From: Thomas Gleixner <tglx@linutronix.de>
To: LKML <linux-kernel@vger.kernel.org>
Cc: Wolfgang Grandegger <wg@grandegger.com>,
Marc Kleine-Budde <mkl@pengutronix.de>,
Markus Pargmann <mpa@pengutronix.de>,
Benedikt Spranger <b.spranger@linutronix.de>,
linux-can@vger.kernel.org, netdev@vger.kernel.org
Subject: [patch 03/12] can: c_can: Make it SMP safe
Date: Tue, 18 Mar 2014 17:19:09 -0000 [thread overview]
Message-ID: <20140318171126.844179673@linutronix.de> (raw)
In-Reply-To: 20140318171007.528610837@linutronix.de
[-- Attachment #1: can-c_can-use-if1-2.patch --]
[-- Type: text/plain, Size: 4418 bytes --]
The hardware has two message control interfaces, but the code only
uses the first one. So on SMP the following can be observed:
CPU0 CPU1
rx_poll()
write IF1 xmit()
write IF1
write IF1
That results in corrupted message object configurations. The TX/RX is
not globally serialized it's only serialized on a core.
Simple solution: Let RX use IF1 and TX use IF2 and all is good.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
---
drivers/net/can/c_can/c_can.c | 36 +++++++++++++++++++++---------------
1 file changed, 21 insertions(+), 15 deletions(-)
Index: linux/drivers/net/can/c_can/c_can.c
===================================================================
--- linux.orig/drivers/net/can/c_can/c_can.c
+++ linux/drivers/net/can/c_can/c_can.c
@@ -133,6 +133,12 @@
#define IF_MCONT_DLC_MASK 0xf
/*
+ * Use IF1 for RX and IF2 for TX
+ */
+#define IF_RX 0
+#define IF_TX 1
+
+/*
* IFx register masks:
* allow easy operation on 16-bit registers when the
* argument is 32-bit instead
@@ -420,7 +426,7 @@ static void c_can_handle_lost_msg_obj(st
priv->write_reg(priv, C_CAN_IFACE(MSGCTRL_REG, iface),
IF_MCONT_CLR_MSGLST);
- c_can_object_put(dev, 0, objno, IF_COMM_CONTROL);
+ c_can_object_put(dev, iface, objno, IF_COMM_CONTROL);
/* create an error msg */
skb = alloc_can_err_skb(dev, &frame);
@@ -551,7 +557,7 @@ static netdev_tx_t c_can_start_xmit(stru
msg_obj_no = get_tx_next_msg_obj(priv);
/* prepare message object for transmission */
- c_can_write_msg_object(dev, 0, frame, msg_obj_no);
+ c_can_write_msg_object(dev, IF_TX, frame, msg_obj_no);
can_put_echo_skb(skb, dev, msg_obj_no - C_CAN_MSG_OBJ_TX_FIRST);
/*
@@ -634,14 +640,14 @@ static void c_can_configure_msg_objects(
/* first invalidate all message objects */
for (i = C_CAN_MSG_OBJ_RX_FIRST; i <= C_CAN_NO_OF_OBJECTS; i++)
- c_can_inval_msg_object(dev, 0, i);
+ c_can_inval_msg_object(dev, IF_RX, i);
/* setup receive message objects */
for (i = C_CAN_MSG_OBJ_RX_FIRST; i < C_CAN_MSG_OBJ_RX_LAST; i++)
- c_can_setup_receive_object(dev, 0, i, 0, 0,
+ c_can_setup_receive_object(dev, IF_RX, i, 0, 0,
(IF_MCONT_RXIE | IF_MCONT_UMASK) & ~IF_MCONT_EOB);
- c_can_setup_receive_object(dev, 0, C_CAN_MSG_OBJ_RX_LAST, 0, 0,
+ c_can_setup_receive_object(dev, IF_RX, C_CAN_MSG_OBJ_RX_LAST, 0, 0,
IF_MCONT_EOB | IF_MCONT_RXIE | IF_MCONT_UMASK);
}
@@ -783,13 +789,13 @@ static void c_can_do_tx(struct net_devic
if (!(val & (1 << (msg_obj_no - 1)))) {
can_get_echo_skb(dev,
msg_obj_no - C_CAN_MSG_OBJ_TX_FIRST);
- c_can_object_get(dev, 0, msg_obj_no, IF_COMM_ALL);
+ c_can_object_get(dev, IF_TX, msg_obj_no, IF_COMM_ALL);
stats->tx_bytes += priv->read_reg(priv,
- C_CAN_IFACE(MSGCTRL_REG, 0))
+ C_CAN_IFACE(MSGCTRL_REG, IF_TX))
& IF_MCONT_DLC_MASK;
stats->tx_packets++;
can_led_event(dev, CAN_LED_EVENT_TX);
- c_can_inval_msg_object(dev, 0, msg_obj_no);
+ c_can_inval_msg_object(dev, IF_TX, msg_obj_no);
} else {
break;
}
@@ -841,13 +847,13 @@ static int c_can_do_rx_poll(struct net_d
while ((msg_obj = ffs(val)) && quota > 0) {
val &= ~BIT(msg_obj - 1);
- c_can_object_get(dev, 0, msg_obj, IF_COMM_ALL &
