[PATCH v3 1/2] cpufreq: powernv: Adding fast_switch for schedutil

[PATCH v3 1/2] cpufreq: powernv: Adding fast_switch for schedutil

[Akshay Adiga]

Adding fast_switch which does light weight operation to set the desired
pstate. Both global and local pstates are set to the same desired pstate.

Signed-off-by: Akshay Adiga <akshay.adiga@linux.vnet.ibm.com>
Reviewed-by: Gautham R. Shenoy <ego@linux.vnet.ibm.com>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
---
Changes from v2 :
- No changes.
Changes from v1 :
- Removed unnecessary check for index out of bound.

 drivers/cpufreq/powernv-cpufreq.c | 20 +++++++++++++++++++-
 1 file changed, 19 insertions(+), 1 deletion(-)

diff --git a/drivers/cpufreq/powernv-cpufreq.c b/drivers/cpufreq/powernv-cpufreq.c
index d3ffde8..4a4380d 100644
--- a/drivers/cpufreq/powernv-cpufreq.c
+++ b/drivers/cpufreq/powernv-cpufreq.c
@@ -752,9 +752,12 @@ static int powernv_cpufreq_cpu_init(struct cpufreq_policy *policy)
 	spin_lock_init(&gpstates->gpstate_lock);
 	ret = cpufreq_table_validate_and_show(policy, powernv_freqs);
 
-	if (ret < 0)
+	if (ret < 0) {
 		kfree(policy->driver_data);
+		return ret;
+	}
 
+	policy->fast_switch_possible = true;
 	return ret;
 }
 
@@ -897,6 +900,20 @@ static void powernv_cpufreq_stop_cpu(struct cpufreq_policy *policy)
 	del_timer_sync(&gpstates->timer);
 }
 
+static unsigned int powernv_fast_switch(struct cpufreq_policy *policy,
+					unsigned int target_freq)
+{
+	int index;
+	struct powernv_smp_call_data freq_data;
+
+	index = cpufreq_table_find_index_dl(policy, target_freq);
+	freq_data.pstate_id = powernv_freqs[index].driver_data;
+	freq_data.gpstate_id = powernv_freqs[index].driver_data;
+	set_pstate(&freq_data);
+
+	return powernv_freqs[index].frequency;
+}
+
 static struct cpufreq_driver powernv_cpufreq_driver = {
 	.name		= "powernv-cpufreq",
 	.flags		= CPUFREQ_CONST_LOOPS,
@@ -904,6 +921,7 @@ static struct cpufreq_driver powernv_cpufreq_driver = {
 	.exit		= powernv_cpufreq_cpu_exit,
 	.verify		= cpufreq_generic_frequency_table_verify,
 	.target_index	= powernv_cpufreq_target_index,
+	.fast_switch	= powernv_fast_switch,
 	.get		= powernv_cpufreq_get,
 	.stop_cpu	= powernv_cpufreq_stop_cpu,
 	.attr		= powernv_cpu_freq_attr,
-- 
2.5.5

[PATCH v3 2/2] cpufreq: powernv: Use PMCR to verify global and local pstate

[Akshay Adiga]

As fast_switch() may get called with interrupt disable mode, we cannot
hold a mutex to update the global_pstate_info. So currently, fast_switch()
does not update the global_pstate_info and it will end up with stale data
whenever pstate is updated through fast_switch().

As the gpstate_timer can fire after fast_switch() has updated the pstates,
the timer handler cannot rely on the cached values of local and global
pstate and needs to read it from the PMCR.

Only gpstate_timer_handler() is affected by the stale cached pstate data
beacause either fast_switch() or target_index() routines will be called
for a given govenor, but gpstate_timer can fire after the governor has
changed to schedutil.

