From b507cc3aceda26eff851230223a62c6fb471573c Mon Sep 17 00:00:00 2001
From: Marek Polacek <mmpolacek@gmail.com>
Date: Mon, 25 Oct 2010 03:44:34 +0200
Subject: powertop: new applet

Signed-off-by: Marek Polacek <mmpolacek@gmail.com>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
---
 libbb/Config.src      |  11 +
 procps/powertop.c     | 886 ++++++++++++++++++++++++++++++++++++++++++++++++++
 util-linux/Config.src |  11 -
 3 files changed, 897 insertions(+), 11 deletions(-)
 create mode 100644 procps/powertop.c

diff --git a/libbb/Config.src b/libbb/Config.src
index 74dc9c549..f6c7a11ea 100644
--- a/libbb/Config.src
+++ b/libbb/Config.src
@@ -43,6 +43,17 @@ config FEATURE_ETC_NETWORKS
 	  a rarely used feature which allows you to use names
 	  instead of IP/mask pairs in route command.
 
+config FEATURE_USE_TERMIOS
+	bool "Use termios to manipulate the screen"
+	default y
+	depends on MORE || TOP || POWERTOP
+	help
+	  This option allows utilities such as 'more' and 'top' to determine
+	  the size of the screen. If you leave this disabled, your utilities
+	  that display things on the screen will be especially primitive and
+	  will be unable to determine the current screen size, and will be
+	  unable to move the cursor.
+
 config FEATURE_EDITING
 	bool "Command line editing"
 	default y
diff --git a/procps/powertop.c b/procps/powertop.c
new file mode 100644
index 000000000..f35aa5c0a
--- /dev/null
+++ b/procps/powertop.c
@@ -0,0 +1,886 @@
+/* vi: set sw=4 ts=4: */
+/*
+ * A mini 'powertop' utility:
+ *   Analyze power consumption on Intel-based laptops.
+ * Based on powertop 1.11.
+ *
+ * Copyright (C) 2010 Marek Polacek <mmpolacek@gmail.com>
+ *
+ * Licensed under GPLv2, see file LICENSE in this source tree.
+ */
+
+//applet:IF_POWERTOP(APPLET(powertop, _BB_DIR_BIN, _BB_SUID_DROP))
+
+//kbuild:lib-$(CONFIG_POWERTOP) += powertop.o
+
+//config:config POWERTOP
+//config:	bool "powertop"
+//config:	default y
+//config:	help
+//config:	  Analyze power consumption on Intel-based laptops
+
+#include "libbb.h"
+
+//#define debug(fmt, ...) fprintf(stderr, fmt, ## __VA_ARGS__)
+#define debug(fmt, ...) ((void)0)
+
+// XXX This should not be here
+#define ENABLE_FEATURE_POWERTOP_PROCIRQ 1
+
+#define DEFAULT_SLEEP      10
+#define DEFAULT_SLEEP_STR "10"
+
+/* Frequency of the ACPI timer */
+#define FREQ_ACPI          3579.545
+#define FREQ_ACPI_1000	   3579545
+
+/* Max filename length of entry in /sys/devices subsystem */
+#define BIG_SYSNAME_LEN    16
+
+typedef unsigned long long ullong;
+
+struct line {
+	char *string;
+	int count;
+	int disk_count;
+};
+
+#if ENABLE_FEATURE_POWERTOP_PROCIRQ
+#define IRQCOUNT		40
+
+struct irqdata {
+	int active;
+	int number;
+	ullong count;
+	char irq_desc[32];
+};
+#endif
+
+struct globals {
+	bool timer_list_read;
+	smallint nostats;
+	int headline;
+	int nlines;
+	int linesize;
+	int maxcstate;
+#if ENABLE_FEATURE_POWERTOP_PROCIRQ
+	int total_interrupt;
+	int interrupt_0;
+	int percpu_hpet_start;
+	int percpu_hpet_end;
+	struct irqdata interrupts[IRQCOUNT];
+#endif
+	unsigned total_cpus;
+	ullong start_usage[8];
+	ullong last_usage[8];
+	ullong start_duration[8];
+	ullong last_duration[8];
+	char cstate_names[8][16];
+	struct line *lines;
+#if ENABLE_FEATURE_USE_TERMIOS
+	struct termios init_settings;
+#endif
+};
+#define G (*ptr_to_globals)
+#define INIT_G() do { \
+	SET_PTR_TO_GLOBALS(xzalloc(sizeof(G))); \
