Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752264AbaFJPrU (ORCPT ); Tue, 10 Jun 2014 11:47:20 -0400 Received: from mail-pa0-f43.google.com ([209.85.220.43]:38891 "EHLO mail-pa0-f43.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750801AbaFJPrS (ORCPT ); Tue, 10 Jun 2014 11:47:18 -0400 Message-ID: <53972883.7070707@gmail.com> Date: Tue, 10 Jun 2014 08:47:15 -0700 From: Dirk Brandewie User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:24.0) Gecko/20100101 Thunderbird/24.5.0 MIME-Version: 1.0 To: Stratos Karafotis , "Rafael J. Wysocki" , Viresh Kumar , Dirk Brandewie CC: dirk.brandewie@gmail.com, "linux-pm@vger.kernel.org" , LKML Subject: Re: [PATCH 3/7] cpufreq: intel_pstate: Add debugfs file stats References: <53962072.6010702@semaphore.gr> In-Reply-To: <53962072.6010702@semaphore.gr> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 06/09/2014 02:00 PM, Stratos Karafotis wrote: > Add stats file in debugfs under driver's parent directory > (pstate_snb) which counts the time in nsecs per requested > P state and the number of times the specific state > was requested. > > The file presents the statistics per logical CPU in the > following format. The time is displayed in msecs: > NAK This adds significantly to the memory footprint to gather information that is available by post processing the perf tracepoint information. The increase isn't horrible on single socket desktop processor machines but gets big with server class machines. One vendor I have talked to considers a machine with 1024 cpus to be a SMALL machine. > CPU0 > P-state Time Count > 16 4882777 23632 > 17 21210 174 > 18 549781 3300 > 19 51171 461 > 20 35487 394 > 21 18173 219 > 22 13752 258 > 23 6048 172 > 24 7754 177 > 25 4587 151 > 26 5465 162 > 27 1432 47 > 28 863 54 > 29 1448 50 > 30 1030 47 > 31 1472 62 > 32 2221 68 > 33 1869 60 > 34 2140 70 > 39 85446 3803 > > ... > > The file can be used for debugging but also for monitoring > various system workloads. > > Also, make the debugfs_parent local as we never remove > the driver's debugfs files. > > Signed-off-by: Stratos Karafotis > --- > drivers/cpufreq/intel_pstate.c | 80 +++++++++++++++++++++++++++++++++++++++++- > 1 file changed, 79 insertions(+), 1 deletion(-) > > diff --git a/drivers/cpufreq/intel_pstate.c b/drivers/cpufreq/intel_pstate.c > index 31e2ae5..3a49269 100644 > --- a/drivers/cpufreq/intel_pstate.c > +++ b/drivers/cpufreq/intel_pstate.c > @@ -86,6 +86,12 @@ struct _pid { > int32_t last_err; > }; > > +struct pstate_stat { > + int pstate; > + u64 time; > + u64 count; > +}; > + > struct cpudata { > int cpu; > > @@ -99,6 +105,7 @@ struct cpudata { > u64 prev_aperf; > u64 prev_mperf; > struct sample sample; > + struct pstate_stat *stats; > }; > > static struct cpudata **all_cpu_data; > @@ -256,9 +263,59 @@ static struct pid_param pid_files[] = { > {NULL, NULL} > }; > > -static struct dentry *debugfs_parent; > +static inline unsigned int stats_state_index(struct cpudata *cpu, int pstate) > +{ > + if (pstate <= cpu->pstate.max_pstate) > + return pstate - cpu->pstate.min_pstate; > + else > + return cpu->pstate.max_pstate - cpu->pstate.min_pstate + 1; > +} > + > +static int stats_debug_show(struct seq_file *m, void *unused) > +{ > + struct cpudata *cpu; > + int i, j, cnt; > + > + get_online_cpus(); > + for_each_online_cpu(i) { > + if (all_cpu_data[i]) > + cpu = all_cpu_data[i]; > + else > + continue; > + > + seq_printf(m, "CPU%u\n", i); > + seq_puts(m, "P-state Time Count\n"); > + > + cnt = cpu->pstate.max_pstate - cpu->pstate.min_pstate + 2; > + for (j = 0; j < cnt; j++) > + seq_printf(m, "%7u %11llu %9llu\n", > + cpu->stats[j].pstate, > + cpu->stats[j].time / USEC_PER_MSEC, > + cpu->stats[j].count); > + > + seq_puts(m, "\n"); > + } > + put_online_cpus(); > + > + return 0; > +} > + > +static int stats_debug_open(struct inode *inode, struct file *file) > +{ > + return single_open(file, stats_debug_show, inode->i_private); > +} > + > +static const struct file_operations fops_stats_pstate = { > + .open = stats_debug_open, > + .read = seq_read, > + .llseek = seq_lseek, > + .release = single_release, > + .owner = THIS_MODULE, > +}; > + > static void intel_pstate_debug_expose_params(void) > { > + struct dentry *debugfs_parent; > int i = 0; > > debugfs_parent = debugfs_create_dir("pstate_snb", NULL); > @@ -270,6 +327,8 @@ static void intel_pstate_debug_expose_params(void) > &fops_pid_param); > i++; > } > + debugfs_create_file("stats", S_IRUSR | S_IRGRP, debugfs_parent, NULL, > + &fops_stats_pstate); > } > > /************************** debugfs end ************************/ > @@ -610,6 +669,7 @@ static inline void intel_pstate_calc_scaled_busy(struct cpudata *cpu) > int32_t core_busy, max_pstate, current_pstate, sample_ratio; > u32 duration_us; > u32 sample_time; > + unsigned int i; > > core_busy = cpu->sample.core_pct_busy; > max_pstate = int_tofp(cpu->pstate.max_pstate); > @@ -626,6 +686,10 @@ static inline void intel_pstate_calc_scaled_busy(struct cpudata *cpu) > } > > cpu->sample.busy_scaled = core_busy; > + > + i = stats_state_index(cpu, cpu->pstate.current_pstate); > + cpu->stats[i].time += duration_us; > + cpu->stats[i].count++; > } > > static inline void intel_pstate_adjust_busy_pstate(struct cpudata *cpu) > @@ -692,6 +756,7 @@ MODULE_DEVICE_TABLE(x86cpu, intel_pstate_cpu_ids); > static int intel_pstate_init_cpu(unsigned int cpunum) > { > struct cpudata *cpu; > + unsigned int i, cnt; > > all_cpu_data[cpunum] = kzalloc(sizeof(struct cpudata), GFP_KERNEL); > if (!all_cpu_data[cpunum]) > @@ -701,6 +766,17 @@ static int intel_pstate_init_cpu(unsigned int cpunum) > > intel_pstate_get_cpu_pstates(cpu); > > + /* cnt equals to number of p-states + 1 (for turbo p-state) */ > + cnt = cpu->pstate.max_pstate - cpu->pstate.min_pstate + 2; > + cpu->stats = kzalloc(sizeof(*cpu->stats) * cnt, GFP_KERNEL); > + if (!cpu->stats) { > + kfree(all_cpu_data[cpunum]); > + return -ENOMEM; > + } > + for (i = 0; i < cnt - 1; i++) > + cpu->stats[i].pstate = cpu->pstate.min_pstate + i; > + cpu->stats[cnt - 1].pstate = cpu->pstate.turbo_pstate; > + > cpu->cpu = cpunum; > > init_timer_deferrable(&cpu->timer); > @@ -779,6 +855,7 @@ static void intel_pstate_stop_cpu(struct cpufreq_policy *policy) > > del_timer_sync(&all_cpu_data[cpu_num]->timer); > intel_pstate_set_pstate(cpu, cpu->pstate.min_pstate); > + kfree(all_cpu_data[cpu_num]->stats); > kfree(all_cpu_data[cpu_num]); > all_cpu_data[cpu_num] = NULL; > } > @@ -980,6 +1057,7 @@ out: > for_each_online_cpu(cpu) { > if (all_cpu_data[cpu]) { > del_timer_sync(&all_cpu_data[cpu]->timer); > + kfree(all_cpu_data[cpu]->stats); > kfree(all_cpu_data[cpu]); > } > } > -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/