-
Type:
Bug
-
Resolution: Done
-
Priority:
Major - P3
-
None
-
Affects Version/s: None
-
Component/s: None
-
Labels:None
-
3
-
Storage Engines 2019-05-06
There have been two recent segfaults in runs of test/random_abort. They both failed on ocelot-aws in Jenkins:
http://build.wiredtiger.com:8080/job/wiredtiger-test-recovery-stress/80835/
http://build.wiredtiger.com:8080/job/wiredtiger-test-recovery-stress/81420/
They are both pretty mysterious. The stack from build 81420 looks like this:
Thread 1 (Thread 0x7f55485d2740 (LWP 23453)):
#0 __wt_checksum_hw (chunk=0x1ebfac06a5a64e, len=0) at ../src/checksum/x86/crc32-x86.c:42
#1 0x000000000056b08d in __log_checksum_match (session=0x7f55485aa960, buf=0x10b17e0, reclen=128) at ../src/log/log.c:54
#2 0x000000000056d491 in __log_open_verify (session=0x7f55485aa960, id=155, fhp=0x7ffe5bd83708, lsnp=0x7ffe5bd836d0, versionp=0x7ffe5bd8369a,
need_salvagep=0x7ffe5bd83685) at ../src/log/log.c:1025
#3 0x0000000000570b2f in __wt_log_scan (session=0x7f55485aa960, lsnp=0x0, flags=9, func=0x4b7c6f <__txn_log_recover>, cookie=0x7ffe5bd837d0)
at ../src/log/log.c:2364
#4 0x00000000004b8bfb in __wt_txn_recover (session=0x7f55485aa960) at ../src/txn/txn_recover.c:705
#5 0x00000000004163e3 in __wt_connection_workers (session=0x7f55485aa010, cfg=0x7ffe5bd838d0) at ../src/conn/conn_open.c:227
#6 0x0000000000411d8c in wiredtiger_open (home=0x0, event_handler=0x0, config=0x5baf17 "log=(recover=on)", connectionp=0x7ffe5bd83f58)
at ../src/conn/conn_api.c:2774
#7 0x0000000000403b9f in main (argc=0, argv=0x7ffe5bd841e8) at ../../../test/csuite/random_abort/main.c:384
The chunk and len are invalid because line 42 is the beginning of the function and those incoming parameters are not yet on the stack where gdb can get them.
Looking in frame 1 the arguments are all valid and everything looks good. So it isn't clear where/why a seg fault happened.
- related to
-
-
- Closed
-