Hi Karen,

I was able to find Dell 1850 server with 82541 LOMs  and installed 2.6.31.1 
kernel on it. With both the in-kernel and the SF driver e1000 8.0.16 I had no 
problems connecting to a Cisco 4948 at GB speeds. So it appears that the issue 
is not generic.

Could you provide the output of:
ethtool ethX
and also
ethtool -d ethX

Preferably after the interface connects at 100F with the driver set to auto.

Are you sure that there are no settings in your configuration files that alter 
the negotiation of the driver (like in /etc/modprobe.conf)?

Thanks,
Emil

________________________________
From: Karen Shaeffer [mailto:[email protected]]
Sent: Wednesday, September 30, 2009 12:19 PM
To: Tantilov, Emil S
Subject: Re: [E1000-devel] 82541GI 1000 Mb link failure

Hi Emil,
I have a Dell Power Edge 1850 here. And the 2.6.28 thru 2.6.31 kernels fail to 
link the Intel 82541GI Gb NIC at 1000 Mb while configured for autoneg. And the 
switch is a Cisco 4948 and also configured for autoneg. It will link if you 
disable autoneg and set the speed at 1000 Mb. But it is a regression, because 
the older kernels did work properly. I believe the failure goes back to the 
2.6.26 kernel, maybe farther back, not sure.

06:07.0 Ethernet controller: Intel Corporation 82541GI Gigabit Ethernet 
Controller (rev 05)
        Subsystem: Dell PRO/1000 MT Network Connection
        Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV+ VGASnoop- ParErr- 
Stepping- SERR+ FastB2B-
        Status: Cap+ 66MHz+ UDF- FastB2B- ParErr- DEVSEL=medium >TAbort- 
<TAbort- <MAbort- >SERR- <PERR-
        Latency: 32 (63750ns min), Cache Line Size: 64 bytes
        Interrupt: pin A routed to IRQ 201
        Region 0: Memory at fe6e0000 (32-bit, non-prefetchable) [size=128K]
        Region 2: I/O ports at ecc0 [size=64]
        Capabilities: [dc] Power Management version 2
                Flags: PMEClk- DSI+ D1- D2- AuxCurrent=0mA 
PME(D0+,D1-,D2-,D3hot+,D3cold+)
                Status: D0 PME-Enable- DSel=0 DScale=1 PME-
        Capabilities: [e4] PCI-X non-bridge device
                Command: DPERE- ERO+ RBC=512 OST=1
                Status: Dev=00:00.0 64bit- 133MHz- SCD- USC- DC=simple 
DMMRBC=2048 DMOST=1 DMCRS=8 RSCEM- 266MHz- 533MHz-

07:08.0 Ethernet controller: Intel Corporation 82541GI Gigabit Ethernet 
Controller (rev 05)
        Subsystem: Dell PRO/1000 MT Network Connection
        Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV+ VGASnoop- ParErr- 
Stepping- SERR+ FastB2B-
        Status: Cap+ 66MHz+ UDF- FastB2B- ParErr- DEVSEL=medium >TAbort- 
<TAbort- <MAbort- >SERR- <PERR-
        Latency: 32 (63750ns min), Cache Line Size: 64 bytes
        Interrupt: pin A routed to IRQ 209
        Region 0: Memory at fe4e0000 (32-bit, non-prefetchable) [size=128K]
        Region 2: I/O ports at dcc0 [size=64]
        Capabilities: [dc] Power Management version 2
                Flags: PMEClk- DSI+ D1- D2- AuxCurrent=0mA 
PME(D0+,D1-,D2-,D3hot+,D3cold+)
                Status: D0 PME-Enable- DSel=0 DScale=1 PME-
        Capabilities: [e4] PCI-X non-bridge device
                Command: DPERE- ERO+ RBC=512 OST=1
                Status: Dev=00:00.0 64bit- 133MHz- SCD- USC- DC=simple 
DMMRBC=2048 DMOST=1 DMCRS=8 RSCEM- 266MHz- 533MHz-

Thanks,
Karen

On Wed, Sep 30, 2009 at 12:06 PM, Tantilov, Emil S 
<[email protected]<mailto:[email protected]>> wrote:
You mean 8.0.16 does not build on 2.6.28? Can you send me the output of the 
error?

As for the issue - I think you're the first to report this bug, so once we root 
cause it we will know for sure as to why and what happened. I have not been 
able to find a system with 82541 LOM, so if you could provide the information I 
asked in the previous email it may be easier for me to come up with repro here 
which would speed up the process of getting it resolved.

