Hal,
I just checked in a new version of uat.c and also the userspace library
(libibat/src/at.c) which should fix this oops.
Thanks, this fixes my oops.
Note that I did not unit test not supplying async completion routine (as
you noticed) so there may be further issues here :-(
I will go back to async mode for now.
This is the safer option right now.
Ok, I now have uDAPL (dtest) up running with IB_CM and AT. :-)
I occassionally see the ib_at_route_by_ip() return a 1 but the ib_route
data is not written out when returned but my separate at_thread that is
waiting on a ib_at_callback_get_timed(-1) seems to pick up the correct
data.
Here is my output from uDAPL.
dapl_ep_connect (0x516c60, {4294967232.4294967208.0.20}, 115D3, 5000000,
0, (nil), 0, 0)
connect: r_SID 71123, pdata (nil), plen 0
connect: conn 0x51a200, cm_id 4
connect: at_route ret=1,Success req_id 0
rt_comp_handler: conn 0x51a200, req_id 0 rec_num 1
rt_comp_handler: SRC GID subnet 0000000000000000 id 0000000000000000
rt_comp_handler: DST GID subnet 0000000000000000 id 0000000000000000
dapl_ep_connect () returns 0x0
3848 Waiting for connect response
dapl_evd_wait (0x5161d0, -1, 1, 0x7ffffffff710, 0x7ffffffff6cc)
dapl_evd_wait: EVD 0x5161d0, CQ (nil), Timeout -1, Threshold 1
Output from at_thread doing the
rt_comp_handler: conn 0x51a200, req_id 0 rec_num 1
rt_comp_handler: SRC GID subnet 00000000000080fe id 9d40000002c90200
rt_comp_handler: DST GID subnet 00000000000080fe id 7140000002c90200
at_thread: callback woke
path_comp_handler: ctxt 0x51a200, req_id 4 rec_num 1
...
Thanks for the help!
-arlin
-- Hal
_______________________________________________
openib-general mailing list
[email protected]
http://openib.org/mailman/listinfo/openib-general
To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
_______________________________________________
openib-general mailing list
[email protected]
http://openib.org/mailman/listinfo/openib-general
To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
