Linux 3.10.37-rt38-rc2

PATCH New Steven Rostedt 1 participant 5 messages Apr 29, 2014

Subsystems (2)

Network drivers Freescale eTSEC Ethernet driver (gianfar)

lore

Share & tools

lore

Steven Rostedt

author

2014-04-29 14:31 UTC #0 lore raw

cover Linux 3.10.37-rt38-rc2

Dear RT Folks,

This is the RT stable review cycle of patch 3.10.37-rt38-rc2.

Please scream at me if I messed something up. Please test the patches too.

The -rc release will be uploaded to kernel.org and will be deleted when the final release is out. This is just a review release (or release candidate).

The pre-releases will not be pushed to the git repository, only the final release is.

If all goes well, this patch will be converted to the next main release on 4/30/2014.

Enjoy,

-- Steve

To build 3.10.37-rt38-rc2 directly, the following patches should be applied:

http://www.kernel.org/pub/linux/kernel/v3.x/linux-3.10.tar.xz

http://www.kernel.org/pub/linux/kernel/v3.x/patch-3.10.37.xz

http://www.kernel.org/pub/linux/kernel/projects/rt/3.10/patch-3.10.37-rt38-rc2.patch.xz

You can also build from 3.10.37-rt37 by applying the incremental patch:

http://www.kernel.org/pub/linux/kernel/projects/rt/3.10/incr/patch-3.10.37-rt37-rt38-rc2.patch.xz

Changes from 3.10.37-rt37:

---

Sebastian Andrzej Siewior (3):
net: gianfar: do not disable interrupts
net: gianfar: do not try to cleanup TX packets if they are not done
rcu: make RCU_BOOST default on RT

Steven Rostedt (Red Hat) (1):

      Linux 3.10.37-rt38-rc2

----

 drivers/net/ethernet/freescale/gianfar.c         | 28 ++++++++++++++----------
 drivers/net/ethernet/freescale/gianfar_ethtool.c |  8 +++----
 drivers/net/ethernet/freescale/gianfar_sysfs.c   | 24 ++++++++++----------
 init/Kconfig                                     |  2 +-
 localversion-rt                                  |  2 +-
 5 files changed, 34 insertions(+), 30 deletions(-)

Steven Rostedt

author

2014-04-29 14:30 UTC #1 lore raw

PATCH 1/4 net: gianfar: do not disable interrupts

3.10.37-rt38-rc2 stable review patch.
If anyone has any objections, please let me know.

------------------

From: Sebastian Andrzej Siewior <[email protected]>

each per-queue lock is taken with spin_lock_irqsave() except in the case where all of them are taken for some kind of serialisation. As an optimisation local_irq_save() is used so that lock_tx_qs() and lock_rx_qs() can use just the spin_lock() variant instead. On RT local_irq_save() behaves differently so we use the nort() variant.
Lockdep screems easily by "ethtool -K eth0 rx off tx off"

What remains is missing lockdep annotation that makes lockdep think lock_tx_qs() may cause a dead lock.

Cc [email protected]

Signed-off-by Sebastian Andrzej Siewior <[email protected]>

Signed-off-by Steven Rostedt <[email protected]>

---
 drivers/net/ethernet/freescale/gianfar.c         | 16 ++++++++--------
 drivers/net/ethernet/freescale/gianfar_ethtool.c |  8 ++++----
 drivers/net/ethernet/freescale/gianfar_sysfs.c   | 24 ++++++++++++------------
 3 files changed, 24 insertions(+), 24 deletions(-)

drivers/net/ethernet/freescale/gianfar.c +8 -8

index 0343a14..5c0efcc 100644
@@ -1274,7 +1274,7 @@ static int gfar_suspend(struct device *dev)
 
 	if (netif_running(ndev)) {
 
-		local_irq_save(flags);
+		local_irq_save_nort(flags);
 		lock_tx_qs(priv);
 		lock_rx_qs(priv);

@@ -1292,7 +1292,7 @@ static int gfar_suspend(struct device *dev)
 
 		unlock_rx_qs(priv);
 		unlock_tx_qs(priv);
-		local_irq_restore(flags);
+		local_irq_restore_nort(flags);
 
 		disable_napi(priv);

