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] [day] [month] [year] [list]
Message-ID: <aLab51BTgvnULBUd@kekkonen.localdomain>
Date: Tue, 2 Sep 2025 10:25:27 +0300
From: Sakari Ailus <sakari.ailus@...ux.intel.com>
To: Brian Norris <briannorris@...omium.org>
Cc: "Rafael J. Wysocki" <rafael@...nel.org>,
	Pavel Machek <pavel@...nel.org>,
	"Rafael J . Wysocki" <rafael.j.wysocki@...el.com>,
	kunit-dev@...glegroups.com, Len Brown <lenb@...nel.org>,
	linux-pm@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 2/3] PM: runtime: Make put{,_sync}() return 1 when
 already suspended

Hi Brian,

On Thu, Aug 28, 2025 at 05:28:27PM -0700, Brian Norris wrote:
> The pm_runtime.h docs say pm_runtime_put() and pm_runtime_put_sync()
> return 1 when already suspended, but this is not true -- they return
> -EAGAIN. On the other hand, pm_runtime_put_sync_suspend() and
> pm_runtime_put_sync_autosuspend() *do* return 1.
> 
> This is an artifact of the fact that the former are built on rpm_idle(),
> whereas the latter are built on rpm_suspend().
> 
> There are precious few pm_runtime_put()/pm_runtime_put_sync() callers
> that check the return code at all, but most of them only log errors, and
> usually only for negative error codes. None of them should be treating
> this as an error, so:
> 
>  * at best, this may fix some case where a driver treats this condition
>    as an error, when it shouldn't;
> 
>  * at worst, this should make no effect; and
> 
>  * somewhere in between, we could potentially clear up non-fatal log
>    messages.
> 
> Fix the pm_runtime_already_suspended_test() while tweaking the behavior.
> The test makes a lot more sense when these all return 1 when the device
> is already suspended:
> 
>     pm_runtime_put(dev);
>     pm_runtime_put_sync(dev);
>     pm_runtime_suspend(dev);
>     pm_runtime_autosuspend(dev);
>     pm_request_autosuspend(dev);
>     pm_runtime_put_sync_autosuspend(dev);
>     pm_runtime_put_autosuspend(dev);
> 
> Signed-off-by: Brian Norris <briannorris@...omium.org>
> ---
> 
>  drivers/base/power/runtime-test.c | 8 ++------
>  drivers/base/power/runtime.c      | 3 +++
>  2 files changed, 5 insertions(+), 6 deletions(-)
> 
> diff --git a/drivers/base/power/runtime-test.c b/drivers/base/power/runtime-test.c
> index 263c28d5fc50..1be18e871f1d 100644
> --- a/drivers/base/power/runtime-test.c
> +++ b/drivers/base/power/runtime-test.c
> @@ -43,15 +43,11 @@ static void pm_runtime_already_suspended_test(struct kunit *test)
>  	KUNIT_EXPECT_EQ(test, 0, pm_runtime_barrier(dev)); /* no wakeup needed */
>  
>  	KUNIT_EXPECT_TRUE(test, pm_runtime_suspended(dev));
> -	/*
> -	 * We never actually left RPM_SUSPENDED, but rpm_idle() still treats
> -	 * this as -EAGAIN / "runtime PM status change ongoing".
> -	 */
> -	KUNIT_EXPECT_EQ(test, -EAGAIN, pm_runtime_put(dev));
> +	KUNIT_EXPECT_EQ(test, 1, pm_runtime_put(dev));
>  
>  	pm_runtime_get_noresume(dev);
>  	KUNIT_EXPECT_TRUE(test, pm_runtime_suspended(dev));
> -	KUNIT_EXPECT_EQ(test, -EAGAIN, pm_runtime_put_sync(dev));
> +	KUNIT_EXPECT_EQ(test, 1, pm_runtime_put_sync(dev));
>  
>  	KUNIT_EXPECT_EQ(test, 1, pm_runtime_suspend(dev));
>  	KUNIT_EXPECT_EQ(test, 1, pm_runtime_autosuspend(dev));
> diff --git a/drivers/base/power/runtime.c b/drivers/base/power/runtime.c
> index 3e84dc4122de..17cf111d16aa 100644
> --- a/drivers/base/power/runtime.c
> +++ b/drivers/base/power/runtime.c
> @@ -498,6 +498,9 @@ static int rpm_idle(struct device *dev, int rpmflags)
>  	if (retval < 0)
>  		;	/* Conditions are wrong. */
>  
> +	else if ((rpmflags & RPM_GET_PUT) && (retval == 1))
> +		;	/* put() is allowed in RPM_SUSPENDED */

Ah, I missed this while reviewing the 3rd patch. Makes sense. Please ignore
my comments regarding the 3rd patch on whether the return value 1 is
applicable.

The latter parentheses are redundant (the former, too, actually, but the
compiler warns so let them be).

> +
>  	/* Idle notifications are allowed only in the RPM_ACTIVE state. */
>  	else if (dev->power.runtime_status != RPM_ACTIVE)
>  		retval = -EAGAIN;

-- 
Kind regards,

Sakari Ailus

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