Re: [3.5/6.x][e1000e] - Regression - Unable to receive packets if MTU == 1500

From: Shawn Starr
Date: Mon Dec 03 2012 - 01:43:20 EST


On Thursday, November 01, 2012 02:01:05 PM Shawn Starr wrote:
> Hello network driver folks,

To follow up, this is fixed now in 3.6.7-5.fc18.x86_64

[38936.525226] e1000e 0000:00:19.0: eth0: changing MTU from 9000 to 1500
[38936.539510] e1000e 0000:00:19.0: Interrupt Throttle Rate turned on
[38939.352053] e1000e: eth0 NIC Link is Up 1000 Mbps Full Duplex, Flow
Control: Rx/Tx

I don't remember which 3.6 -rc fixed it but I saw some adjustments in e1000e
threshold changes in Linus's usual announce -rc emails.

Thanks for fixing! Now onto Fedora 19 (3.7) kernel testing

Shawn.

>
> I recently decided to reinstall my Lenovo W500 laptop and found I wasn't
> able to get DHCP leases, I wasn't able to install over PXE (when getting
> the IP a second time within the OS)
>
> Fedora is currently using kernel-3.6.2-2.fc18.x86_64 (Pre-beta)
>
> The only difference I noticed was in my old install of Fedora, I had added
> MTU=9000 to the interface config while using a 3.5 kernel, not knowing 3.6.x
> seems to break things.
>
> My Cisco DPC3825 Switch/Cable modem has an MTU setting but this is from
> Cable provider <-> WAN interface only. I tried changing to 1500 but by
> default it uses this anyway.
>
> I had noticed tx unit hang errors sometimes, but not when MTU is above 1500.
>
> PCI info:
>
> 00:19.0 Ethernet controller: Intel Corporation 82567LM Gigabit Network
> Connection (rev 03)
> Subsystem: Lenovo Device 20ee
> Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr-
> Stepping- SERR+ FastB2B- DisINTx+
> Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort-
> <TAbort- <MAbort- >SERR- <PERR- INTx-
> Latency: 0
> Interrupt: pin A routed to IRQ 46
> Region 0: Memory at fc000000 (32-bit, non-prefetchable) [size=128K]
> Region 1: Memory at fc025000 (32-bit, non-prefetchable) [size=4K]
> Region 2: I/O ports at 1840 [size=32]
> Capabilities: [c8] Power Management version 2
> Flags: PMEClk- DSI+ D1- D2- AuxCurrent=0mA
> PME(D0+,D1-,D2-,D3hot+,D3cold+)
> Status: D0 NoSoftRst- PME-Enable- DSel=0 DScale=1 PME-
> Capabilities: [d0] MSI: Enable+ Count=1/1 Maskable- 64bit+
> Address: 00000000fee0100c Data: 41a2
> Capabilities: [e0] PCI Advanced Features
> AFCap: TP+ FLR+
> AFCtrl: FLR-
> AFStatus: TP-
> Kernel driver in use: e1000e
>
> e1000e conversation log:
>
> [ 12.653816] e1000e: Intel(R) PRO/1000 Network Driver - 2.0.0-k
> [ 12.653819] e1000e: Copyright(c) 1999 - 2012 Intel Corporation.
> [ 12.653855] e1000e 0000:00:19.0: setting latency timer to 64
> [ 12.653938] e1000e 0000:00:19.0: Interrupt Throttling Rate (ints/sec) set
> to dynamic conservative mode
> [ 12.653980] e1000e 0000:00:19.0: irq 46 for MSI/MSI-X
> [ 12.844121] e1000e 0000:00:19.0: eth0: (PCI Express:2.5GT/s:Width x1)
> 00:22:68:0c:96:e3
> [ 12.844124] e1000e 0000:00:19.0: eth0: Intel(R) PRO/1000 Network
> Connection [ 12.844150] e1000e 0000:00:19.0: eth0: MAC: 7, PHY: 8, PBA
> No: 1008FF-0FF [ 17.459570] e1000e 0000:00:19.0: eth0: changing MTU from
> 1500 to 9000 [ 17.459577] e1000e 0000:00:19.0: Interrupt Throttle Rate
> turned off [ 17.948121] e1000e 0000:00:19.0: irq 46 for MSI/MSI-X
> [ 18.049164] e1000e 0000:00:19.0: irq 46 for MSI/MSI-X
> [ 20.442936] e1000e: eth0 NIC Link is Up 1000 Mbps Full Duplex, Flow
> Control: Rx/Tx
>
> When I set MTU = 1500 (2000 works), it will stop receiving packets, it also
> disables Interrupt Throttling when MTU is set higher.
>
> After testing reasons why it wasn't get a DHCP lease, I decided to bump the
> MTU to 9000 for Jumbo Frames and as soon as I did this, I was able to get a
> lease and the network functioned resumes receiving packets.
>
> Can anyone else report regressions or might know whats going on?
>
> Thanks,
> Shawn.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/