- developers - lists.mariadb.org

Re: [Maria-developers] SQL Parser
by Antony T Curtis 09 May '09

09 May '09

On 9 May 2009, at 10:22, Jim Idle wrote: > Antony T Curtis wrote: >> >> On 9 May 2009, at 00:35, Michael Widenius wrote: >> >>> >>> jim> Indeed. Obviously syntactical changes are just slog really and >>> jim> most of that is ensuring that there is enough coverage in >>> regression >>> jim> tests to exercise all the syntactic differences between T-SQL >>> and >>> jim> MySql, or rather, all the syntax of MySQL. A lot of >>> commonality but >>> jim> still a fair bit of work. What is the definitive language >>> reference >>> jim> for MySQL Marina? I can locate reference manuals and ANSI >>> specs of >>> jim> course, but if you have one particular document in mind, then >>> it would >>> jim> be useful to work off that so I can evaluate the syntactical >>> changes I >>> jim> would need to make. >>> >>> Unfortunately there is no other spec than the sql_yacc.yy file that >>> comes with the MySQL source code. >> >> I have a tool somewhere which can build .dot files from the .yy >> files in the same way how ANTLRworks draws its syntax diagrams.... >> Copies of the tool exist somewhere on the old MySQL internal >> repositories but I am sure I have a backup copy somewhere. > Yes, I added this ability to ANTLR C runtime too. You can see the > effect here: > > http://www.temporal-wave.com/index.php?option=com_psrrun&view=psrrun&Itemid… What I mean was... Feed in the sql_yacc.yy file and output a human understandable syntax diagram. > >> >>> >>> jim> >>> jim> I would like to do this in two steps: >>> jim> >>> jim> a) Replace the mysql_yacc.yy file, with as few changes as >>> possible in >>> jim> the rest of the MySQL source. >>> jim> >>> jim> >>> jim> ANTLR is essentially the same gestalt as bison: one or more >>> source files define the grammar and contain actions in C, >>> triggered at the parse points; These grammar definition files are >>> converted into C source by invoking an external program (in the >>> case of ANTLR it is a Java program) and the resulting .c/.h files >>> are compiled with the rest of the code. >>> jim> >>> jim> I wrote the C output to be machine independent. The idea >>> being that you can generate the .c files on one machine, check >>> that source code in and use it on any other machine without re- >>> generating. This can be useful when targeting systems without Java >>> machines (or ones that don't work very well). So, like many other >>> systems you can have a 'build completely from scratch' option and >>> a 'use pre-generated files option'. >>> jim> >>> jim> In the main then, I suspect that the bigger changes to the >>> MySQL base would be build changes. I say this based upon the >>> assumption that any new parser should build the same AST for a >>> query as the yacc based version does. This means the new parser >>> makes the same calls as the old one and thus the code external to >>> the parser would probably require little, if anything in the way >>> of change (but see b below). >>> >>> That sounds very good. >> >> Not possible to build the same AST as the yacc/bison based parser >> because there is no AST. >> The execution structures are directly built during the parsing phase. > Sure - for AST, you can read execution structures - there didn't > really seem to be any nomenclature but when I discussed this with > some people at Sun a while back, they used the terms AST and tree, > so I just copied them :-) So I think I would use the AST I already > build, to generate the execution structures there currently. Having > an AST means that there is a structure that can be later used for > optimization (should that be any better than optimization the > current execution structures). For a few of us, AST != execution structures .... mostly because the execution structures are very much not "abstract" and were dictated by historical and implementation issues. I have argued years ago that a good AST would allow better optimization later cheaply. For example, right now, there is a lot of work to try to simplify the Item trees for boolean expressions. This work would be better done with ASTs without wasting as much memory. > >> >>> Agree >> >> You will have to implement an all-new AST which then can be used to >> 'compile' the execution structures. >> (this was a minor sticking part of an earlier attempt to replace >> the parser - there were objections to building an AST and adding an >> additional stage to the execution) > Yeah - I just wrote that before I read this paragraph ;-) As I > already have this structure, it can be shown to pretty much zero > overhead compared to query optimization and execution time. A good > AST mostly simplifies rather than complicates things of course. > Whether my current AST needs shaping to optimally generate MySQL > remains to be seen, but it is easy to change if needed. ++Agree The only tricky issues is to try to generate exactly the same structures but in some instances later, there will be scope for actually creating simpler structures than what MySQL currently generates. > >> >>> n more generic structures. That should be easy to define. >>> >>> The MySQL code doesn't rely on bison structures at all (as far as >>> I know). >> >> AFAIK, MySQL does not use any Bison internals as such. > Yes - I could not see any reliance myself. >> A past project from many years ago, I put into MySQL a PCCTS (pre- >> ANTLR) LL parser, basically keeping MySQL's custom lexer but it >> generated Token objects instead suitable for the PCCTS generated >> parser. The project was never completed because integration of >> Stored Procedures became the objective for MySQL 5.0 instead of >> implementation of a new parser. >> >> I have asked Stewart Smith to see if he can help make public that >> old source repository as a Bazaar repository as people may be able >> to learn from it how to replace the parser without major pain. > Sure - any prior work obviously helps, but I think that the lexing > is already good. I might make changes to keyword detection to become > MySql friendly; T-SQL is a historically a bit unsure exactly when > keywords can be used. This would reduce the DFA sizes for the parser. >> >>> >>> Did you get time to do the analyse? >>> >>> jim> >>> jim> A few other things that may be of note: >>> jim> >>> jim> 1) The ANTLR generated code is free threading so long as any >>> action code that you place withing the parser is also free >>> threading. >>> >>> In other words, same as with bison >> >> The ANTLR runtime does, of course, require Java. > This, this would be my C runtime as Java is way too slow for this. > There is no Java required at runtime, only for developers to > generate the C. >> So developers who wish to extend the grammar do have to have the >> Java Runtime installed as well as the ANTLR jars. > For grammar yes as the ANTLR tool runs in Java. In practice this > isn't really a problem I think? Yeah. When I wrote "ANTLR runtime", I meant "ANTLR compiler". > >> >> I too am interested to help if at all possible. > One thing that this would need would be a good regression testing > suite. Right now I have 1300 .sql files that prove the lexer/parser/ > ast, but there would need to be more of these and runtime testing > and so on. There will have to be proper unit tests to show that a certain AST builds the correct required MySQL execution structures (Item trees, JOIN, TABLE_LIST etc etc). The best part of such an effort would be "At last! a formal specification of how the execution structures should look like". This would be exciting and cool if we can (at last) have a sane parser... especially as LL parsers always generate much nicer error messages for the users. Regards, Antony.

1 0

Re: [Maria-developers] SQL Parser
by Michael Widenius 09 May '09

09 May '09

Hi! I fixed the maria-developers address so that everyone else can follow and participate in the discussion. jim> jim> Monty, jim> jim> jim> jim> My T-SQL parser is now complete and a number of people are using jim> jim> the Java, C# and C version. Would you like to have a look at it? jim> jim> Monty> Yes, I would be intrested to know more about this! jim> jim> OK - I have my demo online now. If you visit http://www.temporal-wave.com jim> jim> and take the Online Demos->T-SQL menu option you can enter T-SQL jim> batches and see the parse results. Over the next day or so, I am jim> adding demos of my C# and VB.Net parsers, so if the site goes up and jim> down then just wait a few minutes and try again :-) Once I have jim> evaluated some of the things you mention below then I will let you jim> have a scan at the source code. ok. I will take a look as soon as I have got through the emails that has batched up during the last 3 weeks. jim> jim> I think that I could steal a lot of common syntax from this for a jim> new MySQL parser. There are 1100 regression tests culled from the jim> T-SQL language spec and user examples; on my Linux system the parse jim> time, including building and walking the AST is: jim> jim> jim> jim> jimi(50_64)-clean: wc -l *.sql jim> jim> ... jim> jim> 12923 total jim> jim> jim> jim> jimi(50_64)-clean: time tsqlc . >out jim> jim> 0.15user 0.19system 0:00.34elapsed 100%CPU (0avgtext+0avgdata jim> 0maxresident)k jim> jim> 0inputs+232outputs (0major+102772minor)pagefaults 0swaps jim> jim> jim> Which includes the time taken to read each of the 1100 files from jim> the file system and to print a message for each saying the parse was jim> successful of course. jim> jim> Anyway, if you want to take a look at the code, just let me know :-) jim> jim> Monty> The big question is how much work it would be to jim> replace the MySQLparser with your T-SQL parser. jim> Indeed. Obviously syntactical changes are just slog really and jim> most of that is ensuring that there is enough coverage in regression jim> tests to exercise all the syntactic differences between T-SQL and jim> MySql, or rather, all the syntax of MySQL. A lot of commonality but jim> still a fair bit of work. What is the definitive language reference jim> for MySQL Marina? I can locate reference manuals and ANSI specs of jim> course, but if you have one particular document in mind, then it would jim> be useful to work off that so I can evaluate the syntactical changes I jim> would need to make. Unfortunately there is no other spec than the sql_yacc.yy file that comes with the MySQL source code. jim> jim> I would like to do this in two steps: jim> jim> a) Replace the mysql_yacc.yy file, with as few changes as possible in jim> the rest of the MySQL source. jim> jim> jim> ANTLR is essentially the same gestalt as bison: one or more source files define the grammar and contain actions in C, triggered at the parse points; These grammar definition files are converted into C source by invoking an external program (in the case of ANTLR it is a Java program) and the resulting .c/.h files are compiled with the rest of the code. jim> jim> I wrote the C output to be machine independent. The idea being that you can generate the .c files on one machine, check that source code in and use it on any other machine without re-generating. This can be useful when targeting systems without Java machines (or ones that don't work very well). So, like many other systems you can have a 'build completely from scratch' option and a 'use pre-generated files option'. jim> jim> In the main then, I suspect that the bigger changes to the MySQL base would be build changes. I say this based upon the assumption that any new parser should build the same AST for a query as the yacc based version does. This means the new parser makes the same calls as the old one and thus the code external to the parser would probably require little, if anything in the way of change (but see b below). That sounds very good. jim> b) Optimze MySQL to be even faster, with the new parser at a base. jim> jim> Obviously there are two aspects, being the speed of the parser itself, which I don't think will be an issue, and then what kind of transformations are done on the resulting AST for optimization purposes. If we use exactly the same AST as MySQL does now, then improving MySQL performance is the same problem domain as it is now. However, perhaps there is opportunity for a new parser to provide more information about the query that would aid optimization. Possibilities here include modifications to the AST, external information structures that current code can use when walking the AST, or even a completely new AST. The latter would probably have quite an effect on the front end of the parse->AST->execution phase of course, so I would think that the first step is to try to achieve your step a with the existing AST as output. Agree. jim> Of course maintaining and changing an ANTLR based grammar is somewhat easier than changing bison/yacc, if for no other reason than ANTLR doesn't need much in the way of code assists for correct parsing, whereas Bison quite often does. jim> The a) option is quite important to be able to do easy merges with the jim> main MySQL tree. jim> jim> I have time this week to take a look at this. Mechanically I know it is easy enough, but I need to look at the calls that are made from the current yacc/bison parser and see if there is anything in there that would be awkward. The only things I could think of would be if the existing code relies on internal structures of bison rather than more generic structures. That should be easy to define. The MySQL code doesn't rely on bison structures at all (as far as I know). Did you get time to do the analyse? jim> jim> A few other things that may be of note: jim> jim> 1) The ANTLR generated code is free threading so long as any action code that you place withing the parser is also free threading. In other words, same as with bison jim> 2) The only system calls it uses is {m|c|re}alloc/free, which the C runtime component can re-direct (say to the MySQL overrides). For completely free threading, those calls also need to be free threading and so they are the only limit to threading of course. Sounds perfect. I would *really* like to see a new parser for MariaDB. Anything that we can do to help get this done ? Regards, Monty

