#include "perf.h"
-#include "util/symbol.h"
#include "util/color.h"
+#include "util/session.h"
+#include "util/symbol.h"
+#include "util/thread.h"
#include "util/util.h"
#include <linux/rbtree.h>
#include "util/parse-options.h"
#include "util/parse-events.h"
+#include "util/debug.h"
+
#include <assert.h>
#include <fcntl.h>
#include <stdio.h>
+#include <termios.h>
+#include <unistd.h>
#include <errno.h>
#include <time.h>
static int fd[MAX_NR_CPUS][MAX_COUNTERS];
-static int system_wide = 0;
+static int system_wide = 0;
-static int default_interval = 100000;
+static int default_interval = 0;
-static u64 count_filter = 5;
-static int print_entries = 15;
+static int count_filter = 5;
+static int print_entries;
-static int target_pid = -1;
-static int profile_cpu = -1;
-static int nr_cpus = 0;
-static unsigned int realtime_prio = 0;
-static int group = 0;
+static int target_pid = -1;
+static int inherit = 0;
+static int profile_cpu = -1;
+static int nr_cpus = 0;
+static unsigned int realtime_prio = 0;
+static int group = 0;
static unsigned int page_size;
-static unsigned int mmap_pages = 16;
-static int freq = 0;
-static int verbose = 0;
+static unsigned int mmap_pages = 16;
+static int freq = 1000; /* 1 KHz */
-static char *sym_filter;
-static unsigned long filter_start;
-static unsigned long filter_end;
+static int delay_secs = 2;
+static int zero = 0;
+static int dump_symtab = 0;
-static int delay_secs = 2;
-static int zero;
-static int dump_symtab;
+static bool hide_kernel_symbols = false;
+static bool hide_user_symbols = false;
+static struct winsize winsize;
+
+/*
+ * Source
+ */
+
+struct source_line {
+ u64 eip;
+ unsigned long count[MAX_COUNTERS];
+ char *line;
+ struct source_line *next;
+};
+
+static char *sym_filter = NULL;
+struct sym_entry *sym_filter_entry = NULL;
+static int sym_pcnt_filter = 5;
+static int sym_counter = 0;
+static int display_weighted = -1;
/*
* Symbols
*/
-static u64 min_ip;
-static u64 max_ip = -1ll;
+struct sym_entry_source {
+ struct source_line *source;
+ struct source_line *lines;
+ struct source_line **lines_tail;
+ pthread_mutex_t lock;
+};
struct sym_entry {
struct rb_node rb_node;
struct list_head node;
- unsigned long count[MAX_COUNTERS];
unsigned long snap_count;
double weight;
int skip;
+ u16 name_len;
+ u8 origin;
+ struct map *map;
+ struct sym_entry_source *src;
+ unsigned long count[0];
};
-struct sym_entry *sym_filter_entry;
+/*
+ * Source functions
+ */
+
+static inline struct symbol *sym_entry__symbol(struct sym_entry *self)
+{
+ return ((void *)self) + symbol_conf.priv_size;
+}
+
+static void get_term_dimensions(struct winsize *ws)
+{
+ char *s = getenv("LINES");
+
+ if (s != NULL) {
+ ws->ws_row = atoi(s);
+ s = getenv("COLUMNS");
+ if (s != NULL) {
+ ws->ws_col = atoi(s);
+ if (ws->ws_row && ws->ws_col)
+ return;
+ }
+ }
+#ifdef TIOCGWINSZ
+ if (ioctl(1, TIOCGWINSZ, ws) == 0 &&
+ ws->ws_row && ws->ws_col)
+ return;
+#endif
+ ws->ws_row = 25;
+ ws->ws_col = 80;
+}
+
+static void update_print_entries(struct winsize *ws)
+{
+ print_entries = ws->ws_row;
+
+ if (print_entries > 9)
+ print_entries -= 9;
+}
+
+static void sig_winch_handler(int sig __used)
+{
+ get_term_dimensions(&winsize);
+ update_print_entries(&winsize);
+}
+
+static void parse_source(struct sym_entry *syme)
+{
+ struct symbol *sym;
+ struct sym_entry_source *source;
+ struct map *map;
+ FILE *file;
+ char command[PATH_MAX*2];
+ const char *path;
+ u64 len;
+
+ if (!syme)
+ return;
+
+ if (syme->src == NULL) {
+ syme->src = zalloc(sizeof(*source));
+ if (syme->src == NULL)
+ return;
+ pthread_mutex_init(&syme->src->lock, NULL);
+ }
+
+ source = syme->src;
+
+ if (source->lines) {
+ pthread_mutex_lock(&source->lock);
+ goto out_assign;
+ }
+
+ sym = sym_entry__symbol(syme);
+ map = syme->map;
+ path = map->dso->long_name;
+
+ len = sym->end - sym->start;
+
+ sprintf(command,
+ "objdump --start-address=0x%016Lx "
+ "--stop-address=0x%016Lx -dS %s",
+ map->unmap_ip(map, sym->start),
+ map->unmap_ip(map, sym->end), path);
+
+ file = popen(command, "r");
+ if (!file)