@@ -1334,7 +1334,7 @@ static int gfar_resume(struct device *dev)
 	/* Disable Magic Packet mode, in case something
 	 * else woke us up.
 	 */
-	local_irq_save(flags);
+	local_irq_save_nort(flags);
 	lock_tx_qs(priv);
 	lock_rx_qs(priv);

@@ -1346,7 +1346,7 @@ static int gfar_resume(struct device *dev)
 
 	unlock_rx_qs(priv);
 	unlock_tx_qs(priv);
-	local_irq_restore(flags);
+	local_irq_restore_nort(flags);
 
 	netif_device_attach(ndev);

@@ -2346,7 +2346,7 @@ void gfar_vlan_mode(struct net_device *dev, netdev_features_t features)
 	u32 tempval;
 
 	regs = priv->gfargrp[0].regs;
-	local_irq_save(flags);
+	local_irq_save_nort(flags);
 	lock_rx_qs(priv);
 
 	if (features & NETIF_F_HW_VLAN_CTAG_TX) {

@@ -2379,7 +2379,7 @@ void gfar_vlan_mode(struct net_device *dev, netdev_features_t features)
 	gfar_change_mtu(dev, dev->mtu);
 
 	unlock_rx_qs(priv);
-	local_irq_restore(flags);
+	local_irq_restore_nort(flags);
 }
 
 static int gfar_change_mtu(struct net_device *dev, int new_mtu)

@@ -3258,14 +3258,14 @@ static irqreturn_t gfar_error(int irq, void *grp_id)
 			dev->stats.tx_dropped++;
 			atomic64_inc(&priv->extra_stats.tx_underrun);
 
-			local_irq_save(flags);
+			local_irq_save_nort(flags);
 			lock_tx_qs(priv);
 
 			/* Reactivate the Tx Queues */
 			gfar_write(&regs->tstat, gfargrp->tstat);
 
 			unlock_tx_qs(priv);
-			local_irq_restore(flags);
+			local_irq_restore_nort(flags);
 		}
 		netif_dbg(priv, tx_err, dev, "Transmit Error\n");
 	}

drivers/net/ethernet/freescale/gianfar_ethtool.c +4 -4

index 21cd881..c965c0a 100644
@@ -501,7 +501,7 @@ static int gfar_sringparam(struct net_device *dev,
 		/* Halt TX and RX, and process the frames which
 		 * have already been received
 		 */
-		local_irq_save(flags);
+		local_irq_save_nort(flags);
 		lock_tx_qs(priv);
 		lock_rx_qs(priv);

@@ -509,7 +509,7 @@ static int gfar_sringparam(struct net_device *dev,
 
 		unlock_rx_qs(priv);
 		unlock_tx_qs(priv);
-		local_irq_restore(flags);
+		local_irq_restore_nort(flags);
 
 		for (i = 0; i < priv->num_rx_queues; i++)
 			gfar_clean_rx_ring(priv->rx_queue[i],

@@ -552,7 +552,7 @@ int gfar_set_features(struct net_device *dev, netdev_features_t features)
 		/* Halt TX and RX, and process the frames which
 		 * have already been received
 		 */
-		local_irq_save(flags);
+		local_irq_save_nort(flags);
 		lock_tx_qs(priv);
 		lock_rx_qs(priv);

@@ -560,7 +560,7 @@ int gfar_set_features(struct net_device *dev, netdev_features_t features)
 
 		unlock_tx_qs(priv);
 		unlock_rx_qs(priv);
-		local_irq_restore(flags);
+		local_irq_restore_nort(flags);
 
 		for (i = 0; i < priv->num_rx_queues; i++)
 			gfar_clean_rx_ring(priv->rx_queue[i],

drivers/net/ethernet/freescale/gianfar_sysfs.c +12 ~~-12~~

index acb55af..f0160e67 100644
@@ -68,7 +68,7 @@ static ssize_t gfar_set_bd_stash(struct device *dev,
 		return count;
 
 
-	local_irq_save(flags);
+	local_irq_save_nort(flags);
 	lock_rx_qs(priv);
 
 	/* Set the new stashing value */

@@ -84,7 +84,7 @@ static ssize_t gfar_set_bd_stash(struct device *dev,
 	gfar_write(&regs->attr, temp);
 
 	unlock_rx_qs(priv);
-	local_irq_restore(flags);
+	local_irq_restore_nort(flags);
 
 	return count;
 }

@@ -112,7 +112,7 @@ static ssize_t gfar_set_rx_stash_size(struct device *dev,
 	if (!(priv->device_flags & FSL_GIANFAR_DEV_HAS_BUF_STASHING))
 		return count;
 