2 1

[Maria-developers] bzr commit into MariaDB 5.1, with Maria 1.5:maria branch (knielsen:2698)
by knielsen＠knielsen-hq.org 06 May '09

06 May '09

#At lp:maria 2698 knielsen(a)knielsen-hq.org 2009-05-06 We are now using Valgrind rather than purify, and have for quite some time. Consequently, rename HAVE_purify to HAVE_valgrind, and related changes. Leave some comments about purify when not clear that they apply also to Valgrind. Fix redundant IF_VALGRIND declaration. Misc. small fixes: - Fixes for pool-of-threads patch. - Fixes for push of PBXT storage engine. - mysql-test-run.pl fix. - Fix build problem in compile-pentium64-max. modified: BUILD/SETUP.sh BUILD/build_mccge.sh BUILD/compile-dist BUILD/compile-pentium64-max BUILD/compile-solaris-sparc-purify include/m_string.h include/my_global.h mysql-test/Makefile.am mysql-test/lib/mtr_unique.pm mysql-test/t/pool_of_threads.cnf mysys/mf_qsort.c mysys/my_alloc.c mysys/my_init.c mysys/my_rnd.c mysys/safemalloc.c scripts/mysql_config.pl.in scripts/mysql_config.sh sql/field_conv.cc sql/filesort.cc sql/ha_partition.cc sql/hostname.cc sql/item_timefunc.cc sql/log_event.cc sql/log_event_old.cc sql/my_decimal.h sql/mysqld.cc sql/opt_range.cc sql/opt_range.h sql/records.cc sql/rpl_rli.cc sql/rpl_rli.h sql/set_var.cc sql/slave.cc sql/sql_base.cc sql/sql_binlog.cc sql/sql_class.cc sql/sql_list.h sql/sql_load.cc sql/sql_select.cc sql/table.cc storage/archive/azio.c storage/innobase/buf/buf0buf.c storage/innobase/include/univ.i storage/innobase/srv/srv0start.c storage/maria/ha_maria.cc storage/maria/ma_blockrec.c storage/maria/ma_check.c storage/maria/ma_loghandler.c storage/maria/ma_packrec.c storage/maria/ma_page.c storage/maria/ma_pagecrc.c storage/maria/ma_search.c storage/myisam/mi_check.c storage/myisam/mi_page.c storage/myisam/mi_search.c storage/myisammrg/ha_myisammrg.cc strings/bcmp.c strings/decimal.c strings/strmake.c per-file messages: BUILD/SETUP.sh Rename purify -> valgrind. BUILD/build_mccge.sh Rename purify -> valgrind. BUILD/compile-dist Fix that PBXT was missing in source tarball after `BUILD/compile-dist && make dist` BUILD/compile-pentium64-max Fix a build problem with BUILD/compile-pentium64-max on CentOS/Fedora Core 10 amd64. On these systems, there is libz.so but no libz.a. Finding libz.so, ./configure decides to use system zlib. But since BUILD/compile-pentium64-max builds a fully static binary with -all-static, the link of mysqld fails due to missing libz.a. Fix by using bundled zlib in the build script. BUILD/compile-solaris-sparc-purify Rename purify -> valgrind. include/m_string.h Rename purify -> valgrind. include/my_global.h Rename purify -> valgrind. mysql-test/Makefile.am Fix that PBXT test suite was missing from `make dist` source tarball. mysql-test/lib/mtr_unique.pm Better fix to avoid races when chmod'ing the semaphore file. (Though using chmod 666 shared files in /tmp/ is still not a very good solution). mysql-test/t/pool_of_threads.cnf Fix that test case pool_of_threads fails if run on mysqld with no --with-libevent support. mysys/mf_qsort.c Rename purify -> valgrind. mysys/my_alloc.c Rename purify -> valgrind. mysys/my_init.c Rename purify -> valgrind. mysys/my_rnd.c Rename purify -> valgrind. mysys/safemalloc.c Rename purify -> valgrind. scripts/mysql_config.pl.in Rename purify -> valgrind. scripts/mysql_config.sh Rename purify -> valgrind. sql/field_conv.cc Rename purify -> valgrind. sql/filesort.cc Rename purify -> valgrind. sql/ha_partition.cc Rename purify -> valgrind. sql/hostname.cc Rename purify -> valgrind. sql/item_timefunc.cc Rename purify -> valgrind. sql/log_event.cc Rename purify -> valgrind. sql/log_event_old.cc Rename purify -> valgrind. sql/my_decimal.h Rename purify -> valgrind. sql/mysqld.cc Rename purify -> valgrind. Fix redundant IF_VALGRIND declaration. sql/opt_range.cc Rename purify -> valgrind. sql/opt_range.h Rename purify -> valgrind. sql/records.cc Rename purify -> valgrind. sql/rpl_rli.cc Rename purify -> valgrind. sql/rpl_rli.h Rename purify -> valgrind. sql/set_var.cc Fix missing static declaration on pool_of_threads. sql/slave.cc Rename purify -> valgrind. sql/sql_base.cc Rename purify -> valgrind. sql/sql_binlog.cc Rename purify -> valgrind. sql/sql_class.cc Rename purify -> valgrind. sql/sql_list.h Rename purify -> valgrind. sql/sql_load.cc Rename purify -> valgrind. sql/sql_select.cc Rename purify -> valgrind. sql/table.cc Rename purify -> valgrind. storage/archive/azio.c Rename purify -> valgrind. storage/innobase/buf/buf0buf.c Rename purify -> valgrind. storage/innobase/include/univ.i Rename purify -> valgrind. storage/innobase/srv/srv0start.c Rename purify -> valgrind. storage/maria/ha_maria.cc Rename purify -> valgrind. storage/maria/ma_blockrec.c Rename purify -> valgrind. storage/maria/ma_check.c Rename purify -> valgrind. storage/maria/ma_loghandler.c Rename purify -> valgrind. storage/maria/ma_packrec.c Rename purify -> valgrind. storage/maria/ma_page.c Rename purify -> valgrind. storage/maria/ma_pagecrc.c Rename purify -> valgrind. storage/maria/ma_search.c Rename purify -> valgrind. storage/myisam/mi_check.c Rename purify -> valgrind. storage/myisam/mi_page.c Rename purify -> valgrind. storage/myisam/mi_search.c Rename purify -> valgrind. storage/myisammrg/ha_myisammrg.cc Rename purify -> valgrind. strings/bcmp.c Rename purify -> valgrind. strings/decimal.c Rename purify -> valgrind. strings/strmake.c Rename purify -> valgrind. === modified file 'BUILD/SETUP.sh' --- a/BUILD/SETUP.sh 2009-03-22 12:16:09 +0000 +++ b/BUILD/SETUP.sh 2009-05-06 12:03:24 +0000 @@ -119,7 +119,7 @@ fi # Set flags for various build configurations. # Used in -valgrind builds -valgrind_flags="-USAFEMALLOC -UFORCE_INIT_OF_VARS -DHAVE_purify " +valgrind_flags="-USAFEMALLOC -UFORCE_INIT_OF_VARS -DHAVE_valgrind " valgrind_flags="$valgrind_flags -DMYSQL_SERVER_SUFFIX=-valgrind-max" # # Used in -debug builds === modified file 'BUILD/build_mccge.sh' --- a/BUILD/build_mccge.sh 2008-02-07 23:12:34 +0000 +++ b/BUILD/build_mccge.sh 2009-05-06 12:03:24 +0000 @@ -938,7 +938,7 @@ set_up_ccache() set_valgrind_flags() { if test "x$valgrind_flag" = "xyes" ; then - loc_valgrind_flags="-USAFEMALLOC -UFORCE_INIT_OF_VARS -DHAVE_purify " + loc_valgrind_flags="-USAFEMALLOC -UFORCE_INIT_OF_VARS -DHAVE_valgrind " loc_valgrind_flags="$loc_valgrind_flags -DMYSQL_SERVER_SUFFIX=-valgrind-max" compiler_flags="$compiler_flags $loc_valgrind_flags" fi === modified file 'BUILD/compile-dist' --- a/BUILD/compile-dist 2009-02-19 09:01:25 +0000 +++ b/BUILD/compile-dist 2009-05-06 12:03:24 +0000 @@ -60,8 +60,6 @@ fi # Make sure to enable all features that affect "make dist" # Remember that configure restricts the man pages to the configured features ! ./configure \ - --with-maria-storage-engine \ - --with-embedded-server \ - --with-ndbcluster -$gmake - + --with-plugins=max-no-ndb \ + --with-embedded-server +$gmake -j4 === modified file 'BUILD/compile-pentium64-max' --- a/BUILD/compile-pentium64-max 2007-04-11 12:12:00 +0000 +++ b/BUILD/compile-pentium64-max 2009-05-06 12:03:24 +0000 @@ -4,7 +4,10 @@ path=`dirname $0` . "$path/SETUP.sh" extra_flags="$pentium64_cflags $fast_cflags" -extra_configs="$pentium_configs $max_configs $static_link" +# On CentOS/Fedora Core 10 amd64, there is system libz.so but not +# libz.a, so need to use bundled zlib when building static +# binary. Hence we use --with-zlib-dir=bundled +extra_configs="$pentium_configs $max_configs $static_link --with-zlib-dir=bundled" CC="$CC --pipe" strip=yes === modified file 