Messages in this thread Patch in this message | | | Date | Mon, 29 Mar 2004 21:56:31 +0200 | From | Francois Romieu <> | Subject | Re: usage of RealTek 8169 crashes my Linux system |
| |
Bernd Fuhrmann <silverbanana@gmx.de> : [...] > problem. Linux still crashes after a couple of MBytes have passed the r8169. > > Thanks for your effort. Any other idea how to fix that problem (like > doing things that make r8169 stable while loosing performance)?
The attached patch applies on top of 2.6.5-rc2-mm5 without fuss.
Could you verify that the computer is otherwise stable if you issue a 'dd if=/dev/hda of=/dev/null bs=1024k' for 120~150 seconds ?
An output of lspci -vx/dmesg/lsmod is always welcome btw.
-- Ueimor drivers/net/r8169.c | 32 ++++++++++++++++++-------------- 1 files changed, 18 insertions(+), 14 deletions(-)
diff -puN drivers/net/r8169.c~r8169-break-irq-loop drivers/net/r8169.c --- linux-2.6.5-rc2-mm5/drivers/net/r8169.c~r8169-break-irq-loop 2004-03-29 21:45:07.000000000 +0200 +++ linux-2.6.5-rc2-mm5-fr/drivers/net/r8169.c 2004-03-29 21:45:38.000000000 +0200 @@ -1352,21 +1352,21 @@ rtl8169_tx_interrupt(struct net_device * while (tx_left > 0) { int entry = dirty_tx % NUM_TX_DESC; + struct sk_buff *skb = tp->Tx_skbuff[entry]; + u32 status = le32_to_cpu(tp->TxDescArray[entry].status); - if (!(le32_to_cpu(tp->TxDescArray[entry].status) & OWNbit)) { - struct sk_buff *skb = tp->Tx_skbuff[entry]; + if (status & OWNbit) + break; - /* FIXME: is it really accurate for TxErr ? */ - tp->stats.tx_bytes += skb->len >= ETH_ZLEN ? - skb->len : ETH_ZLEN; - tp->stats.tx_packets++; - rtl8169_unmap_tx_skb(tp->pci_dev, tp->Tx_skbuff + entry, - tp->TxDescArray + entry); - dev_kfree_skb_irq(skb); - tp->Tx_skbuff[entry] = NULL; - dirty_tx++; - tx_left--; - } + /* FIXME: is it really accurate for TxErr ? */ + tp->stats.tx_bytes += skb->len >= ETH_ZLEN ? + skb->len : ETH_ZLEN; + tp->stats.tx_packets++; + rtl8169_unmap_tx_skb(tp->pci_dev, tp->Tx_skbuff + entry, + tp->TxDescArray + entry); + dev_kfree_skb_irq(skb); + dirty_tx++; + tx_left--; } if (tp->dirty_tx != dirty_tx) { @@ -1404,6 +1404,7 @@ rtl8169_rx_interrupt(struct net_device * { unsigned long cur_rx, rx_left; int delta; + int max_try = 8192; assert(dev != NULL); assert(tp != NULL); @@ -1412,7 +1413,7 @@ rtl8169_rx_interrupt(struct net_device * cur_rx = tp->cur_rx; rx_left = NUM_RX_DESC + tp->dirty_rx - cur_rx; - while (rx_left > 0) { + while ((rx_left > 0) && --max_try) { int entry = cur_rx % NUM_RX_DESC; u32 status = le32_to_cpu(tp->RxDescArray[entry].status); @@ -1455,6 +1456,9 @@ rtl8169_rx_interrupt(struct net_device * rx_left--; } + if (!max_try) + printk(KERN_INFO "%s: strangeness in Rx handler", dev->name); + tp->cur_rx = cur_rx; delta = rtl8169_rx_fill(tp, dev, tp->dirty_rx, tp->cur_rx); _
| |