Re: [PATCH 2/6] tty: convert tty_ldisc_ops 'read()' function to take a kernel pointer

[Jiri Slaby]

On 21. 01. 21, 10:00, Greg Kroah-Hartman wrote:
> From: Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx>
>
> The tty line discipline .read() function was passed the final user
> pointer destination as an argument, which doesn't match the 'write()'
> function, and makes it very inconvenient to do a splice method for
> ttys.
>
> This is a conversion to use a kernel buffer instead.
>
> NOTE! It does this by passing the tty line discipline ->read() function
> an additional "cookie" to fill in, and an offset into the cookie data.
>
> The line discipline can fill in the cookie data with its own private
> information, and then the reader will repeat the read until either the
> cookie is cleared or it runs out of data.
>
> The only real user of this is N_HDLC, which can use this to handle big
> packets, even if the kernel buffer is smaller than the whole packet.
>
> Cc: Christoph Hellwig <hch@xxxxxx>
> Cc: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
> Cc: Al Viro <viro@xxxxxxxxxxxxxxxxxx>
> Signed-off-by: Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx>
...
> --- a/drivers/tty/tty_io.c
> +++ b/drivers/tty/tty_io.c
> @@ -833,6 +833,65 @@ static void tty_update_time(struct timespec64 *time)
>   		time->tv_sec = sec;
>   }
>   
> +/*
> + * Iterate on the ldisc ->read() function until we've gotten all
> + * the data the ldisc has for us.
> + *
> + * The "cookie" is something that the ldisc read function can fill
> + * in to let us know that there is more data to be had.
> + *
> + * We promise to continue to call the ldisc until it stops returning
> + * data or clears the cookie. The cookie may be something that the
> + * ldisc maintains state for and needs to free.
> + */
> +static int iterate_tty_read(struct tty_ldisc *ld, struct tty_struct *tty, struct file *file,
> +		char __user *buf, size_t count)
> +{
> +	int retval = 0;
> +	void *cookie = NULL;
> +	unsigned long offset = 0;
> +	char kernel_buf[64];
> +
> +	do {
> +		int size, uncopied;
> +
> +		size = count > sizeof(kernel_buf) ? sizeof(kernel_buf) : count;

Or simply
size = min(count, sizeof(kernel_buf));

> +		size = ld->ops->read(tty, file, kernel_buf, size, &cookie, offset);
> +		if (!size)
> +			break;
> +
> +		/*
> +		 * A ldisc read error return will override any previously copied
> +		 * data (eg -EOVERFLOW from HDLC)
> +		 */
> +		if (size < 0) {
> +			memzero_explicit(kernel_buf, sizeof(kernel_buf));
> +			return size;
> +		}
> +
> +		uncopied = copy_to_user(buf+offset, kernel_buf, size);
> +		size -= uncopied;
> +		offset += size;
> +		count -= size;
> +
> +		/*
> +		 * If the user copy failed, we still need to do another ->read()
> +		 * call if we had a cookie to let the ldisc clear up.
> +		 *
> +		 * But make sure size is zeroed.
> +		 */
> +		if (unlikely(uncopied)) {
> +			count = 0;
> +			retval = -EFAULT;

n_hdlc_tty_read will return EOVERFLOW when size is 0, so this EFAULT is 
never propagated, if I am looking correctly? n_tty seems to be fine 
(returns zero for zeroed size).

> +		}
> +	} while (cookie);
> +
> +	/* We always clear tty buffer in case they contained passwords */
> +	memzero_explicit(kernel_buf, sizeof(kernel_buf));
> +	return offset ? offset : retval;
> +}

thanks,
--
js