'BUILD/compile-solaris-sparc-purify' --- a/BUILD/compile-solaris-sparc-purify 2007-02-26 15:12:40 +0000 +++ b/BUILD/compile-solaris-sparc-purify 2009-05-06 12:03:24 +0000 @@ -37,7 +37,7 @@ make -k maintainer-clean || true path=`dirname $0` . "$path/autorun.sh" -CFLAGS="-g -Wimplicit -Wreturn-type -Wswitch -Wtrigraphs -Wcomment -W -Wchar-subscripts -Wformat -Wimplicit-int -Wparentheses -Wsign-compare -Wwrite-strings -Wunused -DHAVE_purify -DEXTRA_DEBUG -O2" CXX=gcc CXXLD=g++ CXXFLAGS="-g -Wimplicit -Wreturn-type -Wswitch -Wtrigraphs -Wcomment -W -Wchar-subscripts -Wformat -Wparentheses -Wsign-compare -Wwrite-strings -Woverloaded-virtual -Wsign-promo -Wreorder -Wctor-dtor-privacy -Wnon-virtual-dtor -felide-constructors -fno-exceptions -fno-rtti -DHAVE_purify -DEXTRA_DEBUG -O2" ./configure --prefix=/usr/local/mysql --enable-assembler --with-extra-charsets=complex --enable-thread-safe-client --with-embedded-server --with-innodb $EXTRA_CONFIG_FLAGS +CFLAGS="-g -Wimplicit -Wreturn-type -Wswitch -Wtrigraphs -Wcomment -W -Wchar-subscripts -Wformat -Wimplicit-int -Wparentheses -Wsign-compare -Wwrite-strings -Wunused -DHAVE_valgrind -DEXTRA_DEBUG -O2" CXX=gcc CXXLD=g++ CXXFLAGS="-g -Wimplicit -Wreturn-type -Wswitch -Wtrigraphs -Wcomment -W -Wchar-subscripts -Wformat -Wparentheses -Wsign-compare -Wwrite-strings -Woverloaded-virtual -Wsign-promo -Wreorder -Wctor-dtor-privacy -Wnon-virtual-dtor -felide-constructors -fno-exceptions -fno-rtti -DHAVE_valgrind -DEXTRA_DEBUG -O2" ./configure --prefix=/usr/local/mysql --enable-assembler --with-extra-charsets=complex --enable-thread-safe-client --with-embedded-server --with-innodb $EXTRA_CONFIG_FLAGS make -j 4 === modified file 'include/m_string.h' --- a/include/m_string.h 2008-06-26 15:10:11 +0000 +++ b/include/m_string.h 2009-05-06 12:03:24 +0000 @@ -108,7 +108,7 @@ extern const double log_10[309]; #define memcpy_fixed(A,B,C) memcpy((A),(B),(C)) #endif -#if (!defined(USE_BMOVE512) || defined(HAVE_purify)) && !defined(bmove512) +#if (!defined(USE_BMOVE512) || defined(HAVE_valgrind)) && !defined(bmove512) #define bmove512(A,B,C) memcpy(A,B,C) #endif @@ -125,14 +125,14 @@ extern void bzero(uchar * dst,size_t len #if !defined(bcmp) && !defined(HAVE_BCMP) extern size_t bcmp(const uchar *s1,const uchar *s2,size_t len); #endif -#ifdef HAVE_purify +#ifdef HAVE_valgrind extern size_t my_bcmp(const uchar *s1,const uchar *s2,size_t len); #undef bcmp #define bcmp(A,B,C) my_bcmp((A),(B),(C)) -#define bzero_if_purify(A,B) bzero(A,B) +#define bzero_if_valgrind(A,B) bzero(A,B) #else -#define bzero_if_purify(A,B) -#endif /* HAVE_purify */ +#define bzero_if_valgrind(A,B) +#endif /* HAVE_valgrind */ #if defined(_lint) || defined(FORCE_INIT_OF_VARS) #define LINT_INIT_STRUCT(var) bzero(&var, sizeof(var)) /* No uninitialize-warning */ === modified file 'include/my_global.h' --- a/include/my_global.h 2009-04-08 16:55:26 +0000 +++ b/include/my_global.h 2009-05-06 12:03:24 +0000 @@ -563,16 +563,16 @@ int __void__; #define LINT_INIT(var) #endif -#if defined(_lint) || defined(FORCE_INIT_OF_VARS) || defined(HAVE_purify) -#define PURIFY_OR_LINT_INIT(var) var=0 +#if defined(_lint) || defined(FORCE_INIT_OF_VARS) || defined(HAVE_valgrind) +#define VALGRIND_OR_LINT_INIT(var) var=0 #else -#define PURIFY_OR_LINT_INIT(var) +#define VALGRIND_OR_LINT_INIT(var) #endif -#ifdef HAVE_purify -#define IF_PURIFY(A,B) (A) +#ifdef HAVE_valgrind +#define IF_VALGRIND(A,B) (A) #else -#define IF_PURIFY(A,B) (B) +#define IF_VALGRIND(A,B) (B) #endif #if !defined(HAVE_UINT) @@ -1174,7 +1174,7 @@ typedef char bool; /* Ordinary boolean ((uint32) (uchar) (A)[0]))) #define sint4korr(A) (*((const long *) (A))) #define uint2korr(A) (*((const uint16 *) (A))) -#if defined(HAVE_purify) && !defined(_WIN32) +#if defined(HAVE_valgrind) && !defined(_WIN32) #define uint3korr(A) (uint32) (((uint32) ((uchar) (A)[0])) +\ (((uint32) ((uchar) (A)[1])) << 8) +\ (((uint32) ((uchar) (A)[2])) << 16)) @@ -1186,7 +1186,7 @@ typedef char bool; /* Ordinary boolean It means, that you have to provide enough allocated space ! */ #define uint3korr(A) (long) (*((const unsigned int *) (A)) & 0xFFFFFF) -#endif /* HAVE_purify && !_WIN32 */ +#endif /* HAVE_valgrind && !_WIN32 */ #define uint4korr(A) (*((const uint32 *) (A))) #define uint5korr(A) ((ulonglong)(((uint32) ((uchar) (A)[0])) +\ (((uint32) ((uchar) (A)[1])) << 8) +\ === modified file 'mysql-test/Makefile.am' --- a/mysql-test/Makefile.am 2009-02-13 12:25:41 +0000 +++ b/mysql-test/Makefile.am 2009-05-06 12:03:24 +0000 @@ -98,7 +98,8 @@ TEST_DIRS = t r include std_data std_dat suite/stress/include suite/stress/t suite/stress/r \ suite/ndb suite/ndb/t suite/ndb/r \ suite/rpl_ndb suite/rpl_ndb/t suite/rpl_ndb/r \ - suite/parts suite/parts/t suite/parts/r suite/parts/inc + suite/parts suite/parts/t suite/parts/r suite/parts/inc \ + suite/pbxt/t suite/pbxt/r # Used by dist-hook and install-data-local to copy all # test files into either dist or install directory === modified file 'mysql-test/lib/mtr_unique.pm' --- a/mysql-test/lib/mtr_unique.pm 2009-04-08 16:55:26 +0000 +++ b/mysql-test/lib/mtr_unique.pm 2009-05-06 12:03:24 +0000 @@ -62,14 +62,14 @@ sub mtr_get_unique_id($$) { die 'lock file is a symbolic link'; } + my $save_umask= umask(0); open SEM, ">", "$file.sem" or die "can't write to $file.sem"; - chmod 0777, "$file.sem"; flock SEM, LOCK_EX or die "can't lock $file.sem"; if(! -e $file) { open FILE, ">", $file or die "can't create $file"; close FILE; } - chmod 0777, $file; + umask($save_umask); msg("HAVE THE LOCK"); @@ -135,8 +135,8 @@ sub mtr_release_unique_id($) { die 'lock file is a symbolic link'; } + my $save_umask= umask(0); open SEM, ">", "$file.sem" or die "can't write to $file.sem"; - chmod 0777, "$file.sem"; flock SEM, LOCK_EX or die "can't lock $file.sem"; msg("HAVE THE LOCK"); @@ -149,7 +149,7 @@ sub mtr_release_unique_id($) { open FILE, ">", $file or die "can't create $file"; close FILE; } - chmod 0777, "$file.sem"; + umask($save_umask); open FILE, "+<", $file or die "can't open $file"; #select undef,undef,undef,0.2; seek FILE, 0, 0; === modified file 'mysql-test/t/pool_of_threads.cnf' --- a/mysql-test/t/pool_of_threads.cnf 2009-03-18 15:46:32 +0000 +++ b/mysql-test/t/pool_of_threads.cnf 2009-05-06 12:03:24 +0000 @@ -2,8 +2,8 @@ [mysqld.1] test-ignore-wrong-options -thread-handling= pool-of-threads -thread_pool_size= 2 +loose-thread-handling= pool-of-threads +loose-thread_pool_size= 2 extra-port= @mysqld.1.#extra-port extra-max-connections=1 === modified file 'mysys/mf_qsort.c' --- a/mysys/mf_qsort.c 2007-10-17 16:08:58 +0000 +++ b/mysys/mf_qsort.c 2009-05-06 12:03:24 +0000 @@ -108,7 +108,7 @@ qsort_t my_qsort(void *base_ptr, size_t low = (char*) base_ptr; high = low+ size * (count - 1); stack_ptr = stack + 1; -#ifdef HAVE_purify +#ifdef HAVE_valgrind /* The first element in the stack will be accessed for the last POP */ stack[0].low=stack[0].high=0; #endif === modified file 'mysys/my_alloc.c' --- a/mysys/my_alloc.c 2009-01-15 18:11:25 +0000 +++ b/mysys/my_alloc.c 2009-05-06 12:03:24 +0000 @@ -56,7 +56,7 @@ void init_alloc_root(MEM_ROOT *mem_root, mem_root->block_num= 4; /* We shift this with >>2 */ mem_root->first_block_usage= 0; -#if !(defined(HAVE_purify) && defined(EXTRA_DEBUG)) +#if !(defined(HAVE_valgrind) && defined(EXTRA_DEBUG)) if (pre_alloc_size) { if ((mem_root->free= mem_root->pre_alloc= @@ -96,7 +96,7 @@ void reset_root_defaults(MEM_ROOT *mem_r DBUG_ASSERT(alloc_root_inited(mem_root)); mem_root->block_size= block_size - ALLOC_ROOT_MIN_BLOCK_SIZE; -#if !(defined(HAVE_purify) && defined(EXTRA_DEBUG)) +#if !