[PATCH] driver core: avoid klist_remove() on unattached knode_driver

[PATCH] driver core: avoid klist_remove() on unattached knode_driver

[Nguyen Quang Le Kien]

usb_driver_claim_interface() sets dev->driver directly and skips
device_bind_driver() when the interface is not yet registered, so the
device can reach teardown with dev->driver set but knode_driver never
added to the driver's klist_devices. __device_release_driver() then
unconditionally calls klist_remove() on the unattached node, which
dereferences a NULL klist pointer in klist_put() and crashes.

Guard the klist_remove() with klist_node_attached(), mirroring the
existing check in bus_remove_device() for knode_bus.

Reported-by: syzbot+87188222c77c0dbbdb4d@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=87188222c77c0dbbdb4d
Signed-off-by: Nguyen Quang Le Kien <khiemtranzo532001@gmail.com>
---
 drivers/base/dd.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/base/dd.c b/drivers/base/dd.c
index 60c005223..14752a5e5 100644
--- a/drivers/base/dd.c
+++ b/drivers/base/dd.c
@@ -1354,7 +1354,8 @@ static void __device_release_driver(struct device *dev, struct device *parent)
 		device_unbind_cleanup(dev);
 		device_links_driver_cleanup(dev);
 
-		klist_remove(&dev->p->knode_driver);
+		if (klist_node_attached(&dev->p->knode_driver))
+			klist_remove(&dev->p->knode_driver);
 		device_pm_check_callbacks(dev);
 
 		bus_notify(dev, BUS_NOTIFY_UNBOUND_DRIVER);
-- 
2.34.1

[PATCH v2] driver core: avoid klist_remove() on unattached knode_driver

[Nguyen Quang Le Kien]

usb_driver_claim_interface() sets dev->driver directly and skips
device_bind_driver() when the interface is not yet registered, so the
device can reach teardown with dev->driver set but knode_driver never
added to the driver's klist_devices. __device_release_driver() then
unconditionally calls klist_remove() on the unattached node, which
dereferences a NULL klist pointer in klist_put() and crashes.

Only remove the node if the device is actually bound, mirroring the
check device_is_bound() already provides for the driver core. This
matches the existing guard on knode_bus in bus_remove_device().

Reported-by: syzbot+87188222c77c0dbbdb4d@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=87188222c77c0dbbdb4d
Signed-off-by: Nguyen Quang Le Kien <khiemtranzo532001@gmail.com>
---
 drivers/base/dd.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/base/dd.c b/drivers/base/dd.c
index 60c005223..4154b4499 100644
--- a/drivers/base/dd.c
+++ b/drivers/base/dd.c
@@ -1354,7 +1354,8 @@ static void __device_release_driver(struct device *dev, struct device *parent)
 		device_unbind_cleanup(dev);
 		device_links_driver_cleanup(dev);
 
-		klist_remove(&dev->p->knode_driver);
+		if (device_is_bound(dev))
+			klist_remove(&dev->p->knode_driver);
 		device_pm_check_callbacks(dev);
 
 		bus_notify(dev, BUS_NOTIFY_UNBOUND_DRIVER);
-- 
2.34.1

Re: [PATCH v2] driver core: avoid klist_remove() on unattached knode_driver

[Greg KH]

On Thu, Aug 20, 2026 at 02:05:23PM +0800, Nguyen Quang Le Kien wrote:
> usb_driver_claim_interface() sets dev->driver directly and skips
> device_bind_driver() when the interface is not yet registered, so the
> device can reach teardown with dev->driver set but knode_driver never
> added to the driver's klist_devices. __device_release_driver() then
> unconditionally calls klist_remove() on the unattached node, which
> dereferences a NULL klist pointer in klist_put() and crashes.
> 
> Only remove the node if the device is actually bound, mirroring the
> check device_is_bound() already provides for the driver core. This
> matches the existing guard on knode_bus in bus_remove_device().
> 
> Reported-by: syzbot+87188222c77c0dbbdb4d@syzkaller.appspotmail.com
> Closes: https://syzkaller.appspot.com/bug?extid=87188222c77c0dbbdb4d
> Signed-off-by: Nguyen Quang Le Kien <khiemtranzo532001@gmail.com>
> ---
>  drivers/base/dd.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)

What changed from v1?

And why did you send the same patch as Edward Adam Davis <eadavis@qq.com>
just did:
	https://lore.kernel.org/r/tencent_0B76C5676E3DA13640962637444B28743709@qq.com

What is suddenly causing people to care about syzbot bugs for USB?

thanks,

greg k-h

Re: [PATCH v2] driver core: avoid klist_remove() on unattached knode_driver

[Nguyen Quang Le Kien]

Hi Greg,

v1 used klist_node_attached(&dev->p->knode_driver). v1 passed the syzbot
test, but after reviewing the code again I realized klist_node_attached()
isn't really the right check at that level -- it's a raw klist API.
device_is_bound() is what driver core uses to ask whether a device is
bound, and it also handles the NULL dev->p case. So I switched to that
in v2. Nothing else changed.