+ return;
+
+ pthread_mutex_lock(&source->lock);
+ source->lines_tail = &source->lines;
+ while (!feof(file)) {
+ struct source_line *src;
+ size_t dummy = 0;
+ char *c;
+
+ src = malloc(sizeof(struct source_line));
+ assert(src != NULL);
+ memset(src, 0, sizeof(struct source_line));
+
+ if (getline(&src->line, &dummy, file) < 0)
+ break;
+ if (!src->line)
+ break;
+
+ c = strchr(src->line, '\n');
+ if (c)
+ *c = 0;
+
+ src->next = NULL;
+ *source->lines_tail = src;
+ source->lines_tail = &src->next;
+
+ if (strlen(src->line)>8 && src->line[8] == ':') {
+ src->eip = strtoull(src->line, NULL, 16);
+ src->eip = map->unmap_ip(map, src->eip);
+ }
+ if (strlen(src->line)>8 && src->line[16] == ':') {
+ src->eip = strtoull(src->line, NULL, 16);
+ src->eip = map->unmap_ip(map, src->eip);
+ }
+ }
+ pclose(file);
+out_assign:
+ sym_filter_entry = syme;
+ pthread_mutex_unlock(&source->lock);
+}
+
+static void __zero_source_counters(struct sym_entry *syme)
+{
+ int i;
+ struct source_line *line;
+
+ line = syme->src->lines;
+ while (line) {
+ for (i = 0; i < nr_counters; i++)
+ line->count[i] = 0;
+ line = line->next;
+ }
+}
+
+static void record_precise_ip(struct sym_entry *syme, int counter, u64 ip)
+{
+ struct source_line *line;
+
+ if (syme != sym_filter_entry)
+ return;
+
+ if (pthread_mutex_trylock(&syme->src->lock))
+ return;
+
+ if (syme->src == NULL || syme->src->source == NULL)
+ goto out_unlock;
+
+ for (line = syme->src->lines; line; line = line->next) {
+ if (line->eip == ip) {
+ line->count[counter]++;
+ break;
+ }
+ if (line->eip > ip)
+ break;
+ }
+out_unlock:
+ pthread_mutex_unlock(&syme->src->lock);
+}
+
+static void lookup_sym_source(struct sym_entry *syme)
+{
+ struct symbol *symbol = sym_entry__symbol(syme);
+ struct source_line *line;
+ char pattern[PATH_MAX];
+
+ sprintf(pattern, "<%s>:", symbol->name);
+
+ pthread_mutex_lock(&syme->src->lock);
+ for (line = syme->src->lines; line; line = line->next) {
+ if (strstr(line->line, pattern)) {
+ syme->src->source = line;
+ break;
+ }
+ }
+ pthread_mutex_unlock(&syme->src->lock);
+}
+
+static void show_lines(struct source_line *queue, int count, int total)
+{
+ int i;
+ struct source_line *line;
+
+ line = queue;
+ for (i = 0; i < count; i++) {
+ float pcnt = 100.0*(float)line->count[sym_counter]/(float)total;
+
+ printf("%8li %4.1f%%\t%s\n", line->count[sym_counter], pcnt, line->line);
+ line = line->next;
+ }
+}
+
+#define TRACE_COUNT 3
+
+static void show_details(struct sym_entry *syme)
+{
+ struct symbol *symbol;
+ struct source_line *line;
+ struct source_line *line_queue = NULL;
+ int displayed = 0;
+ int line_queue_count = 0, total = 0, more = 0;
+
+ if (!syme)
+ return;
+
+ if (!syme->src->source)
+ lookup_sym_source(syme);
+
+ if (!syme->src->source)
+ return;
+
+ symbol = sym_entry__symbol(syme);
+ printf("Showing %s for %s\n", event_name(sym_counter), symbol->name);
+ printf(" Events Pcnt (>=%d%%)\n", sym_pcnt_filter);
-struct dso *kernel_dso;
+ pthread_mutex_lock(&syme->src->lock);
+ line = syme->src->source;
+ while (line) {
+ total += line->count[sym_counter];
+ line = line->next;
+ }
+
+ line = syme->src->source;
+ while (line) {
+ float pcnt = 0.0;
+
+ if (!line_queue_count)
+ line_queue = line;
+ line_queue_count++;
+
+ if (line->count[sym_counter])
+ pcnt = 100.0 * line->count[sym_counter] / (float)total;
+ if (pcnt >= (float)sym_pcnt_filter) {
+ if (displayed <= print_entries)
+ show_lines(line_queue, line_queue_count, total);
+ else more++;
+ displayed += line_queue_count;
+ line_queue_count = 0;
+ line_queue = NULL;
+ } else if (line_queue_count > TRACE_COUNT) {
+ line_queue = line_queue->next;
+ line_queue_count--;
+ }
+
+ line->count[sym_counter] = zero ? 0 : line->count[sym_counter] * 7 / 8;
+ line = line->next;
+ }
+ pthread_mutex_unlock(&syme->src->lock);
+ if (more)
+ printf("%d lines not displayed, maybe increase display entries [e]\n", more);
+}
/*
- * Symbols will be added here in record_ip and will get out
+ * Symbols will be added here in event__process_sample and will get out
* after decayed.
