linux-serial.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Peter Hurley <peter@hurleysoftware.com>
To: Alan Cox <alan@linux.intel.com>, Jiri Slaby <jslaby@suse.cz>,
	Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: linux-serial@vger.kernel.org, linux-kernel@vger.kernel.org,
	Peter Hurley <peter@hurleysoftware.com>
Subject: [PATCH -next 6/9] tty: Fix ldisc halt sequence on hangup
Date: Tue,  4 Dec 2012 02:07:42 -0500	[thread overview]
Message-ID: <1354604865-10278-7-git-send-email-peter@hurleysoftware.com> (raw)
In-Reply-To: <1354604865-10278-1-git-send-email-peter@hurleysoftware.com>

Flip buffer work cannot be cancelled until all outstanding ldisc
references have been released. Convert the ldisc ref wait into
a full ldisc halt with buffer work cancellation.

Note that the legacy mutex is not held while cancelling.

Signed-off-by: Peter Hurley <peter@hurleysoftware.com>
---
 drivers/tty/tty_ldisc.c | 40 ++++++++++++++++++++--------------------
 1 file changed, 20 insertions(+), 20 deletions(-)

diff --git a/drivers/tty/tty_ldisc.c b/drivers/tty/tty_ldisc.c
index 0d18d1e..c3dec37 100644
--- a/drivers/tty/tty_ldisc.c
+++ b/drivers/tty/tty_ldisc.c
@@ -549,22 +549,31 @@ static int tty_ldisc_halt(struct tty_struct *tty)
 }
 
 /**
- *	tty_ldisc_hangup_wait_idle - wait for the ldisc to become idle
- *	@tty: tty to wait for
- *
- *	Wait for the line discipline to become idle. The discipline must
- *	have been halted for this to guarantee it remains idle.
+ *	tty_ldisc_hangup_halt - halt the line discipline for hangup
+ *	@tty: tty being hung up
  *
+ *	Shut down the line discipline and work queue for the tty device
+ *	being hungup. Clear the TTY_LDISC flag to ensure no further
+ *	references can be obtained, wait for remaining references to be
+ *	released, and cancel pending buffer work to ensure no more
+ *	data is fed to the ldisc.
  *	Caller must hold legacy and ->ldisc_mutex.
  *
  *	NB: tty_set_ldisc() is prevented from changing the ldisc concurrently
  *	with this function when it checks the TTY_HUPPING state.
+ *
+ *	NB: if tty->ldisc is NULL then buffer work does not need to be
+ *	cancelled because it must already have done as a precondition
+ *	of setting tty->ldisc to NULL
+ *
  */
-static bool tty_ldisc_hangup_wait_idle(struct tty_struct *tty)
+static bool tty_ldisc_hangup_halt(struct tty_struct *tty)
 {
 	char cur_n[TASK_COMM_LEN], tty_n[64];
 	long timeout = 3 * HZ;
 
+	clear_bit(TTY_LDISC, &tty->flags);
+
 	if (tty->ldisc) {	/* Not yet closed */
 		tty_unlock(tty);
 
@@ -575,6 +584,10 @@ static bool tty_ldisc_hangup_wait_idle(struct tty_struct *tty)
 				__func__, get_task_comm(cur_n, current),
 				tty_name(tty, tty_n));
 		}
+
+		cancel_work_sync(&tty->port->buf.work);
+		set_bit(TTY_LDISC_HALTED, &tty->flags);
+
 		/* must reacquire both locks and preserve lock order */
 		mutex_unlock(&tty->ldisc_mutex);
 		tty_lock(tty);
@@ -849,24 +862,11 @@ void tty_ldisc_hangup(struct tty_struct *tty)
 	 */
 	mutex_lock(&tty->ldisc_mutex);
 
-	/*
-	 * this is like tty_ldisc_halt, but we need to give up
-	 * the BTM before calling cancel_work_sync, which may
-	 * need to wait for another function taking the BTM
-	 */
-	clear_bit(TTY_LDISC, &tty->flags);
-	tty_unlock(tty);
-	cancel_work_sync(&tty->port->buf.work);
-	set_bit(TTY_LDISC_HALTED, &tty->flags);
-	mutex_unlock(&tty->ldisc_mutex);
-	tty_lock(tty);
-	mutex_lock(&tty->ldisc_mutex);
-
 	/* At this point we have a closed ldisc and we want to
 	   reopen it. We could defer this to the next open but
 	   it means auditing a lot of other paths so this is
 	   a FIXME */
-	if (tty_ldisc_hangup_wait_idle(tty)) {
+	if (tty_ldisc_hangup_halt(tty)) {
 		if (reset == 0) {
 
 			if (!tty_ldisc_reinit(tty, tty->termios.c_line))
-- 
1.8.0

  parent reply	other threads:[~2012-12-04  7:07 UTC|newest]

Thread overview: 17+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-12-04  7:07 [PATCH -next 0/9] tty: Fix buffer work access-after-free Peter Hurley
2012-12-04  7:07 ` [PATCH -next 1/9] tty: WARN if buffer work racing with tty free Peter Hurley
2012-12-04  7:07 ` [PATCH -next 2/9] tty: Add diagnostic for halted line discipline Peter Hurley
2012-12-04  7:07 ` [PATCH -next 3/9] tty: Don't reschedule buffer work while closing Peter Hurley
2012-12-04  7:07 ` [PATCH -next 4/9] tty: Refactor wait for ldisc refs out of tty_ldisc_hangup() Peter Hurley
2012-12-04  7:07 ` [PATCH -next 5/9] tty: Remove unnecessary re-test of ldisc ref count Peter Hurley
2012-12-04  7:07 ` Peter Hurley [this message]
2012-12-04  7:07 ` [PATCH -next 7/9] tty: Strengthen no-subsequent-use guarantee of tty_ldisc_halt() Peter Hurley
2012-12-04  7:07 ` [PATCH -next 8/9] tty: Remove unnecessary buffer work flush Peter Hurley
2012-12-04  7:07 ` [PATCH -next 9/9] tty: Halt both ldiscs concurrently Peter Hurley
2012-12-04  7:40 ` [PATCH -next 0/9] tty: Fix buffer work access-after-free Ilya Zykov
2012-12-04  8:54 ` Alan Cox
2012-12-04 13:58   ` Peter Hurley
2012-12-04 14:30     ` Alan Cox
2012-12-04  9:38 ` Jiri Slaby
2012-12-07  0:57 ` Peter Hurley
2012-12-10 19:00 ` Ilya Zykov

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=1354604865-10278-7-git-send-email-peter@hurleysoftware.com \
    --to=peter@hurleysoftware.com \
    --cc=alan@linux.intel.com \
    --cc=gregkh@linuxfoundation.org \
    --cc=jslaby@suse.cz \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-serial@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;
as well as URLs for NNTP newsgroup(s).