On Sun, 4 Mar 2012 09:48:07 +0000
Simon Wilkinson <simonxwilkin...@gmail.com> wrote:
> On 4 Mar 2012, at 04:34, "Chas Williams (CONTRACTOR)" <c...@cmf.nrl.navy.mil>
> wrote:
>
> > this seems to build correctly on centos 6.2 with the 1_6_x branch.
>
> Any work on RX really needs to be based on the current master tree. There's
> quite a lot of work been done there, in particular in terms of structure
> hiding, that are an important precursor to IPv6.
i agree. but i am merely pointing it out as a reference point. i.e. i
dont think the .xg file is broken. it appears that the split/cleanup
of rx header files along with the accessor routines wasnt quite complete
for rxgen. this seems to fix rxgen problems with rx.simple.
diff --git a/src/rxgen/rpc_parse.c b/src/rxgen/rpc_parse.c
index 0ebaefa..0356c2c 100644
--- a/src/rxgen/rpc_parse.c
+++ b/src/rxgen/rpc_parse.c
@@ -1329,8 +1329,8 @@ cs_ProcTail_setup(definition * defp, int split_flag)
} else {
f_print(fout,
"\t rx_RecordCallStatistics(z_call, \n"
- "\t\t(((afs_uint32)(ntohs(z_call->conn->serviceId) << 16))
|\n"
- "\t\t((afs_uint32)ntohs(z_call->conn->peer->port))),\n");
+
"\t\t(((afs_uint32)(ntohs(rx_ServiceIdOf(rx_ConnectionOf(z_call))) << 16)) |\n"
+
"\t\t((afs_uint32)ntohs(rx_PortOf(rx_PeerOf(rx_ConnectionOf(z_call)))))),\n");
}
f_print(fout, "\t\t%d, %sNO_OF_STAT_FUNCS, 1);\n",
no_of_stat_funcs, PackagePrefix[PackageIndex]);
this is starting to look like lisp. there are two other places in rxgen
that needs fixed as well but this is the minimum fix for troy's problem.
> > besides changing sockaddr_in with sockaddr_in6
>
> The real problem with RX is the use of afs_int32 and friends to represent
> network addresses. That's going to need to be replaced with a more generic
> structure (hopefully a union of various types so we can generalise our
> protocol support further in the future)
yeah sockaddr_storage is that union--unless you plan to run afs over
something that doesnt have sockets.
_______________________________________________
OpenAFS-devel mailing list
OpenAFS-devel@openafs.org
https://lists.openafs.org/mailman/listinfo/openafs-devel
