| Bug #9106 | MySQL coredumps on FreeBSD if the SQL query contains a regexp expression | ||
|---|---|---|---|
| Submitted: | 10 Mar 2005 20:47 | Modified: | 4 May 2005 18:36 |
| Reporter: | Daniel Stefan Haischt | Email Updates: | |
| Status: | Closed | Impact on me: | |
| Category: | MySQL Server: Compiling | Severity: | S1 (Critical) |
| Version: | mysql-5.0.2-alpha (FreeBSD port: mysql-c | OS: | FreeBSD (FreeBSD 5.4-PRERELEASE) |
| Assigned to: | CPU Architecture: | Any | |
[11 Mar 2005 23:35]
Jorge del Conde
Hi, Can you please provide us with the table schema for 'evo_posts' ? Thanks.
[11 Mar 2005 23:56]
Daniel Stefan Haischt
Below you'll find the original DDL that did create the table 'evo_posts':
CREATE TABLE evo_posts (
ID int(10) unsigned NOT NULL auto_increment,
post_author int(4) NOT NULL default '0',
post_issue_date datetime NOT NULL default '0000-00-00 00:00:00',
post_mod_date datetime NOT NULL default '0000-00-00 00:00:00',
post_status enum('published','deprecated','protected','private','draft')
NOT NULL default 'published',
post_locale VARCHAR(20) NOT NULL DEFAULT 'en-EU',
post_content text NOT NULL,
post_title text NOT NULL,
post_urltitle VARCHAR(50) NULL DEFAULT NULL,
post_url VARCHAR(250) NULL DEFAULT NULL,
post_category int(4) NOT NULL default '0',
post_autobr tinyint(4) NOT NULL default '1',
post_flags SET( 'pingsdone', 'imported'),
post_karma int(11) NOT NULL default '0',
post_wordcount int(11) default NULL,
post_comments ENUM('disabled', 'open', 'closed') NOT NULL DEFAULT 'open',
post_renderers VARCHAR(179) NOT NULL default 'default',
PRIMARY KEY post_ID( ID ),
INDEX post_issue_date( post_issue_date ),
INDEX post_category( post_category ),
INDEX post_author( post_author ),
INDEX post_status( post_status ),
UNIQUE post_urltitle( post_urltitle )
)
[12 Mar 2005 0:29]
Daniel Stefan Haischt
I kinda solved the issue: - Initially I did create the database 'b2e' using PhpMyAdmin. If using that database MySQL did coredump while using a select statement that did contain a REGEXP expression. - Today I did drop the b2e database as the user 'b2e' which owns that DB. After that I did re-create the database (create database b2e) as user 'b2e'. - Finally I did re-run the b2e installer and the core dump did not occure any more. No idea what's going on ... - At the very end I did figure out that I did create the very first DB using the following SQL statement 'CREATE DATABASE `b2e` DEFAULT CHARACTER SET utf8 COLLATE utf8_unicode_ci;' However using the statement 'CREATE DATABASE `b2e` DEFAULT CHARACTER SET latin1 COLLATE latin1_german1_ci;' does not cause a mysqld core dump. Anyway I would say MySQL shouldn't coredump if ... - there are issues with the REGEXP - there are issues with wrong database permissions - there is an issue with the character encoding
[14 Apr 2005 23:00]
Bugs System
No feedback was provided for this bug for over a month, so it is being suspended automatically. If you are able to provide the information that was originally requested, please do so and change the status of the bug back to "Open".
[15 Apr 2005 14:44]
Sergei Golubchik
reopened
[4 May 2005 18:36]
Hartmut Holzgraefe
verified with 5.0.2 on Linux, doesn't happen anymore with releases >=5.0.3
Description: I am using MySQL 5.x on FreeBSD 5.4-PRE together with libpthread. These are the knobs I did provide to make (to compile the actual FreeBSD port): FORCE_PKG_REGISTER=yes \ WITH_OPENSSL=yes \ BUILD_OPTIMIZED=yes \ WITH_PROC_SCOPE_PTH=yes If executing the following SQL query, mysqld will quit with a SIGSEGV signal (Actually the SQL statement originates from the b2e blog system). SELECT post_urltitle FROM evo_posts WHERE post_urltitle REGEXP '^erster_eintrag(_[0-9]+)?$' AND ID <> 0; Below is the corresponding backtrace: (gdb) set args --user=mysql --datadir=/var/db/mysql --pid-file=/var/db/mysql/abyssone.abyssworld.de.pid --bind-address=192.168.120.29 (gdb) run Starting program: /usr/ports/databases/mysql50-server/work/mysql-5.0.2-alpha/sql/mysqld --user=mysql --datadir=/var/db/mysql --pid-file=/var/db/mysql/abyssone.abyssworld.de.pid --bind-address=192.168.120.29 050310 19:05:42 InnoDB: Started; log sequence number 0 44764 /usr/ports/databases/mysql50-server/work/mysql-5.0.2-alpha/sql/mysqld: ready for connections. Version: '5.0.2-alpha-debug' socket: '/tmp/mysql.sock' port: 3306 FreeBSD port: mysql-server-5.0.2 Program received signal SIGSEGV, Segmentation fault. [Switching to Thread 3 (LWP 100348)] p_ere (p=0xbf8ed5f0, stop=128) at regcomp.c:866 866 else if (my_islower(charset,ch)) Current language: auto; currently c (gdb) bt full #0 p_ere (p=0xbf8ed5f0, stop=128) at regcomp.c:866 c = 0 '\0' prevback = 188671256 prevfwd = 188671256 conc = 1 first = 1 #1 0x084022e4 in regcomp (preg=0xb3ed478, pattern=0xa5bcb20 "^erster_eintrag(_[0-9]+)?