static ssize_t mcdi_profile_read(struct file *filp,
		char __user *userbuf, size_t count, loff_t *f_pos)
{
	int i, len = 0;
	char *p = dbg_buf;
	unsigned int ratio_raw = 0;
	unsigned int ratio_int = 0;
	unsigned int ratio_fraction = 0;
	unsigned int ratio_dur = 0;
	struct {
		unsigned int sec[DISTRIBUTE_NUM];
		unsigned int total;
	} cnt[2] = {0};

#ifdef MCDI_PWR_SEQ_PROF_BREAKDOWN
	struct mcdi_prof_breakdown *p_prof =
			(struct mcdi_prof_breakdown *)SYSRAM_PROF_PWR_SEQ_BK;
	unsigned int j;
#endif

	mcdi_log(
		"mcdi cpu off    : max_id = 0x%x, avg = %4dus, max = %5dus, cnt = %d\n",
		mcdi_read(CPU_OFF_LATENCY_REG(ID_OFS)),
		mcdi_read(CPU_OFF_LATENCY_REG(AVG_OFS)),
		mcdi_read(CPU_OFF_LATENCY_REG(MAX_OFS)),
		mcdi_read(CPU_OFF_LATENCY_REG(CNT_OFS)));
	mcdi_log(
		"mcdi cpu on     : max_id = 0x%x, avg = %4dus, max = %5dus, cnt = %d\n",
		mcdi_read(CPU_ON_LATENCY_REG(ID_OFS)),
		mcdi_read(CPU_ON_LATENCY_REG(AVG_OFS)),
		mcdi_read(CPU_ON_LATENCY_REG(MAX_OFS)),
		mcdi_read(CPU_ON_LATENCY_REG(CNT_OFS)));
	mcdi_log(
		"mcdi cluster off: max_id = 0x%x, avg = %4dus, max = %5dus, cnt = %d\n",
		mcdi_read(Cluster_OFF_LATENCY_REG(ID_OFS)),
		mcdi_read(Cluster_OFF_LATENCY_REG(AVG_OFS)),
		mcdi_read(Cluster_OFF_LATENCY_REG(MAX_OFS)),
		mcdi_read(Cluster_OFF_LATENCY_REG(CNT_OFS)));
	mcdi_log(
		"mcdi cluster on : max_id = 0x%x, avg = %4dus, max = %5dus, cnt = %d\n",
		mcdi_read(Cluster_ON_LATENCY_REG(ID_OFS)),
		mcdi_read(Cluster_ON_LATENCY_REG(AVG_OFS)),
		mcdi_read(Cluster_ON_LATENCY_REG(MAX_OFS)),
		mcdi_read(Cluster_ON_LATENCY_REG(CNT_OFS)));

	mcdi_log("\n");

	for (i = 0; i < DISTRIBUTE_NUM; i++) {
		cnt[0].sec[i] = mcdi_read(PROF_OFF_CNT_REG(i));
		cnt[1].sec[i] = mcdi_read(PROF_ON_CNT_REG(i));
		cnt[0].total += cnt[0].sec[i];
		cnt[1].total += cnt[1].sec[i];
	}

	cnt[0].total = cnt[0].total ? : 1;
	cnt[1].total = cnt[1].total ? : 1;

	for (i = 0; i < DISTRIBUTE_NUM; i++)
		mcdi_log("pwr off latency (section%d) : %2d%% (%d)\n",
				i,
				(100 * cnt[0].sec[i]) / cnt[0].total,
				cnt[0].sec[i]);

	for (i = 0; i < DISTRIBUTE_NUM; i++)
		mcdi_log("pwr on latency  (section%d) : %2d%% (%d)\n",
				i,
				(100 * cnt[1].sec[i]) / cnt[1].total,
				cnt[1].sec[i]);

	ratio_dur = mcdi_read(SYSRAM_PROF_RARIO_DUR);

	if (ratio_dur == 0)
		ratio_dur = ~0;

	mcdi_log("\nOFF %% (cpu):\n");

	for (i = 0; i < NF_CPU; i++) {
		ratio_raw = 100 * mcdi_read(PROF_CPU_RATIO_REG(i));
		ratio_int = ratio_raw / ratio_dur;
		ratio_fraction = (1000 * (ratio_raw % ratio_dur)) / ratio_dur;

		mcdi_log("%d: %3u.%03u%% (%u)\n",
			i, ratio_int, ratio_fraction, ratio_raw/100);
	}

	mcdi_log("\nOFF %% (cluster):\n");

	for (i = 0; i < NF_CLUSTER; i++) {
		ratio_raw      = 100 * mcdi_read(PROF_CLUSTER_RATIO_REG(i));
		ratio_int      = ratio_raw / ratio_dur;
		ratio_fraction = (1000 * (ratio_raw % ratio_dur)) / ratio_dur;

		mcdi_log("%d: %3u.%03u%% (%u)\n",
			i, ratio_int, ratio_fraction, ratio_raw/100);
	}

	mcdi_log("\nprof cpu = %d, count = %d, state = %d, profile_state=%d\n",
				get_mcdi_profile_cpu(),
				get_mcdi_profile_cnt(),
				mcdi_mbox_read(MCDI_MBOX_PROF_CMD),
				profile_state);
	for (i = 0; i < (NF_MCDI_PROFILE - 1); i++)
		mcdi_log("%s: %u\n", prof_item[i], get_mcdi_profile_sum_us(i));

#ifdef MCDI_PWR_SEQ_PROF_BREAKDOWN
	mcdi_log("\npwr seq (us)\n");
	for (j = 0; j < NF_CLUSTER; j++) {
		mcdi_log("\n---cluster %d---\n", j);
		for (i = 0; i < (MCDI_PROF_BK_NUM); i++) {
			mcdi_log("%s on, cnt : %u, %u\n",
				prof_pwr_seq_item[i],
				p_prof->onoff[0].item[i][j],
				p_prof->onoff[0].count[i][j]);
		}
		mcdi_log("\n");
		for (i = 0; i < (MCDI_PROF_BK_NUM); i++) {
			mcdi_log("%s off, cnt : %u, %u\n",
				prof_pwr_seq_item[i],
				p_prof->onoff[1].item[i][j],
				p_prof->onoff[1].count[i][j]);
		}
	}
#endif
	len = p - dbg_buf;

	return simple_read_from_buffer(userbuf, count, f_pos, dbg_buf, len);
}