+} while (0)
+
+#if ENABLE_FEATURE_USE_TERMIOS
+static void reset_term(void)
+{
+	tcsetattr_stdin_TCSANOW(&G.init_settings);
+}
+
+static void sig_handler(int signo UNUSED_PARAM)
+{
+	reset_term();
+	exit(EXIT_FAILURE);
+}
+#endif
+
+static int write_str_to_file(const char *fname, const char *str)
+{
+	FILE *fp = fopen_for_write(fname);
+	if (!fp)
+		return 1;
+	fputs(str, fp);
+	fclose(fp);
+	return 0;
+}
+
+/* Make it more readable */
+#define start_timer()	write_str_to_file("/proc/timer_stats", "1\n")
+#define stop_timer()	write_str_to_file("/proc/timer_stats", "0\n")
+
+static void NOINLINE clear_lines(void)
+{
+	int i;
+
+	for (i = 0; i < G.headline; i++)
+		free(G.lines[i].string);
+	free(G.lines);
+	G.headline = G.linesize = 0;
+	G.lines = NULL;
+}
+
+static void count_lines(void)
+{
+	int i;
+
+	for (i = 0; i < G.headline; i++)
+		G.nlines += G.lines[i].count;
+}
+
+static int line_compare(const void *p1, const void *p2)
+{
+	const struct line *a = p1;
+	const struct line *b = p2;
+
+	return (b->count + 50 * b->disk_count) - (a->count + 50 * a->disk_count);
+}
+
+static void do_sort(void)
+{
+	qsort(G.lines, G.headline, sizeof(struct line), line_compare);
+}
+
+/*
+ * Save C-state names, usage and duration. Also get maxcstate.
+ * Reads data from /proc.
+ */
+static void read_data(ullong *usage, ullong *duration)
+{
+	DIR *dir;
+	struct dirent *d;
+
+	dir = opendir("/proc/acpi/processor");
+	if (!dir)
+		return;
+
+	while ((d = readdir(dir)) != NULL) {
+		FILE *fp;
+		char buf[192];
+		int level = 0;
+		int len;
+
+		len = strlen(d->d_name);
+		if (len < 3 || len > BIG_SYSNAME_LEN)
+			continue;
+
+		sprintf(buf, "/proc/acpi/processor/%s/power", d->d_name);
+		fp = fopen_for_read(buf);
+		if (!fp)
+			continue;
+
+		while (fgets(buf, sizeof(buf), fp)) {
+			char *p;
+
+			/* Get usage */
+			p = strstr(buf, "age[");
+			if (!p)
+				continue;
+			p += 4;
+			usage[level] += bb_strtoull(p, NULL, 10) + 1;
+
+			/* Get duration */
+			p = strstr(buf, "ation[");
+			if (!p)
+				continue;
+			p += 6;
+			duration[level] += bb_strtoull(p, NULL, 10);
+
+			/* Increment level */
+			level++;
+
+			/* Also update maxcstate */
+			if (level > G.maxcstate)
+				G.maxcstate = level;
+		}
+		fclose(fp);
+	}
+	closedir(dir);
+}
+
+/* Add line and/or update count */
+static void push_line(const char *string, int count)
+{
+	int i;
+
+	if (!string)
+		return;
+
+	/* Loop through entries */
+	for (i = 0; i < G.headline; i++) {
+		if (strcmp(string, G.lines[i].string) == 0) {
+			/* It's already there, only update count */
+			G.lines[i].count += count;
+			return;
+		}
+	}
+
+	G.lines = xrealloc_vector(G.lines, 1, G.headline);
+
+	G.lines[G.headline].string = xstrdup(string);
+	G.lines[G.headline].count = count;
+	G.lines[G.headline].disk_count = 0;
+
+	/* We added a line */
+	G.headline++;
+}
+
+#if ENABLE_FEATURE_POWERTOP_PROCIRQ
+static int percpu_hpet_timer(const char *name)
+{
+	char *p;
+	long hpet_chan;
+
+	/* This is done once */
+	if (!G.timer_list_read) {
+		FILE *fp;
+		char buf[80];
+
+		G.timer_list_read = true;
+		fp = fopen_for_read("/proc/timer_list");
+		if (!fp)
+			return 0;
+
+		while (fgets(buf, sizeof(buf), fp)) {
+			p = strstr(buf, "Clock Event Device: hpet");
+			if (!p)
+				continue;
+			p += sizeof("Clock Event Device: hpet")-1;
+			if (!isdigit(p[0]))
+				continue;
+			hpet_chan = xatoi_positive(p);
+			if (hpet_chan < G.percpu_hpet_start)
