From: "Danilo Krummrich" <dakr@kernel.org>
To: "Johan Hovold" <johan@kernel.org>
Cc: "Alan Stern" <stern@rowland.harvard.edu>,
"Greg Kroah-Hartman" <gregkh@linuxfoundation.org>,
"Rafael J. Wysocki" <rafael@kernel.org>,
"Dmitry Torokhov" <dmitry.torokhov@gmail.com>,
<driver-core@lists.linux.dev>, <linux-usb@vger.kernel.org>,
<linux-kernel@vger.kernel.org>,
"Mario Peter" <mario.peter@leica-geosystems.com>,
<stable@vger.kernel.org>
Subject: Re: [PATCH] driver core: take parent lock on deferred probe
Date: Thu, 15 Oct 2026 02:18:43 +0200 [thread overview]
Message-ID: <DM4ZBLHJLQJB.11A47MPAF74B8@kernel.org> (raw)
In-Reply-To: <DM4X8JU9PXCU.T51SZD8XO7MG@kernel.org>
On Thu Oct 15, 2026 at 12:40 AM CEST, Danilo Krummrich wrote:
> On Wed Oct 14, 2026 at 5:14 PM CEST, Johan Hovold wrote:
>> On Wed, Oct 14, 2026 at 04:42:09PM +0200, Danilo Krummrich wrote:
>>> On Wed Oct 14, 2026 at 10:15 AM CEST, Johan Hovold wrote:
>>
>>> > But it's driver core that needs the extra reference. It seems
>>> > reasonable to require that any subsystem that sets a new
>>> > type->need_parent_lock also makes sure that any device of that type
>>> > holds a reference to its parent. And importantly, it's already the case
>>> > for the only user of this feature today (i.e. USB interfaces).
>>>
>>> Exactly, it is the driver core that needs the extra reference count; for a field
>>> that is owned by the driver core, and for which the driver core manages its
>>> lifetime. If it needs an additional reference count for async or deferred work
>>> it should obtain the reference count for this scope explicitly.
>>
>> As I've mentioned, it's not just for async or deferred probe. It's also
>> needed when a driver gets added or when probe is triggered through
>> sysfs. The bus iterators only takes a reference to the each device and
>> not their parents before calling __driver_attach() etc.
>>
>>> USB interfaces already holding an additional reference makes the proposed
>>> contract work for today's user, but I'd rather not make that prerequisite of the
>>> API. It would push a prerequisite for an implementation detail to the user.
>>
>> I understand your concern, but my reasoning is that this feature is only
>> needed when devices are tightly coupled with their parents and where it
>> therefore probably always make sense to pin the parent.
>>
>> Note that the need_parent_lock flag is now set in the device type struct
>> which also holds a pointer to the release function that drops the
>> reference. So it's pretty easy to check any theoretical further users.
>>
>>> Please, let's not add this API contract. Let's have the driver core deal with
>>> this internally.
>>
>> I can take another look at it, but the alternatives seemed either error
>> prone or a bit icky (e.g. a helper that returns a reference to the
>> parent after checking the dead flag under the device lock).
>
> I thought some more about this, and since it is only the USB device type using
> it it's probably fine for now. Should we ever gain new users of this API (which
> I do not hope), we can still change it and add something like the diff below.
>
> I'd suggest to also add a comment in the USB code, explaining that this
> reference count (also) serves the purpose to fulfill the driver core API
> contract.
I pasted the wrong diff, I meant the one below, but either should work.
