Skip to contentSkip to conversation

macb: avoid lockup when TGO during underrun

PATCH In review Richard Röjfors 2 participants 2 messages Jan 15, 2009 – Jan 19, 2009

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 macb: avoid lockup when TGO during underrun

This patch against linux-2.6.28 solves an issue where macb might run out of TX buffers.

In rare cases when an underrun occur, all macb buffers where consumed and the netif_queue was stopped infinitely. This happens then the TGO (transfer ongoing) bit in the TSR is set (and UND). It seems like clening up after the underrun makes the driver and the macb hardware end up in an inconsistent state. The result of this is that in the following calls to macb_tx no TX buffers are released -> the netif_queue was stopped, and never woken up again.

The solution is to disable the transmitter, if TGO is set, before clening up after the underrun, and re-enable the transmitter when the cleaning up is done.

A patch to avoid lockup in the macb driver in some rare underrun cases.

---

drivers/net/macb.c +8 -0
diff -urpN linux-2.6.28/drivers/net/macb.c linux-2.6.28-macb/drivers/net/macb.c
@@ -321,6 +321,10 @@ static void macb_tx(struct macb *bp)
 		printk(KERN_ERR "%s: TX underrun, resetting buffers\n",
 			bp->dev->name);
 
+		/* Transfer ongoing, disable transmitter, to avoid confusion */
+		if (status & MACB_BIT(TGO))
+			macb_writel(bp, NCR, macb_readl(bp, NCR) & ~MACB_BIT(TE));
+
 		head = bp->tx_head;
 
 		/*Mark all the buffer as used to avoid sending a lost buffer*/
@@ -343,6 +347,10 @@ static void macb_tx(struct macb *bp)
 		}
 
 		bp->tx_head = bp->tx_tail = 0;
+
+		/* Enable the transmitter again */
+		if (status & MACB_BIT(TGO))
+			macb_writel(bp, NCR, macb_readl(bp, NCR) | MACB_BIT(TE));
 	}
 
 	if (!(status & MACB_BIT(COMP)))

--Richard

A patch to avoid lockup in the macb driver in some rare underrun cases.

---

drivers/net/macb.c +8 -0
diff -urpN linux-2.6.28/drivers/net/macb.c linux-2.6.28-macb/drivers/net/macb.c
@@ -321,6 +321,10 @@ static void macb_tx(struct macb *bp)
 		printk(KERN_ERR "%s: TX underrun, resetting buffers\n",
 			bp->dev->name);
 
+		/* Transfer ongoing, disable transmitter, to avoid confusion */
+		if (status & MACB_BIT(TGO))
+			macb_writel(bp, NCR, macb_readl(bp, NCR) & ~MACB_BIT(TE));
+
 		head = bp->tx_head;
 
 		/*Mark all the buffer as used to avoid sending a lost buffer*/
@@ -343,6 +347,10 @@ static void macb_tx(struct macb *bp)
 		}
 
 		bp->tx_head = bp->tx_tail = 0;
+
+		/* Enable the transmitter again */
+		if (status & MACB_BIT(TGO))
+			macb_writel(bp, NCR, macb_readl(bp, NCR) | MACB_BIT(TE));
 	}
 
 	if (!(status & MACB_BIT(COMP)))
maintainer
#1 lore raw

From: Richard Röjfors <[email protected]>
Date: Thu, 15 Jan 2009 19:30:16 +0100

Quoted 7 lines

This patch against linux-2.6.28 solves an issue where macb might run out of TX buffers.

In rare cases when an underrun occur, all macb buffers where consumed and the netif_queue was stopped infinitely. This happens then the TGO (transfer ongoing) bit in the TSR is set (and UND). It seems like clening up after the underrun makes the driver and the macb hardware end up in an inconsistent state. The result of this is that in the following calls to macb_tx no TX buffers are released -> the netif_queue was stopped, and never woken up again.

Quoted 6 lines

The solution is to disable the transmitter, if TGO is set, before clening up after the underrun, and re-enable the transmitter when the cleaning up is done.

A patch to avoid lockup in the macb driver in some rare underrun cases.

Patch applied, thanks Richard.