RoCE mckey hangs in Ubuntu 15.04 (Mellanox)

Bug #1422002 reported by bugproxy
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
librdmacm (Ubuntu)
Fix Released
Undecided
Unassigned

Bug Description

---Problem Description---
mckey test hangs in RoCE.
It does not receive the multicast packets.

---uname output---
Linux powerio-le21 3.18.0-12-generic #13-Ubuntu SMP Thu Jan 29 13:44:26 UTC 2015 ppc64le ppc64le ppc64le GNU/Linux

---Additional Hardware Info---
Need RoCE adapter like Mellanox Connect 3 adapter.

Machine Type = P8

---Steps to Reproduce---
 in one machine run
mckey -m 224.0.0.10 -b 20.20.20.21

and other machine run
mckey -m 224.0.0.10 -b 20.20.20.20 -s

then second machine will see this:
mckey -m 224.0.0.10 -b 20.20.20.20 -s
mckey: starting client
mckey: joining
mckey: joined dgid: ff0e::ffff:e000:a mlid 0x0 sl 0
initiating data transfers
data transfers complete
test complete
return status 0
 but the first machine will just hang waiting for packets:
mckey -m 224.0.0.10 -b 20.20.20.21
mckey: starting server
mckey: joining
mckey: joined dgid: ff0e::ffff:e000:a mlid 0x0 sl 0
receiving data transfers

There is an available patch: https://patchwork.kernel.org/patch/5507181/ to solve this issue.

bugproxy (bugproxy)
tags: added: architecture-ppc64le bugnameltc-121679 severity-medium targetmilestone-inin1504
Luciano Chavez (lnx1138)
affects: ubuntu → librdmacm (Ubuntu)
Revision history for this message
bugproxy (bugproxy) wrote : Comment bridged from LTC Bugzilla

------- Comment From <email address hidden> 2016-02-10 05:52 EDT-------
Hi Canonical Team

Can we get attention on this bug so as to take it to a suitable closure.
please let us know which release having fixed patch for this bug

Thanks for your support

bugproxy (bugproxy)
tags: added: targetmilestone-inin1610
removed: targetmilestone-inin1504
Revision history for this message
Christian Ehrhardt  (paelzer) wrote :

That is:
commit e9a7faf11af94957e5107b40af46c2e329541510
Author: Or Gerlitz <email address hidden>
Date: Wed Dec 17 16:17:34 2014 +0200

    IB/mlx4: Fix wrong usage of IPv4 protocol for multicast attach/detach

    The MLX4_PROT_IB_IPV4 protocol should only be used with RoCEv2 and such.
    Removing this wrong usage allows to run multicast applications over RoCE.

git tag --contains e9a7faf11af94957e5107b40af46c2e329541510 | grep -v rc
v4.0
v4.1
v4.2
v4.3
v4.4
v4.5
v4.6
v4.7

So fixed since Wily:
15.10 Wily Werewolf 4.2
16.04 Xenial Xerus 4.4
16.10 Yakkety Yak 4.8

Changed in librdmacm (Ubuntu):
status: New → Fix Released
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.