lots of bad checksum on igb
sepherosa at gmail.com
Fri May 23 01:10:13 PDT 2014
On Fri, May 23, 2014 at 12:36 AM, k simon <chio1990 at gmail.com> wrote:
> I noticed that Pyun have fixed some bugs about TSO&.CHECKSUM offload
> issues on FreeBSD recently. Does the same issues happend on Dfly ?
I don't aware we have any TSO and TX hardware CSUM issues. And I
don't think our TSO, especially how TCP/IP stack hands TSO info to
drivers and how many TCP segments should be contained in one "TSO
packet" are completely different from FreeBSD's.
> 于 14-4-18 10:12, k simon 写道:
>> And I found lo0 reported bad checksum too.
>> 于 14-4-18 9:38, k simon 写道:
>>> When it report bad checksum, the web pages can not opened. If
>>> disabled the HW checksum, then it acts normally. I think it's a real
>>> problem and I would take a bce card try again.
>>> 于 14-4-18 2:02, Matthew Dillon 写道:
>>>> Bad checksums reported by tcpdump are not necessarily actually bad. What
>>>> is happening is that the checksum field is not updated when HW checksums
>>>> are enabled... the code depends on the HW to tell it if the checksum is
>>>> bad or not (on receive), and depends on the HW to generate the checksum
>>>> (on transmit). So the packet in the mbuf which tcpdump is parsing may
>>>> not have a filled-in checksum field.
>>>> If you are getting actual communications glitches, e.g. pings not
>>>> returning, tcp throughput problems, and so forth, then you might have a
>>>> real problem.
Tomorrow Will Never Die
More information about the Users