+				G.percpu_hpet_start = hpet_chan;
+			if (hpet_chan > G.percpu_hpet_end)
+				G.percpu_hpet_end = hpet_chan;
+		}
+		fclose(fp);
+	}
+
+	p = strstr(name, "hpet");
+	if (!p)
+		return 0;
+
+	p += 4;
+	if (!isdigit(p[0]))
+		return 0;
+
+	hpet_chan = xatoi_positive(p);
+	if (G.percpu_hpet_start <= hpet_chan && hpet_chan <= G.percpu_hpet_end)
+		return 1;
+
+	return 0;
+}
+
+static int update_irq(int irq, ullong count)
+{
+	int unused = IRQCOUNT;
+	int i;
+
+	for (i = 0; i < IRQCOUNT; i++) {
+		if (G.interrupts[i].active && G.interrupts[i].number == irq) {
+			ullong old;
+			old = G.interrupts[i].count;
+			G.interrupts[i].count = count;
+			return count - old;
+		}
+		if (!G.interrupts[i].active && unused > i)
+			unused = i;
+	}
+
+	G.interrupts[unused].active = 1;
+	G.interrupts[unused].count = count;
+	G.interrupts[unused].number = irq;
+
+	return count;
+}
+
+/*
+ * Read /proc/interrupts, save IRQ counts and IRQ description.
+ */
+static void do_proc_irq(void)
+{
+	FILE *fp;
+	char buf[128];
+
+	/* Reset values */
+	G.interrupt_0 = 0;
+	G.total_interrupt = 0;
+
+	fp = xfopen_for_read("/proc/interrupts");
+	while (fgets(buf, sizeof(buf), fp)) {
+		char irq_desc[sizeof("   <kernel IPI> : ") + sizeof(buf)];
+		char *p;
+		const char *name;
+		int nr = -1;
+		ullong count;
+		ullong delta;
+		int special;
+
+		/* Skip header */
+		p = strchr(buf, ':');
+		if (!p)
+			continue;
+		/*  0:  143646045  153901007   IO-APIC-edge      timer
+		 *   ^
+		 */
+		/* Deal with non-maskable interrupts -- make up fake numbers */
+		special = 0;
+		if (buf[0] != ' ' && !isdigit(buf[0])) {
+			if (strncmp(buf, "NMI:", 4) == 0)
+				nr = 20000;
+			if (strncmp(buf, "RES:", 4) == 0)
+				nr = 20001;
+			if (strncmp(buf, "CAL:", 4) == 0)
+				nr = 20002;
+			if (strncmp(buf, "TLB:", 4) == 0)
+				nr = 20003;
+			if (strncmp(buf, "TRM:", 4) == 0)
+				nr = 20004;
+			if (strncmp(buf, "THR:", 4) == 0)
+				nr = 20005;
+			if (strncmp(buf, "SPU:", 4) == 0)
+				nr = 20006;
+			special = 1;
+		} else {
+			/* bb_strtou don't eat leading spaces, using strtoul */
+			nr = strtoul(buf, NULL, 10); /* xato*() wouldn't work */
+		}
+		if (nr == -1)
+			continue;
+
+		p++;
+		/*  0:  143646045  153901007   IO-APIC-edge      timer
+		 *    ^
+		 */
+		/* Count sum of the IRQs */
+		count = 0;
+		while (1) {
+			char *tmp;
+			p = skip_whitespace(p);
+			if (!isdigit(*p))
+				break;
+			count += bb_strtoull(p, &tmp, 10);
+			p = tmp;
+		}
+		/*   0:  143646045  153901007   IO-APIC-edge      timer
+		 * NMI:          1          2   Non-maskable interrupts
+		 *		                ^
+		 */
+		if (!special) {
+			/* Skip to the interrupt name, e.g. 'timer' */
+			p = strchr(p, ' ');
+			if (!p)
+				continue;
+			p = skip_whitespace(p);
+		}
+
+		name = p;
+		strchrnul(name, '\n')[0] = '\0';
+		/* Save description of the interrupt */
+		if (special)
+			sprintf(irq_desc, "   <kernel IPI> : %s", name);
+		else
+			sprintf(irq_desc, "    <interrupt> : %s", name);
+
+		delta = update_irq(nr, count);
+
+		/* Skip per CPU timer interrupts */
+		if (percpu_hpet_timer(name))
+			delta = 0;
+		if (nr > 0 && delta > 0)
+			push_line(irq_desc, delta);
+		if (!nr)
+			G.interrupt_0 = delta;
+		else
+			G.total_interrupt += delta;
+	}
+
+	fclose(fp);
+}
+#endif /* ENABLE_FEATURE_POWERTOP_PROCIRQ */
+
+#ifdef __i386__
+/*
+ * Get information about CPU using CPUID opcode.