(defined(HAVE_valgrind) && defined(EXTRA_DEBUG)) if (pre_alloc_size) { size_t size= pre_alloc_size + ALIGN_SIZE(sizeof(USED_MEM)); @@ -147,7 +147,7 @@ void reset_root_defaults(MEM_ROOT *mem_r void *alloc_root(MEM_ROOT *mem_root, size_t length) { -#if defined(HAVE_purify) && defined(EXTRA_DEBUG) +#if defined(HAVE_valgrind) && defined(EXTRA_DEBUG) reg1 USED_MEM *next; DBUG_ENTER("alloc_root"); DBUG_PRINT("enter",("root: 0x%lx", (long) mem_root)); === modified file 'mysys/my_init.c' --- a/mysys/my_init.c 2009-02-19 09:01:25 +0000 +++ b/mysys/my_init.c 2009-05-06 12:03:24 +0000 @@ -170,7 +170,7 @@ void my_end(int infoflag) { #ifdef HAVE_GETRUSAGE struct rusage rus; -#ifdef HAVE_purify +#ifdef HAVE_valgrind /* Purify assumes that rus is uninitialized after getrusage call */ bzero((char*) &rus, sizeof(rus)); #endif === modified file 'mysys/my_rnd.c' --- a/mysys/my_rnd.c 2007-12-14 08:47:00 +0000 +++ b/mysys/my_rnd.c 2009-05-06 12:03:24 +0000 @@ -26,7 +26,7 @@ void my_rnd_init(struct my_rnd_struct *rand_st, ulong seed1, ulong seed2) { -#ifdef HAVE_purify +#ifdef HAVE_valgrind bzero((char*) rand_st,sizeof(*rand_st)); /* Avoid UMC varnings */ #endif rand_st->max_value= 0x3FFFFFFFL; === modified file 'mysys/safemalloc.c' --- a/mysys/safemalloc.c 2008-10-10 15:28:41 +0000 +++ b/mysys/safemalloc.c 2009-05-06 12:03:24 +0000 @@ -304,7 +304,7 @@ void _myfree(void *ptr, const char *file sf_malloc_count--; pthread_mutex_unlock(&THR_LOCK_malloc); -#ifndef HAVE_purify +#ifndef HAVE_valgrind /* Mark this data as free'ed */ if (!sf_malloc_quick) bfill(ptr, irem->datasize, (pchar) FREE_VAL); === modified file 'scripts/mysql_config.pl.in' --- a/scripts/mysql_config.pl.in 2008-11-10 21:12:15 +0000 +++ b/scripts/mysql_config.pl.in 2009-05-06 12:03:24 +0000 @@ -44,7 +44,7 @@ use strict; my @exclude_cflags = qw/DDBUG_OFF DSAFEMALLOC USAFEMALLOC DSAFE_MUTEX DPEDANTIC_SAFEMALLOC DUNIV_MUST_NOT_INLINE DFORCE_INIT_OF_VARS - DEXTRA_DEBUG DHAVE_purify O O[0-9] xO[0-9] W[-A-Za-z]* + DEXTRA_DEBUG DHAVE_valgrind O O[0-9] xO[0-9] W[-A-Za-z]* Xa xstrconst xc99=none unroll2 ip mp restrict/; === modified file 'scripts/mysql_config.sh' --- a/scripts/mysql_config.sh 2008-11-10 21:12:15 +0000 +++ b/scripts/mysql_config.sh 2009-05-06 12:03:24 +0000 @@ -129,7 +129,7 @@ include="-I$pkgincludedir" # to make --cflags usable for HP C++ (aCC) for remove in DDBUG_OFF DSAFEMALLOC USAFEMALLOC DSAFE_MUTEX \ DPEDANTIC_SAFEMALLOC DUNIV_MUST_NOT_INLINE DFORCE_INIT_OF_VARS \ - DEXTRA_DEBUG DHAVE_purify O 'O[0-9]' 'xO[0-9]' 'W[-A-Za-z]*' \ + DEXTRA_DEBUG DHAVE_valgrind O 'O[0-9]' 'xO[0-9]' 'W[-A-Za-z]*' \ 'mtune=[-A-Za-z0-9]*' 'mcpu=[-A-Za-z0-9]*' 'march=[-A-Za-z0-9]*' \ Xa xstrconst "xc99=none" AC99 \ unroll2 ip mp restrict === modified file 'sql/field_conv.cc' --- a/sql/field_conv.cc 2009-01-31 21:22:44 +0000 +++ b/sql/field_conv.cc 2009-05-06 12:03:24 +0000 @@ -773,7 +773,7 @@ int field_conv(Field *to,Field *from) ((Field_varstring*)from)->length_bytes == ((Field_varstring*)to)->length_bytes)) { // Identical fields -#ifdef HAVE_purify +#ifdef HAVE_valgrind /* This may happen if one does 'UPDATE ... SET x=x' */ if (to->ptr != from->ptr) #endif === modified file 'sql/filesort.cc' --- a/sql/filesort.cc 2009-02-19 09:01:25 +0000 +++ b/sql/filesort.cc 2009-05-06 12:03:24 +0000 @@ -953,13 +953,13 @@ static void make_sortkey(register SORTPA if (addonf->null_bit && field->is_null()) { nulls[addonf->null_offset]|= addonf->null_bit; -#ifdef HAVE_purify +#ifdef HAVE_valgrind bzero(to, addonf->length); #endif } else { -#ifdef HAVE_purify +#ifdef HAVE_valgrind uchar *end= field->pack(to, field->ptr); uint length= (uint) ((to + addonf->length) - end); DBUG_ASSERT((int) length >= 0); === modified file 'sql/ha_partition.cc' --- a/sql/ha_partition.cc 2009-02-19 09:01:25 +0000 +++ b/sql/ha_partition.cc 2009-05-06 12:03:24 +0000 @@ -3548,10 +3548,10 @@ void ha_partition::position(const uchar (ref_length - PARTITION_BYTES_IN_POS)); #ifdef SUPPORTING_PARTITION_OVER_DIFFERENT_ENGINES -#ifdef HAVE_purify +#ifdef HAVE_valgrind bzero(ref + PARTITION_BYTES_IN_POS + ref_length, max_ref_length-ref_length); -#endif /* HAVE_purify */ +#endif /* HAVE_valgrind */ #endif DBUG_VOID_RETURN; } === modified file 'sql/hostname.cc' --- a/sql/hostname.cc 2007-10-11 17:29:09 +0000 +++ b/sql/hostname.cc 2009-05-06 12:03:24 +0000 @@ -175,7 +175,7 @@ char * ip_to_hostname(struct in_addr *in char buff[GETHOSTBYADDR_BUFF_SIZE],buff2[GETHOSTBYNAME_BUFF_SIZE]; int tmp_errno; struct hostent tmp_hostent, tmp_hostent2; -#ifdef HAVE_purify +#ifdef HAVE_valgrind bzero(buff,sizeof(buff)); // Bug in purify #endif if (!(hp=gethostbyaddr_r((char*) in,sizeof(*in), === modified file 'sql/item_timefunc.cc' --- a/sql/item_timefunc.cc 2009-02-15 10:58:34 +0000 +++ b/sql/item_timefunc.cc 2009-05-06 12:03:24 +0000 @@ -290,8 +290,8 @@ static bool extract_date_time(DATE_TIME_ LINT_INIT(strict_week_number); /* Remove valgrind varnings when using gcc 3.3 and -O1 */ - PURIFY_OR_LINT_INIT(strict_week_number_year_type); - PURIFY_OR_LINT_INIT(sunday_first_n_first_week_non_iso); + VALGRIND_OR_LINT_INIT(strict_week_number_year_type); + VALGRIND_OR_LINT_INIT(sunday_first_n_first_week_non_iso); if (!sub_pattern_end) bzero((char*) l_time, sizeof(*l_time)); === modified file 'sql/log_event.cc' --- a/sql/log_event.cc 2009-02-19 09:01:25 +0000 +++ b/sql/log_event.cc 2009-05-06 12:03:24 +0000 @@ -6990,7 +6990,7 @@ int Rows_log_event::do_add_row_data(ucha Don't print debug messages when running valgrind since they can trigger false warnings. */ -#ifndef HAVE_purify +#ifndef HAVE_valgrind DBUG_DUMP("row_data", row_data, min(length, 32)); #endif @@ -7771,7 +7771,7 @@ Table_map_log_event::Table_map_log_event Don't print debug messages when running valgrind since they can trigger false warnings. */ -#ifndef HAVE_purify +#ifndef HAVE_valgrind DBUG_DUMP("event buffer", (uchar*) buf, event_len); #endif @@ -8674,7 +8674,7 @@ int Rows_log_event::find_row(const Relay Don't print debug messages when running valgrind since they can trigger false warnings. */ -#ifndef HAVE_purify +#ifndef HAVE_valgrind DBUG_DUMP("key data", m_key, table->key_info->key_length); #endif @@ -8704,7 +8704,7 @@ int Rows_log_event::find_row(const Relay Don't print debug messages when running valgrind since they can trigger false warnings. */ -#ifndef HAVE_purify +#ifndef HAVE_valgrind DBUG_PRINT("info",("found first matching record")); DBUG_DUMP("record[0]", table->record[0], table->s->reclength); #endif @@ -9067,7 +9067,7 @@ Update_rows_log_event::do_exec_row(const Now we have the right row to update. The old row (the one we're looking for) is in record[1] and the new row is in record[0]. */ -#ifndef HAVE_purify +#ifndef HAVE_valgrind /* Don't print debug messages when running valgrind since they can trigger false warnings. === modified file 'sql/log_event_old.cc' --- a/sql/log_event_old.cc 2009-01-31 21:22:44 +0000 +++ b/sql/log_event_old.cc 2009-05-06 12:03:24 +0000 @@ -724,7 +724,7 @@ static int find_and_fetch_row(TABLE *tab Don't print debug messages when running valgrind since they can trigger false warnings. */ -#ifndef HAVE_purify +#ifndef HAVE_valgrind DBUG_DUMP("table->record[0]", table->record[0], table->s->reclength); DBUG_DUMP("table->record[1]", table->record[1], table->s->reclength); #endif @@ -750,7 +750,7 @@ static int find_and_fetch_row(TABLE *tab Don't print debug messages when running valgrind since they can trigger false warnings. */ -#ifndef HAVE_purify +#ifndef HAVE_valgrind DBUG_DUMP("table->record[0]", table->record[0], table->s->reclength); DBUG_DUMP("table->record[1]", table->record[1], table->s->reclength); #endif @@ -1404,7 +1404,7 @@ int Old_rows_log_event::do_add_row_data( Don't print debug messages when running valgrind since they can trigger false warnings. */ -#ifndef HAVE_purify +#ifndef HAVE_valgrind DBUG_DUMP("row_data", row_data, min(length, 32)); #endif @@ -2353,7 +2353,7 @@ int Old_rows_log_event::find_row(const R Don't print debug messages when running valgrind since they can trigger false warnings. */ -#ifndef HAVE_purify +#ifndef HAVE_valgrind DBUG_DUMP("key data", m_key, table->key_info->key_length); #endif @@ -2383,7 +2383,7 @@ int Old_rows_log_event::find_row(const R Don't print debug messages when running valgrind since they can trigger false warnings. */ -#ifndef HAVE_purify +#ifndef HAVE_valgrind DBUG_PRINT("info",("found first matching record")); DBUG_DUMP("record[0]", table->record[0], table->s->reclength); #endif @@ -2878,7 +2878,7 @@ Update_rows_log_event_old::do_exec_row(c Now we have the right row to update. The old row (the one we're looking for) is in record[1] and the new row is in record[0]. */ -#ifndef HAVE_purify +#ifndef HAVE_valgrind /* Don't print debug messages when running valgrind since they can trigger false warnings. === modified file 'sql/my_decimal.h' --- a/sql/my_decimal.h 2008-05-20 07:38:17 +0000 +++ b/sql/my_decimal.h 2009-05-06 12:03:24 +0000 @@ -99,7 +99,7 @@ public: { len= DECIMAL_BUFF_LENGTH; buf= buffer; -#if !defined (HAVE_purify) && !defined(DBUG_OFF) +#if !defined (HAVE_valgrind) && !defined(DBUG_OFF) /* Set buffer to 'random' value to find wrong buffer usage */ for (uint i= 0; i < DECIMAL_BUFF_LENGTH; i++) buffer[i]= i; === modified file 'sql/mysqld.cc' --- a/sql/mysqld.cc 2009-04-08 21:16:10 +0000 +++ b/sql/mysqld.cc 2009-05-06 12:03:24 +0000 @@ -57,12 +57,6 @@ #define mysqld_charset &my_charset_latin1 -#ifdef HAVE_purify -#define IF_PURIFY(A,B) (A) -#else -#define IF_PURIFY(A,B) (B) -#endif - #if SIZEOF_CHARP == 4 #define MAX_MEM_TABLE_SIZE ~(ulong) 0 #else @@ -76,9 +70,9 @@ char pstack_file_name[80]; #endif /* __linux__ */ -/* We have HAVE_purify below as this speeds up the shutdown of MySQL */ +/* We have HAVE_valgrind below as this speeds up the shutdown of MySQL */ -#if defined(HAVE_DEC_3_2_THREADS) || defined(SIGNALS_DONT_BREAK_READ) || defined(HAVE_purify) && defined(__linux__) +#if defined(HAVE_DEC_3_2_THREADS) || defined(SIGNALS_DONT_BREAK_READ) || defined(HAVE_valgrind) && defined(__linux__) #define HAVE_CLOSE_SERVER_SOCK 1 #endif @@ -6520,7 +6514,7 @@ log and this option does nothing anymore purify. These are not suppressed: instead we disable symlinks option if compiled with valgrind support. */ - IF_PURIFY(0,1), 0, 0, 0, 0, 0}, + IF_VALGRIND(0,1), 0, 0, 0, 0, 0}, {"sysdate-is-now", OPT_SYSDATE_IS_NOW, "Non-default option to alias SYSDATE() to NOW() to make it safe-replicable. Since 5.0, SYSDATE() returns a `dynamic' value different for different invocations, even within the same statement.", (uchar**) &global_system_variables.sysdate_is_now, @@ -6561,7 +6555,7 @@ log and this option does nothing anymore 0, 0, 0, 0, 0}, {"use-symbolic-links", 's', "Enable symbolic link support. Deprecated option; use --symbolic-links instead.", (uchar**) &my_use_symdir, (uchar**) &my_use_symdir, 0, GET_BOOL, NO_ARG, - IF_PURIFY(0,1), 0, 0, 0, 0, 0}, + IF_VALGRIND(0,1), 0, 0, 0, 0, 0}, {"user", 'u', "Run mysqld daemon as user.", 0, 0, 0, GET_STR, REQUIRED_ARG, 0, 0, 0, 0, 0, 0}, {"verbose", 'v', "Used with --help option for detailed help", @@ -7661,7 +7655,7 @@ static int mysql_init_variables(void) /* Things reset to zero */ opt_skip_slave_start= opt_reckless_slave = 0; mysql_home[0]= pidfile_name[0]= log_error_file[0]= 0; -#if defined(HAVE_REALPATH) && !defined(HAVE_purify) && !defined(HAVE_BROKEN_REALPATH) +#if defined(HAVE_REALPATH) && !defined(HAVE_valgrind) && !defined(HAVE_BROKEN_REALPATH) /* We can only test for sub paths if my_symlink.c is using realpath */ myisam_test_invalid_symlink= test_if_data_home_dir; #endif === modified file 'sql/opt_range.cc' --- a/sql/opt_range.cc 2009-02-19 09:01:25 +0000 +++ b/sql/opt_range.cc 2009-05-06 12:03:24 +0000 @@ -8347,7 +8347,7 @@ int QUICK_RANGE_SELECT::reset() multi_range_buff->buffer= mrange_buff; multi_range_buff->buffer_end= mrange_buff + mrange_bufsiz; multi_range_buff->end_of_used_area= mrange_buff; -#ifdef HAVE_purify +#ifdef HAVE_valgrind /* We need this until ndb will use the buffer efficiently (Now ndb stores complete row in here, instead of only the used fields === modified file 'sql/opt_range.h' --- a/sql/opt_range.h 2008-08-25 17:18:22 +0000 +++ b/sql/opt_range.h 2009-05-06 12:03:24 +0000 @@ -44,7 +44,7 @@ class QUICK_RANGE :public Sql_alloc { uint16 min_length,max_length,flag; key_part_map min_keypart_map, // bitmap of used keyparts in min_key max_keypart_map; // bitmap of used keyparts in max_key -#ifdef HAVE_purify +#ifdef HAVE_valgrind uint16 dummy; /* Avoid warnings on 'flag' */ #endif QUICK_RANGE(); /* Full range */ @@ -61,7 +61,7 @@ class QUICK_RANGE :public Sql_alloc { min_keypart_map(min_keypart_map_arg), max_keypart_map(max_keypart_map_arg) { -#ifdef HAVE_purify +#ifdef HAVE_valgrind dummy=0; #endif } === modified file 'sql/records.cc' --- a/sql/records.cc 2008-07-17 18:26:55 +0000 +++ b/sql/records.cc 2009-05-06 12:03:24 +0000 @@ -520,7 +520,7 @@ static int init_rr_cache(THD *thd, READ_ info->struct_length+1, MYF(0)))) DBUG_RETURN(1); -#ifdef HAVE_purify +#ifdef HAVE_valgrind // Avoid warnings in qsort bzero(info->cache,rec_cache_size+info->cache_records* info->struct_length+1); #endif === modified file 'sql/rpl_rli.cc' --- a/sql/rpl_rli.cc 2009-02-19 09:01:25 +0000 +++ b/sql/rpl_rli.cc 2009-05-06 12:03:24 +0000 @@ -34,7 +34,7 @@ Relay_log_info::Relay_log_info() no_storage(FALSE), replicate_same_server_id(::replicate_same_server_id), info_fd(-1), cur_log_fd(-1), save_temporary_tables(0), cur_log_old_open_count(0), group_relay_log_pos(0), event_relay_log_pos(0), -#if HAVE_purify +#if HAVE_valgrind is_fake(FALSE), #endif group_master_log_pos(0), log_space_total(0), ignore_log_space_limit(0), === modified file 'sql/rpl_rli.h' --- a/sql/rpl_rli.h 2008-02-27 17:46:06 +0000 +++ b/sql/rpl_rli.h 2009-05-06 12:03:24 +0000 @@ -154,7 +154,7 @@ public: ulonglong event_relay_log_pos; ulonglong future_event_relay_log_pos; -#ifdef HAVE_purify +#ifdef HAVE_valgrind bool is_fake; /* Mark that this is a fake relay log info structure */ #endif === modified file 'sql/set_var.cc' --- a/sql/set_var.cc 2009-04-08 16:55:26 +0000 +++ b/sql/set_var.cc 2009-05-06 12:03:24 +0000 @@ -642,8 +642,8 @@ static sys_var_long_ptr sys_table_lock_w static sys_var_long_ptr sys_thread_cache_size(&vars, "thread_cache_size", &thread_cache_size); #if HAVE_POOL_OF_THREADS == 1 -sys_var_long_ptr sys_thread_pool_size(&vars, "thread_pool_size", - &thread_pool_size); +static sys_var_long_ptr sys_thread_pool_size(&vars, "thread_pool_size", + &thread_pool_size); #endif static sys_var_thd_enum sys_tx_isolation(&vars, "tx_isolation", &SV::tx_isolation, === modified file 'sql/slave.cc' --- a/sql/slave.cc 2009-02-19 09:01:25 +0000 +++ b/sql/slave.cc 2009-05-06 12:03:24 +0000 @@ -2014,7 +2014,7 @@ int apply_event_and_update_pos(Log_event if (exec_res == 0) { int error= ev->update_pos(rli); -#ifdef HAVE_purify +#ifdef HAVE_valgrind if (!rli->is_fake) #endif { === modified file 'sql/sql_base.cc' --- a/sql/sql_base.cc 2009-02-19 09:01:25 +0000 +++ b/sql/sql_base.cc 2009-05-06 12:03:24 +0000 @@ -3000,7 +3000,7 @@ TABLE *open_table(THD *thd, TABLE_LIST * table->pos_in_table_list= table_list; table_list->updatable= 1; // It is