Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1504431 > unrolled thread

[PATCH 2/2] PM / sleep: don't suspend parent when async child suspend_{noirq,early} fails

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  [PATCH 2/2] PM / sleep: don't suspend parent when async child suspend_{noirq,early} fails Brian Norris <briannorris@chromium.org> - 2016-10-20 02:30 +0200
    Re: [PATCH 2/2] PM / sleep: don't suspend parent when async child  suspend_{noirq,early} fails Brian Norris <briannorris@chromium.org> - 2016-10-20 02:50 +0200
    Re: [PATCH 2/2] PM / sleep: don't suspend parent when async child  suspend_{noirq,early} fails Dmitry Torokhov <dmitry.torokhov@gmail.com> - 2016-10-20 03:00 +0200

#1504431 — [PATCH 2/2] PM / sleep: don't suspend parent when async child suspend_{noirq,early} fails

Consider two devices, A and B, where B is a child of A, and B utilizes
asynchronous suspend (it does not matter whether A is sync or async). If
B fails to suspend_noirq() or suspend_early(), or is interrupted by a
wakeup (pm_wakeup_pending()), then it aborts and sets the async_error
variable. However, device A does not (immediately) check the async_error
variable; it may continue to run its own suspend_noirq()/suspend_early()
callback. This is bad.

We can resolve this problem by checking the async_error flag after
waiting for children to suspend, using the same logic for the noirq and
late suspend cases as we already do for __device_suspend().

It's easy to observe this erroneous behavior by, for example, forcing a
device to sleep a bit in its suspend_noirq() (to ensure the parent is
waiting for the child to complete), then return an error, and watch the
parent suspend_noirq() still get called. (Or similarly, fake a wakeup
event at the right (or is it wrong?) time.)

Fixes: de377b397272 ("PM / sleep: Asynchronous threads for suspend_late")
Fixes: 28b6fd6e3779 ("PM / sleep: Asynchronous threads for suspend_noirq")
Reported-by: Jeffy Chen <jeffy.chen@rock-chips.com>
Signed-off-by: Brian Norris <briannorris@chromium.org>
---
Tested mostly on 4.4, but the code (and seemingly the bug) is mostly untouched
since then. I can try to conjure up a proper upstream test if required.

 drivers/base/power/main.c | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/drivers/base/power/main.c b/drivers/base/power/main.c
index c58563581345..eaf6b53463a5 100644
--- a/drivers/base/power/main.c
+++ b/drivers/base/power/main.c
@@ -1040,6 +1040,9 @@ static int __device_suspend_noirq(struct device *dev, pm_message_t state, bool a
 
 	dpm_wait_for_children(dev, async);
 
+	if (async_error)
+		goto Complete;
+
 	if (dev->pm_domain) {
 		info = "noirq power domain ";
 		callback = pm_noirq_op(&dev->pm_domain->ops, state);
@@ -1187,6 +1190,9 @@ static int __device_suspend_late(struct device *dev, pm_message_t state, bool as
 
 	dpm_wait_for_children(dev, async);
 
+	if (async_error)
+		goto Complete;
+
 	if (dev->pm_domain) {
 		info = "late power domain ";
 		callback = pm_late_early_op(&dev->pm_domain->ops, state);
-- 
2.8.0.rc3.226.g39d4020

[toc] | [next] | [standalone]

#1504441 — Re: [PATCH 2/2] PM / sleep: don't suspend parent when async child suspend_{noirq,early} fails

Ugh, as I hope the patch context makes clear, the subject should be

s/early/late/

as should the body of the commit message.

