| Bug #36428 | MY_MUTEX_INIT_FAST is used before initialization | ||
|---|---|---|---|
| Submitted: | 30 Apr 2008 14:53 | Modified: | 9 Apr 2011 11:03 |
| Reporter: | Sergei Golubchik | Email Updates: | |
| Status: | Closed | Impact on me: | |
| Category: | MySQL Server: General | Severity: | S1 (Critical) |
| Version: | 5.0, 5.1, 6.0 | OS: | Any |
| Assigned to: | CPU Architecture: | Any | |
[30 Apr 2008 14:53]
Sergei Golubchik
[1 May 2008 8:17]
Sveta Smirnova
Thank you for the report. I can not repeat described behavior in my environement. Please indicate your operating system and version of gcc you use.
[14 May 2008 16:15]
Susanne Ebrecht
Bug #36686 is a duplicate of this bug here.
[20 May 2008 10:53]
Sergei Golubchik
Bug#36656 is a duplicate of this bug
[31 May 2008 4:51]
A B
Perhaps I'm missing something, but this problem (as described in bug #36656) completely breaks mysql on multiprocessor FreeBSD 7.0, and it's "Non-critical" ?
[16 Jun 2008 10:53]
Susanne Ebrecht
Just compiled 5.1 bzr tree on FreeBSD 7.0 again and got this:
gcc -g -DSAFE_MUTEX -Wimplicit -Wreturn-type -Wswitch -Wtrigraphs -Wcomment -W -Wchar-subscripts -Wformat -Wparentheses -Wsign-compare -Wwrite-strings -Wunused-function -Wunused-label -Wunused-value -Wunused-variable -Wunused-parameter -DUNIV_MUST_NOT_INLINE -DEXTRA_DEBUG -DFORCE_INIT_OF_VARS -DSAFEMALLOC -DPEDANTIC_SAFEMALLOC -DSAFE_MUTEX -O1 -Wuninitialized -o comp_err comp_err.o -pthread -pthread ../mysys/libmysys.a ../dbug/libdbug.a ../strings/libmystrings.a -lz -pthread -pthread -lcrypt -lm -pthread -pthread
gmake[2]: Leaving directory `/usr/home/miracee/mysqlsrc/mysql51src/extra'
../extra/comp_err \
--charset=../sql/share/charsets \
--out-dir=../sql/share/ \
--header_file=../include/mysqld_error.h \
--name_file=../include/mysqld_ername.h \
--state_file=../include/sql_state.h \
--in_file=../sql/share/errmsg.txt
gmake[1]: *** [../include/mysqld_error.h] Segmentation fault: 11
[16 Jun 2008 15:29]
Susanne Ebrecht
Workaround here on FreeBSD 7.0 is: $ ./configure --with-named-thread-libs=-pthread I didn't test other OS.
[1 Aug 2008 9:25]
Hartmut Holzgraefe
Required patch according to the original reporter (against 6.0bzr here) below, i have no idea though whether this might potentially lead to my_thread_global_init() being called twice and what the consequences of a 2nd call might be (i can only tell that my_thread_global_init doesn't protect itself against being called a second time) === modified file 'mysys/my_init.c' --- mysys/my_init.c 2007-11-08 11:23:08 +0000 +++ mysys/my_init.c 2008-08-01 09:17:28 +0000 @@ -79,6 +79,7 @@ my_bool my_init(void) my_umask_dir= 0700; /* Default umask for new directories */ init_glob_errs(); #if defined(THREAD) && defined(SAFE_MUTEX) + my_thread_global_init(); safe_mutex_global_init(); /* Must be called early */ #endif #if defined(THREAD) && defined(MY_PTHREAD_FASTMUTEX) && !defined(SAFE_MUTEX)
[7 Aug 2008 17:43]
Dan Nelson
my_thread_global_init() is guaranteed to be called a 2nd time, since it's called again just a few lines down in that same function :) This might be a better patch: --- mysql-6.0.5-alpha/mysys/my_init.c 2008-08-07 12:39:07.813993859 -0500 +++ mysql-6.0.5-alpha/mysys/my_init.c 2008-08-07 12:39:07.815991334 -0500 @@ -78,6 +78,8 @@ my_umask= 0660; /* Default umask for new files */ my_umask_dir= 0700; /* Default umask for new directories */ init_glob_errs(); + if (my_thread_global_init()) + return 1; #if defined(THREAD) && defined(SAFE_MUTEX) safe_mutex_global_init(); /* Must be called early */ #endif @@ -89,8 +91,6 @@ #if defined(HAVE_PTHREAD_INIT) pthread_init(); /* Must be called before DBUG_ENTER */ #endif - if (my_thread_global_init()) - return 1; #if !defined( __WIN__) && !defined(__NETWARE__) sigfillset(&my_signals); /* signals blocked by mf_brkhant */ #endif The crash only seems to occur when I build with --with-debug enabled, btw (in my case, gen_lex_hash is what coredumped rather than comp_err).
[12 Jan 2009 19:57]
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/63028 2730 Chad MILLER 2009-01-12 Bug#38364: gen_lex_hash segmentation fault in debug build Bug#36428: MY_MUTEX_INIT_FAST is used before initialization On some thread implementations, we need a fake mutex attri- bute as a placeholder, which we define as a global variable, "my_fast_mutexattr". Well. that must be initialized before used in any mutexes, and the ordering of initializations in the API function my_init() was wrong. Now, put my_thread_global_init(), which initializes the attri- butes that mutexes require.
