Harishreddy01 commented 4 years ago

In sctp_choose_boundspecific_stcb for (laddr = stcb->asoc.last_used_address; laddr; laddr = LIST_NEXT(laddr, sctp_nxt_addr)) { if (laddr->ifa == NULL) { / address has been removed / continue; } if (laddr->action == SCTP_DEL_IP_ADDRESS) { / address is being deleted / continue; } sifa = sctp_is_ifa_addr_preferred(laddr->ifa, dest_is_loop, dest_is_priv, fam); if (sifa == NULL) continue; if (((non_asoc_addr_ok == 0) && (sctp_is_addr_restricted2(stcb, sifa))) || (non_asoc_addr_ok && (sctp_is_addr_restricted2(stcb, sifa)) && (!sctp_is_addr_pending(stcb, sifa)))) { / on the no-no list / continue; } stcb->asoc.last_used_address = laddr; SCTP_TCB_UNLOCK(stcb); atomic_add_int(&sifa->refcount, 1); return (sifa); } if (start_at_beginning == 0) { stcb->asoc.last_used_address = NULL; goto sctp_from_the_top; } My application was crashing when in this part of code, when try to change some config related to sctp in my application and the reloading it. I some times see multiple threads are in this part of code, however they belong to different sctp connections.

tuexen commented 4 years ago

Can you provide a stack trace or a way to reproduce the issue?

Harishreddy01 commented 4 years ago

The stack trace is below:

0 sctp_choose_boundspecific_stcb (net=0x65c6df40, fam=2, non_asoc_addr_ok=0, dest_is_loop=, dest_is_priv=, vrf_id=, ro=0x65c6df98, stcb=0x65c6d800, inp=0x6a48c400)

at /home/acme/cc/Cz8.3.0_integration/acme/bin/sctp/sctp_output.c:2663

1 sctp_source_address_selection (inp=inp@entry=0x6a48c400, stcb=stcb@entry=0x65c6d800, ro=ro@entry=0x3b1a0c58, net=net@entry=0x3b1a0c00, non_asoc_addr_ok=non_asoc_addr_ok@entry=0, vrf_id=vrf_id@entry=0)

at /home/acme/cc/Cz8.3.0_integration/acme/bin/sctp/sctp_output.c:3287

2 0x0000000001de48c5 in sctp_lowlevel_chunk_output (inp=inp@entry=0x6a48c400, stcb=stcb@entry=0x65c6d800, net=net@entry=0x3b1a0c00, to=to@entry=0x3b1a0c60, m=0x2d897cd8, auth_offset=, auth=0x0,

auth_keyid=0, nofragment_flag=1, ecn_ok=0, chk=0x0, out_of_asoc_ok=0, src_port=49175, dest_port=49175, v_tag=2614718190, port=0, so_locked=0, over_addr=0x0)
at /home/acme/cc/Cz8.3.0_integration/acme/bin/sctp/sctp_output.c:3673

3 0x0000000001de63a1 in sctp_med_chunk_output (inp=inp@entry=0x6a48c400, stcb=stcb@entry=0x65c6d800, asoc=asoc@entry=0x65c6d858, num_out=num_out@entry=0x7fffbe2734b0, reason_code=reason_code@entry=0x7fffbe2734b4,

control_only=control_only@entry=0, from_where=<optimized out>, now=<optimized out>, now_filled=<optimized out>, frag_point=<optimized out>, so_locked=<optimized out>)
at /home/acme/cc/Cz8.3.0_integration/acme/bin/sctp/sctp_output.c:8690

4 0x0000000001dea8f2 in sctp_chunk_output (inp=inp@entry=0x6a48c400, stcb=stcb@entry=0x65c6d800, from_where=6, so_locked=0) at /home/acme/cc/Cz8.3.0_integration/acme/bin/sctp/sctp_output.c:10031

5 0x0000000001e0ea94 in sctp_timeout_handler (t=) at /home/acme/cc/Cz8.3.0_integration/acme/bin/sctp/sctputil.c:1723

6 0x0000000001e13b59 in SctpTimedObject::timeout (this=) at /home/acme/cc/Cz8.3.0_integration/acme/bin/sctpd/sctpd.cpp:2022

7 0x0000000002a27237 in TimeoutQueue::process_timeouts (this=0x3b86bb90, tNow=...) at /home/acme/cc/Cz8.3.0_integration/acme/lib/common/timeout.cpp:161

8 0x000000000276de29 in WorkerThread::doWork (this=0x3be30700, tmo=) at /home/acme/cc/Cz8.3.0_integration/acme/lib/common/WorkerThread.cpp:274

9 0x000000000276cb05 in WorkerThread::run (this=0x3be30700) at /home/acme/cc/Cz8.3.0_integration/acme/lib/common/WorkerThread.cpp:160

10 0x00000000024f89b9 in ThreadBase::startThread (pthread=0x3be30700) at /home/acme/cc/Cz8.3.0_integration/aplib/private/common/Thread.cpp:56

11 0x00007ffff69d6a18 in ThreadFunction (arg=0x0) at /home/acme/cc/Cz8.3.0_integration/aplib/private/losel/util/thread.c:395

12 0x00007ffff7bc6dd5 in start_thread () from ./lib64/libpthread.so.0

13 0x00007ffff0cd3f6d in clone () from ./lib64/libc.so.6

Steps to reproduce:

In our application we we try to change sctp related configuration like protocol or multihome address or adding or deleting more sctp interfaces etc and then reload the application ( not reboot) we are observing this crash