Bug 40572

Summary: Intel Gigabit Ethernet 82576 50% packet loss after reboot
Product: Drivers Reporter: Robert Vojcik (vojcik)
Component: NetworkAssignee: drivers_network (drivers_network)
Status: RESOLVED OBSOLETE    
Severity: blocking CC: akpm, alan
Priority: P1    
Hardware: All   
OS: Linux   
Kernel Version: 3.0 Subsystem:
Regression: Yes Bisected commit-id:
Attachments: ethtool -i etho
dmesg
kernel config
lspci

Description Robert Vojcik 2011-08-05 07:07:03 UTC
Hi, 

I have strange problem with Intel dualport Gigabit ehternet card.
Problem appears after 3rd - 5th reboot.

If you ping or make any network traffic you get 50% packet loss. No error messages in logs.
When you make reboot all is ok in next few reboots.

We have eliminated network problems like switches, cables etc. It's software related.

It looks like in kernel 2.6.37 we have the same problem but in 2.6.28.6 everything looks fine.

I attach some files for additional information
Comment 1 Robert Vojcik 2011-08-05 07:07:40 UTC
Created attachment 67582 [details]
ethtool -i etho
Comment 2 Robert Vojcik 2011-08-05 07:09:05 UTC
Created attachment 67592 [details]
dmesg
Comment 3 Robert Vojcik 2011-08-05 07:09:55 UTC
Created attachment 67602 [details]
kernel config
Comment 4 Robert Vojcik 2011-08-05 07:11:00 UTC
Created attachment 67612 [details]
lspci
Comment 5 Andrew Morton 2011-08-23 21:30:56 UTC
(switched to email.  Please respond via emailed reply-to-all, not via the
bugzilla web interface).

On Fri, 5 Aug 2011 07:07:05 GMT
bugzilla-daemon@bugzilla.kernel.org wrote:

> https://bugzilla.kernel.org/show_bug.cgi?id=40572
> 
>            Summary: Intel Gigabit Ethernet 82576 50% packet loss after
>                     reboot
>            Product: Drivers
>            Version: 2.5
>     Kernel Version: 3.0
>           Platform: All
>         OS/Version: Linux
>               Tree: Mainline
>             Status: NEW
>           Severity: blocking
>           Priority: P1
>          Component: Network
>         AssignedTo: drivers_network@kernel-bugs.osdl.org
>         ReportedBy: vojcik@gmail.com
>         Regression: No

I'll change this to "yes".

> 
> Hi, 
> 
> I have strange problem with Intel dualport Gigabit ehternet card.
> Problem appears after 3rd - 5th reboot.
> 
> If you ping or make any network traffic you get 50% packet loss. No error
> messages in logs.
> When you make reboot all is ok in next few reboots.
> 
> We have eliminated network problems like switches, cables etc. It's software
> related.
> 
> It looks like in kernel 2.6.37 we have the same problem but in 2.6.28.6
> everything looks fine.
> 
> I attach some files for additional information
>
Comment 6 Alexander Duyck 2011-08-24 21:54:47 UTC
On 08/23/2011 02:30 PM, Andrew Morton wrote:
> (switched to email.  Please respond via emailed reply-to-all, not via the
> bugzilla web interface).
>
> On Fri, 5 Aug 2011 07:07:05 GMT
> bugzilla-daemon@bugzilla.kernel.org wrote:
>
>> https://bugzilla.kernel.org/show_bug.cgi?id=40572
>>
>>             Summary: Intel Gigabit Ethernet 82576 50% packet loss after
>>                      reboot
>>             Product: Drivers
>>             Version: 2.5
>>      Kernel Version: 3.0
>>            Platform: All
>>          OS/Version: Linux
>>                Tree: Mainline
>>              Status: NEW
>>            Severity: blocking
>>            Priority: P1
>>           Component: Network
>>          AssignedTo: drivers_network@kernel-bugs.osdl.org
>>          ReportedBy: vojcik@gmail.com
>>          Regression: No
> I'll change this to "yes".
>
>> Hi,
>>
>> I have strange problem with Intel dualport Gigabit ehternet card.
>> Problem appears after 3rd - 5th reboot.
>>
>> If you ping or make any network traffic you get 50% packet loss. No error
>> messages in logs.
>> When you make reboot all is ok in next few reboots.
>>
>> We have eliminated network problems like switches, cables etc. It's software
>> related.
>>
>> It looks like in kernel 2.6.37 we have the same problem but in 2.6.28.6
>> everything looks fine.
>>
>> I attach some files for additional information
This type of issue is typically a sign of a hardware problem.  I would 
recommend doing an lspci -vvv for the device in both the working and the 
non-working cases to see if there is any difference between the two.

One thing we have seen in the past is an issue where the PCIe will not 
link at x4 in all cases and will sometimes link at only x1.  When this 
occurs the device does not have enough PCIe bandwidth to handle heavy 
workloads.  You might want to try either reseating the network adapter 
into the slot or moving it from one PCIe slot to another in the system 
as it is possible the PCIe slot it is in may have an issue with one ore 
more of the PCIe lanes.

Thanks,

Alex