*/
static LIST_HEAD(active_symbols);
double weight = sym->snap_count;
int counter;
+ if (!display_weighted)
+ return weight;
+
for (counter = 1; counter < nr_counters-1; counter++)
weight *= sym->count[counter];
static void print_sym_table(void)
{
int printed = 0, j;
- int counter;
+ int counter, snap = !display_weighted ? sym_counter : 0;
float samples_per_sec = samples/delay_secs;
float ksamples_per_sec = (samples-userspace_samples)/delay_secs;
float sum_ksamples = 0.0;
struct sym_entry *syme, *n;
struct rb_root tmp = RB_ROOT;
struct rb_node *nd;
+ int sym_width = 0, dso_width = 0, max_dso_width;
+ const int win_width = winsize.ws_col - 1;
samples = userspace_samples = 0;
pthread_mutex_unlock(&active_symbols_lock);
list_for_each_entry_safe_from(syme, n, &active_symbols, node) {
- syme->snap_count = syme->count[0];
+ syme->snap_count = syme->count[snap];
if (syme->snap_count != 0) {
+
+ if ((hide_user_symbols &&
+ syme->origin == PERF_RECORD_MISC_USER) ||
+ (hide_kernel_symbols &&
+ syme->origin == PERF_RECORD_MISC_KERNEL)) {
+ list_remove_active_sym(syme);
+ continue;
+ }
syme->weight = sym_weight(syme);
rb_insert_active_sym(&tmp, syme);
sum_ksamples += syme->snap_count;
puts(CONSOLE_CLEAR);
- printf(
-"------------------------------------------------------------------------------\n");
+ printf("%-*.*s\n", win_width, win_width, graph_dotted_line);
printf( " PerfTop:%8.0f irqs/sec kernel:%4.1f%% [",
samples_per_sec,
100.0 - (100.0*((samples_per_sec-ksamples_per_sec)/samples_per_sec)));
- if (nr_counters == 1) {
+ if (nr_counters == 1 || !display_weighted) {
printf("%Ld", (u64)attrs[0].sample_period);
if (freq)
printf("Hz ");
printf(" ");
}
- for (counter = 0; counter < nr_counters; counter++) {
+ if (!display_weighted)
+ printf("%s", event_name(sym_counter));
+ else for (counter = 0; counter < nr_counters; counter++) {
if (counter)
printf("/");
printf(", %d CPUs)\n", nr_cpus);
}
- printf("------------------------------------------------------------------------------\n\n");
+ printf("%-*.*s\n", win_width, win_width, graph_dotted_line);
+
+ if (sym_filter_entry) {
+ show_details(sym_filter_entry);
+ return;
+ }
+ /*
+ * Find the longest symbol name that will be displayed
+ */
+ for (nd = rb_first(&tmp); nd; nd = rb_next(nd)) {
+ syme = rb_entry(nd, struct sym_entry, rb_node);
+ if (++printed > print_entries ||
+ (int)syme->snap_count < count_filter)
+ continue;
+
+ if (syme->map->dso->long_name_len > dso_width)
+ dso_width = syme->map->dso->long_name_len;
+
+ if (syme->name_len > sym_width)
+ sym_width = syme->name_len;
+ }
+
+ printed = 0;
+
+ max_dso_width = winsize.ws_col - sym_width - 29;
+ if (dso_width > max_dso_width)
+ dso_width = max_dso_width;
+ putchar('\n');
if (nr_counters == 1)
- printf(" samples pcnt");
+ printf(" samples pcnt");
else
- printf(" weight samples pcnt");
-
- printf(" RIP kernel function\n"
- " ______ _______ _____ ________________ _______________\n\n"
- );
+ printf(" weight samples pcnt");
+
+ if (verbose)
+ printf(" RIP ");
+ printf(" %-*.*s DSO\n", sym_width, sym_width, "function");
+ printf(" %s _______ _____",
+ nr_counters == 1 ? " " : "______");
+ if (verbose)
+ printf(" ________________");
+ printf(" %-*.*s", sym_width, sym_width, graph_line);
+ printf(" %-*.*s", dso_width, dso_width, graph_line);
+ puts("\n");
for (nd = rb_first(&tmp); nd; nd = rb_next(nd)) {
- struct sym_entry *syme = rb_entry(nd, struct sym_entry, rb_node);
- struct symbol *sym = (struct symbol *)(syme + 1);
- char *color = PERF_COLOR_NORMAL;
+ struct symbol *sym;
double pcnt;
- if (++printed > print_entries || syme->snap_count < count_filter)
+ syme = rb_entry(nd, struct sym_entry, rb_node);
+ sym = sym_entry__symbol(syme);
