Hi, I'm attempting to test my driver with IPv6 on a pair of snv_95 machines.
If I simply do 'ifconfig myri10ge0 inet6 plumb' on a pair of machines, and take the inet6 address, I can successfully ping via IPv6 However, telnet / rsh / netperf / etc all hang when attempting to open the connection. If I snoop on the receiver, I see the the initial TCP packet. Everything looks fine, there is no checksum problem, etc. However, the stack never answers, and seems to drop it before it makes it to TCP. If I take a diff of netstat -s before and after, I don't see any error counters, just an increment of ipv6InReceives. Eg: # diff /tmp/a /tmp/b 55c55 < ipv6InReceives = 354 ipv6InHdrErrors = 0 --- > ipv6InReceives = 355 ipv6InHdrErrors = 0 If I dtrace starting at ip_rput_data_v6() (which is where I think IPv6 packets go), I see: CPU FUNCTION 3 -> ip_rput_data_v6 call: 0xffffff01d8e397e8, 0xffffff01d8c03928 0xffffff01e696f960 0xffffff01de98a300 3 -> ire_ctable_lookup_v6 call: 0xffffff01de98a318, 0x0 0x24 0xffffff01fdf22888 3 -> ire_match_args_v6 call: 0xffffff025a751af0, 0xffffff01de98a318 0xffffff025a751b84 0x0 3 -> ire_to_ill call: 0xffffff025a751af0, 0xffffff01de98a318 0xffffff025a751b84 0x0 3 <- ire_to_ill ret : 0x88, 0xffffff01d8c03928 3 <- ire_match_args_v6 ret : 0x477, 0x1 3 <- ire_ctable_lookup_v6 ret : 0x124, 0xffffff025a751af0 3 -> ire_refrele call: 0xffffff025a751af0, 0x1 0x0 0xffffff0008035c80 3 <- ire_refrele ret : 0x36, 0x1 3 -> freemsg call: 0xffffff01e696f960, 0xffffff01eb5e1f60 0xffdd 0x551c 3 -> dblk_lastfree call: 0xffffff01e696f960, 0xffffff01de98a280 0xffdd 0x551c 3 -> kmem_cache_free call: 0xffffff01cb234360, 0xffffff01de98a280 0xffdd 0x551c 3 <- kmem_cache_free ret : 0x1a8, 0x23 3 <- dblk_lastfree ret : 0x6f, 0x23 3 <- freemsg ret : 0x94, 0x23 Does anybody know why the packets might be dropped with no reply? FWIW, doing the same thing on bge0 works just fine. And I've unplumbed inet6 on bge0 to ensure that routes are correct. Thanks, Drew _______________________________________________ networking-discuss mailing list [email protected]
