On 08/28/2014 02:55 PM, Alexander Duyck wrote: > The problem is you would need to be sending the data though a socket > type that can actually request a checksum offload. So for example if > you were actually sending the data through a TCP socket then it can > request such an offload.
Ah, I see -- thanks. > If you are just needing a capture with correct checksums you might > consider disabling the TSO, GRO, LRO, and Tx checksum offloads prior to > capturing the trace. This way the software stack should generate the > checksums prior to sending the frames. Yeah. My plan was somewhat different: I was hoping to be able to leave checksums uncorrected in a slightly modified version of tcpreplay-edit, leaving the checksum computation to the NIC. The checksum computations seem to be the biggest bottleneck in tcpreplay in this setting. Thanks again, and apologies for spamming your list with non-intel specific stuff. Cheers, -C. ------------------------------------------------------------------------------ Slashdot TV. Video for Nerds. Stuff that matters. http://tv.slashdot.org/ _______________________________________________ E1000-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/e1000-devel To learn more about Intel® Ethernet, visit http://communities.intel.com/community/wired
