负载过大时，e1000_watchdog 反复 UP、DOWN 的问题
做 NAT 服务器，当负载过大时，总出现Oct 17 14:56:04 server kernel: e1000: eth0: e1000_watchdog: NIC Link is Down
Oct 17 14:56:08 server kernel: e1000: eth0: e1000_watchdog: NIC Link is Up 100 Mbps Full Duplex
Oct 17 14:56:09 server kernel: e1000: eth0: e1000_watchdog: NIC Link is Down
Oct 17 14:56:15 server kernel: e1000: eth0: e1000_watchdog: NIC Link is Up 100 Mbps Full Duplex
网卡反复 UP、DOWN，甚至导致系统自动重启
这个 e1000_watchdog 是做什么用的？为何有如此现象？如何不让他自动 UP、DOWN ？
负载过大时，e1000_watchdog 反复 UP、DOWN 的问题
static void
e1000_watchdog(unsigned long data)
{
struct e1000_adapter *adapter = (struct e1000_adapter *) data;
struct net_device *netdev = adapter->;netdev;
struct e1000_desc_ring *txdr = &adapter->;tx_ring;
unsigned int i;
uint32_t link;
e1000_check_for_link(&adapter->;hw);
if((adapter->;hw.media_type == e1000_media_type_internal_serdes) &&
!(E1000_READ_REG(&adapter->;hw, TXCW) & E1000_TXCW_ANE))
link = !adapter->;hw.serdes_link_down;
else
link = E1000_READ_REG(&adapter->;hw, STATUS) & E1000_STATUS_LU;
if(link) {
if(!netif_carrier_ok(netdev)) {
e1000_get_speed_and_duplex(&adapter->;hw,
&adapter->;link_speed,
&adapter->;link_duplex);
DPRINTK(LINK, INFO, "NIC Link is Up %d Mbps %s\n",
adapter->;link_speed,
adapter->;link_duplex == FULL_DUPLEX ?
"Full Duplex" : "Half Duplex");
netif_carrier_on(netdev);
netif_wake_queue(netdev);
mod_timer(&adapter->;phy_info_timer, jiffies + 2 * HZ);
adapter->;smartspeed = 0;
}
} else {
if(netif_carrier_ok(netdev)) {
adapter->;link_speed = 0;
adapter->;link_duplex = 0;
DPRINTK(LINK, INFO, "NIC Link is Down\n");
netif_carrier_off(netdev);
netif_stop_queue(netdev);
mod_timer(&adapter->;phy_info_timer, jiffies + 2 * HZ);
}
e1000_smartspeed(adapter);
}
e1000_update_stats(adapter);
adapter->;hw.tx_packet_delta = adapter->;stats.tpt - adapter->;tpt_old;
adapter->;tpt_old = adapter->;stats.tpt;
adapter->;hw.collision_delta = adapter->;stats.colc - adapter->;colc_old;
adapter->;colc_old = adapter->;stats.colc;
adapter->;gorcl = adapter->;stats.gorcl - adapter->;gorcl_old;
adapter->;gorcl_old = adapter->;stats.gorcl;
adapter->;gotcl = adapter->;stats.gotcl - adapter->;gotcl_old;
adapter->;gotcl_old = adapter->;stats.gotcl;
e1000_update_adaptive(&adapter->;hw);
if(!netif_carrier_ok(netdev)) {
if(E1000_DESC_UNUSED(txdr) + 1 < txdr->;count) {
/* We've lost link, so the controller stops DMA,
* but we've got queued Tx work that's never going
* to get done, so reset controller to flush Tx.
* (Do the reset outside of interrupt context). */
schedule_work(&adapter->;tx_timeout_task);
}
}
/* Dynamic mode for Interrupt Throttle Rate (ITR) */
if(adapter->;hw.mac_type >;= e1000_82540 && adapter->;itr == 1) {
/* Symmetric Tx/Rx gets a reduced ITR=2000; Total
* asymmetrical Tx or Rx gets ITR=8000; everyone
* else is between 2000-8000. */
uint32_t goc = (adapter->;gotcl + adapter->;gorcl) / 10000;
uint32_t dif = (adapter->;gotcl >; adapter->;gorcl ?