$", cflags=186002432, charset=0xb162c50) at regcomp.c:177 pa = {next = 0xa5bcb22 "rster_eintrag(_[0-9]+)?$", end = 0xa5bcb3a "", error = 0, strip = 0xb176500, ssize = 40, slen = 2, ncsalloc = 0, g = 0xb162c00, pbegin = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0}, pend = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0}, charset = 0x84cf280} g = (struct re_guts *) 0xb162c00 i = 0 len = 186002512 #2 0x0807feb6 in Item_func_regex::fix_fields (this=0xb3ed408, thd=0x1a, tables=0xb3ed478, ref=0xb3ed704) at sql_string.h:89 buff = "\000\000\000\000\004×\216¿\030×\216¿@\204\020\bg\b\000\000üÖ\216¿\000×\216¿\004×\216¿\000\000\000\000\000\000\000\000\000\000\001", '\0' <repeats 17 times>, "°Ð>\vxÑ>\v\020Ñ>\vhØ\216¿\\\211\020\b\000p\033\vxÑ>\v Ð>\v Ð>\v\r\000\000\000<Ñ>\v\000\000\000\000\000\000\000\000\001\000\000\000\020Ñ>\v\001\000\000\000\000\000\000\000a\224F\ba\224F\001\r\000\000\000 Ð>\v\000\000\000\000\000\000\000\000àØ\034\v\000\000\000\001", '\0' <repeats 236 times>, "\001\000\000\000ðä>\v\000p\033\vÈÚ\216¿Ð`\005\b\000p\033\v°Ð>\vÈÚ\216¿\200a\005\bà"... tmp = {Ptr = 0xbf8ed6d0 "", str_length = 766, Alloced_length = 766, alloced = false, str_charset = 0x84e8440} res = (class String *) 0x84cf280 error = 0 __func__ = "fix_fields" #3 0x0807eb49 in Item_cond::fix_fields (this=0xb3ed660, thd=0xb1b7000, tables=0xb3ed178, ref=0xb3ee4ac) at sql_list.h:323 tmp_table_map = Unhandled dwarf expression opcode 0x93 How to repeat: Compile MySQL on the following FreeBSD system ... FreeBSD abyssone.abyssworld.de 5.4-PRERELEASE FreeBSD 5.4-PRERELEASE #4: Fri Mar 4 15:10:39 CET 2005 root@abyssone.abyssworld.de:/usr/obj/usr/src/sys/ABYSSONE i386 Some paths: /usr/local/bin/perl /usr/bin/make /usr/local/bin/gmake /usr/bin/gcc /usr/bin/cc GCC: Using built-in specs. Configured with: FreeBSD/i386 system compiler Thread model: posix gcc version 3.4.2 [FreeBSD] 20040728 Compilation info: CC='cc' CFLAGS='-O -pipe -O2 -O3 -fno-omit-frame-pointer -fno-gcse' CXX='c++' CXXFLAGS='-O -pipe -O2 -O3 -fno-omit-frame-pointer -fno-gcse -O2 -felide-constructors -fno-exceptions -fno-rtti -O -pipe -O2 -O3 -fno-omit-frame-pointer -fno-gcse -felide-constructors -fno-rtti -fno-exceptions' LDFLAGS=' -rpath=/usr/local/lib' ASFLAGS='' LIBC: -r--r--r-- 1 root wheel 883816 Mar 3 17:16 /lib/libc.so.5 -r--r--r-- 1 root wheel 1876040 Mar 3 17:16 /usr/lib/libc.a lrwxr-xr-x 1 root wheel 14 Mar 3 17:16 /usr/lib/libc.so -> /lib/libc.so.5 lrwxr-xr-x 1 root wheel 16 Dec 1 13:13 /usr/lib/libc.so.4 -> /usr/lib/libc.so ... With the following compiler ... C compiler: cc (GCC) 3.4.2 [FreeBSD] 20040728 C++ compiler: c++ (GCC) 3.4.2 [FreeBSD] 20040728 ... and the following configure options ... ./configure '--localstatedir=/var/db/mysql' '--with-debug' '--without-readline' '--without-libedit' '--without-bench' '--without-extra-tools' '--with-mysqlfs' '--with-vio' '--with-low-memory' '--with-comment=FreeBSD port: mysql-client-5.0.2' '--enable-thread-safe-client' '--with-openssl=/usr/local' '--enable-assembler' '--with-berkeley-db' '--with-named-thread-libs=-pthread' '--without-server' '--prefix=/usr/local' '--build=i386-portbld-freebsd5.4' 'LDFLAGS= -rpath=/usr/local/lib' 'CFLAG S=-O -pipe -O2 -O3 -fno-omit-frame-pointer -fno-gcse' 'CXX=c++' 'build_alias=i386-portbld-freebsd5.4' 'CC=cc' 'CXXFLAGS=-O -pipe -O2 -O3 -fno-omit-frame-pointer -fno-gcse -O2 -felide-constructors -fno-exceptions -fno-rtti -O -pipe -O2 -O3 -fno-omit-frame-pointer -fno-gcse -felide-constructors -fno-rtti -fno-exceptions' I did also try to use (without a difference) ... --with-collation=utf8_general_ci \ --with-extra-charsets=all \ --with-charset=utf8 ... because I thought there might be an issue if using latin1. Suggested fix: None.