2020-06-14 03:36:41

by Aditya Pakki

[permalink] [raw]
Subject: [PATCH] usb: core: fix reference count leak in usb_port_resume

usb_port_resume() calls pm_runtime_get_sync() that increments
the reference counter. In case of failure, decrement the reference
count and return the error.

Signed-off-by: Aditya Pakki <[email protected]>
---
drivers/usb/core/hub.c | 1 +
1 file changed, 1 insertion(+)

diff --git a/drivers/usb/core/hub.c b/drivers/usb/core/hub.c
index b1e14beaac5f..a9231f27144e 100644
--- a/drivers/usb/core/hub.c
+++ b/drivers/usb/core/hub.c
@@ -3542,6 +3542,7 @@ int usb_port_resume(struct usb_device *udev, pm_message_t msg)
if (status < 0) {
dev_dbg(&udev->dev, "can't resume usb port, status %d\n",
status);
+ pm_runtime_put_sync(&port_dev->dev);
return status;
}
}
--
2.25.1


2020-06-14 13:44:14

by Alan Stern

[permalink] [raw]
Subject: Re: [PATCH] usb: core: fix reference count leak in usb_port_resume

On Sat, Jun 13, 2020 at 10:33:53PM -0500, Aditya Pakki wrote:
> usb_port_resume() calls pm_runtime_get_sync() that increments
> the reference counter. In case of failure, decrement the reference
> count and return the error.
>
> Signed-off-by: Aditya Pakki <[email protected]>
> ---
> drivers/usb/core/hub.c | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/drivers/usb/core/hub.c b/drivers/usb/core/hub.c
> index b1e14beaac5f..a9231f27144e 100644
> --- a/drivers/usb/core/hub.c
> +++ b/drivers/usb/core/hub.c
> @@ -3542,6 +3542,7 @@ int usb_port_resume(struct usb_device *udev, pm_message_t msg)
> if (status < 0) {
> dev_dbg(&udev->dev, "can't resume usb port, status %d\n",
> status);
> + pm_runtime_put_sync(&port_dev->dev);

This is wrong; you need to do test_and_clear_bit(port1,
hub->child_usage_bits) before calling pm_runtime_put_sync(). Otherwise
the child_usage_bits value will get out of sync with the port's runtime
status.

Alan Stern