Re: [patch] goldfish: locking bugs in goldfish_pipe_read_write()

[Christoffer Dall <christoffer.dall@linaro.org>]

On Tue, Feb 02, 2016 at 12:48:09PM +0300, Dan Carpenter wrote:
> We recently messed up the error handling here.  We can return with the
> pipe->lock held or sometimes we unlock twice by mistake.
> 
> Fixes: 2f3be88237a3 ('goldfish_pipe: Pin pages to memory while copying and other cleanups')
> Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
> 
> diff --git a/drivers/platform/goldfish/goldfish_pipe.c b/drivers/platform/goldfish/goldfish_pipe.c
> index e3fab9a..839df4a 100644
> --- a/drivers/platform/goldfish/goldfish_pipe.c
> +++ b/drivers/platform/goldfish/goldfish_pipe.c
> @@ -313,7 +313,7 @@ static ssize_t goldfish_pipe_read_write(struct file *filp, char __user *buffer,
>  				     !is_write, 0, &page, NULL);
>  		up_read(&current->mm->mmap_sem);
>  		if (ret < 0)
> -			return ret;
> +			break;
>  
>  		if (dev->version) {
>  			/* Device version 1 or newer (qemu-android) expects the
> @@ -400,22 +400,16 @@ static ssize_t goldfish_pipe_read_write(struct file *filp, char __user *buffer,
>  		while (test_bit(wakeBit, &pipe->flags)) {
>  			if (wait_event_interruptible(
>  					pipe->wake_queue,
> -					!test_bit(wakeBit, &pipe->flags))) {
> -				ret = -ERESTARTSYS;
> -				break;
> -			}
> -
> -			if (test_bit(BIT_CLOSED_ON_HOST, &pipe->flags)) {
> -				ret = -EIO;
> -				break;
> -			}
> +					!test_bit(wakeBit, &pipe->flags)))
> +				return -ERESTARTSYS;
> +
> +			if (test_bit(BIT_CLOSED_ON_HOST, &pipe->flags))
> +				return -EIO;
>  		}
>  
>  		/* Try to re-acquire the lock */
> -		if (mutex_lock_interruptible(&pipe->lock)) {
> -			ret = -ERESTARTSYS;
> -			break;
> -		}
> +		if (mutex_lock_interruptible(&pipe->lock))
> +			return -ERESTARTSYS;
>  	}
>  	mutex_unlock(&pipe->lock);
>  

yeah, that was pretty broken, thanks for fixing this.

Reviewed-by: Christoffer Dall <christoffer.dall@linaro.org>