+
+ if (++printed > print_entries || (int)syme->snap_count < count_filter)
continue;
pcnt = 100.0 - (100.0 * ((sum_ksamples - syme->snap_count) /
sum_ksamples));
- /*
- * We color high-overhead entries in red, mid-overhead
- * entries in green - and keep the low overhead places
- * normal:
- */
- if (pcnt >= 5.0) {
- color = PERF_COLOR_RED;
- } else {
- if (pcnt >= 0.5)
- color = PERF_COLOR_GREEN;
+ if (nr_counters == 1 || !display_weighted)
+ printf("%20.2f ", syme->weight);
+ else
+ printf("%9.1f %10ld ", syme->weight, syme->snap_count);
+
+ percent_color_fprintf(stdout, "%4.1f%%", pcnt);
+ if (verbose)
+ printf(" %016llx", sym->start);
+ printf(" %-*.*s", sym_width, sym_width, sym->name);
+ printf(" %-*.*s\n", dso_width, dso_width,
+ dso_width >= syme->map->dso->long_name_len ?
+ syme->map->dso->long_name :
+ syme->map->dso->short_name);
+ }
+}
+
+static void prompt_integer(int *target, const char *msg)
+{
+ char *buf = malloc(0), *p;
+ size_t dummy = 0;
+ int tmp;
+
+ fprintf(stdout, "\n%s: ", msg);
+ if (getline(&buf, &dummy, stdin) < 0)
+ return;
+
+ p = strchr(buf, '\n');
+ if (p)
+ *p = 0;
+
+ p = buf;
+ while(*p) {
+ if (!isdigit(*p))
+ goto out_free;
+ p++;
+ }
+ tmp = strtoul(buf, NULL, 10);
+ *target = tmp;
+out_free:
+ free(buf);
+}
+
+static void prompt_percent(int *target, const char *msg)
+{
+ int tmp = 0;
+
+ prompt_integer(&tmp, msg);
+ if (tmp >= 0 && tmp <= 100)
+ *target = tmp;
+}
+
+static void prompt_symbol(struct sym_entry **target, const char *msg)
+{
+ char *buf = malloc(0), *p;
+ struct sym_entry *syme = *target, *n, *found = NULL;
+ size_t dummy = 0;
+
+ /* zero counters of active symbol */
+ if (syme) {
+ pthread_mutex_lock(&syme->src->lock);
+ __zero_source_counters(syme);
+ *target = NULL;
+ pthread_mutex_unlock(&syme->src->lock);
+ }
+
+ fprintf(stdout, "\n%s: ", msg);
+ if (getline(&buf, &dummy, stdin) < 0)
+ goto out_free;
+
+ p = strchr(buf, '\n');
+ if (p)
+ *p = 0;
+
+ pthread_mutex_lock(&active_symbols_lock);
+ syme = list_entry(active_symbols.next, struct sym_entry, node);
+ pthread_mutex_unlock(&active_symbols_lock);
+
+ list_for_each_entry_safe_from(syme, n, &active_symbols, node) {
+ struct symbol *sym = sym_entry__symbol(syme);
+
+ if (!strcmp(buf, sym->name)) {
+ found = syme;
+ break;
}
+ }
- if (nr_counters == 1)
- printf("%20.2f - ", syme->weight);
- else
- printf("%9.1f %10ld - ", syme->weight, syme->snap_count);
+ if (!found) {
+ fprintf(stderr, "Sorry, %s is not active.\n", sym_filter);
+ sleep(1);
+ return;
+ } else
+ parse_source(found);
+
+out_free:
+ free(buf);
+}
+
+static void print_mapped_keys(void)
+{
+ char *name = NULL;
+
+ if (sym_filter_entry) {
+ struct symbol *sym = sym_entry__symbol(sym_filter_entry);
+ name = sym->name;
+ }
+
+ fprintf(stdout, "\nMapped keys:\n");
+ fprintf(stdout, "\t[d] display refresh delay. \t(%d)\n", delay_secs);
+ fprintf(stdout, "\t[e] display entries (lines). \t(%d)\n", print_entries);
- color_fprintf(stdout, color, "%4.1f%%", pcnt);
- printf(" - %016llx : %s\n", sym->start, sym->name);
+ if (nr_counters > 1)
+ fprintf(stdout, "\t[E] active event counter. \t(%s)\n", event_name(sym_counter));
+
+ fprintf(stdout, "\t[f] profile display filter (count). \t(%d)\n", count_filter);
+
+ if (symbol_conf.vmlinux_name) {
+ fprintf(stdout, "\t[F] annotate display filter (percent). \t(%d%%)\n", sym_pcnt_filter);
+ fprintf(stdout, "\t[s] annotate symbol. \t(%s)\n", name?: "NULL");
+ fprintf(stdout, "\t[S] stop annotation.\n");
+ }
+
+ if (nr_counters > 1)
+ fprintf(stdout, "\t[w] toggle display weighted/count[E]r. \t(%d)\n", display_weighted ? 1 : 0);
+
+ fprintf(stdout,
+ "\t[K] hide kernel_symbols symbols. \t(%s)\n",
