linux-kernel - [PATCH 2/2] PM: s2idle: Fully prevent the system from entering s2idle when cpuidle isn't supported

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Tue, 11 Jul 2023 14:54:22 +0900
From:   Kazuki Hashimoto <kazukih0205@...il.com>
To:     "Rafael J. Wysocki" <rafael@...nel.org>,
        Daniel Lezcano <daniel.lezcano@...aro.org>,
        Ingo Molnar <mingo@...hat.com>,
        Peter Zijlstra <peterz@...radead.org>,
        Len Brown <len.brown@...el.com>, Pavel Machek <pavel@....cz>
Cc:     linux-pm@...r.kernel.org, linux-kernel@...r.kernel.org,
        "Rafael J. Wysocki" <rafael.j.wysocki@...el.com>,
        Hector Martin <marcan@...can.st>,
        Kazuki Hashimoto <kazukih0205@...il.com>
Subject: [PATCH 2/2] PM: s2idle: Fully prevent the system from entering
 s2idle when cpuidle isn't supported

In order for systems to properly enter s2idle, we need functions both in
the idle subsystem (such as call_cpuidle_s2idle()) and the suspend subsystem
to be executed.

s2idle got blocked in the idle subsystem on platforms without cpuidle after
commit ef2b22ac540c ("cpuidle / sleep: Use broadcast timer for states that stop
local timer"). However, the suspend subsystem doesn't have this, which can cause
the suspend subsystem to begin entering s2idle behind the idle subsystem's back,
which in turn can cause the system to enter s2idle even though all the functions
necessary for s2idle hasn't been executed, breaking the system
(e.g. ClOCK_MONOTONIC keeps ticking during suspend even though it's not supposed
to).

Prevent the system from entering s2idle when cpuidle isn't supported in the
suspend subsystem as well.

Fixes: ef2b22ac540c ("cpuidle / sleep: Use broadcast timer for states that stop local timer")
Signed-off-by: Kazuki Hashimoto <kazukih0205@...il.com>
---
 kernel/power/main.c    | 12 +++++++++---
 kernel/power/suspend.c |  5 +++++
 2 files changed, 14 insertions(+), 3 deletions(-)

diff --git a/kernel/power/main.c b/kernel/power/main.c
index f6425ae3e8b0..82fedcf6032d 100644
--- a/kernel/power/main.c
+++ b/kernel/power/main.c
@@ -174,6 +174,8 @@ static ssize_t mem_sleep_show(struct kobject *kobj, struct kobj_attribute *attr,
 	for (i = PM_SUSPEND_MIN; i < PM_SUSPEND_MAX; i++) {
 		if (i >= PM_SUSPEND_MEM && cxl_mem_active())
 			continue;
+		if (i == PM_SUSPEND_TO_IDLE && cpuidle_not_available())
+			continue;
 		if (mem_sleep_states[i]) {
 			const char *label = mem_sleep_states[i];
 
@@ -226,11 +228,15 @@ static ssize_t mem_sleep_store(struct kobject *kobj, struct kobj_attribute *attr
 	}
 
 	state = decode_suspend_state(buf, n);
-	if (state < PM_SUSPEND_MAX && state > PM_SUSPEND_ON)
+	if (state == PM_SUSPEND_TO_IDLE && cpuidle_not_available())
+		goto err;
+	if (state < PM_SUSPEND_MAX && state > PM_SUSPEND_ON) {
 		mem_sleep_current = state;
-	else
-		error = -EINVAL;
+		goto out;
+	}
 
+ err:
+	error = -EINVAL;
  out:
 	pm_autosleep_unlock();
 	return error ? error : n;
diff --git a/kernel/power/suspend.c b/kernel/power/suspend.c
index fa3bf161d13f..02cc76c9109e 100644
--- a/kernel/power/suspend.c
+++ b/kernel/power/suspend.c
@@ -556,6 +556,11 @@ static int enter_state(suspend_state_t state)
 
 	trace_suspend_resume(TPS("suspend_enter"), state, true);
 	if (state == PM_SUSPEND_TO_IDLE) {
+		if (cpuidle_not_available()) {
+			pr_warn("s2idle is unsupported when cpuidle is unavailable");
+			return -EINVAL;
+		}
+
 #ifdef CONFIG_PM_DEBUG
 		if (pm_test_level != TEST_NONE && pm_test_level <= TEST_CPUS) {
 			pr_warn("Unsupported test mode for suspend to idle, please choose none/freezer/devices/platform.\n");

-- 
2.41.0