* [PATCH v2] PM / wakeup: Register wakeup class kobj after device is added
@ 2019-08-19 15:06 Stephen Boyd
2019-08-19 17:04 ` Tri Vo
0 siblings, 1 reply; 3+ messages in thread
From: Stephen Boyd @ 2019-08-19 15:06 UTC (permalink / raw)
To: Rafael J. Wysocki; +Cc: linux-kernel, linux-pm, Qian Cai, Tri Vo
The device_set_wakeup_enable() function can be called on a device that
hasn't been registered with device_add() yet. This allows the device to
be in a state where wakeup is enabled for it but the device isn't
published to userspace in sysfs yet.
After commit 986845e747af ("PM / wakeup: Show wakeup sources stats in
sysfs"), calling device_set_wakeup_enable() will fail for a device that
hasn't been registered with the driver core via device_add(). This is
because we try to create sysfs entries for the device and associate a
wakeup class kobject with it before the device has been registered.
Let's follow a similar approach that device_set_wakeup_capable() takes
here and register the wakeup class either from
device_set_wakeup_enable() when the device is already registered, or
from dpm_sysfs_add() when the device is being registered with the driver
core via device_add().
Fixes: 986845e747af ("PM / wakeup: Show wakeup sources stats in sysfs")
Reported-by: Qian Cai <cai@lca.pw>
Cc: Qian Cai <cai@lca.pw>
Cc: Tri Vo <trong@android.com>
Signed-off-by: Stephen Boyd <swboyd@chromium.org>
---
Changes from v1:
* Export wakeup_source_sysfs_add/remove stubs
* New function to check if we should add the device from
dpm_sysfs_add()
drivers/base/power/power.h | 9 +++++++++
drivers/base/power/sysfs.c | 10 +++++++++-
drivers/base/power/wakeup.c | 10 ++++++----
include/linux/pm_wakeup.h | 10 ++++++++++
4 files changed, 34 insertions(+), 5 deletions(-)
diff --git a/drivers/base/power/power.h b/drivers/base/power/power.h
index 57b1d1d88c8e..22a1533ec56b 100644
--- a/drivers/base/power/power.h
+++ b/drivers/base/power/power.h
@@ -156,5 +156,14 @@ static inline void device_pm_init(struct device *dev)
extern int wakeup_source_sysfs_add(struct device *parent,
struct wakeup_source *ws);
extern void wakeup_source_sysfs_remove(struct wakeup_source *ws);
+#else /* !CONFIG_PM_SLEEP */
+
+static inline int wakeup_source_sysfs_add(struct device *parent,
+ struct wakeup_source *ws)
+{
+ return 0;
+}
+
+static inline void wakeup_source_sysfs_remove(struct wakeup_source *ws) {}
#endif /* CONFIG_PM_SLEEP */
diff --git a/drivers/base/power/sysfs.c b/drivers/base/power/sysfs.c
index 1b9c281cbe41..1468d03ae9fb 100644
--- a/drivers/base/power/sysfs.c
+++ b/drivers/base/power/sysfs.c
@@ -5,6 +5,7 @@
#include <linux/export.h>
#include <linux/pm_qos.h>
#include <linux/pm_runtime.h>
+#include <linux/pm_wakeup.h>
#include <linux/atomic.h>
#include <linux/jiffies.h>
#include "power.h"
@@ -661,14 +662,21 @@ int dpm_sysfs_add(struct device *dev)
if (rc)
goto err_runtime;
}
+ if (!device_has_wakeup_dev(dev)) {
+ rc = wakeup_source_sysfs_add(dev, dev->power.wakeup);
+ if (rc)
+ goto err_wakeup;
+ }
if (dev->power.set_latency_tolerance) {
rc = sysfs_merge_group(&dev->kobj,
&pm_qos_latency_tolerance_attr_group);
if (rc)
- goto err_wakeup;
+ goto err_wakeup_source;
}
return 0;
+ err_wakeup_source:
+ wakeup_source_sysfs_remove(dev->power.wakeup);
err_wakeup:
sysfs_unmerge_group(&dev->kobj, &pm_wakeup_attr_group);
err_runtime:
diff --git a/drivers/base/power/wakeup.c b/drivers/base/power/wakeup.c
index f7925820b5ca..5817b51d2b15 100644
--- a/drivers/base/power/wakeup.c
+++ b/drivers/base/power/wakeup.c
@@ -220,10 +220,12 @@ struct wakeup_source *wakeup_source_register(struct device *dev,
ws = wakeup_source_create(name);
if (ws) {
- ret = wakeup_source_sysfs_add(dev, ws);
- if (ret) {
- wakeup_source_free(ws);
- return NULL;
+ if (!dev || device_is_registered(dev)) {
+ ret = wakeup_source_sysfs_add(dev, ws);
+ if (ret) {
+ wakeup_source_free(ws);
+ return NULL;
+ }
}
wakeup_source_add(ws);
}
diff --git a/include/linux/pm_wakeup.h b/include/linux/pm_wakeup.h
index 661efa029c96..986f797a8b26 100644
--- a/include/linux/pm_wakeup.h
+++ b/include/linux/pm_wakeup.h
@@ -79,6 +79,11 @@ static inline bool device_may_wakeup(struct device *dev)
