2018-02-22 17:40:20

by Colin King

Subject: [PATCH] usbip: vudc: fix null pointer dereference on udc->lock

From: Colin Ian King <[email protected]>

Currently the driver attempts to spin lock on udc->lock before a NULL
pointer check is performed on udc, hence there is a potential null
pointer dereference on udc->lock. Fix this by moving the null check
on udc before the lock occurs.

Fixes: ea6873a45a22 ("usbip: vudc: Add SysFS infrastructure for VUDC")
Signed-off-by: Colin Ian King <[email protected]>
---
drivers/usb/usbip/vudc_sysfs.c | 8 ++++++--
1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/drivers/usb/usbip/vudc_sysfs.c b/drivers/usb/usbip/vudc_sysfs.c
index d86f72bbbb91..6dcd3ff655c3 100644
--- a/drivers/usb/usbip/vudc_sysfs.c
+++ b/drivers/usb/usbip/vudc_sysfs.c
@@ -105,10 +105,14 @@ static ssize_t usbip_sockfd_store(struct device *dev, struct device_attribute *a
if (rv != 0)
return -EINVAL;

+ if (!udc) {
+ dev_err(dev, "no device");
+ return -ENODEV;
+ }
spin_lock_irqsave(&udc->lock, flags);
/* Don't export what we don't have */
- if (!udc || !udc->driver || !udc->pullup) {
- dev_err(dev, "no device or gadget not bound");
+ if (!udc->driver || !udc->pullup) {
+ dev_err(dev, "gadget not bound");
ret = -ENODEV;
goto unlock;
}
--
2.15.1

2018-02-26 16:42:17

by Shuah Khan

Subject: Re: [PATCH] usbip: vudc: fix null pointer dereference on udc->lock

On 02/22/2018 10:39 AM, Colin King wrote:
> From: Colin Ian King <[email protected]>
>
> Currently the driver attempts to spin lock on udc->lock before a NULL
> pointer check is performed on udc, hence there is a potential null
> pointer dereference on udc->lock. Fix this by moving the null check
> on udc before the lock occurs.
>
> Fixes: ea6873a45a22 ("usbip: vudc: Add SysFS infrastructure for VUDC")
> Signed-off-by: Colin Ian King <[email protected]>
> ---
> drivers/usb/usbip/vudc_sysfs.c | 8 ++++++--
> 1 file changed, 6 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/usb/usbip/vudc_sysfs.c b/drivers/usb/usbip/vudc_sysfs.c
> index d86f72bbbb91..6dcd3ff655c3 100644
> --- a/drivers/usb/usbip/vudc_sysfs.c
> +++ b/drivers/usb/usbip/vudc_sysfs.c
> @@ -105,10 +105,14 @@ static ssize_t usbip_sockfd_store(struct device *dev, struct device_attribute *a
> if (rv != 0)
> return -EINVAL;
>
> + if (!udc) {
> + dev_err(dev, "no device");
> + return -ENODEV;
> + }
> spin_lock_irqsave(&udc->lock, flags);
> /* Don't export what we don't have */
> - if (!udc || !udc->driver || !udc->pullup) {
> - dev_err(dev, "no device or gadget not bound");
> + if (!udc->driver || !udc->pullup) {
> + dev_err(dev, "gadget not bound");
> ret = -ENODEV;
> goto unlock;
> }
>

Thanks for the patch. Looks good to me.

Acked-by: Shuah Khan <[email protected]>

thanks,
-- Shuah

2018-02-27 13:27:01

by Krzysztof Opasiak

Subject: Re: [PATCH] usbip: vudc: fix null pointer dereference on udc->lock

On 02/26/2018 05:40 PM, Shuah Khan wrote:
> On 02/22/2018 10:39 AM, Colin King wrote:
>> From: Colin Ian King <[email protected]>
>>
>> Currently the driver attempts to spin lock on udc->lock before a NULL
>> pointer check is performed on udc, hence there is a potential null
>> pointer dereference on udc->lock. Fix this by moving the null check
>> on udc before the lock occurs.
>>
>> Fixes: ea6873a45a22 ("usbip: vudc: Add SysFS infrastructure for VUDC")
>> Signed-off-by: Colin Ian King <[email protected]>
>> ---
>> drivers/usb/usbip/vudc_sysfs.c | 8 ++++++--
>> 1 file changed, 6 insertions(+), 2 deletions(-)
>>
>> diff --git a/drivers/usb/usbip/vudc_sysfs.c b/drivers/usb/usbip/vudc_sysfs.c
>> index d86f72bbbb91..6dcd3ff655c3 100644
>> --- a/drivers/usb/usbip/vudc_sysfs.c
>> +++ b/drivers/usb/usbip/vudc_sysfs.c
>> @@ -105,10 +105,14 @@ static ssize_t usbip_sockfd_store(struct device *dev, struct device_attribute *a
>> if (rv != 0)
>> return -EINVAL;
>>
>> + if (!udc) {
>> + dev_err(dev, "no device");
>> + return -ENODEV;
>> + }
>> spin_lock_irqsave(&udc->lock, flags);
>> /* Don't export what we don't have */
>> - if (!udc || !udc->driver || !udc->pullup) {
>> - dev_err(dev, "no device or gadget not bound");
>> + if (!udc->driver || !udc->pullup) {
>> + dev_err(dev, "gadget not bound");
>> ret = -ENODEV;
>> goto unlock;
>> }
>>
>
> Thanks for the patch. Looks good to me.
>
> Acked-by: Shuah Khan <[email protected]>

Reviewed-by: Krzysztof Opasiak <[email protected]>

but you could fix also a similar bug one one function above
(dev_desc_read()) ;)

Best regards,
--
Krzysztof Opasiak
Samsung R&D Institute Poland
Samsung Electronics