+ hide_kernel_symbols ? "yes" : "no");
+ fprintf(stdout,
+ "\t[U] hide user symbols. \t(%s)\n",
+ hide_user_symbols ? "yes" : "no");
+ fprintf(stdout, "\t[z] toggle sample zeroing. \t(%d)\n", zero ? 1 : 0);
+ fprintf(stdout, "\t[qQ] quit.\n");
+}
+
+static int key_mapped(int c)
+{
+ switch (c) {
+ case 'd':
+ case 'e':
+ case 'f':
+ case 'z':
+ case 'q':
+ case 'Q':
+ case 'K':
+ case 'U':
+ return 1;
+ case 'E':
+ case 'w':
+ return nr_counters > 1 ? 1 : 0;
+ case 'F':
+ case 's':
+ case 'S':
+ return symbol_conf.vmlinux_name ? 1 : 0;
+ default:
+ break;
+ }
+
+ return 0;
+}
+
+static void handle_keypress(int c)
+{
+ if (!key_mapped(c)) {
+ struct pollfd stdin_poll = { .fd = 0, .events = POLLIN };
+ struct termios tc, save;
+
+ print_mapped_keys();
+ fprintf(stdout, "\nEnter selection, or unmapped key to continue: ");
+ fflush(stdout);
+
+ tcgetattr(0, &save);
+ tc = save;
+ tc.c_lflag &= ~(ICANON | ECHO);
+ tc.c_cc[VMIN] = 0;
+ tc.c_cc[VTIME] = 0;
+ tcsetattr(0, TCSANOW, &tc);
+
+ poll(&stdin_poll, 1, -1);
+ c = getc(stdin);
+
+ tcsetattr(0, TCSAFLUSH, &save);
+ if (!key_mapped(c))
+ return;
+ }
+
+ switch (c) {
+ case 'd':
+ prompt_integer(&delay_secs, "Enter display delay");
+ if (delay_secs < 1)
+ delay_secs = 1;
+ break;
+ case 'e':
+ prompt_integer(&print_entries, "Enter display entries (lines)");
+ if (print_entries == 0) {
+ sig_winch_handler(SIGWINCH);
+ signal(SIGWINCH, sig_winch_handler);
+ } else
+ signal(SIGWINCH, SIG_DFL);
+ break;
+ case 'E':
+ if (nr_counters > 1) {
+ int i;
+
+ fprintf(stderr, "\nAvailable events:");
+ for (i = 0; i < nr_counters; i++)
+ fprintf(stderr, "\n\t%d %s", i, event_name(i));
+
+ prompt_integer(&sym_counter, "Enter details event counter");
+
+ if (sym_counter >= nr_counters) {
+ fprintf(stderr, "Sorry, no such event, using %s.\n", event_name(0));
+ sym_counter = 0;
+ sleep(1);
+ }
+ } else sym_counter = 0;
+ break;
+ case 'f':
+ prompt_integer(&count_filter, "Enter display event count filter");
+ break;
+ case 'F':
+ prompt_percent(&sym_pcnt_filter, "Enter details display event filter (percent)");
+ break;
+ case 'K':
+ hide_kernel_symbols = !hide_kernel_symbols;
+ break;
+ case 'q':
+ case 'Q':
+ printf("exiting.\n");
+ if (dump_symtab)
+ dsos__fprintf(stderr);
+ exit(0);
+ case 's':
+ prompt_symbol(&sym_filter_entry, "Enter details symbol");
+ break;
+ case 'S':
+ if (!sym_filter_entry)
+ break;
+ else {
+ struct sym_entry *syme = sym_filter_entry;
+
+ pthread_mutex_lock(&syme->src->lock);
+ sym_filter_entry = NULL;
+ __zero_source_counters(syme);
+ pthread_mutex_unlock(&syme->src->lock);
+ }
+ break;
+ case 'U':
+ hide_user_symbols = !hide_user_symbols;
+ break;
+ case 'w':
+ display_weighted = ~display_weighted;
+ break;
+ case 'z':
+ zero = ~zero;
+ break;
+ default:
+ break;
}
}
static void *display_thread(void *arg __used)
{
struct pollfd stdin_poll = { .fd = 0, .events = POLLIN };
- int delay_msecs = delay_secs * 1000;
+ struct termios tc, save;
+ int delay_msecs, c;
- printf("PerfTop refresh period: %d seconds\n", delay_secs);
+ tcgetattr(0, &save);
+ tc = save;
+ tc.c_lflag &= ~(ICANON | ECHO);
+ tc.c_cc[VMIN] = 0;
+ tc.c_cc[VTIME] = 0;
+
+repeat:
+ delay_msecs = delay_secs * 1000;
+ tcsetattr(0, TCSANOW, &tc);
+ /* trash return*/
+ getc(stdin);
do {
print_sym_table();
} while (!poll(&stdin_poll, 1, delay_msecs) == 1);
- printf("key pressed - exiting.\n");
- exit(0);
+ c = getc(stdin);
+ tcsetattr(0, TCSAFLUSH, &save);
+
+ handle_keypress(c);
+ goto repeat;
return NULL;
}
"enter_idle",
"exit_idle",
"mwait_idle",
+ "mwait_idle_with_hints",
+ "poll_idle",
"ppc64_runlatch_off",
"pseries_dedicated_idle_sleep",
NULL
};
-static int symbol_filter(struct dso *self, struct symbol *sym)