Thanks,
Emil

________________________________
From: Karen Shaeffer [mailto:[email protected]<mailto:[email protected]>]
Sent: Wednesday, September 30, 2009 11:59 AM

To: Tantilov, Emil S
Subject: Re: [E1000-devel] 82541GI 1000 Mb link failure

Hi Emil,
Well I am only interested in the 2.6.28 kernel. And it fails to compile for me. 
I hacked up your Makefile so see the source, but it fails with a lot of compile 
time errors.

I am just wondering how in the world Intel could leave this broken in the 
kernel source code driver. Someone must have lost their mind at Intel. This 
situation is not what one has come to expect from Intel with respect to Linux.
Karen

On Wed, Sep 30, 2009 at 11:05 AM, Tantilov, Emil S 
<[email protected]<mailto:[email protected]>> wrote:
The e1000 8.0.16 driver is a standalone (out of tree) driver that should 
compile on most Linux kernels. If you look at the kcompat files you will see 
there is lots of code to help with building on all kinds of kernels.

Could you provide some more information about your system:

1.  ethtool -e eth0
2. lspci -vvv
3. dmesg after loading the driver
4. cat /proc/interrupts

Thanks,
Emil

________________________________
From: Karen Shaeffer [mailto:[email protected]<mailto:[email protected]>]
Sent: Wednesday, September 30, 2009 11:02 AM
To: Tantilov, Emil S
Subject: Re: [E1000-devel] 82541GI 1000 Mb link failure

Hi Emil,
One question. Which canonical 2.6.* kernel versions is the e1000-8.0.16 driver 
compatible with? Do you folks know?

Thanks,
Karen

On Tue, Sep 29, 2009 at 5:06 PM, Karen Shaeffer 
<[email protected]<mailto:[email protected]>> wrote:
Hi Emil,
I did download the e1000-8.0.16 driver from the intel.com<http://intel.com> 
site. And I set E1000_MASTER_SLAVE = 1. This compiled without problems against 
my 2.6.28.9 kernel. But it did not resolve the problem. This driver failed to 
link at 1000 Mb and ended up linked at 100 Mb.

It is interesting the the 2.6.16 e1000 driver does link at 1000 Mb. Can't 
understand how folks let this get broken. Still working towards a solution.

Thanks,
Karen


On Tue, Sep 29, 2009 at 2:15 PM, Tantilov, Emil S 
<[email protected]<mailto:[email protected]>> wrote:
Hi Karen,

Sorry to hear you're having issues with the e1000 adapter. I don't have an 
82541 based LOM in the lab, all the 82541 NICs I tested seem to work fine.

Could you try the out-of-tree e1000 driver from 
e1000.sf.net<http://e1000.sf.net>?

Thanks,
Emil

-----Original Message-----
From: Karen Shaeffer [mailto:[email protected]<mailto:[email protected]>]
Sent: Tuesday, September 29, 2009 12:22 PM
To: [email protected]<mailto:[email protected]>
Subject: [E1000-devel] 82541GI 1000 Mb link failure

Hi,
I am running a vanilla 2.6.28.9 kernel on a server with the Intel 82541GI
NICs integrated onto the system board. The problem is the NICs fail to link
at 1000 Mb, when the switch is set to autonegotiate and the NIC is set to
autonegotiate and running this kernel. This failure did not happen with an
older kernel. (2.6.16)

This is a Cisco 4948 switch I am testing with, but this NIC with this kernel
has failed to link with many switches at 1000 Mb. I put some debugging
statements in the E1000 driver. There are no errors from the PHY, but the
link fails due to the time allotted for link expiring. I increased the
allotted time for the link by 3X, but this had no effect. I set the
E1000_MASTER_SLAVE = 1, for e1000_ms_force_master mode, but even this fails
to help.

the 2.6.28.9 e1000 driver:
# ethtool -i eth0
driver: e1000
version: 7.3.21-k3-NAPI
firmware-version: N/A
bus-info: 0000:06:07.0

The 2.6.16 e1000 driver: (does not suffer the failure)
# ethtool -i eth0
driver: e1000
version: 7.1.9-k4-NAPI
firmware-version: N/A
bus-info: 0000:06:07.0

Any advice on how to resolve this 1000 Mb link failure is appreciated.

Thanks,
Karen




------------------------------------------------------------------------------
Come build with us! The BlackBerry&reg; Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay 
ahead of the curve. Join us from November 9&#45;12, 2009. Register now&#33;
http://p.sf.net/sfu/devconf
_______________________________________________
E1000-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/e1000-devel

Reply via email to