[14 Jan 2009 16:03]
Chad MILLER
Queued to 5.1-bugteam and 6.0-bugteam.
[15 Jan 2009 6:25]
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/63317 2733 He Zhenxing 2009-01-15 [merge] auto merge
[15 Jan 2009 6:37]
Bugs System
Pushed into 5.1.31 (revid:joro@sun.com-20090115053147-tx1oapthnzgvs1ro) (version source revid:chad@mysql.com-20090113155022-3ievlj5a3073o2iw) (merge vers: 5.1.31) (pib:6)
[16 Jan 2009 11:52]
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/63438 2969 Georgi Kodinov 2009-01-16 Reverted the 6.0 port of fixes for bug#38364 and bug#36428: lot of test suite failures.
[19 Jan 2009 11:29]
Bugs System
Pushed into 5.1.31-ndb-6.2.17 (revid:tomas.ulin@sun.com-20090119095303-uwwvxiibtr38djii) (version source revid:tomas.ulin@sun.com-20090115073240-1wanl85vlvw2she1) (merge vers: 5.1.31-ndb-6.2.17) (pib:6)
[19 Jan 2009 13:07]
Bugs System
Pushed into 5.1.31-ndb-6.3.21 (revid:tomas.ulin@sun.com-20090119104956-guxz190n2kh31fxl) (version source revid:tomas.ulin@sun.com-20090119104956-guxz190n2kh31fxl) (merge vers: 5.1.31-ndb-6.3.21) (pib:6)
[19 Jan 2009 16:12]
Bugs System
Pushed into 5.1.31-ndb-6.4.1 (revid:tomas.ulin@sun.com-20090119144033-4aylstx5czzz88i5) (version source revid:tomas.ulin@sun.com-20090119144033-4aylstx5czzz88i5) (merge vers: 5.1.31-ndb-6.4.1) (pib:6)
[20 Jan 2009 14:30]
Chad MILLER
Needs work in 6.0 at least.
[20 Jan 2009 15:18]
Dan Nelson
The patch worked in mysql-6.0.8, but fails in 6.0.9 to new code in my_thread_global_init(). In mysql-6.0.9-alpha/mysys/my_thr_init.c:122, this function is called:
my_pthread_mutex_init(&THR_LOCK_threads, MY_MUTEX_INIT_FAST, "THR_LOCK_threads", MYF_NO_DEADLOCK_DETECTION);
, but MY_MUTEX_INIT_FAST isn't initialized until line 140 in the same function (MY_MUTEX_INIT_FAST is "&my_fast_mutexattr"). The following patch (applied along with the backed-out patch) lets me build 6.0.9:
--- mysys/my_thr_init.c 2009-01-19 17:36:27.871697542 -0600
+++ mysys/my_thr_init.c 2009-01-19 17:36:27.873689710 -0600
@@ -118,15 +118,6 @@
}
#endif /* TARGET_OS_LINUX */
- /* Mutex used by my_thread_init() and after my_thread_destroy_mutex() */
- my_pthread_mutex_init(&THR_LOCK_threads, MY_MUTEX_INIT_FAST,
- "THR_LOCK_threads", MYF_NO_DEADLOCK_DETECTION);
- my_pthread_mutex_init(&THR_LOCK_malloc, MY_MUTEX_INIT_FAST,
- "THR_LOCK_malloc", MYF_NO_DEADLOCK_DETECTION);
-
- if (my_thread_init())
- return 1;
-
#ifdef PTHREAD_ADAPTIVE_MUTEX_INITIALIZER_NP
/*
Set mutex type to "fast" a.k.a "adaptive"
@@ -168,6 +159,16 @@
#ifndef HAVE_GETHOSTBYNAME_R
pthread_mutex_init(&LOCK_gethostbyname_r,MY_MUTEX_INIT_SLOW);
#endif
+
+ /* Mutex used by my_thread_init() and after my_thread_destroy_mutex() */
+ my_pthread_mutex_init(&THR_LOCK_threads, MY_MUTEX_INIT_FAST,
+ "THR_LOCK_threads", MYF_NO_DEADLOCK_DETECTION);
+ my_pthread_mutex_init(&THR_LOCK_malloc, MY_MUTEX_INIT_FAST,
+ "THR_LOCK_malloc", MYF_NO_DEADLOCK_DETECTION);
+
+ if (my_thread_init())
+ return 1;
+
return 0;
}
[20 Jan 2009 18:58]
Bugs System
Pushed into 6.0.10-alpha (revid:joro@sun.com-20090119171328-2hemf2ndc1dxl0et) (version source revid:timothy.smith@sun.com-20090116165151-xtp5e4z6qsmxyvy0) (merge vers: 6.0.10-alpha) (pib:6)
[26 Jan 2009 16:08]
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/64045 2980 Chad MILLER 2009-01-26 Bug#36428: MY_MUTEX_INIT_FAST is used before initialization Bug#38364: gen_lex_hash segmentation fault in debug build Fix for v6.0 uniquely. Initialze mutex attributes before doing any other thread operations.
[26 Jan 2009 18:28]
Chad MILLER
Queued to 6.0-bugteam tree.
[4 Feb 2009 11:16]
Bugs System
Pushed into 6.0.10-alpha (revid:kostja@sun.com-20090204104420-mw1i2u9lum4bxjo6) (version source revid:chad@mysql.com-20090126155607-n0j3zbmgbfepnmmo) (merge vers: 6.0.10-alpha) (pib:6)
[13 Feb 2009 20:37]
Paul DuBois
Noted in 6.0.10 changelog. comp_err sometimes crashed due to improper mutex use.
