Received-SPF: pass (google.com: domain of linux-kernel+bounces-117494-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45d1:ec00::1 as permitted sender) client-ip=2604:1380:45d1:ec00::1;
From: "Rafael J. Wysocki" <rjw@rjwysocki.net>
To: Linux PM <linux-pm@vger.kernel.org>
Cc: LKML <linux-kernel@vger.kernel.org>,
 Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Subject:
 [Update][PATCH v1.1 4/5] cpufreq: intel_pstate: Get rid of unnecessary
 READ_ONCE() annotations
Date: Mon, 25 Mar 2024 17:45:30 +0100
Message-ID: <12420098.O9o76ZdvQC@kreacher>
In-Reply-To: <2184891.irdbgypaU6@kreacher>
References: <12409658.O9o76ZdvQC@kreacher> <2184891.irdbgypaU6@kreacher>
Precedence: bulk
MIME-Version: 1.0
Content-Transfer-Encoding: 7Bit
Content-Type: text/plain; charset="UTF-8"

From: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

Drop two redundant checks involving READ_ONCE() from notify_hwp_interrupt()
and make it check hwp_active without READ_ONCE() which is not necessary,
because that variable is only set once during the early initialization of
the driver.

In order to make that clear, annotate hwp_active with __ro_after_init.

Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
---

v1 -> v1.1:
   * There are two redundant checks in notify_hwp_interrupt() that can
     be dropped, so drop them both.

---
 drivers/cpufreq/intel_pstate.c |   23 +++--------------------
 1 file changed, 3 insertions(+), 20 deletions(-)

Index: linux-pm/drivers/cpufreq/intel_pstate.c
===================================================================
--- linux-pm.orig/drivers/cpufreq/intel_pstate.c
+++ linux-pm/drivers/cpufreq/intel_pstate.c
@@ -292,7 +292,7 @@ struct pstate_funcs {
 
 static struct pstate_funcs pstate_funcs __read_mostly;
 
-static int hwp_active __read_mostly;
+static bool hwp_active __ro_after_init;
 static int hwp_mode_bdw __read_mostly;
 static bool per_cpu_limits __read_mostly;
 static bool hwp_boost __read_mostly;
@@ -1640,7 +1640,7 @@ void notify_hwp_interrupt(void)
 	unsigned long flags;
 	u64 value;
 
-	if (!READ_ONCE(hwp_active) || !boot_cpu_has(X86_FEATURE_HWP_NOTIFY))
+	if (!hwp_active || !boot_cpu_has(X86_FEATURE_HWP_NOTIFY))
 		return;
 
 	rdmsrl_safe(MSR_HWP_STATUS, &value);
@@ -1652,23 +1652,6 @@ void notify_hwp_interrupt(void)
 	if (!cpumask_test_cpu(this_cpu, &hwp_intr_enable_mask))
 		goto ack_intr;
 
-	/*
-	 * Currently we never free all_cpu_data. And we can't reach here
-	 * without this allocated. But for safety for future changes, added
-	 * check.
-	 */
-	if (unlikely(!READ_ONCE(all_cpu_data)))
-		goto ack_intr;
-
-	/*
-	 * The free is done during cleanup, when cpufreq registry is failed.
-	 * We wouldn't be here if it fails on init or switch status. But for
-	 * future changes, added check.
-	 */
-	cpudata = READ_ONCE(all_cpu_data[this_cpu]);
-	if (unlikely(!cpudata))
-		goto ack_intr;
-
 	schedule_delayed_work(&cpudata->hwp_notify_work, msecs_to_jiffies(10));
 
 	spin_unlock_irqrestore(&hwp_notify_lock, flags);
@@ -3464,7 +3447,7 @@ static int __init intel_pstate_init(void
 		 * deal with it.
 		 */
 		if ((!no_hwp && boot_cpu_has(X86_FEATURE_HWP_EPP)) || hwp_forced) {
-			WRITE_ONCE(hwp_active, 1);
+			hwp_active = true;
 			hwp_mode_bdw = id->driver_data;
 			intel_pstate.attr = hwp_cpufreq_attrs;
 			intel_cpufreq.attr = hwp_cpufreq_attrs;