return dev->power.can_wakeup && !!dev->power.wakeup;
}
+static inline bool device_has_wakeup_dev(struct device *dev)
+{
+ return dev->power.wakeup && !!dev->power.wakeup->dev;
+}
+
static inline void device_set_wakeup_path(struct device *dev)
{
dev->power.wakeup_path = true;
@@ -165,6 +170,11 @@ static inline bool device_may_wakeup(struct device *dev)
return dev->power.can_wakeup && dev->power.should_wakeup;
}
+static inline bool device_has_wakeup_dev(struct device *dev)
+{
+ return false;
+}
+
static inline void device_set_wakeup_path(struct device *dev) {}
static inline void __pm_stay_awake(struct wakeup_source *ws) {}
base-commit: 0c3d3d648b3ed72b920a89bc4fd125e9b7aa5f23
--
Sent by a computer through tubes
^ permalink raw reply related [flat|nested] 3+ messages in thread
* Re: [PATCH v2] PM / wakeup: Register wakeup class kobj after device is added
2019-08-19 15:06 [PATCH v2] PM / wakeup: Register wakeup class kobj after device is added Stephen Boyd
@ 2019-08-19 17:04 ` Tri Vo
2019-08-19 17:15 ` Stephen Boyd
0 siblings, 1 reply; 3+ messages in thread
From: Tri Vo @ 2019-08-19 17:04 UTC (permalink / raw)
To: Stephen Boyd; +Cc: Rafael J. Wysocki, LKML, Linux PM, Qian Cai
On Mon, Aug 19, 2019 at 8:06 AM Stephen Boyd <swboyd@chromium.org> wrote:
>
> The device_set_wakeup_enable() function can be called on a device that
> hasn't been registered with device_add() yet. This allows the device to
> be in a state where wakeup is enabled for it but the device isn't
> published to userspace in sysfs yet.
>
> After commit 986845e747af ("PM / wakeup: Show wakeup sources stats in
> sysfs"), calling device_set_wakeup_enable() will fail for a device that
> hasn't been registered with the driver core via device_add(). This is
> because we try to create sysfs entries for the device and associate a
> wakeup class kobject with it before the device has been registered.
> Let's follow a similar approach that device_set_wakeup_capable() takes
> here and register the wakeup class either from
> device_set_wakeup_enable() when the device is already registered, or
> from dpm_sysfs_add() when the device is being registered with the driver
> core via device_add().
>
> Fixes: 986845e747af ("PM / wakeup: Show wakeup sources stats in sysfs")
> Reported-by: Qian Cai <cai@lca.pw>
> Cc: Qian Cai <cai@lca.pw>
> Cc: Tri Vo <trong@android.com>
> Signed-off-by: Stephen Boyd <swboyd@chromium.org>
> ---
>
> Changes from v1:
> * Export wakeup_source_sysfs_add/remove stubs
> * New function to check if we should add the device from
> dpm_sysfs_add()
>
> drivers/base/power/power.h | 9 +++++++++
> drivers/base/power/sysfs.c | 10 +++++++++-
> drivers/base/power/wakeup.c | 10 ++++++----
> include/linux/pm_wakeup.h | 10 ++++++++++
> 4 files changed, 34 insertions(+), 5 deletions(-)
>
> diff --git a/drivers/base/power/power.h b/drivers/base/power/power.h
> index 57b1d1d88c8e..22a1533ec56b 100644
> --- a/drivers/base/power/power.h
> +++ b/drivers/base/power/power.h
> @@ -156,5 +156,14 @@ static inline void device_pm_init(struct device *dev)
> extern int wakeup_source_sysfs_add(struct device *parent,
> struct wakeup_source *ws);
> extern void wakeup_source_sysfs_remove(struct wakeup_source *ws);
> +#else /* !CONFIG_PM_SLEEP */
> +
> +static inline int wakeup_source_sysfs_add(struct device *parent,
> + struct wakeup_source *ws)
> +{
> + return 0;
> +}
> +
> +static inline void wakeup_source_sysfs_remove(struct wakeup_source *ws) {}
>
> #endif /* CONFIG_PM_SLEEP */
> diff --git a/drivers/base/power/sysfs.c b/drivers/base/power/sysfs.c
> index 1b9c281cbe41..1468d03ae9fb 100644
> --- a/drivers/base/power/sysfs.c
> +++ b/drivers/base/power/sysfs.c
> @@ -5,6 +5,7 @@
> #include <linux/export.h>
> #include <linux/pm_qos.h>
> #include <linux/pm_runtime.h>
> +#include <linux/pm_wakeup.h>
> #include <linux/atomic.h>
> #include <linux/jiffies.h>
> #include "power.h"
> @@ -661,14 +662,21 @@ int dpm_sysfs_add(struct device *dev)
> if (rc)
> goto err_runtime;
> }
> + if (!device_has_wakeup_dev(dev)) {
This evaluates to true if dev->power.wakeup is NULL, which will result
in a null pointer dereference later in wakeup_source_sysfs_add().