+ */
+static void cpuid(unsigned int *eax, unsigned int *ebx, unsigned int *ecx,
+				  unsigned int *edx)
+{
+	/* EAX value specifies what information to return */
+	__asm__(
+		"	pushl %%ebx\n"     /* Save EBX */
+		"	cpuid\n"
+		"	movl %%ebx, %1\n"  /* Save content of EBX */
+		"	popl %%ebx\n"      /* Restore EBX */
+		: "=a"(*eax), /* Output */
+		  "=r"(*ebx),
+		  "=c"(*ecx),
+		  "=d"(*edx)
+		: "0"(*eax),  /* Input */
+		  "1"(*ebx),
+		  "2"(*ecx),
+		  "3"(*edx)
+		/* No clobbered registers */
+	);
+}
+#endif
+
+static void NOINLINE print_intel_cstates(void)
+{
+#ifdef __i386__
+	int bios_table[8] = { 0 };
+	int nbios = 0;
+	DIR *cpudir;
+	struct dirent *d;
+	int i;
+	unsigned eax, ebx, ecx, edx;
+
+	cpudir = opendir("/sys/devices/system/cpu");
+	if (!cpudir)
+		return;
+
+	/* Loop over cpuN entries */
+	while ((d = readdir(cpudir)) != NULL) {
+		DIR *dir;
+		int len;
+		char fname[sizeof("/sys/devices/system/cpu//cpuidle//desc") + 2*BIG_SYSNAME_LEN];
+
+		len = strlen(d->d_name);
+		if (len < 3 || len > BIG_SYSNAME_LEN)
+			continue;
+
+		if (!isdigit(d->d_name[3]))
+			continue;
+
+		len = sprintf(fname, "/sys/devices/system/cpu/%s/cpuidle", d->d_name);
+		dir = opendir(fname);
+		if (!dir)
+			continue;
+
+		/*
+		 * Every C-state has its own stateN directory, that
+		 * contains a `time' and a `usage' file.
+		 */
+		while ((d = readdir(dir)) != NULL) {
+			FILE *fp;
+			char buf[64];
+			int n;
+
+			n = strlen(d->d_name);
+			if (n < 3 || n > BIG_SYSNAME_LEN)
+				continue;
+
+			sprintf(fname + len, "/%s/desc", d->d_name);
+			fp = fopen_for_read(fname);
+			if (fp) {
+				char *p = fgets(buf, sizeof(buf), fp);
+				fclose(fp);
+				if (!p)
+					break;
+				p = strstr(p, "MWAIT ");
+				if (p) {
+					int pos;
+					p += sizeof("MWAIT ") - 1;
+					pos = (bb_strtoull(p, NULL, 16) >> 4) + 1;
+					if (pos >= ARRAY_SIZE(bios_table))
+						continue;
+					bios_table[pos]++;
+					nbios++;
+				}
+			}
+		}
+		closedir(dir);
+	}
+	closedir(cpudir);
+
+	if (!nbios)
+		return;
+
+	eax = 5;
+	ebx = ecx = edx = 0;
+	cpuid(&eax, &ebx, &ecx, &edx);
+	if (!edx || !(ecx & 1))
+		return;
+
+	printf("Your CPU supports the following C-states: ");
+	i = 0;
+	while (edx) {
+		if (edx & 7)
+			printf("C%u ", i);
+		edx >>= 4;
+		i++;
+	}
+	bb_putchar('\n');
+
+	/* Print BIOS C-States */
+	printf("Your BIOS reports the following C-states: ");
+	for (i = 0; i < 8; i++)
+		if (bios_table[i])
+			printf("C%u ", i);
+
+	bb_putchar('\n');
+#endif
+}
+
+static void print_header(void)
+{
+	printf(
+		/* Clear the screen */
+		"\033[H\033[J"
+		/* Print the header */
+		"\033[7m%.*s\033[0m", 79, "PowerTOP (C) 2007 Intel Corporation\n"
+	);
+}
+
+static void show_cstates(char cstate_lines[][64])
+{
+	int i;
+
+	for (i = 0; i < 10; i++)
+		if ((cstate_lines[i][0]))
+			printf("%s", cstate_lines[i]);
+}
+
+static void show_timerstats(int nostats)
+{
+	unsigned lines;
+
+	/* Get terminal height */
+	get_terminal_width_height(STDOUT_FILENO, NULL, &lines);
+
