store_avro_kafka segfault/sigabort on Ctrl-C

[Christopher J. Morrone]

When I ctrl-c (SIGINT) an ldmsd that is running in the foreground (ldmsd -F), I get a segfault or sigabort under the avro_kafka store cleanup. Here is a backtrace that I captured:

(gdb) bt
#0  0x0000155554831aff in raise () from /lib64/libc.so.6
#1  0x0000155554804ea5 in abort () from /lib64/libc.so.6
#2  0x0000155554874097 in __libc_message () from /lib64/libc.so.6
#3  0x000015555487b4ec in malloc_printerr () from /lib64/libc.so.6
#4  0x000015555487cdec in _int_free () from /lib64/libc.so.6
#5  0x000015554bb9e181 in url_list_clear () from /lib64/libserdes.so.1
#6  0x000015554bb9cefd in serdes_conf_destroy () from /lib64/libserdes.so.1
#7  0x000015555542488d in close_store (_sh=0x1554f00012b0) at store_avro_kafka.c:431
#8  0x000000000042314f in ldmsd_store_close (store=<optimized out>, sh=<optimized out>) at ldmsd.h:897
#9  strgp_close (strgp=strgp@entry=0x47a390) at ldmsd_strgp.c:507
#10 0x00000000004246ad in ldmsd_strgp_close () at ldmsd_strgp.c:846
#11 0x000000000040cefc in cleanup (x=x@entry=0, reason=reason@entry=0x42b0f8 "signal to exit caught") at ldmsd.c:487
#12 0x000000000040d0a7 in cleanup_sa (signal=<optimized out>, info=0x7fffffffd1b0, arg=<optimized out>) at ldmsd.c:598
#13 <signal handler called>
#14 0x00001555548dd9b8 in nanosleep () from /lib64/libc.so.6
#15 0x000015555490af58 in usleep () from /lib64/libc.so.6
#16 0x000000000040b77a in main (argc=10, argv=0x7fffffffd958) at ldmsd.c:2509

I suppose it could also be a libserdes bug, but my first guess is something store_avro_kafka does.

[Christopher J. Morrone]

I believe this one is fixed by #1149.