USB: gadget: udc: Fix use after free bug in udc_plat_remove due to race condition

In udc_plat_probe, &udc->drd_work is bound with
udc_drd_work. udc_drd_work may be called by
usbd_connect_notify to start the work.

Besides, there is a invoking chain:
udc_plat_probe
->udc_probe
->usb_add_gadget_udc_release
->usb_add_gadget

It will add a new gadget to the udc class driver
 list. In usb_add_gadget, it uses usb_udc_release
 as its release function, which will kfree(udc)
 to when destroying the gadget.

If we remove the module which will call udc_plat_remove
  to make cleanup, there may be a unfinished work.
The possible sequence is as follows:

Fix it by finishing the work before cleanup in the udc_plat_remove

Fixes: 1b9f35adb0ff ("usb: gadget: udc: Add Synopsys UDC Platform driver")
Signed-off-by: Zheng Wang <zyytlz.wz@163.com>
---
 drivers/usb/gadget/udc/snps_udc_plat.c | 1 +
 1 file changed, 1 insertion(+)

On Fri, Mar 10, 2023 at 03:00:39PM +0800, Zheng Wang wrote:
> In udc_plat_probe, &udc->drd_work is bound with
> udc_drd_work. udc_drd_work may be called by
> usbd_connect_notify to start the work.
> 
> Besides, there is a invoking chain:
> udc_plat_probe
> ->udc_probe
> ->usb_add_gadget_udc_release
> ->usb_add_gadget
> 
> It will add a new gadget to the udc class driver
>  list. In usb_add_gadget, it uses usb_udc_release
>  as its release function, which will kfree(udc)
>  to when destroying the gadget.
> 
> If we remove the module which will call udc_plat_remove
>   to make cleanup, there may be a unfinished work.
> The possible sequence is as follows:
> 
> Fix it by finishing the work before cleanup in the udc_plat_remove
> 
> Fixes: 1b9f35adb0ff ("usb: gadget: udc: Add Synopsys UDC Platform driver")
> Signed-off-by: Zheng Wang <zyytlz.wz@163.com>
> ---
>  drivers/usb/gadget/udc/snps_udc_plat.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/drivers/usb/gadget/udc/snps_udc_plat.c b/drivers/usb/gadget/udc/snps_udc_plat.c
> index 8bbb89c80348..6228e178cc0a 100644
> --- a/drivers/usb/gadget/udc/snps_udc_plat.c
> +++ b/drivers/usb/gadget/udc/snps_udc_plat.c
> @@ -230,6 +230,7 @@ static int udc_plat_remove(struct platform_device *pdev)
>  	struct udc *dev;
>  
>  	dev = platform_get_drvdata(pdev);
> +	cancel_delayed_work_sync(&dev->drd_work);
>  
>  	usb_del_gadget_udc(&dev->gadget);
>  	/* gadget driver must not be registered */
> -- 
> 2.25.1
> 

Please test this to verify that it actually works.

thanks,

greg k-h

Greg KH <gregkh@linuxfoundation.org> 于2023年3月24日周五 00:20写道：
>
> On Fri, Mar 10, 2023 at 03:00:39PM +0800, Zheng Wang wrote:
> > In udc_plat_probe, &udc->drd_work is bound with
> > udc_drd_work. udc_drd_work may be called by
> > usbd_connect_notify to start the work.
> >
> > Besides, there is a invoking chain:
> > udc_plat_probe
> > ->udc_probe
> > ->usb_add_gadget_udc_release
> > ->usb_add_gadget
> >
> > It will add a new gadget to the udc class driver
> >  list. In usb_add_gadget, it uses usb_udc_release
> >  as its release function, which will kfree(udc)
> >  to when destroying the gadget.
> >
> > If we remove the module which will call udc_plat_remove
> >   to make cleanup, there may be a unfinished work.
> > The possible sequence is as follows:
> >
> > Fix it by finishing the work before cleanup in the udc_plat_remove
> >
> > Fixes: 1b9f35adb0ff ("usb: gadget: udc: Add Synopsys UDC Platform driver")
> > Signed-off-by: Zheng Wang <zyytlz.wz@163.com>
> > ---
> >  drivers/usb/gadget/udc/snps_udc_plat.c | 1 +
> >  1 file changed, 1 insertion(+)
> >
> > diff --git a/drivers/usb/gadget/udc/snps_udc_plat.c b/drivers/usb/gadget/udc/snps_udc_plat.c
> > index 8bbb89c80348..6228e178cc0a 100644
> > --- a/drivers/usb/gadget/udc/snps_udc_plat.c
> > +++ b/drivers/usb/gadget/udc/snps_udc_plat.c
> > @@ -230,6 +230,7 @@ static int udc_plat_remove(struct platform_device *pdev)
> >       struct udc *dev;
> >
> >       dev = platform_get_drvdata(pdev);
> > +     cancel_delayed_work_sync(&dev->drd_work);
> >
> >       usb_del_gadget_udc(&dev->gadget);
> >       /* gadget driver must not be registered */
> > --
> > 2.25.1
> >
>
> Please test this to verify that it actually works.
>

Hi,

Sorry for my late reply. I will try to simulate a device in qemu to test.

Best regards,
Zheng

> thanks,
>
> greg k-h