+ c_can_object_get(dev, IF_RX, msg_obj, IF_COMM_ALL &
~IF_COMM_TXRQST);
msg_ctrl_save = priv->read_reg(priv,
- C_CAN_IFACE(MSGCTRL_REG, 0));
+ C_CAN_IFACE(MSGCTRL_REG, IF_RX));
if (msg_ctrl_save & IF_MCONT_MSGLST) {
- c_can_handle_lost_msg_obj(dev, 0, msg_obj);
+ c_can_handle_lost_msg_obj(dev, IF_RX, msg_obj);
num_rx_pkts++;
quota--;
continue;
@@ -860,19 +866,19 @@ static int c_can_do_rx_poll(struct net_d
continue;
/* read the data from the message object */
- c_can_read_msg_object(dev, 0, msg_ctrl_save);
+ c_can_read_msg_object(dev, IF_RX, msg_ctrl_save);
if (msg_obj < C_CAN_MSG_RX_LOW_LAST)
- c_can_mark_rx_msg_obj(dev, 0,
+ c_can_mark_rx_msg_obj(dev, IF_RX,
msg_ctrl_save, msg_obj);
else if (msg_obj > C_CAN_MSG_RX_LOW_LAST)
/* activate this msg obj */
- c_can_activate_rx_msg_obj(dev, 0,
+ c_can_activate_rx_msg_obj(dev, IF_RX,
msg_ctrl_save, msg_obj);
else if (msg_obj == C_CAN_MSG_RX_LOW_LAST)
/* activate all lower message objects */
c_can_activate_all_lower_rx_msg_obj(dev,
- 0, msg_ctrl_save);
+ IF_RX, msg_ctrl_save);
num_rx_pkts++;
quota--;
next prev parent reply other threads:[~2014-03-18 17:19 UTC|newest]
Thread overview: 33+ messages / expand[flat|nested] mbox.gz Atom feed top
2014-03-18 17:19 [patch 00/12] can: c_can: Fix a series of serious bugs and improve the performance Thomas Gleixner
2014-03-18 17:19 ` [patch 02/12] can: c_can: Fix hardware raminit function Thomas Gleixner
2014-03-18 18:38 ` Marc Kleine-Budde
2014-03-18 22:15 ` Thomas Gleixner
2014-03-19 6:37 ` Oliver Hartkopp
2014-03-19 9:22 ` Thomas Gleixner
2014-03-18 17:19 ` [patch 01/12] can: c_can: Wait for CONTROL_INIT to be cleared Thomas Gleixner
2014-03-18 18:11 ` Marc Kleine-Budde
2014-03-18 18:19 ` Thomas Gleixner
2014-03-18 17:19 ` Thomas Gleixner [this message]
2014-03-18 18:46 ` [patch 03/12] can: c_can: Make it SMP safe Marc Kleine-Budde
2014-03-18 19:40 ` Thomas Gleixner
2014-03-18 17:19 ` [patch 05/12] can: c_can: Fix the lost message handling Thomas Gleixner
2014-03-18 17:19 ` [patch 04/12] can: c_can: Fix buffer ordering for real Thomas Gleixner
2014-03-18 17:19 ` [patch 06/12] can: c_can: Remove braindamaged EOB exit Thomas Gleixner
2014-03-18 17:19 ` [patch 07/12] can: c_can: Provide protection in the xmit path Thomas Gleixner
2014-03-18 17:19 ` [patch 08/12] can: c_can: Makethe code readable Thomas Gleixner
2014-03-18 17:37 ` Joe Perches
2014-03-18 18:23 ` Thomas Gleixner
2014-03-18 18:27 ` [patch 08/12 V2] " Thomas Gleixner
2014-03-18 17:19 ` [patch 09/12] can: c_can: Reduce register access for real Thomas Gleixner
2014-03-18 17:19 ` [patch 11/12] can: c_can: Simplify TX interrupt cleanup Thomas Gleixner
2014-03-18 17:19 ` [patch 10/12] can: c_can: Store dlc private Thomas Gleixner
2014-03-18 17:19 ` [patch 12/12] can: c_can: Avoid led toggling for every packet Thomas Gleixner
2014-03-18 20:18 ` can: c_can: Reduce interrupt load by 50% Thomas Gleixner
2014-03-18 20:35 ` Joe Perches
2014-03-18 20:43 ` Thomas Gleixner
2014-03-18 21:27 ` Joe Perches
2014-03-31 22:35 ` [patch 00/12] can: c_can: Fix a series of serious bugs and improve the performance Thomas Gleixner
2014-04-01 8:09 ` Marc Kleine-Budde
2014-04-01 9:07 ` Thomas Gleixner
2014-04-01 9:09 ` Marc Kleine-Budde
2014-04-01 21:29 ` Marc Kleine-Budde
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=20140318171126.844179673@linutronix.de \
--to=tglx@linutronix.de \
--cc=b.spranger@linutronix.de \
--cc=linux-can@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=mkl@pengutronix.de \
--cc=mpa@pengutronix.de \
--cc=netdev@vger.kernel.org \
--cc=wg@grandegger.com \
/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).