+static int symbol_filter(struct map *map, struct symbol *sym)
{
- static int filter_match;
struct sym_entry *syme;
const char *name = sym->name;
int i;
strstr(name, "_text_end"))
return 1;
- syme = dso__sym_priv(self, sym);
+ syme = symbol__priv(sym);
+ syme->map = map;
+ syme->src = NULL;
+ if (!sym_filter_entry && sym_filter && !strcmp(name, sym_filter))
+ sym_filter_entry = syme;
+
for (i = 0; skip_symbols[i]; i++) {
if (!strcmp(skip_symbols[i], name)) {
syme->skip = 1;
}
}
- if (filter_match == 1) {
- filter_end = sym->start;
- filter_match = -1;
- if (filter_end - filter_start > 10000) {
- fprintf(stderr,
- "hm, too large filter symbol <%s> - skipping.\n",
- sym_filter);
- fprintf(stderr, "symbol filter start: %016lx\n",
- filter_start);
- fprintf(stderr, " end: %016lx\n",
- filter_end);
- filter_end = filter_start = 0;
- sym_filter = NULL;
- sleep(1);
- }
- }
-
- if (filter_match == 0 && sym_filter && !strcmp(name, sym_filter)) {
- filter_match = 1;
- filter_start = sym->start;
- }
-
+ if (!syme->skip)
+ syme->name_len = strlen(sym->name);
return 0;
}
-static int parse_symbols(void)
+static void event__process_sample(const event_t *self,
+ struct perf_session *session, int counter)
{
- struct rb_node *node;
- struct symbol *sym;
-
- kernel_dso = dso__new("[kernel]", sizeof(struct sym_entry));
- if (kernel_dso == NULL)
- return -1;
-
- if (dso__load_kernel(kernel_dso, NULL, symbol_filter, 1) != 0)
- goto out_delete_dso;
-
- node = rb_first(&kernel_dso->syms);
- sym = rb_entry(node, struct symbol, rb_node);
- min_ip = sym->start;
-
- node = rb_last(&kernel_dso->syms);
- sym = rb_entry(node, struct symbol, rb_node);
- max_ip = sym->end;
-
- if (dump_symtab)
- dso__fprintf(kernel_dso, stderr);
-
- return 0;
-
-out_delete_dso:
- dso__delete(kernel_dso);
- kernel_dso = NULL;
- return -1;
-}
-
-#define TRACE_COUNT 3
-
-/*
- * Binary search in the histogram table and record the hit:
- */
-static void record_ip(u64 ip, int counter)
-{
- struct symbol *sym = dso__find_symbol(kernel_dso, ip);
-
- if (sym != NULL) {
- struct sym_entry *syme = dso__sym_priv(kernel_dso, sym);
+ u64 ip = self->ip.ip;
+ struct sym_entry *syme;
+ struct addr_location al;
+ u8 origin = self->header.misc & PERF_RECORD_MISC_CPUMODE_MASK;
- if (!syme->skip) {
- syme->count[counter]++;
- pthread_mutex_lock(&active_symbols_lock);
- if (list_empty(&syme->node) || !syme->node.next)
- __list_insert_active_sym(syme);
- pthread_mutex_unlock(&active_symbols_lock);
+ switch (origin) {
+ case PERF_RECORD_MISC_USER:
+ if (hide_user_symbols)
return;
- }
+ break;
+ case PERF_RECORD_MISC_KERNEL:
+ if (hide_kernel_symbols)
+ return;
+ break;
+ default:
+ return;
}
- samples--;
+ if (event__preprocess_sample(self, session, &al, symbol_filter) < 0 ||
+ al.sym == NULL || al.filtered)
+ return;
+
+ syme = symbol__priv(al.sym);
+ if (!syme->skip) {
+ syme->count[counter]++;
+ syme->origin = origin;
+ record_precise_ip(syme, counter, ip);
+ pthread_mutex_lock(&active_symbols_lock);
+ if (list_empty(&syme->node) || !syme->node.next)
+ __list_insert_active_sym(syme);
+ pthread_mutex_unlock(&active_symbols_lock);
+ if (origin == PERF_RECORD_MISC_USER)
+ ++userspace_samples;
+ ++samples;
+ }
}
-static void process_event(u64 ip, int counter, int user)
+static int event__process(event_t *event, struct perf_session *session)
{
- samples++;
-
- if (user) {
- userspace_samples++;
- return;
+ switch (event->header.type) {
+ case PERF_RECORD_COMM:
+ event__process_comm(event, session);
+ break;
+ case PERF_RECORD_MMAP:
+ event__process_mmap(event, session);
+ break;
+ default:
+ break;
}
- record_ip(ip, counter);
+ return 0;
}
struct mmap_data {
static unsigned int mmap_read_head(struct mmap_data *md)