diff --git a/drivers/base/base.h b/drivers/base/base.h
index 3c36a2128f88..b3730d030f5f 100644
--- a/drivers/base/base.h
+++ b/drivers/base/base.h
@@ -161,8 +161,9 @@ bool bus_is_registered(const struct bus_type *bus);
int bus_add_driver(struct device_driver *drv);
void bus_remove_driver(struct device_driver *drv);
+struct device *device_get_parent_for_lock(struct device *dev);
void device_release_driver_internal(struct device *dev, const struct device_driver *drv,
- struct device *parent);
+ bool lock_parent);
void driver_detach(const struct device_driver *drv);
void driver_deferred_probe_del(struct device *dev);
diff --git a/drivers/base/bus.c b/drivers/base/bus.c
index fd2257f73253..9975f250d22e 100644
--- a/drivers/base/bus.c
+++ b/drivers/base/bus.c
@@ -637,11 +637,13 @@ void bus_probe_device(struct device *dev)
void bus_probe_device_deferred(struct device *dev)
{
- if (dev->type && dev->type->need_parent_lock)
- device_lock(dev->parent);
+ struct device *parent __free(put_device) = device_get_parent_for_lock(dev);
+
+ if (parent)
+ device_lock(parent);
__bus_probe_device(dev);
- if (dev->type && dev->type->need_parent_lock)
- device_unlock(dev->parent);
+ if (parent)
+ device_unlock(parent);
}
/**
@@ -866,11 +868,13 @@ static int __must_check bus_rescan_devices_helper(struct device *dev,
int ret = 0;
if (!dev->driver) {
- if (dev->type && dev->type->need_parent_lock)
- device_lock(dev->parent);
+ struct device *parent __free(put_device) = device_get_parent_for_lock(dev);
+
+ if (parent)
+ device_lock(parent);
ret = device_attach(dev);
- if (dev->type && dev->type->need_parent_lock)
- device_unlock(dev->parent);
+ if (parent)
+ device_unlock(parent);
}
return ret < 0 ? ret : 0;
}
diff --git a/drivers/base/core.c b/drivers/base/core.c
index 5daa88fa724a..3a924c4e194e 100644
--- a/drivers/base/core.c
+++ b/drivers/base/core.c
@@ -1676,8 +1676,7 @@ void device_links_unbind_consumers(struct device *dev)
device_links_write_unlock();
- device_release_driver_internal(consumer, NULL,
- consumer->parent);
+ device_release_driver_internal(consumer, NULL, true);
put_device(consumer);
goto start;
}
diff --git a/drivers/base/dd.c b/drivers/base/dd.c
index 52123251965d..6ff8d0653769 100644
--- a/drivers/base/dd.c
+++ b/drivers/base/dd.c
@@ -1030,6 +1030,30 @@ static int __device_attach_driver(struct device_driver *drv, void *_data)
return ret == 0;
}
+/*
+ * Pin the parent before taking its lock. A reference to dev alone does not
+ * keep its parent alive after device_del().
+ *
+ * device_del() marks the device dead under its lock before dropping the
+ * registration reference to the parent. A dead device may still need to be
+ * unbound; while dev->driver is set, device_del() cannot have completed the
+ * unbind and dropped that reference either.
+ *
+ * The caller must hold a reference to dev and put the returned parent after
+ * its last access, including any temporary unlock/relock during unbind.
+ */
+struct device *device_get_parent_for_lock(struct device *dev)
+{
+ if (!dev->type || !dev->type->need_parent_lock)
+ return NULL;
+
+ guard(device)(dev);
+ if (dev->p->dead && !dev->driver)
+ return NULL;
+
+ return get_device(dev->parent);
+}
+
/*
* __device_driver_lock - acquire locks needed to manipulate dev->drv
* @dev: Device we will update driver info for
@@ -1065,13 +1089,14 @@ static void __device_driver_unlock(struct device *dev, struct device *parent)
static void __device_attach_async_helper(void *_dev, async_cookie_t cookie)
{
struct device *dev = _dev;
+ struct device *parent __free(put_device) = device_get_parent_for_lock(dev);
struct device_attach_data data = {
.dev = dev,
.check_async = true,
.want_async = true,
};
- __device_driver_lock(dev, dev->parent);
+ __device_driver_lock(dev, parent);
/*
* Check if device has already been removed or claimed. This may
@@ -1093,7 +1118,7 @@ static void __device_attach_async_helper(void *_dev, async_cookie_t cookie)
if (dev->parent)
pm_runtime_put(dev->parent);
out_unlock:
- __device_driver_unlock(dev, dev->parent);
+ __device_driver_unlock(dev, parent);
put_device(dev);
}
@@ -1205,11 +1230,12 @@ void device_initial_probe(struct device *dev)
*/
int device_driver_attach(const struct device_driver *drv, struct device *dev)
