mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Johan.Adolfsson@axis.com
To: <linux-kernel@vger.kernel.org>
Subject: Safer flush_to_ldisc()
Date: Wed, 30 Jul 2003 12:16:04 +0200	[thread overview]
Message-ID: <179001c35683$95ba1290$e2070d0a@pcjohana> (raw)

There is currently (and has always been it seems) a problem
with flush_to_ldisc().
The below diff is against 2.4.20 and contains some stuff made in 2.5,
as well, but the important thing is to set TTY_DONT_FLIP
while processing the flip buffer.
Otherwise, on slow systems or systems with high load and
fast flipping (high baudrates) a new flip might start before
the next flip has been processed, resulting in that characters
could be lost or perhaps arrive out of order.
A driver can write data to a flip buffer that really isn't processed yet.

At least that is what I think happens...
Does it make sense?

/Johan



--- linux/drivers/char/tty_io.c    2 Dec 2002 08:13:09 -0000       1.16
+++ linux/drivers/char/tty_io.c    30 Jul 2003 10:06:04 -0000
@@ -1910,24 +1910,23 @@ static void flush_to_ldisc(void *private
        int             count;
        unsigned long flags;

-       if (test_bit(TTY_DONT_FLIP, &tty->flags)) {
+       /* Have TTY_DONT_FLIP set while processing the flip buffer */
+       if (test_and_set_bit(TTY_DONT_FLIP, &tty->flags)) {
+               /* Already set, process later. */
                queue_task(&tty->flip.tqueue, &tq_timer);
                return;
        }
+       save_flags(flags); cli();
        if (tty->flip.buf_num) {
                cp = tty->flip.char_buf + TTY_FLIPBUF_SIZE;
                fp = tty->flip.flag_buf + TTY_FLIPBUF_SIZE;
-               tty->flip.buf_num = 0;
-
-               save_flags(flags); cli();
+               tty->flip.buf_num = 0;
                tty->flip.char_buf_ptr = tty->flip.char_buf;
                tty->flip.flag_buf_ptr = tty->flip.flag_buf;
        } else {
                cp = tty->flip.char_buf;
                fp = tty->flip.flag_buf;
-               tty->flip.buf_num = 1;
-
-               save_flags(flags); cli();
+               tty->flip.buf_num = 1;
                tty->flip.char_buf_ptr = tty->flip.char_buf +
TTY_FLIPBUF_SIZE;
                tty->flip.flag_buf_ptr = tty->flip.flag_buf +
TTY_FLIPBUF_SIZE;
        }
@@ -1936,6 +1935,7 @@ static void flush_to_ldisc(void *private
        restore_flags(flags);

        tty->ldisc.receive_buf(tty, cp, fp, count);
+       clear_bit(TTY_DONT_FLIP, &tty->flags);
 }


             reply	other threads:[~2003-07-30 10:17 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2003-07-30 10:16 Johan.Adolfsson [this message]
2003-08-20  8:12 ` Johan.Adolfsson

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to='179001c35683$95ba1290$e2070d0a@pcjohana' \
    --to=johan.adolfsson@axis.com \
    --cc=linux-kernel@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®