Joe Perches wrote:
ib_gid's can be print'd using the new %p6 facility
Joe, Harvey
One thing which I'd like to find an easy way to get rid of, is the wrong printing of IPoIB devices HW address under bonding. The bonding driver uses the DECLARE_MAC_BUF and print_mac() way so a possible solution would be to enhance this two to work with variable HW address lengths, but this would have the price of changing the 600 or so references to print_mac in the code, etc. A possible solution would be to just change the bonding driver to use some other macro/api, but I believe that the posted patches could not serve for that purpose as is, since the IPoIB HW address is actually more then a GID, its 20 bytes whose lower 16 are a GID. Printing only the GID portion of the HW address would be better then the current situation, so we can take that approach as well.
Or. _______________________________________________ general mailing list [email protected] http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