-	local_irq_save(flags);
+	local_irq_save_nort(flags);
 	lock_rx_qs(priv);
 
 	if (length > priv->rx_buffer_size)

@@ -140,7 +140,7 @@ static ssize_t gfar_set_rx_stash_size(struct device *dev,
 
 out:
 	unlock_rx_qs(priv);
-	local_irq_restore(flags);
+	local_irq_restore_nort(flags);
 
 	return count;
 }

@@ -171,7 +171,7 @@ static ssize_t gfar_set_rx_stash_index(struct device *dev,
 	if (!(priv->device_flags & FSL_GIANFAR_DEV_HAS_BUF_STASHING))
 		return count;
 
-	local_irq_save(flags);
+	local_irq_save_nort(flags);
 	lock_rx_qs(priv);
 
 	if (index > priv->rx_stash_size)

@@ -189,7 +189,7 @@ static ssize_t gfar_set_rx_stash_index(struct device *dev,
 
 out:
 	unlock_rx_qs(priv);
-	local_irq_restore(flags);
+	local_irq_restore_nort(flags);
 
 	return count;
 }

@@ -219,7 +219,7 @@ static ssize_t gfar_set_fifo_threshold(struct device *dev,
 	if (length > GFAR_MAX_FIFO_THRESHOLD)
 		return count;
 
-	local_irq_save(flags);
+	local_irq_save_nort(flags);
 	lock_tx_qs(priv);
 
 	priv->fifo_threshold = length;

@@ -230,7 +230,7 @@ static ssize_t gfar_set_fifo_threshold(struct device *dev,
 	gfar_write(&regs->fifo_tx_thr, temp);
 
 	unlock_tx_qs(priv);
-	local_irq_restore(flags);
+	local_irq_restore_nort(flags);
 
 	return count;
 }

@@ -259,7 +259,7 @@ static ssize_t gfar_set_fifo_starve(struct device *dev,
 	if (num > GFAR_MAX_FIFO_STARVE)
 		return count;
 
-	local_irq_save(flags);
+	local_irq_save_nort(flags);
 	lock_tx_qs(priv);
 
 	priv->fifo_starve = num;

@@ -270,7 +270,7 @@ static ssize_t gfar_set_fifo_starve(struct device *dev,
 	gfar_write(&regs->fifo_tx_starve, temp);
 
 	unlock_tx_qs(priv);
-	local_irq_restore(flags);
+	local_irq_restore_nort(flags);
 
 	return count;
 }

@@ -300,7 +300,7 @@ static ssize_t gfar_set_fifo_starve_off(struct device *dev,
 	if (num > GFAR_MAX_FIFO_STARVE_OFF)
 		return count;
 
-	local_irq_save(flags);
+	local_irq_save_nort(flags);
 	lock_tx_qs(priv);
 
 	priv->fifo_starve_off = num;

@@ -311,7 +311,7 @@ static ssize_t gfar_set_fifo_starve_off(struct device *dev,
 	gfar_write(&regs->fifo_tx_starve_shutoff, temp);
 
 	unlock_tx_qs(priv);
-	local_irq_restore(flags);
+	local_irq_restore_nort(flags);
 
 	return count;
 }

Signature · 1 line

1.8.5.3

Steven Rostedt

author

2014-04-29 14:31 UTC #2 lore raw

PATCH 2/4 net: gianfar: do not try to cleanup TX packets if they are not done

In reply to Steven Rostedt

3.10.37-rt38-rc2 stable review patch.
If anyone has any objections, please let me know.

------------------

From: Sebastian Andrzej Siewior <[email protected]>

What I observe is that the TX queue is not empty and does not make any progress. gfar_clean_tx_ring() does not clean up the packet because it is not completed yet.
The root cause is that the DMA engine did not start yet (it was preempted before doing so) and that dumb loop, loops until that packet is gone.
This is broken since c233cf4 ("gianfar: Fix tx napi polling").

What remains are spurious interrupts if CPU0 cleans up TX packages and CPU1 returns with IRQ_NONE.

