On 8 September 2016 at 21:23, Ben Pfaff <b...@ovn.org> wrote:
> On Tue, Sep 06, 2016 at 10:59:19AM -0700, Joe Stringer wrote:
>> In the new test I'm seeing OVSDB segfault:
>>
>> #0 0x00000000004313ad in ovsdb_datum_compare_3way (a=0x10ebaa0, b=0x0,
>> type=0x10c29e8) at ../lib/ovsdb-data.c:1626
>> 1626 if (a->n != b->n) {
>> (gdb) bt
>> #0 0x00000000004313ad in ovsdb_datum_compare_3way (a=0x10ebaa0, b=0x0,
>> type=0x10c29e8) at ../lib/ovsdb-data.c:1626
>> #1 0x0000000000431388 in ovsdb_datum_equals (a=0x10ebaa0, b=0x0,
>> type=0x10c29e8)
>> at ../lib/ovsdb-data.c:1616
>> #2 0x00000000004117cd in update_monitor_row_data (mt=0x10eab60,
>> row=0x10eba40,
>> data=0x0) at ../ovsdb/monitor.c:310
>> #3 0x0000000000413796 in ovsdb_monitor_changes_update (old=0x0,
>> new=0x10eba40,
>> mt=0x10eab60, changes=0x10ec630) at ../ovsdb/monitor.c:1255
>> #4 0x00000000004139ec in ovsdb_monitor_change_cb (old=0x0, new=0x10eba40,
>> changed=0x10eb908, aux_=0x7ffe4c7803c0) at ../ovsdb/monitor.c:1339
>> #5 0x000000000041c2c1 in ovsdb_txn_for_each_change (txn=0x10ebdd0,
>> cb=0x4138be <ovsdb_monitor_change_cb>, aux=0x7ffe4c7803c0)
>> at ../ovsdb/transaction.c:906
>> #6 0x000000000041431f in ovsdb_monitor_commit (replica=0x10ec060,
>> txn=0x10ebdd0,
>> durable=false) at ../ovsdb/monitor.c:1553
>> #7 0x000000000041c124 in ovsdb_txn_commit_ (txn=0x10ebdd0, durable=false)
>> at ../ovsdb/transaction.c:868
>> #8 0x000000000041c219 in ovsdb_txn_commit (txn=0x10ebdd0, durable=false)
>> at ../ovsdb/transaction.c:893
>> #9 0x0000000000415b3f in process_notification (table_updates=0x10c3380,
>> db=0x10c3220) at ../ovsdb/replication.c:576
>> #10 0x0000000000414b5e in replication_run () at ../ovsdb/replication.c:185
>> #11 0x00000000004058a1 in main_loop (jsonrpc=0x10be2b0,
>> all_dbs=0x7ffe4c7806a0,
>> unixctl=0x10e9b60, remotes=0x7ffe4c780740, run_process=0x0,
>> exiting=0x7ffe4c7806ff, is_backup=0x7ffe4c78070f)
>> at ../ovsdb/ovsdb-server.c:198
>> #12 0x00000000004062d8 in main (argc=1, argv=0x7ffe4c7808f8)
>> at ../ovsdb/ovsdb-server.c:429
>
> Hmm, I don't see it here. I did find a memory leak (and sent a patch).
>
> Can you run this test under valgrind, e.g.:
> make check-valgrind TESTSUITEFLAGS='-k get-exclude-tables'
> and see what you get?
That seems fine to me.
I actually meant this test:
1866: ovsdb-server/active-backup-role-switching FAILED
(ovsdb-server.at:1321)
Doing tests with "-k replication" will trigger this and a small number
of other tests that are irrelevant to this issue.
When I run this particular test under valgrind, I found a memory leak
as well as an invalid read with the same stack trace as above.
I sent a short series to address these:
http://openvswitch.org/pipermail/dev/2016-September/079314.html
_______________________________________________
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