not derived table nor non-updatable VIEW table->clear_column_bitmaps(); -#if !defined(DBUG_OFF) && !defined(HAVE_purify) +#if !defined(DBUG_OFF) && !defined(HAVE_valgrind) /* Fill record with random values to find bugs where we access fields without first reading them. @@ -7363,7 +7363,7 @@ int setup_wild(THD *thd, TABLE_LIST *tab /* make * substituting permanent */ SELECT_LEX *select_lex= thd->lex->current_select; select_lex->with_wild= 0; -#ifdef HAVE_purify +#ifdef HAVE_valgrind if (&select_lex->item_list != &fields) // Avoid warning #endif select_lex->item_list= fields; === modified file 'sql/sql_binlog.cc' --- a/sql/sql_binlog.cc 2009-01-09 12:49:24 +0000 +++ b/sql/sql_binlog.cc 2009-05-06 12:03:24 +0000 @@ -59,7 +59,7 @@ void mysql_client_binlog_statement(THD* if (!thd->rli_fake) { thd->rli_fake= new Relay_log_info; -#ifdef HAVE_purify +#ifdef HAVE_valgrind thd->rli_fake->is_fake= TRUE; #endif have_fd_event= FALSE; @@ -95,7 +95,7 @@ void mysql_client_binlog_statement(THD* char const *endptr= 0; int bytes_decoded= base64_decode(strptr, coded_len, buf, &endptr); -#ifndef HAVE_purify +#ifndef HAVE_valgrind /* This debug printout should not be used for valgrind builds since it will read from unassigned memory. @@ -188,7 +188,7 @@ void mysql_client_binlog_statement(THD* bufptr += event_len; DBUG_PRINT("info",("ev->get_type_code()=%d", ev->get_type_code())); -#ifndef HAVE_purify +#ifndef HAVE_valgrind /* This debug printout should not be used for valgrind builds since it will read from unassigned memory. === modified file 'sql/sql_class.cc' --- a/sql/sql_class.cc 2009-03-12 22:27:35 +0000 +++ b/sql/sql_class.cc 2009-05-06 12:03:24 +0000 @@ -3480,7 +3480,7 @@ int THD::binlog_update_row(TABLE* table, Don't print debug messages when running valgrind since they can trigger false warnings. */ -#ifndef HAVE_purify +#ifndef HAVE_valgrind DBUG_DUMP("before_record", before_record, table->s->reclength); DBUG_DUMP("after_record", after_record, table->s->reclength); DBUG_DUMP("before_row", before_row, before_size); === modified file 'sql/sql_list.h' --- a/sql/sql_list.h 2009-01-15 18:11:25 +0000 +++ b/sql/sql_list.h 2009-05-06 12:03:24 +0000 @@ -43,7 +43,7 @@ public: static void operator delete[](void *ptr, MEM_ROOT *mem_root) { /* never called */ } static void operator delete[](void *ptr, size_t size) { TRASH(ptr, size); } -#ifdef HAVE_purify +#ifdef HAVE_valgrind bool dummy; inline Sql_alloc() :dummy(0) {} inline ~Sql_alloc() {} === modified file 'sql/sql_load.cc' --- a/sql/sql_load.cc 2009-01-31 21:22:44 +0000 +++ b/sql/sql_load.cc 2009-05-06 12:03:24 +0000 @@ -581,7 +581,7 @@ read_fixed_length(THD *thd, COPY_INFO &i } it.rewind(); uchar *pos=read_info.row_start; -#ifdef HAVE_purify +#ifdef HAVE_valgrind read_info.row_end[0]=0; #endif === modified file 'sql/sql_select.cc' --- a/sql/sql_select.cc 2009-04-08 16:55:26 +0000 +++ b/sql/sql_select.cc 2009-05-06 12:03:24 +0000 @@ -1818,7 +1818,7 @@ JOIN::exec() curr_join->having= curr_join->tmp_having= 0; // Allready done /* Change sum_fields reference to calculated fields in tmp_table */ -#ifdef HAVE_purify +#ifdef HAVE_valgrind if (curr_join != this) #endif curr_join->all_fields= *curr_all_fields; @@ -1840,7 +1840,7 @@ JOIN::exec() fields_list.elements, all_fields)) DBUG_VOID_RETURN; } -#ifdef HAVE_purify +#ifdef HAVE_valgrind if (curr_join != this) #endif { @@ -1994,7 +1994,7 @@ JOIN::exec() tmp_fields_list2, tmp_all_fields2, fields_list.elements, tmp_all_fields1)) DBUG_VOID_RETURN; -#ifdef HAVE_purify +#ifdef HAVE_valgrind /* Some GCCs use memcpy() for struct assignment, even for x=x. GCC bug 19410: http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19410 @@ -2059,7 +2059,7 @@ JOIN::exec() tmp_table_param.save_copy_field= curr_join->tmp_table_param.copy_field; tmp_table_param.save_copy_field_end= curr_join->tmp_table_param.copy_field_end; -#ifdef HAVE_purify +#ifdef HAVE_valgrind if (curr_join != this) #endif { @@ -3848,7 +3848,7 @@ update_ref_and_keys(THD *thd, DYNAMIC_AR continue; } -#ifdef HAVE_purify +#ifdef HAVE_valgrind /* Valgrind complains about overlapped memcpy when save_pos==use. */ if (save_pos != use) #endif @@ -15138,7 +15138,7 @@ setup_copy_fields(THD *thd, TMP_TABLE_PA DBUG_ASSERT (param->field_count > (uint) (copy - copy_start)); copy->set(tmp, item->result_field); item->result_field->move_field(copy->to_ptr,copy->to_null_ptr,1); -#ifdef HAVE_purify +#ifdef HAVE_valgrind copy->to_ptr[copy->from_length]= 0; #endif copy++; === modified file 'sql/table.cc' --- a/sql/table.cc 2009-02-19 09:01:25 +0000 +++ b/sql/table.cc 2009-05-06 12:03:24 +0000 @@ -1694,7 +1694,7 @@ int open_table_from_share(THD *thd, TABL outparam->record[1]= outparam->record[0]; // Safety } -#ifdef HAVE_purify +#ifdef HAVE_valgrind /* We need this because when we read var-length rows, we are not updating bytes after end of varchar @@ -1916,7 +1916,7 @@ partititon_err: } } -#if defined(HAVE_purify) && !defined(DBUG_OFF) +#if defined(HAVE_valgrind) && !defined(DBUG_OFF) bzero((char*) bitmaps, bitmap_size*3); #endif === modified file 'storage/archive/azio.c' --- a/storage/archive/azio.c 2009-04-08 16:55:26 +0000 +++ b/storage/archive/azio.c 2009-05-06 12:03:24 +0000 @@ -63,7 +63,7 @@ void read_header(azio_stream *s, unsigne */ static void *az_allocator(void *dummy, uInt items, uInt size) { - return my_malloc((size_t)items*(size_t)size, IF_PURIFY(MY_ZEROFILL, MYF(0))); + return my_malloc((size_t)items*(size_t)size, IF_VALGRIND(MY_ZEROFILL, MYF(0))); } static void az_free(void *dummy, void *address) === modified file 'storage/innobase/buf/buf0buf.c' --- a/storage/innobase/buf/buf0buf.c 2008-08-20 00:37:41 +0000 +++ b/storage/innobase/buf/buf0buf.c 2009-05-06 12:03:24 +0000 @@ -756,7 +756,7 @@ buf_pool_init( /* Wipe contents of frame to eliminate a Purify warning */ -#ifdef HAVE_purify +#ifdef HAVE_valgrind memset(block->frame, '\0', UNIV_PAGE_SIZE); #endif if (srv_use_awe) { === modified file 'storage/innobase/include/univ.i' --- a/storage/innobase/include/univ.i 2008-03-27 01:40:45 +0000 +++ b/storage/innobase/include/univ.i 2009-05-06 12:03:24 +0000 @@ -107,7 +107,7 @@ operations (very slow); also UNIV_DEBUG #define UNIV_BTR_DEBUG /* check B-tree links */ #define UNIV_LIGHT_MEM_DEBUG /* light memory debugging */ -#ifdef HAVE_purify +#ifdef HAVE_valgrind /* The following sets all new allocated memory to zero before use: this can be used to eliminate unnecessary Purify warnings, but note that it also masks many bugs Purify could detect. For detailed Purify analysis it === modified file 'storage/innobase/srv/srv0start.c' --- a/storage/innobase/srv/srv0start.c 2008-12-14 19:21:24 +0000 +++ b/storage/innobase/srv/srv0start.c 2009-05-06 12:03:24 +0000 @@ -104,7 +104,7 @@ static char* srv_monitor_file_name; /* Avoid warnings when using purify */ -#ifdef HAVE_purify +#ifdef HAVE_valgrind static int inno_bcmp(register const char *s1, register const char *s2, register uint len) { === modified file 'storage/maria/ha_maria.cc' --- a/storage/maria/ha_maria.cc 2009-02-19 09:01:25 +0000 +++ b/storage/maria/ha_maria.cc 2009-05-06 12:03:24 +0000 @@ -3166,7 +3166,7 @@ static int ha_maria_init(void *p) ma_checkpoint_init(checkpoint_interval); maria_multi_threaded= maria_in_ha_maria= TRUE; -#if defined(HAVE_REALPATH) && !defined(HAVE_purify) && !defined(HAVE_BROKEN_REALPATH) +#if defined(HAVE_REALPATH) && !defined(HAVE_valgrind) && !defined(HAVE_BROKEN_REALPATH) /* We can only test for sub paths if my_symlink.c is using realpath */ maria_test_invalid_symlink= test_if_data_home_dir; #endif === modified file 