Linux CAN drivers development
 help / color / mirror / Atom feed
From: Alexander Stein <alexander.stein@systec-electronic.com>
To: Pavel Pisa <pisa@cmp.felk.cvut.cz>
Cc: "Rostislav Lisový" <lisovy@gmail.com>,
	"Michal Sojka" <sojkam1@fel.cvut.cz>,
	"Oliver Hartkopp" <oliver.hartkopp@volkswagen.de>,
	linux-can@vger.kernel.org,
	"Alexander Stein" <alexander.stein@systec-electronic.com>
Subject: [PATCH 02/11] sllin: If length is unknown read until timer occurs or break is received
Date: Mon, 16 Dec 2013 08:54:05 +0100	[thread overview]
Message-ID: <1387180454-1738-3-git-send-email-alexander.stein@systec-electronic.com> (raw)
In-Reply-To: <1387180454-1738-1-git-send-email-alexander.stein@systec-electronic.com>

If the length is unknown stop waiting for new data until the tty rx queue
is empty once is not working on fast systems where sllin_receive_buf is
called for each byte received. Instead wait up to 8 bytes, when the
timeout occurs or when a new break is received.

Signed-off-by: Alexander Stein <alexander.stein@systec-electronic.com>
---
 sllin/sllin.c | 54 ++++++++++++++++++++++++++++++++++++++++++++----------
 1 file changed, 44 insertions(+), 10 deletions(-)

diff --git a/sllin/sllin.c b/sllin/sllin.c
index e8aca08..36f23a5 100644
--- a/sllin/sllin.c
+++ b/sllin/sllin.c
@@ -670,6 +670,23 @@ static int sllin_rx_validate(struct sllin *sl)
 	return 0;
 }
 
+static void sllin_slave_finish_rx_msg(struct sllin *sl)
+{
+	if (sllin_rx_validate(sl) == -1) {
+		netdev_dbg(sl->dev, "sllin: RX validation failed.\n");
+		sllin_report_error(sl, LIN_ERR_CHECKSUM);
+	} else {
+		/* Send CAN non-RTR frame with data */
+		netdev_dbg(sl->dev, "sllin: sending NON-RTR CAN frame with LIN payload.");
+		sll_bump(sl); /* send packet to the network layer */
+	}
+	/* Prepare for reception of new header */
+	sl->rx_cnt = 0;
+	sl->rx_expect = SLLIN_BUFF_ID + 1;
+	sl->rx_len_unknown = false; /* We do know exact length of the header */
+	sl->header_received = false;
+}
+
 static void sllin_slave_receive_buf(struct tty_struct *tty,
 			      const unsigned char *cp, char *fp, int count)
 {
@@ -681,6 +698,20 @@ static void sllin_slave_receive_buf(struct tty_struct *tty,
 	/* Read the characters out of the buffer */
 	while (count--) {
 		if (fp && *fp++) {
+			/*
+			 * If we don't know the length of the current message
+			 * we received the break of the next message.
+			 * Evaluate the previous one before continuing
+			 */
+			if (sl->rx_len_unknown == true)
+			{
+				hrtimer_cancel(&sl->rx_timer);
+				sllin_slave_finish_rx_msg(sl);
+
+				set_bit(SLF_RXEVENT, &sl->flags);
+				wake_up(&sl->kwt_wq);
+			}
+
 			netdev_dbg(sl->dev, "sllin_slave_receive_buf char 0x%02x ignored "
 				"due marker 0x%02x, flags 0x%lx\n",
 				*cp, *(fp-1), sl->flags);
@@ -736,25 +767,23 @@ static void sllin_slave_receive_buf(struct tty_struct *tty,
 
 			sl->header_received = true;
 
+			hrtimer_start(&sl->rx_timer,
+				ktime_add(ktime_get(), sl->rx_timer_timeout),
+				HRTIMER_MODE_ABS);
 			sll_send_rtr(sl);
 			continue;
 		}
 
 		/* Response received */
 		if ((sl->header_received == true) &&
-			((sl->rx_cnt >= sl->rx_expect) ||
-			((sl->rx_len_unknown == true) && (count == 0)))) {
+			((sl->rx_cnt >= sl->rx_expect))) {
 