{
- struct perf_counter_mmap_page *pc = md->base;
+ struct perf_event_mmap_page *pc = md->base;
int head;
head = pc->data_head;
return head;
}
-struct timeval last_read, this_read;
-
-static void mmap_read_counter(struct mmap_data *md)
+static void perf_session__mmap_read_counter(struct perf_session *self,
+ struct mmap_data *md)
{
unsigned int head = mmap_read_head(md);
unsigned int old = md->prev;
unsigned char *data = md->base + page_size;
int diff;
- gettimeofday(&this_read, NULL);
-
/*
* If we're further behind than half the buffer, there's a chance
* the writer will bite our tail and mess up the samples under us.
*/
diff = head - old;
if (diff > md->mask / 2 || diff < 0) {
- struct timeval iv;
- unsigned long msecs;
-
- timersub(&this_read, &last_read, &iv);
- msecs = iv.tv_sec*1000 + iv.tv_usec/1000;
-
- fprintf(stderr, "WARNING: failed to keep up with mmap data."
- " Last read %lu msecs ago.\n", msecs);
+ fprintf(stderr, "WARNING: failed to keep up with mmap data.\n");
/*
* head points to a known good entry, start there.
old = head;
}
- last_read = this_read;
-
for (; old != head;) {
- struct ip_event {
- struct perf_event_header header;
- u64 ip;
- u32 pid, target_pid;
- };
- struct mmap_event {
- struct perf_event_header header;
- u32 pid, target_pid;
- u64 start;
- u64 len;
- u64 pgoff;
- char filename[PATH_MAX];
- };
-
- typedef union event_union {
- struct perf_event_header header;
- struct ip_event ip;
- struct mmap_event mmap;
- } event_t;
-
event_t *event = (event_t *)&data[old & md->mask];
event_t event_copy;
event = &event_copy;
}
+ if (event->header.type == PERF_RECORD_SAMPLE)
+ event__process_sample(event, self, md->counter);
+ else
+ event__process(event, self);
old += size;
-
- if (event->header.type == PERF_EVENT_SAMPLE) {
- int user =
- (event->header.misc & PERF_EVENT_MISC_CPUMODE_MASK) == PERF_EVENT_MISC_USER;
- process_event(event->ip.ip, md->counter, user);
- }
}
md->prev = old;
static struct pollfd event_array[MAX_NR_CPUS * MAX_COUNTERS];
static struct mmap_data mmap_array[MAX_NR_CPUS][MAX_COUNTERS];
-static void mmap_read(void)
+static void perf_session__mmap_read(struct perf_session *self)
{
int i, counter;
for (i = 0; i < nr_cpus; i++) {
for (counter = 0; counter < nr_counters; counter++)
- mmap_read_counter(&mmap_array[i][counter]);
+ perf_session__mmap_read_counter(self, &mmap_array[i][counter]);
}
}
static void start_counter(int i, int counter)
{
- struct perf_counter_attr *attr;
- unsigned int cpu;
+ struct perf_event_attr *attr;
+ int cpu;
cpu = profile_cpu;
if (target_pid == -1 && profile_cpu == -1)
attr = attrs + counter;
attr->sample_type = PERF_SAMPLE_IP | PERF_SAMPLE_TID;
- attr->freq = freq;
+
+ if (freq) {
+ attr->sample_type |= PERF_SAMPLE_PERIOD;
+ attr->freq = 1;
+ attr->sample_freq = freq;
+ }
+
+ attr->inherit = (cpu < 0) && inherit;
+ attr->mmap = 1;
try_again:
- fd[i][counter] = sys_perf_counter_open(attr, target_pid, cpu, group_fd, 0);
+ fd[i][counter] = sys_perf_event_open(attr, target_pid, cpu, group_fd, 0);
if (fd[i][counter] < 0) {
int err = errno;
- if (err == EPERM)
+ if (err == EPERM || err == EACCES)
die("No permission - are you root?\n");
/*
* If it's cycles then fall back to hrtimer
printf("\n");
error("perfcounter syscall returned with %d (%s)\n",
fd[i][counter], strerror(err));
- die("No CONFIG_PERF_COUNTERS=y kernel support configured?\n");
+ die("No CONFIG_PERF_EVENTS=y kernel support configured?\n");
exit(-1);
}
assert(fd[i][counter] >= 0);
pthread_t thread;
int i, counter;
int ret;
+ /*
+ * FIXME: perf_session__new should allow passing a O_MMAP, so that all this
+ * mmap reading, etc is encapsulated in it. Use O_WRONLY for now.
