On 06/22/2017 12:39 PM, Mike Christie wrote:
> On 06/22/2017 01:21 AM, tang.wenji@xxxxxxxxxx wrote:
>> From: Wenji Tang <tang.wenji@xxxxxxxxxx>
>>
>> In the tcmu_open function does not increase the reference count of dev, but in the tcmu_release function to decrease its reference count. When the user oepn/close device many times, the corresponding dev device is illegally released
>>
>> Signed-off-by: Wenji Tang <tang.wenji@xxxxxxxxxx>
>> ---
>> drivers/target/target_core_user.c | 2 ++
>> 1 file changed, 2 insertions(+)
>>
>> diff --git a/drivers/target/target_core_user.c b/drivers/target/target_core_user.c
>> index beb5f09..376a290 100644
>> --- a/drivers/target/target_core_user.c
>> +++ b/drivers/target/target_core_user.c
>> @@ -1142,6 +1142,8 @@ static int tcmu_open(struct uio_info *info, struct inode *inode)
>> udev->inode = inode;
>>
>> pr_debug("open\n");
>> +
>> + kref_get(&udev->kref);
>>
>
> Patch looks correct. I am removing the tcmu based refcounting in these
I take that back.
We would need to implement a uio close function and do a matching
kref_put in there right?
> patches though:
>
My patchset removed the refcounting because it assumed userspace used it
like how tcmu-runner did.
Are you using a different app that will open the uio device multiple
times? Will it have multiple users opening it at the same time?
> https://www.spinics.net/lists/target-devel/msg15657.html
>
> For refcounting on the uio side and to fix a leak in the uio code, I am
> working on doing uio refcounting similar to these patches began to do:
>
> https://marc.info/?t=142686341100003&r=1&w=2
> --
> To unsubscribe from this list: send the line "unsubscribe target-devel" in
> the body of a message to majordomo@xxxxxxxxxxxxxxx
> More majordomo info at http://vger.kernel.org/majordomo-info.html
>
--
To unsubscribe from this list: send the line "unsubscribe target-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