{
+ struct device *parent __free(put_device) = device_get_parent_for_lock(dev);
int ret;
- __device_driver_lock(dev, dev->parent);
+ __device_driver_lock(dev, parent);
ret = __driver_probe_device(drv, dev);
- __device_driver_unlock(dev, dev->parent);
+ __device_driver_unlock(dev, parent);
/* also return probe errors as normal negative errnos */
if (ret > 0)
@@ -1223,14 +1249,15 @@ EXPORT_SYMBOL_GPL(device_driver_attach);
static void __driver_attach_async_helper(void *_dev, async_cookie_t cookie)
{
struct device *dev = _dev;
+ struct device *parent __free(put_device) = device_get_parent_for_lock(dev);
const struct device_driver *drv;
int ret;
- __device_driver_lock(dev, dev->parent);
+ __device_driver_lock(dev, parent);
drv = dev->p->async_driver;
dev->p->async_driver = NULL;
ret = driver_probe_device(drv, dev);
- __device_driver_unlock(dev, dev->parent);
+ __device_driver_unlock(dev, parent);
dev_dbg(dev, "driver %s async attach completed: %d\n", drv->name, ret);
@@ -1296,9 +1323,11 @@ static int __driver_attach(struct device *dev, void *data)
return 0;
}
- __device_driver_lock(dev, dev->parent);
+ struct device *parent __free(put_device) = device_get_parent_for_lock(dev);
+
+ __device_driver_lock(dev, parent);
driver_probe_device(drv, dev);
- __device_driver_unlock(dev, dev->parent);
+ __device_driver_unlock(dev, parent);
return 0;
}
@@ -1372,8 +1401,11 @@ static void __device_release_driver(struct device *dev, struct device *parent)
void device_release_driver_internal(struct device *dev,
const struct device_driver *drv,
- struct device *parent)
+ bool lock_parent)
{
+ struct device *parent __free(put_device) = lock_parent ?
+ device_get_parent_for_lock(dev) : NULL;
+
__device_driver_lock(dev, parent);
if (!drv || drv == dev->driver)
@@ -1400,7 +1432,7 @@ void device_release_driver(struct device *dev)
* within their ->remove callback for the same device, they
* will deadlock right here.
*/
- device_release_driver_internal(dev, NULL, NULL);
+ device_release_driver_internal(dev, NULL, false);
}
EXPORT_SYMBOL_GPL(device_release_driver);
@@ -1413,7 +1445,7 @@ EXPORT_SYMBOL_GPL(device_release_driver);
*/
void device_driver_detach(struct device *dev)
{
- device_release_driver_internal(dev, NULL, dev->parent);
+ device_release_driver_internal(dev, NULL, true);
}
/**
@@ -1440,7 +1472,7 @@ void driver_detach(const struct device_driver *drv)
dev = dev_prv->device;
get_device(dev);
spin_unlock(&drv->p->klist_devices.k_lock);
- device_release_driver_internal(dev, drv, dev->parent);
+ device_release_driver_internal(dev, drv, true);
put_device(dev);
}
}
diff --git a/include/linux/device.h b/include/linux/device.h
index 7c7682e4dafe..7604060ff99b 100644
--- a/include/linux/device.h
+++ b/include/linux/device.h
@@ -85,8 +85,7 @@ int subsys_virtual_register(const struct bus_type *subsys,
* the DEVTYPE variable.
*
* @need_parent_lock: When probing or removing a device of this type,
- * driver core should lock the device's parent. Note that
- * the device must hold a reference to its parent.
+ * driver core should lock the device's parent.
*/
struct device_type {
const char *name;
prev parent reply other threads:[~2026-10-15 0:18 UTC|newest]
Thread overview: 12+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-10-13 14:06 [PATCH] driver core: take parent lock on deferred probe Johan Hovold
2026-10-13 17:43 ` Danilo Krummrich
2026-10-13 19:11 ` Alan Stern
2026-10-13 21:01 ` Danilo Krummrich
2026-10-14 2:38 ` Alan Stern
2026-10-14 14:30 ` Danilo Krummrich
2026-10-14 14:51 ` Johan Hovold
2026-10-14 8:15 ` Johan Hovold
2026-10-14 14:42 ` Danilo Krummrich
2026-10-14 15:14 ` Johan Hovold
2026-10-14 22:40 ` Danilo Krummrich
2026-10-15 0:18 ` Danilo Krummrich [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=DM4ZBLHJLQJB.11A47MPAF74B8@kernel.org \
--to=dakr@kernel.org \
--cc=dmitry.torokhov@gmail.com \
--cc=driver-core@lists.linux.dev \
--cc=gregkh@linuxfoundation.org \
--cc=johan@kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-usb@vger.kernel.org \
--cc=mario.peter@leica-geosystems.com \
--cc=rafael@kernel.org \
--cc=stable@vger.kernel.org \
--cc=stern@rowland.harvard.edu \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox