From mboxrd@z Thu Jan 1 00:00:00 1970 From: Mark Langsdorf Subject: [PATCH 3/6 v4] cpufreq: tolerate inexact values when collecting stats Date: Wed, 7 Nov 2012 12:32:43 -0600 Message-ID: <1352313166-28980-4-git-send-email-mark.langsdorf@calxeda.com> References: <1351631056-25938-1-git-send-email-mark.langsdorf@calxeda.com> <1352313166-28980-1-git-send-email-mark.langsdorf@calxeda.com> Return-path: In-Reply-To: <1352313166-28980-1-git-send-email-mark.langsdorf@calxeda.com> Sender: cpufreq-owner@vger.kernel.org To: linux-kernel@vger.kernel.org, cpufreq@vger.kernel.org, linux-pm@vger.kernel.org Cc: Mark Langsdorf , MyungJoo Ham List-Id: linux-pm@vger.kernel.org When collecting stats, if a frequency doesn't match the table, go through the table again with both the search frequency and table values shifted left by 10 bits. Signed-off-by: Mark Langsdorf Cc: MyungJoo Ham --- Changes from v3, v2 None Changes from v1: Implemented a simple round-up algorithm instead of the over/under method that could cause errors on Intel processors with boost mode. drivers/cpufreq/cpufreq_stats.c | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/drivers/cpufreq/cpufreq_stats.c b/drivers/cpufreq/cpufreq_stats.c index 3998316..30aee36 100644 --- a/drivers/cpufreq/cpufreq_stats.c +++ b/drivers/cpufreq/cpufreq_stats.c @@ -161,6 +161,9 @@ static int freq_table_get_index(struct cpufreq_stats *stat, unsigned int freq) for (index = 0; index < stat->max_state; index++) if (stat->freq_table[index] == freq) return index; + for (index = 0; index < stat->max_state; index++) + if ((stat->freq_table[index] >> 10) == (freq >> 10)) + return index; return -1; } @@ -251,6 +254,8 @@ static int cpufreq_stats_create_table(struct cpufreq_policy *policy, spin_lock(&cpufreq_stats_lock); stat->last_time = get_jiffies_64(); stat->last_index = freq_table_get_index(stat, policy->cur); + if (stat->last_index > stat->max_state) + stat->last_index = stat->max_state - 1; spin_unlock(&cpufreq_stats_lock); cpufreq_cpu_put(data); return 0; -- 1.7.11.7