[Bug 291105] Intel i350 not receiving arp replies

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=291105

Dave Hayes <[email protected]> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|Affects Only Me             |Affects Some People

--- Comment #3 from Dave Hayes <[email protected]> ---
More data on this bug. I've confirmed that igb0 and igb2, both on different
switches (both physically and model) are not receiving ethernet broadcast
frames. I've done this by verifying other hosts on the same switches do receive
broadcast traffic (so it's there to be received, e.g. arp requests). 

Igb1 and Igb3 (also on different switches from igb0 and igb2) work fine. 

Doing some debugging, I noticed this oddity:

> vmstat -i | grep igb0
irq61: igb0:rxq1                    3766          1
irq62: igb0:rxq2                   15734          3
irq63: igb0:rxq3                   19954          4
irq64: igb0:rxq4                    2355          0
irq65: igb0:rxq5                    5472          1
irq66: igb0:rxq6                   17782          3
irq67: igb0:rxq7                    1385          0
irq68: igb0:aq                        12          0
> vmstat -i | grep igb1
irq69: igb1:rxq0                  640055        118
irq70: igb1:rxq1                   84127         16
irq71: igb1:rxq2                   37207          7
irq72: igb1:rxq3                   73880         14
irq73: igb1:rxq4                   80836         15
irq74: igb1:rxq5                   33638          6
irq75: igb1:rxq6                   47593          9
irq76: igb1:rxq7                   66655         12
irq77: igb1:aq                         4          0

Isn't there supposed to be an rxq0 for igb0? The same difference can be seen
for igb2 vs igb3. 

Also look at this:

> sysctl dev.igb.3.iflib | grep rxq | grep cpu
dev.igb.3.iflib.rxq7.cpu: 30
dev.igb.3.iflib.rxq6.cpu: 28
dev.igb.3.iflib.rxq5.cpu: 26
dev.igb.3.iflib.rxq4.cpu: 24
dev.igb.3.iflib.rxq3.cpu: 22
dev.igb.3.iflib.rxq2.cpu: 20
dev.igb.3.iflib.rxq1.cpu: 18
dev.igb.3.iflib.rxq0.cpu: 16
> sysctl dev.igb.2.iflib | grep rxq | grep cpu
dev.igb.2.iflib.rxq7.cpu: 14
dev.igb.2.iflib.rxq6.cpu: 12
dev.igb.2.iflib.rxq5.cpu: 10
dev.igb.2.iflib.rxq4.cpu: 8
dev.igb.2.iflib.rxq3.cpu: 6
dev.igb.2.iflib.rxq2.cpu: 4
dev.igb.2.iflib.rxq1.cpu: 2
dev.igb.2.iflib.rxq0.cpu: 0
> sysctl dev.igb.1.iflib | grep rxq | grep cpu
dev.igb.1.iflib.rxq7.cpu: 30
dev.igb.1.iflib.rxq6.cpu: 28
dev.igb.1.iflib.rxq5.cpu: 26
dev.igb.1.iflib.rxq4.cpu: 24
dev.igb.1.iflib.rxq3.cpu: 22
dev.igb.1.iflib.rxq2.cpu: 20
dev.igb.1.iflib.rxq1.cpu: 18
dev.igb.1.iflib.rxq0.cpu: 16
> sysctl dev.igb.0.iflib | grep rxq | grep cpu
dev.igb.0.iflib.rxq7.cpu: 14
dev.igb.0.iflib.rxq6.cpu: 12
dev.igb.0.iflib.rxq5.cpu: 10
dev.igb.0.iflib.rxq4.cpu: 8
dev.igb.0.iflib.rxq3.cpu: 6
dev.igb.0.iflib.rxq2.cpu: 4
dev.igb.0.iflib.rxq1.cpu: 2
dev.igb.0.iflib.rxq0.cpu: 0

Does any of this data help to get a fix? :)

-- 
You are receiving this mail because:
You are the assignee for the bug.