Hi,
just recently we bought DNA enabled driver for e1000e (Intel 82571EB),
and decided to implement pf_ring as part of our test solution.
But we have strange compile issues on Dell poweredge 1950 (2x Intel Dual
Core 3.0).
Operating system installed is Gentoo Linux 64 bit, compiled kernel is
2.6.39-r3. When we compile it it finishes normally, then I can load
pf_ring with:
| insmod pf_ring.ko transparent_mode=2 enable_tx_capture=0 quick_mode=1|
and it loads normally:
cat /proc/net/pf_ring/info
PF_RING Version : 4.7.3 ($Revision: $)
Ring slots : 4096
Slot version : 13
Capture TX : No [RX only]
IP Defragment : No
Socket Mode : Quick
Transparent mode : No (mode 2)
Total rings : 0
Total plugins : 0
Then trying to insert e1000e driver spawns this:
[ 882.456375] e1000e: Intel(R) PRO/1000 Network Driver - 1.3.10a-NAPI
[ 882.456378] e1000e: Copyright(c) 1999 - 2011 Intel Corporation.
[ 882.456410] e1000e 0000:0b:00.0: Disabling ASPM L1
[ 882.456425] e1000e 0000:0b:00.0: PCI INT A -> GSI 16 (level, low) ->
IRQ 16
[ 882.456438] e1000e 0000:0b:00.0: setting latency timer to 64
[ 882.456485] e1000e 0000:0b:00.0: PCI INT A disabled
[ 882.456489] e1000e: probe of 0000:0b:00.0 failed with error -5
[ 882.456494] e1000e 0000:0b:00.1: Disabling ASPM L1
[ 882.456507] e1000e 0000:0b:00.1: PCI INT B -> GSI 17 (level, low) ->
IRQ 17
[ 882.456514] e1000e 0000:0b:00.1: setting latency timer to 64
[ 882.456549] e1000e 0000:0b:00.1: PCI INT B disabled
[ 882.456552] e1000e: probe of 0000:0b:00.1 failed with error -5
But if I compile the same driver on my home machine and just transfer it
to the server:
[ 1720.874465] e1000e: Intel(R) PRO/1000 Network Driver - 1.3.10a-NAPI
[ 1720.874468] e1000e: Copyright(c) 1999 - 2011 Intel Corporation.
[ 1720.874500] e1000e 0000:0b:00.0: Disabling ASPM L1
[ 1720.874515] e1000e 0000:0b:00.0: PCI INT A -> GSI 16 (level, low) ->
IRQ 16
[ 1720.874529] e1000e 0000:0b:00.0: setting latency timer to 64
[ 1720.874636] e1000e 0000:0b:00.0: irq 106 for MSI/MSI-X
[ 1721.029675] e1000e 0000:0b:00.0: eth2: (PCI Express:2.5GB/s:Width x4)
00:15:17:3d:14:a2
[ 1721.029677] e1000e 0000:0b:00.0: eth2: Intel(R) PRO/1000 Network
Connection
[ 1721.029749] e1000e 0000:0b:00.0: eth2: MAC: 1, PHY: 4, PBA No: C57721-005
[ 1721.029758] e1000e 0000:0b:00.1: Disabling ASPM L1
[ 1721.029765] e1000e 0000:0b:00.1: PCI INT B -> GSI 17 (level, low) ->
IRQ 17
[ 1721.029775] e1000e 0000:0b:00.1: setting latency timer to 64
[ 1721.029859] e1000e 0000:0b:00.1: irq 107 for MSI/MSI-X
[ 1721.038652] net.sh used greatest stack depth: 3768 bytes left
[ 1721.184663] e1000e 0000:0b:00.1: eth3: (PCI Express:2.5GB/s:Width x4)
00:15:17:3d:14:a3
[ 1721.184665] e1000e 0000:0b:00.1: eth3: Intel(R) PRO/1000 Network
Connection
[ 1721.184741] e1000e 0000:0b:00.1: eth3: MAC: 1, PHY: 4, PBA No: C57721-005
[ 1733.298130] [DNA] Enabled DNA on eth2 (len=65536)
[ 1733.298214] e1000e 0000:0b:00.0: irq 106 for MSI/MSI-X
[ 1733.349077] e1000e 0000:0b:00.0: irq 106 for MSI/MSI-X
[ 1733.350240] ADDRCONF(NETDEV_UP): eth2: link is not ready
[ 1735.427902] e1000e: eth2 NIC Link is Up 1000 Mbps Full Duplex, Flow
Control: Rx
[ 1735.428689] ADDRCONF(NETDEV_CHANGE): eth2: link becomes ready
[ 1735.626133] [DNA] Enabled DNA on eth3 (len=65536)
[ 1735.626213] e1000e 0000:0b:00.1: irq 107 for MSI/MSI-X
[ 1735.677083] e1000e 0000:0b:00.1: irq 107 for MSI/MSI-X
[ 1735.678220] ADDRCONF(NETDEV_UP): eth3: link is not ready
[ 1737.727880] e1000e: eth3 NIC Link is Up 1000 Mbps Full Duplex, Flow
Control: None
[ 1737.728683] ADDRCONF(NETDEV_CHANGE): eth3: link becomes ready
so I can normally load DNA driver and it seems to work.
Then we tested using pfcount application and it shows 350Mbit of traffic
completely normally and 0% packet loss, but then headache starts:
when I compile libpcap included with pf_ring and tcpdump, tcpdump
segfaults with following:
[ 9554.546293] test[19674]: segfault at 0 ip 00007f1475c88f7e sp
00007fffd0152100 error 6 in libpcap.so.1.1.1[7f1475c5d000+46000
Tried compiling same thing on my own machine, and just transfering data
remote, but that doesnt help, tried running locally no segfaults. I
though OK let's change NIC, didn't helped, both kernel and OS and
packages are the same, different things are CPU and few more additional
packages on my home computer. gcc, g++, make all are the same version.
Also pfcount_multichannel segfaults immediatelly.
Output from pfcount:
=========================
Absolute Stats: [2095523 pkts rcvd][0 pkts dropped]
Total Pkts=2095523/Dropped=0.0 %
2'095'523 pkts - 491'997'002 bytes [190'769.20 pkt/sec - 358.32 Mbit/sec]
=========================
Actual Stats: 166996 pkts [983.14 ms][169'860.18 pkt/sec]
=========================
Should I try some compile switches, I am slowly loosing ideas?
Tnx in advance,
J
_______________________________________________
Ntop-misc mailing list
[email protected]
http://listgateway.unipi.it/mailman/listinfo/ntop-misc
