[Bug 234766] em(4) Intel 82579LM regression on Supermicro X9SCM-F

[bugzilla-noreply at freebsd.org]

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=234766

freebsd at coombscloud.com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |freebsd at coombscloud.com

--- Comment #7 from freebsd at coombscloud.com ---
I'm seeing similar behavior on a different Intel chip:

$ freebsd-version -kr
12.0-RELEASE-p4
12.0-RELEASE-p4

$ pciconf -lv
...
em0 at pci0:0:31:6:        class=0x020000 card=0x86721043 chip=0x15b88086 rev=0x31
hdr=0x00
    vendor     = 'Intel Corporation'
    device     = 'Ethernet Connection (2) I219-V'
    class      = network
    subclass   = ethernet

I will suddenly get the following kernel messages (the avail and pidx values
differ at each failure):
   kernel: em0: TX(0) desc avail = 42, pidx = 208
   kernel: em0: link state changed to DOWN

Followed by a constant stream of the following messages:
   kernel: em0: TX(0) desc avail = 1024, pidx = 0

This seems to be triggered by periods of high network activity. For example,
another computer has a backup task set to run once daily, which backs up a
SAMBA share hosted by this computer. This network failure often happens shortly
after that task begins.

I used 11.2-RELASE on this computer for some time and never encountered this
behavior with it.

I'll give net/intel-em-kmod a try.

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