Details
-
Bug
-
Status: Closed (View Workflow)
-
Major
-
Resolution: Fixed
-
10.1(EOL), 10.2(EOL), 10.3(EOL), 10.4(EOL), 10.5, 10.6, 10.7(EOL), 10.8(EOL)
Description
Each of these testcases leads to UBSAN issues in strings/decimal.c, each with individual stack traces:
SELECT RIGHT('a', -10000000000000000000); |
SELECT LPAD (0,-18446744073709551615,0); |
SELECT RPAD (0,-18446744073709551615,0); |
SELECT LOCATE (0,0,-18446744073709551615); |
SELECT INSERT (0,18446744073709551616,1,0); |
SELECT HEX(COLUMN_CREATE (1,99999999999999999999999999999 AS INT)); |
SELECT COLUMN_GET (COLUMN_CREATE (1,99999999999999999999999999999 AS DECIMAL),1 AS INT); |
SELECT 0 + (10101010101010101010101010101010101010101010101010101010101010101<<4); |
These issues and others like it significantly affect UBSAN testing. Please fix asap.
Setup (though ASAN can likely be left off):
Compiled with GCC >=7.5.0 (I use GCC 9.3.0) and:
|
-DWITH_ASAN=ON -DWITH_ASAN_SCOPE=ON -DWITH_UBSAN=ON -DWITH_RAPID=OFF -DWSREP_LIB_WITH_ASAN=ON
|
Set before execution:
|
export UBSAN_OPTIONS=print_stacktrace=1
|
Attachments
Issue Links
- relates to
-
MDEV-33734 sql_sequence.next fails with UBSAN in buildbot and outside
- Closed
-
MDEV-25454 Make MariaDB server UBSAN safe
- Confirmed