+	/* We don't have whole terminal just for timerstats */
+	lines -= 12;
+
+	if (!nostats) {
+		int i, n = 0;
+
+		puts("\nTop causes for wakeups:");
+		for (i = 0; i < G.headline; i++) {
+			if ((G.lines[i].count > 0 || G.lines[i].disk_count > 0)
+			 && n++ < lines
+			) {
+				char c = ' ';
+				if (G.lines[i].disk_count)
+					c = 'D';
+				printf(" %5.1f%% (%5.1f)%c  %s\n",
+						G.lines[i].count * 100.0 / G.nlines,
+						G.lines[i].count * 1.0 / DEFAULT_SLEEP, c,
+						G.lines[i].string);
+			}
+		}
+	} else {
+		bb_putchar('\n');
+		bb_error_msg("no stats available; run as root or"
+				" enable the cpufreq_stats module");
+	}
+}
+
+//usage:#define powertop_trivial_usage
+//usage:       ""
+//usage:#define powertop_full_usage "\n\n"
+//usage:       "Analyze power consumption on Intel-based laptops\n"
+
+int powertop_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
+int powertop_main(int UNUSED_PARAM argc, char UNUSED_PARAM **argv)
+{
+	ullong cur_usage[8];
+	ullong cur_duration[8];
+	char cstate_lines[12][64];
+	char buf[128];
+#if ENABLE_FEATURE_USE_TERMIOS
+	struct termios new_settings;
+	struct pollfd pfd[1];
+
+	pfd[0].fd = 0;
+	pfd[0].events = POLLIN;
+#endif
+
+	INIT_G();
+
+#if ENABLE_FEATURE_POWERTOP_PROCIRQ
+	G.percpu_hpet_start = INT_MAX;
+	G.percpu_hpet_end = INT_MIN;
+#endif
+
+	/* Print warning when we don't have superuser privileges */
+	if (geteuid() != 0)
+		bb_error_msg("run as root to collect enough information");
+
+#if ENABLE_FEATURE_USE_TERMIOS
+	/* So we don't forget to reset term settings */
+	atexit(reset_term);
+#endif
+
+	/* Get number of CPUs */
+	G.total_cpus = get_cpu_count();
+
+	printf("Collecting data for "DEFAULT_SLEEP_STR" seconds\n");
+
+#if ENABLE_FEATURE_USE_TERMIOS
+	tcgetattr(0, (void *)&G.init_settings);
+	memcpy(&new_settings, &G.init_settings, sizeof(new_settings));
+
+	/* Turn on unbuffered input, turn off echoing */
+	new_settings.c_lflag &= ~(ISIG | ICANON | ECHO | ECHONL);
+
+	bb_signals(BB_FATAL_SIGS, sig_handler);
+	tcsetattr_stdin_TCSANOW(&new_settings);
+#endif
+
+#if ENABLE_FEATURE_POWERTOP_PROCIRQ
+	/* Collect initial data */
+	do_proc_irq();
+	do_proc_irq();
+#endif
+
+	/* Read initial usage and duration */
+	read_data(&G.start_usage[0], &G.start_duration[0]);
+
+	/* Copy them to "last" */
+	memcpy(G.last_usage, G.start_usage, sizeof(G.last_usage));
+	memcpy(G.last_duration, G.start_duration, sizeof(G.last_duration));
+
+	/* Display C-states */
+	print_intel_cstates();
+
+	if (stop_timer())
+		G.nostats = 1;
+
+	/* The main loop */
+	for (;;) {
+		double maxsleep = 0.0;
+		ullong totalticks, totalevents;
+		int i;
+		FILE *fp;
+		double newticks;
+
+		if (start_timer())
+			G.nostats = 1;
+
+#if !ENABLE_FEATURE_USE_TERMIOS
+		sleep(DEFAULT_SLEEP);
+#else
+		if (safe_poll(pfd, 1, DEFAULT_SLEEP * 1000) > 0) {
+			unsigned char c;
+			if (safe_read(STDIN_FILENO, &c, 1) != 1)
+				break; /* EOF/error */
+			if (c == G.init_settings.c_cc[VINTR])
+				break; /* ^C */
+			if ((c | 0x20) == 'q')
