(tok)))
			continue;

		if (sd->entry->collapse)
			sort__need_collapse = 1;

		if (sd->entry == &sort_parent) {
			int ret = regcomp(&parent_regex, parent_pattern, REG_EXTENDED);
			if (ret) {
				char err[BUFSIZ];

				regerror(ret, &parent_regex, err, sizeof(err));
				fprintf(stderr, "Invalid regex: %s\n%s",
					parent_pattern, err);
				exit(-1);
			}
			sort__has_parent = 1;
		}

		list_add_tail(&sd->entry->list, &hist_entry__sort_list);
		sd->taken = 1;

		return 0;
	}

	return -ESRCH;
}

static int64_t
hist_entry__cmp(struct hist_entry *left, struct hist_entry *right)
{
	struct sort_entry *se;
	int64_t cmp = 0;

	list_for_each_entry(se, &hist_entry__sort_list, list) {
		cmp = se->cmp(left, right);
		if (cmp)
			break;
	}

	return cmp;
}

static int64_t
hist_entry__collapse(struct hist_entry *left, struct hist_entry *right)
{
	struct sort_entry *se;
	int64_t cmp = 0;

	list_for_each_entry(se, &hist_entry__sort_list, list) {
		int64_t (*f)(struct hist_entry *, struct hist_entry *);

		f = se->collapse ?: se->cmp;

		cmp = f(left, right);
		if (cmp)
			break;
	}

	return cmp;
}

static size_t
callchain__fprintf(FILE *fp, struct callchain_node *self, u64 total_samples)
{
	struct callchain_list *chain;
	size_t ret = 0;

	if (!self)
		return 0;

	ret += callchain__fprintf(fp, self->parent, total_samples);


	list_for_each_entry(chain, &self->val, list)
		ret += fprintf(fp, "                %p\n", (void *)chain->ip);

	return ret;
}

static size_t
hist_entry_callchain__fprintf(FILE *fp, struct hist_entry *self,
			      u64 total_samples)
{
	struct rb_node *rb_node;
	struct callchain_node *chain;
	size_t ret = 0;

	rb_node = rb_first(&self->sorted_chain);
	while (rb_node) {
		double percent;

		chain = rb_entry(rb_node, struct callchain_node, rb_node);
		percent = chain->hit * 100.0 / total_samples;
		ret += fprintf(fp, "           %6.2f%%\n", percent);
		ret += callchain__fprintf(fp, chain, total_samples);
		ret += fprintf(fp, "\n");
		rb_node = rb_next(rb_node);
	}

	return ret;
}


static size_t
hist_entry__fprintf(FILE *fp, struct hist_entry *self, u64 total_samples)
{
	struct sort_entry *se;
	size_t ret;

	if (exclude_other && !self->parent)
		return 0;

	if (total_samples) {
		double percent = self->count * 100.0 / total_samples;
		char *color = PERF_COLOR_NORMAL;

		/*
		 * We color high-overhead entries in red, mid-overhead
		 * entries in green - and keep the low overhead places
		 * normal:
		 */
		if (percent >= 5.0) {
			color = PERF_COLOR_RED;
		} else {
			if (percent >= 0.5)
				color = PERF_COLOR_GREEN;
		}

		ret = color_fprintf(fp, color, "   %6.2f%%",
				(self->count * 100.0) / total_samples);
	} else
		ret = fprintf(fp, "%12Ld ", self->count);

	list_for_each_entry(se, &hist_entry__sort_list, list) {
		if (exclude_other && (se == &sort_parent))
			continue;

		fprintf(fp, "  ");
		ret += se->print(fp, self);
	}

	ret += fprintf(fp, "\n");

	if (callchain)
		hist_entry_callchain__fprintf(fp, self, total_samples);

	return ret;
}

/*
 *
 */

static struct symbol *
resolve_symbol(struct thread *thread, struct map **mapp,
	       struct dso **dsop, u64 *ipp)
{
	struct dso *dso = dsop ? *dsop : NULL;
	struct map *map = mapp ? *mapp : NULL;
	u64 ip = *ipp;

	if (!thread)
		return NULL;

	if (dso)
		goto got_dso;

	if (map)
		goto got_map;

	map = thread__find_map(thread, ip);
	if (map != NULL) {
		if (mapp)
			*mapp = map;
got_map:
		ip = map->map_ip(map, ip);

		dso = map->dso;
	} else {
		/*
		 * If this is outside of all known maps,
		 * and is a negative address, try to look it
		 * up in the kernel dso, as it might be a
		 * vsyscall (which executes in user-mode):
		 */
		if ((long long)ip < 0)
		dso = kernel_dso;
	}
	dprintf(" ...... dso: %s\n", dso ? dso->name : "<not found>");
	dprintf(" ...... map: %Lx -> %Lx\n", *ipp, ip);
	*ipp  = ip;

	if (dsop)
		*dsop = dso;

	if (!dso)
		return NULL;
got_dso:
	return dso->find_symbol(dso, ip);
}

static int call__match(struct symbol *sym)
{
	if (sym->name && !regexec(&parent_regex, sym->name, 0, NULL, 0))
		return 1;

	return 0;
}

/*
 * collect histogram counts
 */

static int
hist_entry__add(struct thread *thread, struct map *map, struct dso *dso,
		struct symbol *sym, u64 ip, struct ip_callchain *chain,
		char level, u64 count)
{
	struct rb_node **p = &hist.rb_node;
	struct rb_node *parent = NULL;
	struct hist_entry *he;
	struct hist_entry entry = {
		.thread	= thread,
		.map	= map,
		.dso	= dso,
		.sym	= sym,
		.ip	= ip,
		.level	= level,
		.count	= count,
		.parent = NULL,
		.sorted_chain = RB_ROOT
	};
	int cmp;

	if (sort__has_parent && chain) {
		u64 context = PERF_CONTEXT_MAX;
		int i;

		for (i = 0; i < chain->nr; i++) {
			u64 ip = chain->ips[i];
			struct dso *dso = NULL;
			struct symbol *sym;

			if (ip >= PERF_CONTEXT_MAX) {
				context = ip;
				continue;
			}

			switch (context) {
			case PERF_CONTEXT_KERNEL:
				dso = kernel_dso;
				break;
			default:
				break;
			}

			sym = resolve_symbol(thread, NULL, &dso, &ip);

			if (sym && call__match(sym)) {
				entry.parent = sym;
				break;
			}
		}
	}

	while (*p != NULL) {
		parent = *p;
		he = rb_entry(parent, struct hist_entry, rb_node);

		cmp = hist_entry__cmp(&entry, he);

		if (!cmp) {
			he->count += count;
			if (callchain)
				append_chain(&he->callchain, chain);
			return 0;
		}

		if (cmp < 0)
			p = &(*p)->rb_left;
		else
			p = &(*p)->rb_right;
	}