adapter->;gotcl - adapter->;gorcl :
adapter->;gorcl - adapter->;gotcl) / 10000;
uint32_t itr = goc >; 0 ? (dif * 6000 / goc + 2000) : 8000;
E1000_WRITE_REG(&adapter->;hw, ITR, 1000000000 / (itr * 256));
}
/* Cause software interrupt to ensure rx ring is cleaned */
E1000_WRITE_REG(&adapter->;hw, ICS, E1000_ICS_RXDMT0);
/* Early detection of hung controller */
i = txdr->;next_to_clean;
if(txdr->;buffer_info.dma &&
time_after(jiffies, txdr->;buffer_info.time_stamp + HZ) &&
!(E1000_READ_REG(&adapter->;hw, STATUS) & E1000_STATUS_TXOFF))
netif_stop_queue(netdev);
/* Reset the timer */
mod_timer(&adapter->;watchdog_timer, jiffies + 2 * HZ);
}
[ 本帖最后由 platinum 于 2005-11-3 14:24 编辑 ]
负载过大时，e1000_watchdog 反复 UP、DOWN 的问题
if((adapter->;hw.media_type == e1000_media_type_internal_serdes) &&
!(E1000_READ_REG(&adapter->;hw, TXCW) & E1000_TXCW_ANE))
link = !adapter->;hw.serdes_link_down;
else
link = E1000_READ_REG(&adapter->;hw, STATUS) & E1000_STATUS_LU;
这段话什么意思？什么情况会导致 link = 0 ？
如果删除了 else 后面的东西重新编译驱动，会不会有副作用？
负载过大时，e1000_watchdog 反复 UP、DOWN 的问题
watchdog这个好像是检测到坏桢认为可能硬件出现故障复位硬件，不过e1000吃不住流量的情况应该不大，估计是线制作的不合格导致大流量时坏桢严重，或者有电气干扰等。负载过大时，e1000_watchdog 反复 UP、DOWN 的问题
对了，ifconfig看一下是不是相应网卡出现很多errors的包负载过大时，e1000_watchdog 反复 UP、DOWN 的问题
很奇怪呢，RX-ERR RX-DRP RX-OVR，这三项都是 0是这样的，这台机器上还运行其他服务，CPU 一直比较高，会不会是没有过多的 CPU 来负责处理数据包造成的？
还有，我如果修改源代码，注释掉硬件复位那段代码可不可以？会不会有副作用？
负载过大时，e1000_watchdog 反复 UP、DOWN 的问题
cat /proc/interrupts看下是否和这相似，注释掉硬件复位那段代码没问题，CPU0 CPU1 CPU2 CPU3
0:347275720347745526340396987340996245 IO-APIC-edgetimer
1: 284 789 878 556 IO-APIC-edgei8042
8: 0 0 0 1 IO-APIC-edgertc
9: 0 0 0 0 IO-APIC-levelacpi
14: 1 0 20 19 IO-APIC-edgeide0
15: 87104 399199 5819 328406 IO-APIC-edgelibata
169: 0 0 0 0 IO-APIC-leveluhci_hcd
177: 0 0 0 0 IO-APIC-leveluhci_hcd
185: 0 0 0 0 IO-APIC-levelehci_hcd
209: 186220 5466568 211081 5950435 IO-APIC-levelioc0
217: 27184131 0 0 157 IO-APIC-leveleth0
225: 0 0 51742411 29 IO-APIC-leveleth1
NMI: 0 0 0 0
LOC: 1376397708 1376397707 1376397706 1376397705
ERR: 0
MIS: 0
负载过大时，e1000_watchdog 反复 UP、DOWN 的问题
CPU0
0: 5208209 IO-APIC-edgetimer
1: 2 IO-APIC-edgekeyboard
4: 30 IO-APIC-edgeserial
8: 1 IO-APIC-edgertc
9: 0 IO-APIC-levelacpi
15: 108463 IO-APIC-edgeide1
21: 15700851 IO-APIC-leveleth0
22: 15716884 IO-APIC-leveleth1
NMI: 0
LOC: 5208446
ERR: 0
MIS: 0
这样正常吗？
负载过大时，e1000_watchdog 反复 UP、DOWN 的问题
不会是负载过大的原因.我们单位做iptv,负荷比你重多了吧,也是e1000负载过大时，e1000_watchdog 反复 UP、DOWN 的问题
e1000 是千兆的网卡, 100M应该没事呀?watchdog 好象是用来检测什么东东的一个计时, 我觉得是不是你的 switch 受不了呀? 因为 switch 如果不行了, 网卡一样会认为网络 down 呀.