Cc [email protected]

Signed-off-by Sebastian Andrzej Siewior <[email protected]>

[ added return howmany; ]

Signed-off-by Steven Rostedt <[email protected]>

---
 drivers/net/ethernet/freescale/gianfar.c | 12 ++++++++----
 1 file changed, 8 insertions(+), 4 deletions(-)

drivers/net/ethernet/freescale/gianfar.c +8 -4

index 5c0efcc..b87a8c9 100644
@@ -132,7 +132,6 @@ static int gfar_poll(struct napi_struct *napi, int budget);
 static void gfar_netpoll(struct net_device *dev);
 #endif
 int gfar_clean_rx_ring(struct gfar_priv_rx_q *rx_queue, int rx_work_limit);
-static void gfar_clean_tx_ring(struct gfar_priv_tx_q *tx_queue);
 static void gfar_process_frame(struct net_device *dev, struct sk_buff *skb,
 			       int amount_pull, struct napi_struct *napi);
 void gfar_halt(struct net_device *dev);

@@ -2475,7 +2474,7 @@ static void gfar_align_skb(struct sk_buff *skb)
 }
 
 /* Interrupt Handler for Transmit complete */
-static void gfar_clean_tx_ring(struct gfar_priv_tx_q *tx_queue)
+static int gfar_clean_tx_ring(struct gfar_priv_tx_q *tx_queue)
 {
 	struct net_device *dev = tx_queue->dev;
 	struct netdev_queue *txq;

@@ -2575,6 +2574,7 @@ static void gfar_clean_tx_ring(struct gfar_priv_tx_q *tx_queue)
 	tx_queue->dirty_tx = bdp;
 
 	netdev_tx_completed_queue(txq, howmany, bytes_sent);
+	return howmany;
 }
 
 static void gfar_schedule_cleanup(struct gfar_priv_grp *gfargrp)

@@ -2856,10 +2856,14 @@ static int gfar_poll(struct napi_struct *napi, int budget)
 			tx_queue = priv->tx_queue[i];
 			/* run Tx cleanup to completion */
 			if (tx_queue->tx_skbuff[tx_queue->skb_dirtytx]) {
-				gfar_clean_tx_ring(tx_queue);
-				has_tx_work = 1;
+				int ret;
+
+				ret = gfar_clean_tx_ring(tx_queue);
+				if (ret)
+					has_tx_work++;
 			}
 		}
+		work_done += has_tx_work;
 
 		for_each_set_bit(i, &gfargrp->rx_bit_map, priv->num_rx_queues) {
 			/* skip queue if not active */

Signature · 1 line

1.8.5.3

Steven Rostedt

author

2014-04-29 14:31 UTC #3 lore raw

PATCH 3/4 rcu: make RCU_BOOST default on RT

In reply to Steven Rostedt

3.10.37-rt38-rc2 stable review patch.
If anyone has any objections, please let me know.

------------------

From: Sebastian Andrzej Siewior <[email protected]>

Since it is no longer invoked from the softirq people run into OOM more often if the priority of the RCU thread is too low. Making boosting default on RT should help in those case and it can be switched off if someone knows better.

Cc [email protected]

Signed-off-by Sebastian Andrzej Siewior <[email protected]>

Signed-off-by Steven Rostedt <[email protected]>

---
 init/Kconfig | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

init/Kconfig +1 -1

index 6c3a4fd..bd3612d 100644
@@ -604,7 +604,7 @@ config TREE_RCU_TRACE
 config RCU_BOOST
 	bool "Enable RCU priority boosting"
 	depends on RT_MUTEXES && PREEMPT_RCU
-	default n
+	default y if PREEMPT_RT_FULL
 	help
 	  This option boosts the priority of preempted RCU readers that
 	  block the current preemptible RCU grace period for too long.

Signature · 1 line

1.8.5.3

Steven Rostedt

author

2014-04-29 14:31 UTC #4 lore raw

PATCH 4/4 Linux 3.10.37-rt38-rc2

In reply to Steven Rostedt

3.10.37-rt38-rc2 stable review patch.
If anyone has any objections, please let me know.

------------------

From: "Steven Rostedt (Red Hat)" <[email protected]>

---
 localversion-rt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

localversion-rt +1 -1

index a3b2408..43245dc 100644
@@ -1 +1 @@
--rt37
+-rt38-rc2

Signature · 1 line

1.8.5.3