From: Giovanni Gherdovich <ggherdovich@suse.cz>
Date: Tue, 17 Dec 2019 10:23:09 +0100
Subject: cpufreq: intel_pstate: Revert upstream changes to iowait boosting
Patch-mainline: Never, upstream favours power consumption over performance
References: bsc#1131437

Revert the following two upstream changes:

b8bd1581aa6110eb234c0d424eccd3f32d7317e6 ("cpufreq: intel_pstate: Rework
iowait boosting to be less aggressive")
8e3b403954507eb74ad241dc3750443ccc9ee40a ("cpufreq: intel_pstate: Fix up
iowait_boost computation")

as they negatively affect benchmarks that depend on iowait boosting such as
dbench (table below). This regression was found with the Impera automated
bisection tool.

dbench results on hardy2:
				  initial                  first
				good-v5.0           bad-b8bd1581

    Amean      1         22.27 (   0.00%)       36.94 ( -65.82%)
    Amean      2         24.67 (   0.00%)       39.91 ( -61.79%)
    Amean      4         38.13 (   0.00%)       47.25 ( -23.92%)
    Amean      8         60.89 (   0.00%)       64.44 (  -5.82%)
    Amean      16       106.21 (   0.00%)      109.80 (  -3.38%)
    Amean      32       200.53 (   0.00%)      215.62 (  -7.53%)
    Amean      64       433.18 (   0.00%)      452.34 (  -4.42%)
    Amean      128     1103.80 (   0.00%)     1206.91 (  -9.34%)

    Stddev     1          3.69 (   0.00%)        2.11 (  42.98%)
    Stddev     2          3.71 (   0.00%)        3.83 (  -3.15%)
    Stddev     4          6.95 (   0.00%)        6.38 (   8.20%)
    Stddev     8         12.13 (   0.00%)       12.33 (  -1.69%)
    Stddev     16        20.71 (   0.00%)       21.35 (  -3.08%)
    Stddev     32        40.66 (   0.00%)       44.32 (  -8.99%)
    Stddev     64        80.36 (   0.00%)       77.75 (   3.25%)
    Stddev     128      199.79 (   0.00%)      210.81 (  -5.52%)

Signed-off-by: Giovanni Gherdovich <ggherdovich@suse.cz>
---
 drivers/cpufreq/intel_pstate.c |   36 ++++++++++++++++++------------------
 1 file changed, 18 insertions(+), 18 deletions(-)

--- a/drivers/cpufreq/intel_pstate.c
+++ b/drivers/cpufreq/intel_pstate.c
@@ -48,8 +48,6 @@
 #define int_tofp(X) ((int64_t)(X) << FRAC_BITS)
 #define fp_toint(X) ((X) >> FRAC_BITS)
 
-#define ONE_EIGHTH_FP ((int64_t)1 << (FRAC_BITS - 3))
-
 #define EXT_BITS 6
 #define EXT_FRAC_BITS (EXT_BITS + FRAC_BITS)
 #define fp_ext_toint(X) ((X) >> EXT_FRAC_BITS)
@@ -2100,14 +2098,17 @@ static inline int32_t get_avg_pstate(str
 static inline int32_t get_target_pstate(struct cpudata *cpu)
 {
 	struct sample *sample = &cpu->sample;
-	int32_t busy_frac;
+	int32_t busy_frac, boost;
 	int target, avg_pstate;
 
 	busy_frac = div_fp(sample->mperf << cpu->aperf_mperf_shift,
 			   sample->tsc);
 
-	if (busy_frac < cpu->iowait_boost)
-		busy_frac = cpu->iowait_boost;
+	boost = cpu->iowait_boost;
+	cpu->iowait_boost >>= 1;
+
+	if (busy_frac < boost)
+		busy_frac = boost;
 
 	sample->busy_scaled = busy_frac * 100;
 
@@ -2184,30 +2185,29 @@ static void intel_pstate_update_util(str
 	if (smp_processor_id() != cpu->cpu)
 		return;
 
-	delta_ns = time - cpu->last_update;
 	if (flags & SCHED_CPUFREQ_IOWAIT) {
-		/* Start over if the CPU may have been idle. */
-		if (delta_ns > TICK_NSEC) {
-			cpu->iowait_boost = ONE_EIGHTH_FP;
-		} else if (cpu->iowait_boost >= ONE_EIGHTH_FP) {
-			cpu->iowait_boost <<= 1;
-			if (cpu->iowait_boost > int_tofp(1))
-				cpu->iowait_boost = int_tofp(1);
-		} else {
-			cpu->iowait_boost = ONE_EIGHTH_FP;
-		}
+		cpu->iowait_boost = int_tofp(1);
+		cpu->last_update = time;
+		/*
+		 * The last time the busy was 100% so P-state was max anyway
+		 * so avoid overhead of computation.
+		 */
+		if (fp_toint(cpu->sample.busy_scaled) == 100)
+			return;
+
+		goto set_pstate;
 	} else if (cpu->iowait_boost) {
 		/* Clear iowait_boost if the CPU may have been idle. */
+		delta_ns = time - cpu->last_update;
 		if (delta_ns > TICK_NSEC)
 			cpu->iowait_boost = 0;
-		else
-			cpu->iowait_boost >>= 1;
 	}
 	cpu->last_update = time;
 	delta_ns = time - cpu->sample.time;
 	if ((s64)delta_ns < INTEL_PSTATE_SAMPLING_INTERVAL)
 		return;
 
+set_pstate:
 	if (intel_pstate_sample(cpu, time))
 		intel_pstate_adjust_pstate(cpu);
 }