Bug 1452102

Assigning this to Nithya, as she is working on this.

RCA:

(gdb) bt
#0  0x00007fd6bcfbdfb8 in dht_selfheal_dir_setattr (frame=0x7fd6a8000e50, loc=0x7fd6a8000f68, stbuf=0x7fd6a8000ff8, valid=-1, layout=0x7fd6b8003500)
    at /home/jenkins/root/workspace/centos6-regression/xlators/cluster/dht/src/dht-selfheal.c:1180
#1  0x00007fd6bcfbf994 in dht_selfheal_dir_mkdir (frame=0x7fd6a8000e50, loc=0x7fd6a8000f68, layout=0x7fd6b8003500, force=0)
    at /home/jenkins/root/workspace/centos6-regression/xlators/cluster/dht/src/dht-selfheal.c:1512
#2  0x00007fd6bcfc162a in dht_selfheal_directory (frame=0x7fd6a8000e50, dir_cbk=0x7fd6bcfd1b3c <dht_lookup_selfheal_cbk>, loc=0x7fd6a8000f68, layout=0x7fd6b8003500)
    at /home/jenkins/root/workspace/centos6-regression/xlators/cluster/dht/src/dht-selfheal.c:2167
#3  0x00007fd6bcfd4d98 in dht_lookup_dir_cbk (frame=0x7fd6a8000e50, cookie=0x7fd6b800a860, this=0x7fd6b800cb50, op_ret=0, op_errno=22, inode=0x7fd6a8000c50, 
    stbuf=0x7fd6bdeba840, xattr=0x7fd6b801c700, postparent=0x7fd6bdeba7d0) at /home/jenkins/root/workspace/centos6-regression/xlators/cluster/dht/src/dht-common.c:928
#4  0x00007fd6bd2868ec in client3_3_lookup_cbk (req=0x7fd6b8001cf0, iov=0x7fd6b8001d30, count=1, myframe=0x7fd6b8003870)
    at /home/jenkins/root/workspace/centos6-regression/xlators/protocol/client/src/client-rpc-fops.c:2867
#5  0x00007fd6ca90984d in rpc_clnt_handle_reply (clnt=0x7fd6b8021ee0, pollin=0x7fd6b8001ac0)
    at /home/jenkins/root/workspace/centos6-regression/rpc/rpc-lib/src/rpc-clnt.c:778
#6  0x00007fd6ca909e17 in rpc_clnt_notify (trans=0x7fd6b80220b0, mydata=0x7fd6b8021f10, event=RPC_TRANSPORT_MSG_RECEIVED, data=0x7fd6b8001ac0)
    at /home/jenkins/root/workspace/centos6-regression/rpc/rpc-lib/src/rpc-clnt.c:971
#7  0x00007fd6ca905dac in rpc_transport_notify (this=0x7fd6b80220b0, event=RPC_TRANSPORT_MSG_RECEIVED, data=0x7fd6b8001ac0)
    at /home/jenkins/root/workspace/centos6-regression/rpc/rpc-lib/src/rpc-transport.c:538
#8  0x00007fd6bf6eb58a in socket_event_poll_in (this=0x7fd6b80220b0, notify_handled=_gf_true)
    at /home/jenkins/root/workspace/centos6-regression/rpc/rpc-transport/socket/src/socket.c:2315
#9  0x00007fd6bf6ebbd5 in socket_event_handler (fd=12, idx=6, gen=1, data=0x7fd6b80220b0, poll_in=1, poll_out=0, poll_err=0)
    at /home/jenkins/root/workspace/centos6-regression/rpc/rpc-transport/socket/src/socket.c:2467
#10 0x00007fd6cabb4d9e in event_dispatch_epoll_handler (event_pool=0x119ffc0, event=0x7fd6bdebae70)
    at /home/jenkins/root/workspace/centos6-regression/libglusterfs/src/event-epoll.c:572
#11 0x00007fd6cabb50a0 in event_dispatch_epoll_worker (data=0x11e70b0) at /home/jenkins/root/workspace/centos6-regression/libglusterfs/src/event-epoll.c:648
#12 0x00007fd6c9e1caa1 in start_thread () from ./lib64/libpt