+ */
+ struct perf_session *session = perf_session__new(NULL, O_WRONLY, false);
+ if (session == NULL)
+ return -ENOMEM;
+
+ if (perf_session__create_kernel_maps(session) < 0) {
+ pr_err("Problems creating kernel maps\n");
+ return -1;
+ }
+
+ if (target_pid != -1)
+ event__synthesize_thread(target_pid, event__process, session);
+ else
+ event__synthesize_threads(event__process, session);
for (i = 0; i < nr_cpus; i++) {
group_fd = -1;
/* Wait for a minimal set of events before starting the snapshot */
poll(event_array, nr_poll, 100);
- mmap_read();
+ perf_session__mmap_read(session);
if (pthread_create(&thread, NULL, display_thread, NULL)) {
printf("Could not create display thread.\n");
while (1) {
int hits = samples;
- mmap_read();
+ perf_session__mmap_read(session);
if (hits == samples)
ret = poll(event_array, nr_poll, 100);
"system-wide collection from all CPUs"),
OPT_INTEGER('C', "CPU", &profile_cpu,
"CPU to profile on"),
+ OPT_STRING('k', "vmlinux", &symbol_conf.vmlinux_name,
+ "file", "vmlinux pathname"),
+ OPT_BOOLEAN('K', "hide_kernel_symbols", &hide_kernel_symbols,
+ "hide kernel symbols"),
OPT_INTEGER('m', "mmap-pages", &mmap_pages,
"number of mmap data pages"),
OPT_INTEGER('r', "realtime", &realtime_prio,
"only display functions with more events than this"),
OPT_BOOLEAN('g', "group", &group,
"put the counters into a counter group"),
- OPT_STRING('s', "sym-filter", &sym_filter, "pattern",
- "only display symbols matchig this pattern"),
+ OPT_BOOLEAN('i', "inherit", &inherit,
+ "child tasks inherit counters"),
+ OPT_STRING('s', "sym-annotate", &sym_filter, "symbol name",
+ "symbol to annotate - requires -k option"),
OPT_BOOLEAN('z', "zero", &zero,
"zero history across updates"),
OPT_INTEGER('F', "freq", &freq,
"profile at this frequency"),
OPT_INTEGER('E', "entries", &print_entries,
"display this many functions"),
+ OPT_BOOLEAN('U', "hide_user_symbols", &hide_user_symbols,
+ "hide user symbols"),
OPT_BOOLEAN('v', "verbose", &verbose,
"be more verbose (show counter open errors, etc)"),
OPT_END()
if (argc)
usage_with_options(top_usage, options);
- if (freq) {
- default_interval = freq;
- freq = 1;
- }
-
/* CPU and PID are mutually exclusive */
if (target_pid != -1 && profile_cpu != -1) {
printf("WARNING: PID switch overriding CPU\n");
if (!nr_counters)
nr_counters = 1;
+ symbol_conf.priv_size = (sizeof(struct sym_entry) +
+ (nr_counters + 1) * sizeof(unsigned long));
+ if (symbol_conf.vmlinux_name == NULL)
+ symbol_conf.try_vmlinux_path = true;
+ if (symbol__init() < 0)
+ return -1;
+
if (delay_secs < 1)
delay_secs = 1;
- parse_symbols();
+ parse_source(sym_filter_entry);
+
+ /*
+ * User specified count overrides default frequency.
+ */
+ if (default_interval)
+ freq = 0;
+ else if (freq) {
+ default_interval = freq;
+ } else {
+ fprintf(stderr, "frequency and count are zero, aborting\n");
+ exit(EXIT_FAILURE);
+ }
/*
* Fill in the ones not specifically initialized via -c:
if (target_pid != -1 || profile_cpu != -1)
nr_cpus = 1;
+ get_term_dimensions(&winsize);
+ if (print_entries == 0) {
+ update_print_entries(&winsize);
+ signal(SIGWINCH, sig_winch_handler);
+ }
+
return __cmd_top();
}
Code Review / linux-2.6.git / blobdiff