-			sll_bump(sl);
+			hrtimer_cancel(&sl->rx_timer);
 			netdev_dbg(sl->dev, "Received LIN header & LIN response. "
 					"rx_cnt = %u, rx_expect = %u\n", sl->rx_cnt,
 					sl->rx_expect);
+			sllin_slave_finish_rx_msg(sl);
 
-			/* Prepare for reception of new header */
-			sl->rx_cnt = 0;
-			sl->rx_expect = SLLIN_BUFF_ID + 1;
-			sl->rx_len_unknown = false; /* We do know exact length of the header */
-			sl->header_received = false;
 		}
 	}
 }
@@ -905,8 +934,13 @@ static enum hrtimer_restart sllin_rx_timeout_handler(struct hrtimer *hrtimer)
 {
 	struct sllin *sl = container_of(hrtimer, struct sllin, rx_timer);
 
-	sllin_report_error(sl, LIN_ERR_RX_TIMEOUT);
-	set_bit(SLF_TMOUTEVENT, &sl->flags);
+	if (sl->lin_master) {
+		sllin_report_error(sl, LIN_ERR_RX_TIMEOUT);
+		set_bit(SLF_TMOUTEVENT, &sl->flags);
+	} else {
+		sllin_slave_finish_rx_msg(sl);
+		set_bit(SLF_RXEVENT, &sl->flags);
+	}
 	wake_up(&sl->kwt_wq);
 
 	return HRTIMER_NORESTART;
-- 
1.8.3.2


  parent reply	other threads:[~2013-12-16  7:55 UTC|newest]

Thread overview: 21+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2013-12-10 17:16 [RFC] WIP-patches on sllin Alexander Stein
2013-12-10 17:16 ` [PATCH 1/6] sllin: Replace pr_* with netdev_* where approriate Alexander Stein
2013-12-10 17:16 ` [PATCH 2/6] sllin: Make local functions static Alexander Stein
2013-12-10 17:16 ` [PATCH 3/6] sllin: slave: don't forget to wait for the checksum Alexander Stein
2013-12-10 17:16 ` [PATCH 4/6] sllin: reorder functions Alexander Stein
2013-12-10 17:16 ` [PATCH 5/6] sllin: If length is unknown read until timer occurs or break is received Alexander Stein
2013-12-10 17:16 ` [PATCH 6/6] sllin: Add first support for LIN slave task on slave node Alexander Stein
2013-12-13  1:00 ` [RFC] WIP-patches on sllin Pavel Pisa
2013-12-16  7:46   ` Alexander Stein
2013-12-16  7:54     ` [RFC v2] " Alexander Stein
2013-12-16  7:54       ` [PATCH 01/11] sllin: reorder functions Alexander Stein
2013-12-16  7:54       ` Alexander Stein [this message]
2013-12-16  7:54       ` [PATCH 03/11] sllin: Add first support for LIN slave task on slave node Alexander Stein
2013-12-16  7:54       ` [PATCH 04/11] sllin: Add version dependent access to termios in tty_struct Alexander Stein
2013-12-16  7:54       ` [PATCH 05/11] sllin: Fix checksum mode decision Alexander Stein
2013-12-16  7:54       ` [PATCH 06/11] sllin: dlc in cache > 0 is the expected LIN frame length Alexander Stein
2013-12-16  7:54       ` [PATCH 07/11] sllin: Do not set SLF_MSGEVENT flag Alexander Stein
2013-12-16  7:54       ` [PATCH 08/11] sllin: signal timeout if there are less than configured characters Alexander Stein
2013-12-16  7:54       ` [PATCH 09/11] sllin: Don't fiddle with netdev queue and skb in slave receive mode Alexander Stein
2013-12-16  7:54       ` [PATCH 10/11] sllin: Add LIN ID to timeout frame Alexander Stein
2013-12-16  7:54       ` [PATCH 11/11] sllin: Evaluate message upon break only if received at least a LIN ID Alexander Stein

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=1387180454-1738-3-git-send-email-alexander.stein@systec-electronic.com \
    --to=alexander.stein@systec-electronic.com \
    --cc=linux-can@vger.kernel.org \
    --cc=lisovy@gmail.com \
    --cc=oliver.hartkopp@volkswagen.de \
    --cc=pisa@cmp.felk.cvut.cz \
    --cc=sojkam1@fel.cvut.cz \
    /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