| Bug #38566 | Falcon crash in RecordVersion::thaw at RecordVersion.cpp:421 | ||
|---|---|---|---|
| Submitted: | 5 Aug 2008 11:21 | Modified: | 8 Jan 2009 17:19 |
| Reporter: | Philip Stoev | Email Updates: | |
| Status: | Closed | Impact on me: | |
| Category: | MySQL Server: Falcon storage engine | Severity: | S1 (Critical) |
| Version: | 6.0-falcon | OS: | Any |
| Assigned to: | Christopher Powers | CPU Architecture: | Any |
| Tags: | F_CHILL THAW | ||
[5 Aug 2008 11:55]
Philip Stoev
To reproduce this bug, please clone the mysql-test-extra-6.0 repository and execute $ cd mysql-test-extra-6.0/mysqltest/gentest $ perl runall.pl \ --basedir=/path/to/6.0-falcon/ \ --engine=falcon \ --grammar=conf/chill_thaw.yy \ --threads=20 --queries=100000 \ --mysqld=--falcon_debug_mask=7167 \ --mysqld=--falcon_index_chill_threshold=4K \ --mysqld=--falcon_record_chill_threshold=4K The crash should happen within less than a mintute, after a few dozen chill/thaw operations. During the Falcon Boston meeting, it was agreed that 4K is an acceptable value for those parameters, that is why I am testing with it. Values at least up to 32K will also produce crashes. The required grammar file is part of the mysql-test-extra-6.0 repository and does not have to be obtained separately. If the chill options are set to their default values, no crash occurs.
[10 Aug 2008 13:00]
Philip Stoev
This bug is also observed with --threads=1, that is, in a non-concurrent environment.
[19 Aug 2008 21:01]
Kevin Lewis
Related to falcon_record_chill_threshold=4K
[27 Aug 2008 7:11]
Bugs System
A patch for this bug has been committed. After review, it may be pushed to the relevant source trees for release in the next version. You can access the patch from: http://lists.mysql.com/commits/52651 2801 Christopher Powers 2008-08-27 Bug #38566 Falcon crash in RecordVersion::thaw at RecordVersion.cpp:421 Removed unnecessary log messages Modified assertions to accommodate race condition where two threads thaw the same record
[8 Jan 2009 17:19]
MC Brown
A noter has been added to the 6.0.8 changelog: Executing a light INSERT and UPDATE workload with falcon_index_chill_threshold set to 4K and falcon_record_chill_threshold set to to 4K, MySQL could crash
Description: When executing an insert/update workload with falcon_index_chill_threshold=4K and falcon_record_chill_threshold=4K, Falcon crashed as follows: #0 0x00110416 in __kernel_vsyscall () #1 0x00581c78 in pthread_kill () from /lib/libpthread.so.0 #2 0x087bdb43 in my_write_core (sig=11) at stacktrace.c:307 #3 0x082acd2b in handle_segfault (sig=11) at mysqld.cc:2657 #4 <signal handler called> #5 0x085b54dc in RecordVersion::thaw (this=0xaf23e7d8) at RecordVersion.cpp:421 #6 0x085302b9 in Table::fetchForUpdate (this=0xb7406e68, transaction=0xb7206518, source=0xaf23e7d8, usingIndex=false) at Table.cpp:3513 #7 0x0851bd42 in StorageDatabase::nextRow (this=0xb707d130, storageTable=0xb7474c20, recordNumber=1002, lockForUpdate=true) at StorageDatabase.cpp:286 #8 0x08521eae in StorageTable::next (this=0xb7474c20, recordNumber=1000, lockForUpdate=true) at StorageTable.cpp:169 #9 0x0851572a in StorageInterface::rnd_next (this=0xa2575e8, buf=0xa2577b0 "") at ha_falcon.cpp:586 #10 0x083c8808 in rr_sequential (info=0xa706b33c) at records.cc:366 #11 0x0835fc35 in mysql_update (thd=0x9f43668, table_list=0xa276118, fields=@0x9f44b00, values=@0x9f44d30, conds=0xa2767c0, order_num=0, order=0x0, limit=18446744073709550615, handle_duplicates=DUP_ERROR, ignore=false) at sql_update.cc:573 #12 0x082bf817 in mysql_execute_command (thd=0x9f43668) at sql_parse.cc:2846 #13 0x082c63e1 in mysql_parse (thd=0x9f43668, inBuf=0xa275df0 "UPDATE E AS X SET `int_key` = 206 WHERE X . `int_nokey` > 176", length=61, found_semicolon=0xa706c270) at sql_parse.cc:5809 #14 0x082c6e55 in dispatch_command (command=COM_QUERY, thd=0x9f43668, packet=0xa22b751 "UPDATE E AS X SET `int_key` = 206 WHERE X . `int_nokey` > 176", packet_length=61) at sql_parse.cc:1050 #15 0x082c8145 in do_command (thd=0x9f43668) at sql_parse.cc:723 #16 0x082b5e23 in handle_one_connection (arg=0x9f43668) at sql_connect.cc:1153 #17 0x0057d32f in start_thread () from /lib/libpthread.so.0 #18 0x0049a27e in clone () from /lib/libc.so.6 The crash is here: 416 } 417 } 418 419 if (bytesRestored <= 0) 420 Log::debug("RecordVersion::thaw: writePending %d, was %d, recordFetched %d, data %p\n", 421 trans->writePending, wasWritePending, recordFetched, data.record); <<<<<<<<<<<<<< HERE 422 423 ASSERT(bytesRestored > 0 || data.record == NULL); 424 state = recData; 425 (gdb) print trans $2 = (class Transaction *) 0x0 How to repeat: If this is repeatable, a test case will be uploaded shortly. Suggested fix: If a non-null trans is important for the functioning of this code, there should have been a separate assertion checking for it, so that the crash is not when the error message is being printed.