I think the condition you want to check for is the one you pointed out
in previous patch.
if (dev->power.wakeup && !dev->power.wakeup->dev)
> + rc = wakeup_source_sysfs_add(dev, dev->power.wakeup);
> + if (rc)
> + goto err_wakeup;
> + }
> if (dev->power.set_latency_tolerance) {
> rc = sysfs_merge_group(&dev->kobj,
> &pm_qos_latency_tolerance_attr_group);
> if (rc)
> - goto err_wakeup;
> + goto err_wakeup_source;
> }
> return 0;
>
> + err_wakeup_source:
> + wakeup_source_sysfs_remove(dev->power.wakeup);
> err_wakeup:
> sysfs_unmerge_group(&dev->kobj, &pm_wakeup_attr_group);
> err_runtime:
> diff --git a/drivers/base/power/wakeup.c b/drivers/base/power/wakeup.c
> index f7925820b5ca..5817b51d2b15 100644
> --- a/drivers/base/power/wakeup.c
> +++ b/drivers/base/power/wakeup.c
> @@ -220,10 +220,12 @@ struct wakeup_source *wakeup_source_register(struct device *dev,
>
> ws = wakeup_source_create(name);
> if (ws) {
> - ret = wakeup_source_sysfs_add(dev, ws);
> - if (ret) {
> - wakeup_source_free(ws);
> - return NULL;
> + if (!dev || device_is_registered(dev)) {
> + ret = wakeup_source_sysfs_add(dev, ws);
> + if (ret) {
> + wakeup_source_free(ws);
> + return NULL;
> + }
> }
> wakeup_source_add(ws);
> }
> diff --git a/include/linux/pm_wakeup.h b/include/linux/pm_wakeup.h
> index 661efa029c96..986f797a8b26 100644
> --- a/include/linux/pm_wakeup.h
> +++ b/include/linux/pm_wakeup.h
> @@ -79,6 +79,11 @@ static inline bool device_may_wakeup(struct device *dev)
> return dev->power.can_wakeup && !!dev->power.wakeup;
> }
>
> +static inline bool device_has_wakeup_dev(struct device *dev)
> +{
> + return dev->power.wakeup && !!dev->power.wakeup->dev;
nit: use "!!" on both (or neither) inputs for consistency.
> +}
> +
> static inline void device_set_wakeup_path(struct device *dev)
> {
> dev->power.wakeup_path = true;
> @@ -165,6 +170,11 @@ static inline bool device_may_wakeup(struct device *dev)
> return dev->power.can_wakeup && dev->power.should_wakeup;
> }
>
> +static inline bool device_has_wakeup_dev(struct device *dev)
> +{
> + return false;
> +}
> +
> static inline void device_set_wakeup_path(struct device *dev) {}
>
> static inline void __pm_stay_awake(struct wakeup_source *ws) {}
>
> base-commit: 0c3d3d648b3ed72b920a89bc4fd125e9b7aa5f23
> --
> Sent by a computer through tubes
>
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH v2] PM / wakeup: Register wakeup class kobj after device is added
2019-08-19 17:04 ` Tri Vo
@ 2019-08-19 17:15 ` Stephen Boyd
0 siblings, 0 replies; 3+ messages in thread
From: Stephen Boyd @ 2019-08-19 17:15 UTC (permalink / raw)
To: Tri Vo; +Cc: Rafael J. Wysocki, LKML, Linux PM, Qian Cai
Quoting Tri Vo (2019-08-19 10:04:56)
> On Mon, Aug 19, 2019 at 8:06 AM Stephen Boyd <swboyd@chromium.org> wrote:
> >
> > The device_set_wakeup_enable() function can be called on a device that
> > hasn't been registered with device_add() yet. This allows the device to
> > be in a state where wakeup is enabled for it but the device isn't
> > published to userspace in sysfs yet.