(gdb) f 0
#0  0x00007fd6bcfbdfb8 in dht_selfheal_dir_setattr (frame=0x7fd6a8000e50, loc=0x7fd6a8000f68, stbuf=0x7fd6a8000ff8, valid=-1, layout=0x7fd6b8003500)
    at /home/jenkins/root/workspace/centos6-regression/xlators/cluster/dht/src/dht-selfheal.c:1180
1180	                        gf_msg_trace (this->name, 0,
(gdb) l
1175	                gf_uuid_copy (loc->gfid, local->gfid);
1176	
1177	        local->call_cnt = missing_attr;
1178	        for (i = 0; i < layout->cnt; i++) {
1179	                if (layout->list[i].err == -1) {
1180	                        gf_msg_trace (this->name, 0,
1181	                                      "%s: setattr on subvol %s, gfid = %s",
1182	                                      loc->path, layout->list[i].xlator->name,
1183	                                      uuid_utoa(loc->gfid));
1184	
1185	                        STACK_WIND (frame, dht_selfheal_dir_setattr_cbk,
1186	                                    layout->list[i].xlator,
1187	                                    layout->list[i].xlator->fops->setattr,
1188	                                    loc, stbuf, valid, NULL);




(gdb) p i
$2 = 60
(gdb) p layout->list[i].xlator->name
Cannot access memory at address 0x0
(gdb) p layout
$3 = (dht_layout_t *) 0x7fd6b8003500
(gdb) p *layout
$4 = {spread_cnt = -1379869184, cnt = 8378078, preset = 25198592, commit_hash = 8378040, gen = 41216, type = 8377856, ref = 20480, search_unhashed = _gf_false, 
  list = 0x7fd6b8003520}

(gdb) p *frame
$5 = {root = 0x100000000, parent = 0x100000001, frames = {next = 0x7fd6a8000ea8, prev = 0x7fd6a8000eb0}, local = 0x0, this = 0x0, ret = 0x0, ref_count = 0, lock = {
    spinlock = 0, mutex = {__data = {__lock = 0, __count = 0, __owner = 0, __nusers = 0, __kind = 0, __spins = 0, __list = {__prev = 0x7fd6a8000eb0, __next = 0x0}}, 
      __size = '\000' <repeats 24 times>, "\260\016\000\250\326\177\000\000\000\000\000\000\000\000\000", __align = 0}}, cookie = 0x0, complete = _gf_false, 
  op = GF_FOP_NULL, begin = {tv_sec = 140559918309968, tv_usec = 140559918311376}, end = {tv_sec = 1, tv_usec = 0}, wind_from = 0x0, wind_to = 0x0, unwind_from = 0x0, 
  unwind_to = 0x0}



It looks like both frame and layout have been freed and layout is still being accessed. I am unable to reproduce the problem on my local setup/

However, accessing layout->cnt in the for loop is not a good idea so replacing it with a local variable which I think should fix this problem.

REVIEW: https://review.gluster.org/17343 (cluster/dht: Fix crash in dht_selfheal_dir_setattr) posted (#1) for review on master by N Balachandran (nbalacha)

REVIEW: https://review.gluster.org/17343 (cluster/dht: Fix crash in dht_selfheal_dir_setattr) posted (#2) for review on master by N Balachandran (nbalacha)

REVIEW: https://review.gluster.org/17343 (cluster/dht: Fix crash in dht_selfheal_dir_setattr) posted (#3) for review on master by N Balachandran (nbalacha)

COMMIT: https://review.gluster.org/17343 committed in master by Shyamsundar Ranganathan (srangana) 
------
commit 17784aaa311494e4538c616f02bf95477ae781bc
Author: N Balachandran <nbalacha>
Date:   Fri May 19 15:22:12 2017 +0530

    cluster/dht: Fix crash in dht_selfheal_dir_setattr
    
    Use a local variable to store the call cnt used in the
    for loop for the STACK_WIND so as not to access local
    which may be freed by STACK_UNWIND after all fops return.
    
    Change-Id: I24f49b6dbd29a2b706e388e2f6d5196c0f80afc5
    BUG: 1452102
    Signed-off-by: N Balachandran <nbalacha>
    Reviewed-on: https://review.gluster.org/17343
    Smoke: Gluster Build System <jenkins.org>
    NetBSD-regression: NetBSD Build System <jenkins.org>
    Reviewed-by: Shyamsundar Ranganathan <srangana>
    CentOS-regression: Gluster Build System <jenkins.org>

This bug is getting closed because a release has been made available that should address the reported issue. In case the problem is still not fixed with glusterfs-3.12.0, please open a new bug report.

glusterfs-3.12.0 has been announced on the Gluster mailinglists [1], packages for several distributions should become available in the near future. Keep an eye on the Gluster Users mailinglist [2] and the update infrastructure for your distribution.

[1] http://lists.gluster.org/pipermail/announce/2017-September/000082.html
[2] https://www.gluster.org/pipermail/gluster-users/