On Mon, Apr 10, 2017 at 12:53:37AM +0200, Samuel Thibault wrote:
> Hello,
>
> We need to add comments about the synchronization, added inline:
>
> Okash Khawaja, on lun. 03 avril 2017 21:21:29 +0100, wrote:
> > +static int spk_ttyio_receive_buf2(struct tty_struct *tty,
> > + const unsigned char *cp, char *fp, int count)
> > +{
> > + struct spk_ldisc_data *ldisc_data = (struct spk_ldisc_data *)tty->disc_data;
> > +
> > + if (spk_ttyio_synth->read_buff_add) {
> > + int i;
> > + for (i = 0; i < count; i++)
> > + spk_ttyio_synth->read_buff_add(cp[i]);
> > +
> > + return count;
> > + }
> > +
> > + if (ldisc_data->buf_free == 0)
> /* ttyio_in will tty_schedule_flip */
> > + return 0;
> > +
> /* Make sure the consumer has read buf before we have seen
> * buf_free == 1 and overwrite buf */
> > + mb();
> > +
> > + ldisc_data->buf = cp[0];
> > + ldisc_data->buf_free--;
>
> Don't decrement, set to 0. Decrementing would make the reader think
> that there is potential for multi-character support, but with such a
> lightweight synchronization there can't be. The fact that buf_free is 1
> allows the tty part to set it to 0, but not more.
How about changing the type to bool and switching between true and false
only?
_______________________________________________
Speakup mailing list
Speakup@xxxxxxxxxxxxxxxxx
http://linux-speakup.org/cgi-bin/mailman/listinfo/speakup
