Chris Caputo wrote:
> Just a heads-up on a possible e1000 or e1000e driver issue on linux
> kernel
> 2.6.27.10 while running IPTraf 3.0.0.
>
> I was playing around with this program. The different options, such
> as "IP traffic monitor" seemed to work fine. When I tried the "LAN
> Station
> Monitor" option the server crashed and rebooted. No kernel messages
> were
> seen.
>
> When the server rebooted, I tried IPTraf's "IP traffic monitor" (all
> interfaces) again. This time I noticed the following in my dmesg
> output:
Can you try rebuilding the e1000/e1000e driver with NAPI enabled?
e1000:
# make CFLAGS_EXTRA=-DE1000_NAPI install
e1000e:
# make CFLAGS_EXTRA=-DE1000E_NAPI install
not sure, but it seems that you're running in "non-napi" mode, which is
generally not recommended. dmesg will tell you which way the driver is
configured, usually.
That said, I'm almost positive that there is some kind of an imbalance for
e1000e's pci_map/unmap calls, which is triggering IOMMU's to freak out.
> [ 1702.975867] ------------[ cut here ]------------
> [ 1702.975952] WARNING: at kernel/softirq.c:136
> local_bh_enable+0x3c/0x94() [ 1702.976040] Modules linked in:
> w83627hf_wdt e1000 e1000e [last unloaded: microcode] [ 1702.976190]
> Pid: 0, comm: swapper Not tainted 2.6.27.10 #1 [ 1702.976269]
> [ 1702.976269] Call Trace:
> [ 1702.976386] <IRQ> [<ffffffff8022e1a2>] warn_on_slowpath+0x58/0x7f
> [ 1702.976475] [<ffffffff8033073c>] ?
> swiotlb_map_single_attrs+0x3b/0xb4 [ 1702.976562]
> [<ffffffff8021ceac>] ? swiotlb_map_single_phys+0x0/0x18 [
> 1702.976647] [<ffffffff8032fed9>] ?
> swiotlb_dma_mapping_error+0x18/0x25 [ 1702.976734]
> [<ffffffff8021ceac>] ? swiotlb_map_single_phys+0x0/0x18 [
> 1702.976828] [<ffffffffa0006b88>] ?
> e1000_alloc_rx_buffers+0x78/0x2a0 [e1000e] [ 1702.976962]
> [<ffffffff80232c48>] local_bh_enable+0x3c/0x94 [ 1702.977045]
> [<ffffffff804b2572>] packet_rcv+0xb7/0x2c9 [ 1702.977124]
> [<ffffffff803eb2f3>] netif_nit_deliver+0x6d/0x90 [ 1702.977207]
> [<ffffffff804c6c68>] __vlan_hwaccel_rx+0x7b/0x193 [ 1702.977302]
> [<ffffffffa00297b3>] e1000_clean_rx_irq+0x4d3/0x4e0 [e1000] [
> 1702.977395] [<ffffffffa00280b8>] e1000_intr+0x328/0x720 [e1000] [
> 1702.977479] [<ffffffff80242ca6>] ? hrtimer_get_next_event+0xa3/0xb3
> [ 1702.977566] [<ffffffff802118a9>] ? read_tsc+0x9/0x1c [
> 1702.977643] [<ffffffff80242a2d>] ? ktime_get_ts+0x49/0x4e [
> 1702.977725] [<ffffffff8025516e>] handle_IRQ_event+0x27/0x57 [
> 1702.977806] [<ffffffff802567e0>] handle_fasteoi_irq+0x98/0xd3 [
> 1702.977889] [<ffffffff8020e6e5>] do_IRQ+0x6d/0xd4 [ 1702.977965]
> [<ffffffff8020b7b1>] ret_from_intr+0x0/0xa [ 1702.978043] <EOI>
> [<ffffffff802123cb>] ? mwait_idle+0x45/0x4a [ 1702.978128]
> [<ffffffff80209be6>] ? enter_idle+0x22/0x24 [ 1702.978206]
> [<ffffffff8020a0a1>] ? cpu_idle+0x88/0xae [ 1702.978286]
> [<ffffffff804ee956>] ? start_secondary+0x157/0x15c [ 1702.978367] [
> 1702.978423] ---[ end trace 6f172ff6a7c67481 ]---
>
> Maybe the above is related to the crash...
>
> kernel/softirq.c:136 is:
>
> 134: static inline void _local_bh_enable_ip(unsigned long ip)
> 135: {
> 136: WARN_ON_ONCE(in_irq() || irqs_disabled());
>
> Hardware details below.
>
> Thanks,
> Chris
------------------------------------------------------------------------------
_______________________________________________
E1000-devel mailing list
E1000-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/e1000-devel