+				break;
+		}
+#endif
+
+		if (stop_timer())
+			G.nostats = 1;
+
+		clear_lines();
+#if ENABLE_FEATURE_POWERTOP_PROCIRQ
+		do_proc_irq();
+#endif
+
+		/* Clear the stats */
+		memset(cur_duration, 0, sizeof(cur_duration));
+		memset(cur_usage, 0, sizeof(cur_usage));
+
+		/* Read them */
+		read_data(&cur_usage[0], &cur_duration[0]);
+
+		totalticks = totalevents = 0;
+
+		/* Count totalticks and totalevents */
+		for (i = 0; i < 8; i++) {
+			if (cur_usage[i]) {
+				totalticks += cur_duration[i] - G.last_duration[i];
+				totalevents += cur_usage[i] - G.last_usage[i];
+			}
+		}
+
+		/* Show title bar */
+		print_header();
+
+		/* Clear C-state lines */
+		memset(&cstate_lines, 0, sizeof(cstate_lines));
+
+		if (totalevents == 0 && G.maxcstate <= 1) {
+			/* This should not happen */
+			sprintf(cstate_lines[5], "< Detailed C-state information is not "
+				"available.>\n");
+		} else {
+			double slept;
+			double percentage;
+
+			newticks = G.total_cpus * DEFAULT_SLEEP * FREQ_ACPI_1000 - totalticks;
+
+			/* Handle rounding errors: do not display negative values */
+			if (newticks < 0)
+				newticks = 0;
+
+			sprintf(cstate_lines[0], "Cn\t          Avg residency\n");
+			percentage = newticks * 100.0 / (G.total_cpus * DEFAULT_SLEEP * FREQ_ACPI_1000);
+			sprintf(cstate_lines[1], "C0 (cpu running)        (%4.1f%%)\n",
+				percentage);
+
+			/* Compute values for individual C-states */
+			for (i = 0; i < 8; i++) {
+				if (cur_usage[i]) {
+					slept = (cur_duration[i] - G.last_duration[i])
+						/ (cur_usage[i] - G.last_usage[i] + 0.1) / FREQ_ACPI;
+					percentage = (cur_duration[i] - G.last_duration[i]) * 100
+						/ (G.total_cpus * DEFAULT_SLEEP * FREQ_ACPI_1000);
+
+					if (!G.cstate_names[i][0])
+						sprintf(G.cstate_names[i], "C%u", i + 1);
+					sprintf(cstate_lines[i + 2], "%s\t%5.1fms (%4.1f%%)\n",
+						G.cstate_names[i], slept, percentage);
+					if (maxsleep < slept)
+						maxsleep = slept;
+				}
+			}
+		}
+
+		/* Display C-states */
+		show_cstates(cstate_lines);
+
+		/* Do timer_stats info */
+		buf[0] = '\0';
+		totalticks = 0;
+
+		fp = NULL;
+		if (!G.nostats)
+			fp = fopen_for_read("/proc/timer_stats");
+		if (fp) {
+			while (fgets(buf, sizeof(buf), fp)) {
+				const char *count, *process, *func;
+				char line[512];
+				int cnt = 0;
+				bool defferable = false;
+				char *p;
+				int j = 0;
+
+/* Find char ' ', then eat remaining spaces */
+#define ADVANCE(p) do {           \
+	(p) = strchr((p), ' ');   \
+	if (!(p))                 \
+		continue;         \
+	*(p) = '\0';              \
+	(p)++;                    \
+	(p) = skip_whitespace(p); \
+} while (0)
+
+				if (strstr(buf, "total events"))
+					break;
+
+				while (isspace(buf[j]))
+					j++;
+
+				count = &buf[j];
+				p = (char *)count;
+
+				/* Skip PID */
+				p = strchr(p, ',');
+				if (!p)
+					continue;
+				*p = '\0';
+				p++;
+
+				p = skip_whitespace(p);
+
+				/* Get process */
+				ADVANCE(p);
+				process = p;
+
+				/* Get function */