> >
> > After commit 986845e747af ("PM / wakeup: Show wakeup sources stats in
> > sysfs"), calling device_set_wakeup_enable() will fail for a device that
> > hasn't been registered with the driver core via device_add(). This is
> > because we try to create sysfs entries for the device and associate a
> > wakeup class kobject with it before the device has been registered.
> > Let's follow a similar approach that device_set_wakeup_capable() takes
> > here and register the wakeup class either from
> > device_set_wakeup_enable() when the device is already registered, or
> > from dpm_sysfs_add() when the device is being registered with the driver
> > core via device_add().
> >
> > Fixes: 986845e747af ("PM / wakeup: Show wakeup sources stats in sysfs")
> > Reported-by: Qian Cai <cai@lca.pw>
> > Cc: Qian Cai <cai@lca.pw>
> > Cc: Tri Vo <trong@android.com>
> > Signed-off-by: Stephen Boyd <swboyd@chromium.org>
> > ---
> >
> > Changes from v1:
> > * Export wakeup_source_sysfs_add/remove stubs
> > * New function to check if we should add the device from
> > dpm_sysfs_add()
> >
> > drivers/base/power/power.h | 9 +++++++++
> > drivers/base/power/sysfs.c | 10 +++++++++-
> > drivers/base/power/wakeup.c | 10 ++++++----
> > include/linux/pm_wakeup.h | 10 ++++++++++
> > 4 files changed, 34 insertions(+), 5 deletions(-)
> >
> > diff --git a/drivers/base/power/power.h b/drivers/base/power/power.h
> > index 57b1d1d88c8e..22a1533ec56b 100644
> > --- a/drivers/base/power/power.h
> > +++ b/drivers/base/power/power.h
> > @@ -156,5 +156,14 @@ static inline void device_pm_init(struct device *dev)
> > extern int wakeup_source_sysfs_add(struct device *parent,
> > struct wakeup_source *ws);
> > extern void wakeup_source_sysfs_remove(struct wakeup_source *ws);
> > +#else /* !CONFIG_PM_SLEEP */
> > +
> > +static inline int wakeup_source_sysfs_add(struct device *parent,
> > + struct wakeup_source *ws)
> > +{
> > + return 0;
> > +}
> > +
> > +static inline void wakeup_source_sysfs_remove(struct wakeup_source *ws) {}
> >
> > #endif /* CONFIG_PM_SLEEP */
> > diff --git a/drivers/base/power/sysfs.c b/drivers/base/power/sysfs.c
> > index 1b9c281cbe41..1468d03ae9fb 100644
> > --- a/drivers/base/power/sysfs.c
> > +++ b/drivers/base/power/sysfs.c
> > @@ -5,6 +5,7 @@
> > #include <linux/export.h>
> > #include <linux/pm_qos.h>
> > #include <linux/pm_runtime.h>
> > +#include <linux/pm_wakeup.h>
> > #include <linux/atomic.h>
> > #include <linux/jiffies.h>
> > #include "power.h"
> > @@ -661,14 +662,21 @@ int dpm_sysfs_add(struct device *dev)
> > if (rc)
> > goto err_runtime;
> > }
> > + if (!device_has_wakeup_dev(dev)) {
>
> This evaluates to true if dev->power.wakeup is NULL, which will result
> in a null pointer dereference later in wakeup_source_sysfs_add().
>
> I think the condition you want to check for is the one you pointed out
> in previous patch.
>
> if (dev->power.wakeup && !dev->power.wakeup->dev)
Aha thanks. I need to wrapper it because CONFIG_PM_SLEEP is the one that
defines the power path.
>
> > + rc = wakeup_source_sysfs_add(dev, dev->power.wakeup);
But also, dev->power.wakeup doesn't exist with !CONFIG_PM_SLEEP.
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2019-08-19 17:15 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2019-08-19 15:06 [PATCH v2] PM / wakeup: Register wakeup class kobj after device is added Stephen Boyd
2019-08-19 17:04 ` Tri Vo
2019-08-19 17:15 ` Stephen Boyd
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).