On Wed, Oct 19, 2016 at 05:26:10PM -0700, Brian Norris wrote:
> Consider two devices, A and B, where B is a child of A, and B utilizes
> asynchronous suspend (it does not matter whether A is sync or async). If
> B fails to suspend_noirq() or suspend_early(), or is interrupted by a

s/early/late/

> wakeup (pm_wakeup_pending()), then it aborts and sets the async_error
> variable. However, device A does not (immediately) check the async_error
> variable; it may continue to run its own suspend_noirq()/suspend_early()

s/early/late/

> callback. This is bad.
> 
> We can resolve this problem by checking the async_error flag after
> waiting for children to suspend, using the same logic for the noirq and
> late suspend cases as we already do for __device_suspend().
> 
> It's easy to observe this erroneous behavior by, for example, forcing a
> device to sleep a bit in its suspend_noirq() (to ensure the parent is
> waiting for the child to complete), then return an error, and watch the
> parent suspend_noirq() still get called. (Or similarly, fake a wakeup
> event at the right (or is it wrong?) time.)
> 
> Fixes: de377b397272 ("PM / sleep: Asynchronous threads for suspend_late")
> Fixes: 28b6fd6e3779 ("PM / sleep: Asynchronous threads for suspend_noirq")
> Reported-by: Jeffy Chen <jeffy.chen@rock-chips.com>
> Signed-off-by: Brian Norris <briannorris@chromium.org>

If the patch is otherwise acceptable, feel free to make the above edits.
Or I can fix them up if I send v2.

Sorry for the noise,
Brian

[toc] | [prev] | [next] | [standalone]

#1504446 — Re: [PATCH 2/2] PM / sleep: don't suspend parent when async child suspend_{noirq,early} fails

On Wed, Oct 19, 2016 at 05:26:10PM -0700, Brian Norris wrote:
> Consider two devices, A and B, where B is a child of A, and B utilizes
> asynchronous suspend (it does not matter whether A is sync or async). If
> B fails to suspend_noirq() or suspend_early(), or is interrupted by a
> wakeup (pm_wakeup_pending()), then it aborts and sets the async_error
> variable. However, device A does not (immediately) check the async_error
> variable; it may continue to run its own suspend_noirq()/suspend_early()
> callback. This is bad.
> 
> We can resolve this problem by checking the async_error flag after
> waiting for children to suspend, using the same logic for the noirq and
> late suspend cases as we already do for __device_suspend().
> 
> It's easy to observe this erroneous behavior by, for example, forcing a
> device to sleep a bit in its suspend_noirq() (to ensure the parent is
> waiting for the child to complete), then return an error, and watch the
> parent suspend_noirq() still get called. (Or similarly, fake a wakeup
> event at the right (or is it wrong?) time.)
> 
> Fixes: de377b397272 ("PM / sleep: Asynchronous threads for suspend_late")
> Fixes: 28b6fd6e3779 ("PM / sleep: Asynchronous threads for suspend_noirq")
> Reported-by: Jeffy Chen <jeffy.chen@rock-chips.com>
> Signed-off-by: Brian Norris <briannorris@chromium.org>

Makes sense to me.

Reviewed-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>

> ---
> Tested mostly on 4.4, but the code (and seemingly the bug) is mostly untouched
> since then. I can try to conjure up a proper upstream test if required.
> 
>  drivers/base/power/main.c | 6 ++++++
>  1 file changed, 6 insertions(+)
> 
> diff --git a/drivers/base/power/main.c b/drivers/base/power/main.c
> index c58563581345..eaf6b53463a5 100644
> --- a/drivers/base/power/main.c
> +++ b/drivers/base/power/main.c
> @@ -1040,6 +1040,9 @@ static int __device_suspend_noirq(struct device *dev, pm_message_t state, bool a
>  
>  	dpm_wait_for_children(dev, async);
>  
> +	if (async_error)
> +		goto Complete;
> +
>  	if (dev->pm_domain) {
>  		info = "noirq power domain ";
>  		callback = pm_noirq_op(&dev->pm_domain->ops, state);
> @@ -1187,6 +1190,9 @@ static int __device_suspend_late(struct device *dev, pm_message_t state, bool as
>  
>  	dpm_wait_for_children(dev, async);
>  
> +	if (async_error)
> +		goto Complete;
> +
>  	if (dev->pm_domain) {
>  		info = "late power domain ";
>  		callback = pm_late_early_op(&dev->pm_domain->ops, state);
> -- 
> 2.8.0.rc3.226.g39d4020
> 

-- 
Dmitry

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web