+				ADVANCE(p);
+				func = p;
+
+				if (strcmp(process, "swapper") == 0
+				 && strcmp(func, "hrtimer_start_range_ns (tick_sched_timer)\n") == 0
+				) {
+					process = "[kernel scheduler]";
+					func = "Load balancing tick";
+				}
+
+				if (strcmp(process, "insmod") == 0)
+					process = "[kernel module]";
+				if (strcmp(process, "modprobe") == 0)
+					process = "[kernel module]";
+				if (strcmp(process, "swapper") == 0)
+					process = "[kernel core]";
+
+				p = strchr(p, '\n');
+
+				if (strncmp(func, "tick_nohz_", 10) == 0)
+					continue;
+				if (strncmp(func, "tick_setup_sched_timer", 20) == 0)
+					continue;
+				if (strcmp(process, "powertop") == 0)
+					continue;
+
+				if (p)
+					*p = '\0';
+
+				cnt = bb_strtoull(count, &p, 10);
+				while (*p != 0) {
+					if (*p++ == 'D')
+						defferable = true;
+				}
+				if (defferable)
+					continue;
+
+				if (strchr(process, '['))
+					sprintf(line, "%s %s", process, func);
+				else
+					sprintf(line, "%s", process);
+				push_line(line, cnt);
+			}
+			fclose(fp);
+		}
+
+#if ENABLE_FEATURE_POWERTOP_PROCIRQ
+		if (strstr(buf, "total events")) {
+			int n = bb_strtoull(buf, NULL, 10) / G.total_cpus;
+
+			if (totalevents == 0) {
+				/* No C-state info available, use timerstats */
+				totalevents = n * G.total_cpus + G.total_interrupt;
+				if (n < 0)
+					totalevents += G.interrupt_0 - n;
+			}
+			if (n > 0 && n < G.interrupt_0)
+				push_line("[extra timer interrupt]", G.interrupt_0 - n);
+		}
+#endif
+		if (totalevents)
+			printf("\n\033[1mWakeups-from-idle per second : %4.1f\tinterval:"
+				"%ds\n\033[0m",
+				(double)totalevents / DEFAULT_SLEEP / G.total_cpus, DEFAULT_SLEEP);
+
+		count_lines();
+		do_sort();
+
+		show_timerstats(G.nostats);
+
+		fflush(stdout);
+
+		/* Clear the stats */
+		memset(cur_duration, 0, sizeof(cur_duration));
+		memset(cur_usage, 0, sizeof(cur_usage));
+
+		/* Get new values */
+		read_data(&cur_usage[0], &cur_duration[0]);
+
+		/* Save them */
+		memcpy(G.last_usage, cur_usage, sizeof(G.last_usage));
+		memcpy(G.last_duration, cur_duration, sizeof(G.last_duration));
+	} /* for (;;) */
+
+	bb_putchar('\n');
+
+	return EXIT_SUCCESS;
+}
diff --git a/util-linux/Config.src b/util-linux/Config.src
index 19b309e57..c71b4de38 100644
--- a/util-linux/Config.src
+++ b/util-linux/Config.src
@@ -486,17 +486,6 @@ config MORE
 	  you will probably find this utility very helpful. If you don't have
 	  any need to reading text files, you can leave this disabled.
 
-config FEATURE_USE_TERMIOS
-	bool "Use termios to manipulate the screen"
-	default y
-	depends on MORE || TOP
-	help
-	  This option allows utilities such as 'more' and 'top' to determine
-	  the size of the screen. If you leave this disabled, your utilities
-	  that display things on the screen will be especially primitive and
-	  will be unable to determine the current screen size, and will be
-	  unable to move the cursor.
-
 config MOUNT
 	bool "mount"
 	default y
-- 
cgit v1.2.3-55-g6feb