As for syzbot -- it's just where I find kernel bugs, and it can test my
patches for me. That's all.

Thanks,
Kien

Re: [PATCH v2] driver core: avoid klist_remove() on unattached knode_driver

[Greg Kroah-Hartman]

On Thu, Aug 20, 2026 at 02:56:06PM +0800, Nguyen Quang Le Kien wrote:
> Hi Greg,

Please provide email context when responding, some of us get 1000+
emails a day.

> v1 used klist_node_attached(&dev->p->knode_driver). v1 passed the syzbot
> test, but after reviewing the code again I realized klist_node_attached()
> isn't really the right check at that level -- it's a raw klist API.

Then please document this in the proper place, for v2, as required.

thanks,

greg k-h

[PATCH v3] driver core: avoid klist_remove() on unattached knode_driver

[Nguyen Quang Le Kien]

usb_driver_claim_interface() sets dev->driver directly and skips
device_bind_driver() when the interface is not yet registered, so the
device can reach teardown with dev->driver set but knode_driver never
added to the driver's klist_devices. __device_release_driver() then
unconditionally calls klist_remove() on the unattached node, which
dereferences a NULL klist pointer in klist_put() and crashes.

Only remove the node if the device is actually bound. Use
device_is_bound() rather than klist_node_attached() directly: the
latter is a raw klist API and does not NULL-check dev->p, while
device_is_bound() is the standard bound-state check used throughout
driver core (driver_bound(), __device_attach()).

Fixes: 94e7b1c5ff20 ("[PATCH] Add a klist to struct device_driver for the devices bound to it.")
Reported-by: syzbot+87188222c77c0dbbdb4d@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=87188222c77c0dbbdb4d
Signed-off-by: Nguyen Quang Le Kien <khiemtranzo532001@gmail.com>
---
 drivers/base/dd.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/base/dd.c b/drivers/base/dd.c
index 60c005223..4154b4499 100644
--- a/drivers/base/dd.c
+++ b/drivers/base/dd.c
@@ -1354,7 +1354,8 @@ static void __device_release_driver(struct device *dev, struct device *parent)
 		device_unbind_cleanup(dev);
 		device_links_driver_cleanup(dev);
 
-		klist_remove(&dev->p->knode_driver);
+		if (device_is_bound(dev))
+			klist_remove(&dev->p->knode_driver);
 		device_pm_check_callbacks(dev);
 
 		bus_notify(dev, BUS_NOTIFY_UNBOUND_DRIVER);
-- 
2.34.1

Re: [PATCH v3] driver core: avoid klist_remove() on unattached knode_driver

[Greg Kroah-Hartman]

On Thu, Aug 20, 2026 at 03:45:38PM +0800, Nguyen Quang Le Kien wrote:
> usb_driver_claim_interface() sets dev->driver directly and skips
> device_bind_driver() when the interface is not yet registered, so the
> device can reach teardown with dev->driver set but knode_driver never
> added to the driver's klist_devices. __device_release_driver() then
> unconditionally calls klist_remove() on the unattached node, which
> dereferences a NULL klist pointer in klist_put() and crashes.
> 
> Only remove the node if the device is actually bound. Use
> device_is_bound() rather than klist_node_attached() directly: the
> latter is a raw klist API and does not NULL-check dev->p, while
> device_is_bound() is the standard bound-state check used throughout
> driver core (driver_bound(), __device_attach()).
> 
> Fixes: 94e7b1c5ff20 ("[PATCH] Add a klist to struct device_driver for the devices bound to it.")
> Reported-by: syzbot+87188222c77c0dbbdb4d@syzkaller.appspotmail.com
> Closes: https://syzkaller.appspot.com/bug?extid=87188222c77c0dbbdb4d
> Signed-off-by: Nguyen Quang Le Kien <khiemtranzo532001@gmail.com>
> ---
>  drivers/base/dd.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/base/dd.c b/drivers/base/dd.c
> index 60c005223..4154b4499 100644
> --- a/drivers/base/dd.c
> +++ b/drivers/base/dd.c
> @@ -1354,7 +1354,8 @@ static void __device_release_driver(struct device *dev, struct device *parent)
>  		device_unbind_cleanup(dev);
>  		device_links_driver_cleanup(dev);
>  
> -		klist_remove(&dev->p->knode_driver);
> +		if (device_is_bound(dev))
> +			klist_remove(&dev->p->knode_driver);
>  		device_pm_check_callbacks(dev);
>  
>  		bus_notify(dev, BUS_NOTIFY_UNBOUND_DRIVER);
> -- 
> 2.34.1
> 
> 

Hi,

This is the friendly patch-bot of Greg Kroah-Hartman.  You have sent him
a patch that has triggered this response.  He used to manually respond
to these common problems, but in order to save his sanity (he kept
writing the same thing over and over, yet to different people), I was
created.  Hopefully you will not take offence and will fix the problem
in your patch and resubmit it so that it can be accepted into the Linux
kernel tree.

