All of lore.kernel.org
 help / color / mirror / Atom feed
From: Huaitong Han <huaitong.han@intel.com>
To: jbeulich@suse.com
Cc: Huaitong Han <huaitong.han@intel.com>, xen-devel@lists.xen.org
Subject: [V5] x86/cpuidle: get accurate C0 value with xenpm tool
Date: Thu, 14 May 2015 13:23:19 +0800	[thread overview]
Message-ID: <1431580999-7886-1-git-send-email-huaitong.han@intel.com> (raw)

When checking the ACPI funciton of C-status, after 100 seconds sleep,
the sampling value of C0 status from the xenpm tool decreases.
Because C0=NOW()-C1-C2-C3-C4, when NOW() value is during idle time,
NOW() value is bigger than last C-status update time, and C0 value
is also bigger than ture value. if margin of the second error cannot
make up for margin of the first error, the value of C0 would decrease.

Signed-off-by: Huaitong Han <huaitong.han@intel.com>

---
ChangeLog:
V5:
Ticks clock souce may be acpi_pm, so use common funciton "ticks_elapsed".
Taking every "tick_to_ns" outside spin_lock.
Spliting the "for" loop.

V4:
delete pointless initializers and hard tabs.

V3:
1.Don't use tick_to_ns inside lock in print_acpi_power.
2.Use 08 padding in printk.
3.Merge two "for" circulation into one for coding style.

V2:
C0 = last_cx_update_time-C1-C2-C3-C4, but last_cx_update_time is not now,
so the C0 value is stale, NOW-last_update_time should be calculated.
C[current_cx_stat]+=NOW-last_update_time, so the CX value is fresh.

V1:
Initial patch
---

diff --git a/xen/arch/x86/acpi/cpu_idle.c b/xen/arch/x86/acpi/cpu_idle.c
index e639c99..07ee3a2 100644
--- a/xen/arch/x86/acpi/cpu_idle.c
+++ b/xen/arch/x86/acpi/cpu_idle.c
@@ -254,9 +254,11 @@ static char* acpi_cstate_method_name[] =
 
 static void print_acpi_power(uint32_t cpu, struct acpi_processor_power *power)
 {
-    uint32_t i, idle_usage = 0;
-    uint64_t res, idle_res = 0;
-    u32 usage;
+    uint64_t idle_res = 0, idle_usage = 0;
+    uint64_t last_state_update_tick, current_tick, current_stime;
+    uint64_t usage[ACPI_PROCESSOR_MAX_POWER] = { 0 };
+    uint64_t res_tick[ACPI_PROCESSOR_MAX_POWER] = { 0 };
+    unsigned int i;
     u8 last_state_idx;
 
     printk("==cpu%d==\n", cpu);
@@ -264,28 +266,37 @@ static void print_acpi_power(uint32_t cpu, struct acpi_processor_power *power)
     printk("active state:\t\tC%d\n", last_state_idx);
     printk("max_cstate:\t\tC%d\n", max_cstate);
     printk("states:\n");
-    
+
+    spin_lock_irq(&power->stat_lock);
+    current_tick = cpuidle_get_tick();
+    current_stime = NOW();
     for ( i = 1; i < power->count; i++ )
     {
-        spin_lock_irq(&power->stat_lock);	
-        res = tick_to_ns(power->states[i].time);
-        usage = power->states[i].usage;
-        spin_unlock_irq(&power->stat_lock);
+        res_tick[i] = power->states[i].time;
+        usage[i] = power->states[i].usage;
+    }
+    last_state_update_tick = power->last_state_update_tick;
+    spin_unlock_irq(&power->stat_lock);
+
+    res_tick[last_state_idx] += ticks_elapsed(last_state_update_tick, current_tick);
+    usage[last_state_idx]++;
 
-        idle_usage += usage;
-        idle_res += res;
+    for ( i = 1; i < power->count; i++ )
+    {
+        idle_usage += usage[i];
+        idle_res += tick_to_ns(res_tick[i]);
 
         printk((last_state_idx == i) ? "   *" : "    ");
         printk("C%d:\t", i);
         printk("type[C%d] ", power->states[i].type);
         printk("latency[%03d] ", power->states[i].latency);
-        printk("usage[%08d] ", usage);
+        printk("usage[%08"PRIu64"] ", usage[i]);
         printk("method[%5s] ", acpi_cstate_method_name[power->states[i].entry_method]);
-        printk("duration[%"PRId64"]\n", res);
+        printk("duration[%"PRIu64"]\n", tick_to_ns(res_tick[i]));
     }
     printk((last_state_idx == 0) ? "   *" : "    ");
-    printk("C0:\tusage[%08d] duration[%"PRId64"]\n",
-           idle_usage, NOW() - idle_res);
+    printk("C0:\tusage[%08"PRIu64"] duration[%"PRIu64"]\n",
+           usage[0] + idle_usage, current_stime - idle_res);
 
     print_hw_residencies(cpu);
 }
@@ -329,7 +340,7 @@ static uint64_t acpi_pm_ticks_elapsed(uint64_t t1, uint64_t t2)
 }
 
 uint64_t (*__read_mostly cpuidle_get_tick)(void) = get_acpi_pm_tick;
-static uint64_t (*__read_mostly ticks_elapsed)(uint64_t, uint64_t)
+uint64_t (*__read_mostly ticks_elapsed)(uint64_t, uint64_t)
     = acpi_pm_ticks_elapsed;
 
 /*
@@ -486,6 +497,17 @@ bool_t errata_c6_eoi_workaround(void)
     return (fix_needed && cpu_has_pending_apic_eoi());
 }
 
+void update_last_cx_stat(struct acpi_processor_power *power,
+                         struct acpi_processor_cx *cx, uint64_t ticks)
+{
+    ASSERT(!local_irq_is_enabled());
+
+    spin_lock(&power->stat_lock);
+    power->last_state = cx;
+    power->last_state_update_tick = ticks;
+    spin_unlock(&power->stat_lock);
+}
+
 void update_idle_stats(struct acpi_processor_power *power,
                        struct acpi_processor_cx *cx,
                        uint64_t before, uint64_t after)
@@ -501,6 +523,8 @@ void update_idle_stats(struct acpi_processor_power *power,
         power->last_residency = tick_to_ns(sleep_ticks) / 1000UL;
         cx->time += sleep_ticks;
     }
+    power->last_state = &power->states[0];
+    power->last_state_update_tick = after;
 
     spin_unlock(&power->stat_lock);
 }
@@ -557,7 +581,6 @@ static void acpi_processor_idle(void)
     if ( (cx->type == ACPI_STATE_C3) && errata_c6_eoi_workaround() )
         cx = power->safe_state;
 
-    power->last_state = cx;
 
     /*
      * Sleep:
@@ -574,6 +597,7 @@ static void acpi_processor_idle(void)
             t1 = cpuidle_get_tick();
             /* Trace cpu idle entry */
             TRACE_4D(TRC_PM_IDLE_ENTRY, cx->idx, t1, exp, pred);
+            update_last_cx_stat(power, cx, t1);
             /* Invoke C2 */
             acpi_idle_do_entry(cx);
             /* Get end time (ticks) */
@@ -602,7 +626,7 @@ static void acpi_processor_idle(void)
         t1 = cpuidle_get_tick();
         /* Trace cpu idle entry */
         TRACE_4D(TRC_PM_IDLE_ENTRY, cx->idx, t1, exp, pred);
-
+        update_last_cx_stat(power, cx, t1);
         /*
          * disable bus master
          * bm_check implies we need ARB_DIS
@@ -1172,7 +1196,10 @@ int pmstat_get_cx_stat(uint32_t cpuid, struct pm_cx_stat *stat)
 {
     struct acpi_processor_power *power = processor_powers[cpuid];
     uint64_t idle_usage = 0, idle_res = 0;
-    uint64_t usage[ACPI_PROCESSOR_MAX_POWER], res[ACPI_PROCESSOR_MAX_POWER];
+    uint64_t last_state_update_tick, current_stime, current_tick;
+    uint64_t usage[ACPI_PROCESSOR_MAX_POWER] = { 0 };
+    uint64_t res_ticks[ACPI_PROCESSOR_MAX_POWER] = { 0 };
+    uint64_t res[ACPI_PROCESSOR_MAX_POWER] = { 0 };
     unsigned int i, nr, nr_pc = 0, nr_cc = 0;
 
     if ( power == NULL )
@@ -1185,7 +1212,6 @@ int pmstat_get_cx_stat(uint32_t cpuid, struct pm_cx_stat *stat)
         return 0;
     }
 
-    stat->last = power->last_state ? power->last_state->idx : 0;
     stat->idle_time = get_cpu_idle_time(cpuid);
     nr = min(stat->nr, power->count);
 
@@ -1193,9 +1219,12 @@ int pmstat_get_cx_stat(uint32_t cpuid, struct pm_cx_stat *stat)
     if ( pm_idle_save == NULL )
     {
         stat->nr = 2;
+        stat->last = power->last_state ? power->last_state->idx : 0;
 
         usage[1] = idle_usage = 1;
         res[1] = idle_res = stat->idle_time;
+
+        current_stime = NOW();
     }
     else
     {
@@ -1203,13 +1232,25 @@ int pmstat_get_cx_stat(uint32_t cpuid, struct pm_cx_stat *stat)
 
         stat->nr = power->count;
 
+        spin_lock_irq(&power->stat_lock);
+        current_tick = cpuidle_get_tick();
+        current_stime = NOW();
         for ( i = 1; i < nr; i++ )
         {
-            spin_lock_irq(&power->stat_lock);
             usage[i] = power->states[i].usage;
-            res[i] = tick_to_ns(power->states[i].time);
-            spin_unlock_irq(&power->stat_lock);
+            res_ticks[i] = power->states[i].time;
+        }
+        last_state_update_tick = power->last_state_update_tick;
+        stat->last = power->last_state ? power->last_state->idx : 0;
+        spin_unlock_irq(&power->stat_lock);
+
+        usage[stat->last]++;
+        res_ticks[stat->last] +=
+            ticks_elapsed(last_state_update_tick, current_tick);
 
+        for ( i = 1; i < nr; i++ )
+        {
+            res[i] = tick_to_ns(res_ticks[i]);
             idle_usage += usage[i];
             idle_res += res[i];
         }
@@ -1242,8 +1283,8 @@ int pmstat_get_cx_stat(uint32_t cpuid, struct pm_cx_stat *stat)
 #undef PUT_xC
     }
 
-    usage[0] = idle_usage;
-    res[0] = NOW() - idle_res;
+    usage[0] += idle_usage;
+    res[0] = current_stime - idle_res;
 
     if ( copy_to_guest(stat->triggers, usage, nr) ||
          copy_to_guest(stat->residencies, res, nr) )
diff --git a/xen/arch/x86/cpu/mwait-idle.c b/xen/arch/x86/cpu/mwait-idle.c
index 6dd5822..596d32f 100644
--- a/xen/arch/x86/cpu/mwait-idle.c
+++ b/xen/arch/x86/cpu/mwait-idle.c
@@ -536,7 +536,6 @@ static void mwait_idle(void)
 		return;
 	}
 
-	power->last_state = cx;
 	eax = cx->address;
 	cstate = ((eax >> MWAIT_SUBSTATE_SIZE) & MWAIT_CSTATE_MASK) + 1;
 
@@ -554,6 +553,8 @@ static void mwait_idle(void)
 
 	before = cpuidle_get_tick();
 	TRACE_4D(TRC_PM_IDLE_ENTRY, cx->type, before, exp, pred);
+	/* Now in CX */
+	update_last_cx_stat(power, cx, before);
 
 	if (cpu_is_haltable(cpu))
 		mwait_idle_with_hints(eax, MWAIT_ECX_INTERRUPT_BREAK);
@@ -565,15 +566,13 @@ static void mwait_idle(void)
 	TRACE_6D(TRC_PM_IDLE_EXIT, cx->type, after,
 		irq_traced[0], irq_traced[1], irq_traced[2], irq_traced[3]);
 
+	/* Now back in C0 */
 	update_idle_stats(power, cx, before, after);
 	local_irq_enable();
 
 	if (!(lapic_timer_reliable_states & (1 << cstate)))
 		lapic_timer_on();
 
-	/* Now back in C0. */
-	power->last_state = &power->states[0];
-
 	sched_tick_resume();
 	cpufreq_dbs_timer_resume();
 
diff --git a/xen/include/asm-x86/cpuidle.h b/xen/include/asm-x86/cpuidle.h
index 4d70677..4807a85 100644
--- a/xen/include/asm-x86/cpuidle.h
+++ b/xen/include/asm-x86/cpuidle.h
@@ -15,6 +15,7 @@ extern void (*lapic_timer_off)(void);
 extern void (*lapic_timer_on)(void);
 
 extern uint64_t (*cpuidle_get_tick)(void);
+extern uint64_t (*ticks_elapsed)(uint64_t, uint64_t);
 
 int mwait_idle_init(struct notifier_block *);
 int cpuidle_init_cpu(unsigned int cpu);
@@ -23,6 +24,8 @@ void acpi_dead_idle(void);
 void trace_exit_reason(u32 *irq_traced);
 void update_idle_stats(struct acpi_processor_power *,
                        struct acpi_processor_cx *, uint64_t, uint64_t);
+void update_last_cx_stat(struct acpi_processor_power *,
+                         struct acpi_processor_cx *, uint64_t);
 
 /*
  * vcpu is urgent if vcpu is polling event channel
diff --git a/xen/include/xen/cpuidle.h b/xen/include/xen/cpuidle.h
index b7b9e8c..342f4fe 100644
--- a/xen/include/xen/cpuidle.h
+++ b/xen/include/xen/cpuidle.h
@@ -66,6 +66,7 @@ struct acpi_processor_power
     struct acpi_processor_cx *last_state;
     struct acpi_processor_cx *safe_state;
     void *gdata; /* governor specific data */
+    u64 last_state_update_tick;
     u32 last_residency;
     u32 count;
     spinlock_t stat_lock;
-- 
1.9.1

             reply	other threads:[~2015-05-14  5:23 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-05-14  5:23 Huaitong Han [this message]
2015-05-19  9:01 ` [V5] x86/cpuidle: get accurate C0 value with xenpm tool Jan Beulich
2015-05-20  2:42   ` Han, Huaitong
2015-05-20  8:36     ` Han, Huaitong
2015-05-20 10:07       ` Jan Beulich

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1431580999-7886-1-git-send-email-huaitong.han@intel.com \
    --to=huaitong.han@intel.com \
    --cc=jbeulich@suse.com \
    --cc=xen-devel@lists.xen.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.