Signed-off-by: Akshay Adiga <akshay.adiga@linux.vnet.ibm.com>
Reviewed-by: Gautham R. Shenoy <ego@linux.vnet.ibm.com>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
---
Changes from v2 :
- Added generic macros GET_LPSTATE and GET_GPSTATE
 instead of making it specific to PMCR.
Changes from v1 :
- Corrected Commit message
- Type cast pstate values read from PMCR to type s8
- Added Macros to get local and global pstates from PMCR

 drivers/cpufreq/powernv-cpufreq.c | 36 ++++++++++++++++++++++++++----------
 1 file changed, 26 insertions(+), 10 deletions(-)

diff --git a/drivers/cpufreq/powernv-cpufreq.c b/drivers/cpufreq/powernv-cpufreq.c
index 4a4380d..c82304b 100644
--- a/drivers/cpufreq/powernv-cpufreq.c
+++ b/drivers/cpufreq/powernv-cpufreq.c
@@ -42,6 +42,10 @@
 #define PMSR_PSAFE_ENABLE	(1UL << 30)
 #define PMSR_SPR_EM_DISABLE	(1UL << 31)
 #define PMSR_MAX(x)		((x >> 32) & 0xFF)
+#define LPSTATE_SHIFT		48
+#define GPSTATE_SHIFT		56
+#define GET_LPSTATE(x)		(((x) >> LPSTATE_SHIFT) & 0xFF)
+#define GET_GPSTATE(x)		(((x) >> GPSTATE_SHIFT) & 0xFF)
 
 #define MAX_RAMP_DOWN_TIME				5120
 /*
@@ -592,7 +596,8 @@ void gpstate_timer_handler(unsigned long data)
 {
 	struct cpufreq_policy *policy = (struct cpufreq_policy *)data;
 	struct global_pstate_info *gpstates = policy->driver_data;
-	int gpstate_idx;
+	int gpstate_idx, lpstate_idx;
+	unsigned long val;
 	unsigned int time_diff = jiffies_to_msecs(jiffies)
 					- gpstates->last_sampled_time;
 	struct powernv_smp_call_data freq_data;
@@ -600,21 +605,36 @@ void gpstate_timer_handler(unsigned long data)
 	if (!spin_trylock(&gpstates->gpstate_lock))
 		return;
 
+	/*
+	 * If PMCR was last updated was using fast_swtich then
+	 * We may have wrong in gpstate->last_lpstate_idx
+	 * value. Hence, read from PMCR to get correct data.
+	 */
+	val = get_pmspr(SPRN_PMCR);
+	freq_data.gpstate_id = (s8)GET_GPSTATE(val);
+	freq_data.pstate_id = (s8)GET_LPSTATE(val);
+	if (freq_data.gpstate_id  == freq_data.pstate_id) {
+		reset_gpstates(policy);
+		spin_unlock(&gpstates->gpstate_lock);
+		return;
+	}
+
 	gpstates->last_sampled_time += time_diff;
 	gpstates->elapsed_time += time_diff;
-	freq_data.pstate_id = idx_to_pstate(gpstates->last_lpstate_idx);
 
-	if ((gpstates->last_gpstate_idx == gpstates->last_lpstate_idx) ||
-	    (gpstates->elapsed_time > MAX_RAMP_DOWN_TIME)) {
+	if (gpstates->elapsed_time > MAX_RAMP_DOWN_TIME) {
 		gpstate_idx = pstate_to_idx(freq_data.pstate_id);
 		reset_gpstates(policy);
 		gpstates->highest_lpstate_idx = gpstate_idx;
 	} else {
+		lpstate_idx = pstate_to_idx(freq_data.pstate_id);
 		gpstate_idx = calc_global_pstate(gpstates->elapsed_time,
 						 gpstates->highest_lpstate_idx,
-						 gpstates->last_lpstate_idx);
+						 lpstate_idx);
 	}
-
+	freq_data.gpstate_id = idx_to_pstate(gpstate_idx);
+	gpstates->last_gpstate_idx = gpstate_idx;
+	gpstates->last_lpstate_idx = lpstate_idx;
 	/*
 	 * If local pstate is equal to global pstate, rampdown is over
 	 * So timer is not required to be queued.
@@ -622,10 +642,6 @@ void gpstate_timer_handler(unsigned long data)
 	if (gpstate_idx != gpstates->last_lpstate_idx)
 		queue_gpstate_timer(gpstates);
 
-	freq_data.gpstate_id = idx_to_pstate(gpstate_idx);
-	gpstates->last_gpstate_idx = pstate_to_idx(freq_data.gpstate_id);
-	gpstates->last_lpstate_idx = pstate_to_idx(freq_data.pstate_id);
-
 	spin_unlock(&gpstates->gpstate_lock);
 
 	/* Timer may get migrated to a different cpu on cpu hot unplug */
-- 
2.5.5