You are receiving this message because of the following common error(s)
as indicated below:

- This looks like a new version of a previously submitted patch, but you
  did not list below the --- line any changes from the previous version.
  Please read the section entitled "The canonical patch format" in the
  kernel file, Documentation/process/submitting-patches.rst for what
  needs to be done here to properly describe this.

- You have marked a patch with a "Fixes:" tag for a commit that is in an
  older released kernel, yet you do not have a cc: stable line in the
  signed-off-by area at all, which means that the patch will not be
  applied to any older kernel releases.  To properly fix this, please
  follow the documented rules in the
  Documentation/process/stable-kernel-rules.rst file for how to resolve
  this.

- You have sent patches very very quickly, with no time to have others
  review them.  Slow down, there is no rush here, and no deadlines.

If you wish to discuss this problem further, or you have questions about
how to resolve this issue, please feel free to respond to this email and
Greg will reply once he has dug out from the pending patches received
from other developers.

thanks,

greg k-h's patch email bot

[PATCH v4] driver core: avoid klist_remove() on unattached knode_driver

[Nguyen Quang Le Kien]

usb_driver_claim_interface() sets dev->driver directly and skips
device_bind_driver() when the interface is not yet registered, so the
device can reach teardown with dev->driver set but knode_driver never
added to the driver's klist_devices. __device_release_driver() then
unconditionally calls klist_remove() on the unattached node, which
dereferences a NULL klist pointer in klist_put() and crashes.

Only remove the node if the device is actually bound. Use
device_is_bound() rather than klist_node_attached() directly: the
latter is a raw klist API and does not NULL-check dev->p, while
device_is_bound() is the standard bound-state check used throughout
driver core (driver_bound(), __device_attach()).

Fixes: 94e7b1c5ff20 ("[PATCH] Add a klist to struct device_driver for the devices bound to it.")
Reported-by: syzbot+87188222c77c0dbbdb4d@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=87188222c77c0dbbdb4d
Cc: stable@vger.kernel.org
Signed-off-by: Nguyen Quang Le Kien <khiemtranzo532001@gmail.com>
---
Changes in v4:
- add a changelog below the --- line, as required by submitting-patches.rst
- add Cc: stable@vger.kernel.org, Fixes: points to a released kernel

Changes in v3:
- document in the commit message why device_is_bound() is used instead
  of klist_node_attached()
- add Fixes: tag

Changes in v2:
- use device_is_bound() instead of klist_node_attached() as the bound
  check, per review
---
 drivers/base/dd.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/base/dd.c b/drivers/base/dd.c
index 60c005223..4154b4499 100644
--- a/drivers/base/dd.c
+++ b/drivers/base/dd.c
@@ -1354,7 +1354,8 @@ static void __device_release_driver(struct device *dev, struct device *parent)
 		device_unbind_cleanup(dev);
 		device_links_driver_cleanup(dev);
 
-		klist_remove(&dev->p->knode_driver);
+		if (device_is_bound(dev))
+			klist_remove(&dev->p->knode_driver);
 		device_pm_check_callbacks(dev);
 
 		bus_notify(dev, BUS_NOTIFY_UNBOUND_DRIVER);
-- 
2.34.1

Re: [PATCH v4] driver core: avoid klist_remove() on unattached knode_driver

[Greg KH]

On Thu, Aug 20, 2026 at 04:45:57PM +0800, Nguyen Quang Le Kien wrote:
> usb_driver_claim_interface() sets dev->driver directly and skips
> device_bind_driver() when the interface is not yet registered, so the
> device can reach teardown with dev->driver set but knode_driver never
> added to the driver's klist_devices. __device_release_driver() then
> unconditionally calls klist_remove() on the unattached node, which
> dereferences a NULL klist pointer in klist_put() and crashes.
> 
> Only remove the node if the device is actually bound. Use
> device_is_bound() rather than klist_node_attached() directly: the
> latter is a raw klist API and does not NULL-check dev->p, while
> device_is_bound() is the standard bound-state check used throughout
> driver core (driver_bound(), __device_attach()).
> 
> Fixes: 94e7b1c5ff20 ("[PATCH] Add a klist to struct device_driver for the devices bound to it.")
> Reported-by: syzbot+87188222c77c0dbbdb4d@syzkaller.appspotmail.com
> Closes: https://syzkaller.appspot.com/bug?extid=87188222c77c0dbbdb4d
> Cc: stable@vger.kernel.org
> Signed-off-by: Nguyen Quang Le Kien <khiemtranzo532001@gmail.com>
> ---
> Changes in v4:
> - add a changelog below the --- line, as required by submitting-patches.rst
> - add Cc: stable@vger.kernel.org, Fixes: points to a released kernel

Please slow down.  As my bot said, there is no rush, no deadline, and
it's the middle of the merge window and we can't even do anything with
this anyway.

At the least, wait a week between patch revisions.  If you wish to see
patches reviewed faster, then help out with actual reviews of patches
from others.  Constantly resending stuff like this doesn't make anyone
want to review this at all.

thanks,

greg k-h