'storage/maria/ma_blockrec.c' --- a/storage/maria/ma_blockrec.c 2009-04-08 08:47:30 +0000 +++ b/storage/maria/ma_blockrec.c 2009-05-06 12:03:24 +0000 @@ -1594,7 +1594,7 @@ static void make_empty_page(MARIA_HA *in bzero(buff, PAGE_HEADER_SIZE); -#if !defined(DONT_ZERO_PAGE_BLOCKS) || defined(HAVE_purify) +#if !defined(DONT_ZERO_PAGE_BLOCKS) || defined(HAVE_valgrind) /* We zero the rest of the block to avoid getting old memory information to disk and to allow the file to be compressed better if archived. === modified file 'storage/maria/ma_check.c' --- a/storage/maria/ma_check.c 2009-02-19 09:01:25 +0000 +++ b/storage/maria/ma_check.c 2009-05-06 12:03:24 +0000 @@ -4522,7 +4522,7 @@ static int sort_key_read(MARIA_SORT_PARA sort_param->current_filepos, 0); sort_param->real_key_length= int_key.data_length + int_key.ref_length; -#ifdef HAVE_purify +#ifdef HAVE_valgrind bzero(key+sort_param->real_key_length, (sort_param->key_length-sort_param->real_key_length)); #endif @@ -4567,7 +4567,7 @@ static int sort_maria_ft_key_read(MARIA_ sort_param->current_filepos); sort_param->real_key_length= int_key.data_length + int_key.ref_length; -#ifdef HAVE_purify +#ifdef HAVE_valgrind if (sort_param->key_length > sort_param->real_key_length) bzero(key+sort_param->real_key_length, (sort_param->key_length-sort_param->real_key_length)); @@ -5104,7 +5104,7 @@ static int sort_get_next_record(MARIA_SO llstr(sort_param->pos,llbuff)); continue; } -#ifdef HAVE_purify +#ifdef HAVE_valgrind bzero(sort_param->rec_buff + block_info.rec_len, share->base.extra_rec_buff_size); #endif === modified file 'storage/maria/ma_loghandler.c' --- a/storage/maria/ma_loghandler.c 2009-03-17 19:05:01 +0000 +++ b/storage/maria/ma_loghandler.c 2009-05-06 12:03:24 +0000 @@ -6194,12 +6194,12 @@ my_bool translog_write_record(LSN *lsn, { uint i; uint len= 0; -#ifdef HAVE_purify +#ifdef HAVE_valgrind ha_checksum checksum= 0; #endif for (i= TRANSLOG_INTERNAL_PARTS; i < part_no; i++) { -#ifdef HAVE_purify +#ifdef HAVE_valgrind /* Find unitialized bytes early */ checksum+= my_checksum(checksum, parts_data[i].str, parts_data[i].length); === modified file 'storage/maria/ma_packrec.c' --- a/storage/maria/ma_packrec.c 2009-01-09 04:23:25 +0000 +++ b/storage/maria/ma_packrec.c 2009-05-06 12:03:24 +0000 @@ -256,7 +256,7 @@ static my_bool _ma_read_pack_info(MARIA_ (uint) (share->pack.header_length-sizeof(header)), MYF(MY_NABP))) goto err2; -#ifdef HAVE_purify +#ifdef HAVE_valgrind /* Zero bytes accessed by fill_buffer */ bzero(disk_cache + (share->pack.header_length-sizeof(header)), share->base.extra_rec_buff_size); === modified file 'storage/maria/ma_page.c' --- a/storage/maria/ma_page.c 2008-09-01 17:31:40 +0000 +++ b/storage/maria/ma_page.c 2009-05-06 12:03:24 +0000 @@ -200,7 +200,7 @@ my_bool _ma_write_keypage(MARIA_PAGE *pa /* Verify that keynr is correct */ DBUG_ASSERT(_ma_get_keynr(share, buff) == page->keyinfo->key_nr); -#if defined(EXTRA_DEBUG) && defined(HAVE_purify) && defined(NOT_ANYMORE) +#if defined(EXTRA_DEBUG) && defined(HAVE_valgrind) && defined(NOT_ANYMORE) { /* This is here to catch uninitialized bytes */ uint length= page->size; === modified file 'storage/maria/ma_pagecrc.c' --- a/storage/maria/ma_pagecrc.c 2008-03-04 11:58:21 +0000 +++ b/storage/maria/ma_pagecrc.c 2009-05-06 12:03:24 +0000 @@ -318,7 +318,7 @@ my_bool maria_page_filler_set_none(uchar __attribute__((unused)), uchar *data_ptr __attribute__((unused))) { -#ifdef HAVE_purify +#ifdef HAVE_valgrind int4store_aligned(page + ((MARIA_SHARE *)data_ptr)->block_size - CRC_SIZE, 0); #endif === modified file 'storage/maria/ma_search.c' --- a/storage/maria/ma_search.c 2009-01-12 11:12:00 +0000 +++ b/storage/maria/ma_search.c 2009-05-06 12:03:24 +0000 @@ -2172,7 +2172,7 @@ int _ma_calc_bin_pack_key_length(const M s_temp->totlength= key_length= (int_key->data_length + int_key->ref_length+ nod_flag); -#ifdef HAVE_purify +#ifdef HAVE_valgrind s_temp->n_length= s_temp->n_ref_length=0; /* For valgrind */ #endif s_temp->key=key; === modified file 'storage/myisam/mi_check.c' --- a/storage/myisam/mi_check.c 2009-02-19 09:01:25 +0000 +++ b/storage/myisam/mi_check.c 2009-05-06 12:03:24 +0000 @@ -3147,7 +3147,7 @@ static int sort_key_read(MI_SORT_PARAM * (info->s->rec_reflength+ _mi_make_key(info, sort_param->key, (uchar*) key, sort_param->record, sort_param->filepos)); -#ifdef HAVE_purify +#ifdef HAVE_valgrind bzero(key+sort_param->real_key_length, (sort_param->key_length-sort_param->real_key_length)); #endif @@ -3187,7 +3187,7 @@ static int sort_ft_key_read(MI_SORT_PARA sort_param->real_key_length=(info->s->rec_reflength+ _ft_make_key(info, sort_param->key, key, wptr++, sort_param->filepos)); -#ifdef HAVE_purify +#ifdef HAVE_valgrind if (sort_param->key_length > sort_param->real_key_length) bzero(key+sort_param->real_key_length, (sort_param->key_length-sort_param->real_key_length)); === modified file 'storage/myisam/mi_page.c' --- a/storage/myisam/mi_page.c 2008-08-28 12:43:44 +0000 +++ b/storage/myisam/mi_page.c 2009-05-06 12:03:24 +0000 @@ -86,7 +86,7 @@ int _mi_write_keypage(register MI_INFO * if ((length=keyinfo->block_length) > IO_SIZE*2 && info->state->key_file_length != page+length) length= ((mi_getint(buff)+IO_SIZE-1) & (uint) ~(IO_SIZE-1)); -#ifdef HAVE_purify +#ifdef HAVE_valgrind { length=mi_getint(buff); bzero((uchar*) buff+length,keyinfo->block_length-length); === modified file 'storage/myisam/mi_search.c' --- a/storage/myisam/mi_search.c 2008-08-28 18:52:23 +0000 +++ b/storage/myisam/mi_search.c 2009-05-06 12:03:24 +0000 @@ -1749,7 +1749,7 @@ _mi_calc_bin_pack_key_length(MI_KEYDEF * uint length,key_length,ref_length; s_temp->totlength=key_length=_mi_keylength(keyinfo,key)+nod_flag; -#ifdef HAVE_purify +#ifdef HAVE_valgrind s_temp->n_length= s_temp->n_ref_length=0; /* For valgrind */ #endif s_temp->key=key; === modified file 'storage/myisammrg/ha_myisammrg.cc' --- a/storage/myisammrg/ha_myisammrg.cc 2008-12-04 00:36:55 +0000 +++ b/storage/myisammrg/ha_myisammrg.cc 2009-05-06 12:03:24 +0000 @@ -837,7 +837,7 @@ int ha_myisammrg::info(uint flag) { if (table->s->key_parts && mrg_info.rec_per_key) { -#ifdef HAVE_purify +#ifdef HAVE_valgrind /* valgrind may be unhappy about it, because optimizer may access values between file->keys and table->key_parts, that will be uninitialized. === modified file 'strings/bcmp.c' --- a/strings/bcmp.c 2007-05-10 09:59:39 +0000 +++ b/strings/bcmp.c 2009-05-06 12:03:24 +0000 @@ -24,7 +24,7 @@ #include <my_global.h> #include "m_string.h" -#ifdef HAVE_purify +#ifdef HAVE_valgrind #undef bcmp #undef HAVE_BCMP #endif @@ -50,7 +50,7 @@ uint len; /* 0 <= len <= 65535 */ #else -#ifndef HAVE_purify +#ifndef HAVE_valgrind size_t bcmp(register const uchar *s1,register const uchar *s2, register size_t len) #else === modified file 'strings/decimal.c' --- a/strings/decimal.c 2008-08-15 19:49:43 +0000 +++ b/strings/decimal.c 2009-05-06 12:03:24 +0000 @@ -144,7 +144,7 @@ static double scaler1[]= { 1.0, 10.0, 1e2, 1e3, 1e4, 1e5, 1e6, 1e7, 1e8, 1e9 }; -#ifdef HAVE_purify +#ifdef HAVE_valgrind #define sanity(d) DBUG_ASSERT((d)->len > 0) #else #define sanity(d) DBUG_ASSERT((d)->len >0 && ((d)->buf[0] | \ === modified file 'strings/strmake.c' --- a/strings/strmake.c 2008-04-28 16:24:05 +0000 +++ b/strings/strmake.c 2009-05-06 12:03:24 +0000 @@ -51,11 +51,11 @@ char *strmake(register char *dst, regist */ if (length) { -#ifdef HAVE_purify +#ifdef HAVE_valgrind dst[length-1]= 'Z'; #else bfill(dst, length-1, (int) 'Z'); -#endif /* HAVE_purify */ +#endif /* HAVE_valgrind */ } #endif /* EXTRA_DEBUG */ return dst-1;

1 0

[Maria-developers] bzr commit into MariaDB 5.1, with Maria 1.5:maria branch (knielsen:2699)
by knielsen＠knielsen-hq.org 06 May '09

06 May '09

#At lp:maria 2699 knielsen(a)knielsen-hq.org 2009-04-27 We are now using Valgrind rather than purify, and have for quite some time. Consequently, rename HAVE_purify to HAVE_valgrind, and related changes. Leave some comments about purify when not clear that they apply also to Valgrind. Fix redundant IF_VALGRIND declaration. modified: BUILD/SETUP.sh BUILD/build_mccge.sh BUILD/compile-solaris-sparc-purify include/m_string.h include/my_global.h mysys/mf_qsort.c mysys/my_alloc.c mysys/my_init.c mysys/my_rnd.c mysys/safemalloc.c scripts/mysql_config.pl.in scripts/mysql_config.sh sql/field_conv.cc sql/filesort.cc sql/ha_partition.cc sql/hostname.cc sql/item_timefunc.cc sql/log_event.cc sql/log_event_old.cc sql/my_decimal.h sql/mysqld.cc sql/opt_range.cc sql/opt_range.h sql/records.cc sql/rpl_rli.cc sql/rpl_rli.h sql/slave.cc sql/sql_base.cc sql/sql_binlog.cc sql/sql_class.cc sql/sql_list.h sql/sql_load.cc sql/sql_select.cc sql/table.cc storage/archive/azio.c storage/innobase/buf/buf0buf.c storage/innobase/include/univ.i storage/innobase/srv/srv0start.c storage/maria/ha_maria.cc storage/maria/ma_blockrec.c storage/maria/ma_check.c storage/maria/ma_loghandler.c storage/maria/ma_packrec.c storage/maria/ma_page.c storage/maria/ma_pagecrc.c storage/maria/ma_search.c storage/myisam/mi_check.c storage/myisam/mi_page.c storage/myisam/mi_search.c storage/myisammrg/ha_myisammrg.cc strings/bcmp.c strings/decimal.c strings/strmake.c === modified file 'BUILD/SETUP.sh' --- a/BUILD/SETUP.sh 2009-03-22 12:16:09 +0000 +++ b/BUILD/SETUP.sh 2009-04-27 18:01:20 +0000 @@ -119,7 +119,7 @@ fi # Set flags for various build configurations. # Used in -valgrind builds -valgrind_flags="-USAFEMALLOC -UFORCE_INIT_OF_VARS -DHAVE_purify " +valgrind_flags="-USAFEMALLOC -UFORCE_INIT_OF_VARS -DHAVE_valgrind " valgrind_flags="$valgrind_flags -DMYSQL_SERVER_SUFFIX=-valgrind-max" # # Used in -debug builds === modified file 'BUILD/build_mccge.sh' --- a/BUILD/build_mccge.sh 2008-02-07 23:12:34 +0000 +++ b/BUILD/build_mccge.sh 2009-04-27 18:01:20 +0000 @@ -938,7 +938,7 @@ set_up_ccache() set_valgrind_flags() { if test "x$valgrind_flag" = "xyes" ; then - loc_valgrind_flags="-USAFEMALLOC -UFORCE_INIT_OF_VARS -DHAVE_purify " + loc_valgrind_flags="-USAFEMALLOC -UFORCE_INIT_OF_VARS -DHAVE_valgrind " loc_valgrind_flags="$loc_valgrind_flags -DMYSQL_SERVER_SUFFIX=-valgrind-max" compiler_flags="$compiler_flags $loc_valgrind_flags" fi === modified file 'BUILD/compile-solaris-sparc-purify' --- a/BUILD/compile-solaris-sparc-purify 2007-02-26 15:12:40 +0000 +++ b/BUILD/compile-solaris-sparc-purify 2009-04-27 18:01:20 +0000 @@ -37,7 +37,7 @@ make -k maintainer-clean || true path=`dirname $0` . "$path/autorun.sh" -CFLAGS="-g -Wimplicit -Wreturn-type -Wswitch -Wtrigraphs -Wcomment -W -Wchar-subscripts -Wformat -Wimplicit-int -Wparentheses -Wsign-compare -Wwrite-strings -Wunused -DHAVE_purify -DEXTRA_DEBUG -O2" CXX=gcc CXXLD=g++ CXXFLAGS="-g -Wimplicit -Wreturn-type -Wswitch -Wtrigraphs -Wcomment -W -Wchar-subscripts -Wformat -Wparentheses -Wsign-compare -Wwrite-strings -Woverloaded-virtual -Wsign-promo -Wreorder -Wctor-dtor-privacy -Wnon-virtual-dtor -felide-constructors -fno-exceptions -fno-rtti -DHAVE_purify -DEXTRA_DEBUG -O2" ./configure --prefix=/usr/local/mysql --enable-assembler --with-extra-charsets=complex --enable-thread-safe-client --with-embedded-server --with-innodb $EXTRA_CONFIG_FLAGS +CFLAGS="-g -Wimplicit -Wreturn-type -Wswitch -Wtrigraphs -Wcomment -W -Wchar-subscripts -Wformat -Wimplicit-int -Wparentheses -Wsign-compare -Wwrite-strings -Wunused -DHAVE_valgrind -DEXTRA_DEBUG -O2" CXX=gcc CXXLD=g++ CXXFLAGS="-g -Wimplicit -Wreturn-type -Wswitch -Wtrigraphs -Wcomment -W -Wchar-subscripts -Wformat -Wparentheses -Wsign-compare -Wwrite-strings -Woverloaded-virtual -Wsign-promo -Wreorder -Wctor-dtor-privacy -Wnon-virtual-dtor -felide-constructors -fno-exceptions -fno-rtti -DHAVE_valgrind -DEXTRA_DEBUG -O2" ./configure --prefix=/usr/local/mysql --enable-assembler --with-extra-charsets=complex --enable-thread-safe-client --with-embedded-server --with-innodb $EXTRA_CONFIG_FLAGS make -j 4 === modified file 'include/m_string.h' --- a/include/m_string.h 2008-06-26 15:10:11 +0000 +++ b/include/m_string.h 2009-04-27 18:01:20 +0000 @@ -108,7 +108,7 @@ extern const double log_10[309]; #define memcpy_fixed(A,B,C) memcpy((A),(B),(C)) #endif -#if (!defined(USE_BMOVE512) || defined(HAVE_purify)) && !defined(bmove512) +#if (!defined(USE_BMOVE512) || defined(HAVE_valgrind)) && !defined(bmove512) #define bmove512(A,B,C) memcpy(A,B,C) #endif @@ -125,14 +125,14 @@ extern void bzero(uchar * dst,size_t len #if !defined(bcmp) && !defined(HAVE_BCMP) extern size_t bcmp(const uchar *s1,const uchar *s2,size_t len); #endif -#ifdef HAVE_purify +#ifdef HAVE_valgrind extern size_t my_bcmp(const uchar *s1,const uchar *s2,size_t len); #undef bcmp #define bcmp(A,B,C) my_bcmp((A),(B),(C)) -#define bzero_if_purify(A,B) bzero(A,B) +#define bzero_if_valgrind(A,B) bzero(A,B) #else -#define bzero_if_purify(A,B) -#endif /* HAVE_purify */ +#define bzero_if_valgrind(A,B) +#endif /* HAVE_valgrind */ #if defined(_lint) || defined(FORCE_INIT_OF_VARS) #define LINT_INIT_STRUCT(var) bzero(&var, sizeof(var)) /* No uninitialize-warning */ === modified file 'include/my_global.h' --- a/include/my_global.h 2009-04-08 16:55:26 +0000 +++ b/include/my_global.h 2009-04-27 18:01:20 +0000 @@ -563,16 +563,16 @@ int __void__; #define LINT_INIT(var) #endif -#if defined(_lint) || defined(FORCE_INIT_OF_VARS) || defined(HAVE_purify) -#define PURIFY_OR_LINT_INIT(var) var=0 +#if defined(_lint) || defined(FORCE_INIT_OF_VARS) || defined(HAVE_valgrind) +#define VALGRIND_OR_LINT_INIT(var) var=0 #else -#define PURIFY_OR_LINT_INIT(var) +#define VALGRIND_OR_LINT_INIT(var) #endif -#ifdef HAVE_purify -#define IF_PURIFY(A,B) (A) +#ifdef HAVE_valgrind +#define IF_VALGRIND(A,B) (A) #else -#define IF_PURIFY(A,B) (B) +#define IF_VALGRIND(A,B) (B) #endif #if !defined(HAVE_UINT) @@ -1174,7 +1174,7 @@ typedef char bool; /* Ordinary boolean ((uint32) (uchar) (A)[0]))) #define sint4korr(A) (*((const long *) (A))) #define uint2korr(A) (*((const uint16 *) (A))) -#if defined(HAVE_purify) && !defined(_WIN32) +#if defined(HAVE_valgrind) && !defined(_WIN32) #define uint3korr(A) (uint32) (((uint32) ((uchar) (A)[0])) +\ (((uint32) ((uchar) (A)[1])) << 8) +\ (((uint32) ((uchar) (A)[2])) << 16)) @@ -1186,7 +1186,7 @@ typedef char bool; /* Ordinary boolean It means, that you have to provide enough allocated space ! */ #define uint3korr(A) (long) (*((const unsigned int *) (A)) & 0xFFFFFF) -#endif /* HAVE_purify && !_WIN32 */ +#endif /* HAVE_valgrind && !_WIN32 */ #define uint4korr(A) (*((const uint32 *) (A))) #define uint5korr(A) ((ulonglong)(((uint32) ((uchar) (A)[0])) +\ (((uint32) ((uchar) (A)[1])) << 8) +\ === modified file 'mysys/mf_qsort.c' --- a/mysys/mf_qsort.c 2007-10-17 16:08:58 +0000 +++ b/mysys/mf_qsort.c 2009-04-27 18:01:20 +0000 @@ -108,7 +108,7 @@ qsort_t my_qsort(void *base_ptr, size_t low = (char*) base_ptr; high = low+ size * (count - 1); stack_ptr = stack + 1; -#ifdef HAVE_purify +#ifdef HAVE_valgrind /* The first element in the stack will be accessed for the last POP */ stack[0].low=stack[0].high=0; #endif === modified file 'mysys/my_alloc.c' --- a/mysys/my_alloc.c 2009-01-15 18:11:25 +0000 +++ b/mysys/my_alloc.c 2009-04-27 18:01:20 +0000 @@ -56,7 +56,7 @@ void init_alloc_root(MEM_ROOT *mem_root, mem_root->block_num= 4; /* We shift this with >>2 */ mem_root->first_block_usage= 0; -#if !(defined(HAVE_purify) && defined(EXTRA_DEBUG)) +#if !(defined(HAVE_valgrind) && defined(EXTRA_DEBUG)) if (pre_alloc_size) { if ((mem_root->free= mem_root->pre_alloc= @@ -96,7 +96,7 @@ void reset_root_defaults(MEM_ROOT *mem_r DBUG_ASSERT(alloc_root_inited(mem_root)); mem_root->block_size= block_size - ALLOC_ROOT_MIN_BLOCK_SIZE; -#if !(defined(HAVE_purify) && defined(EXTRA_DEBUG)) +#if !(defined(HAVE_valgrind) && defined(EXTRA_DEBUG)) if (pre_alloc_size) { size_t size= pre_alloc_size + ALIGN_SIZE(sizeof(USED_MEM)); @@ -147,7 +147,7 @@ void reset_root_defaults(MEM_ROOT *mem_r void *alloc_root(MEM_ROOT *mem_root, size_t length) { -#if defined(HAVE_purify) && defined(EXTRA_DEBUG) +#if defined(HAVE_valgrind) && defined(EXTRA_DEBUG) reg1 USED_MEM *next; DBUG_ENTER("alloc_root"); DBUG_PRINT("enter",("root: 0x%lx", (long) mem_root)); === modified file 'mysys/my_init.c' --- a/mysys/my_init.c 2009-02-19 09:01:25 +0000 +++ b/mysys/my_init.c 2009-04-27 18:01:20 +0000 @@ -170,7 +170,7 @@ void my_end(int infoflag) { #ifdef HAVE_GETRUSAGE struct rusage rus; -#ifdef HAVE_purify +#ifdef HAVE_valgrind /* Purify assumes that rus is uninitialized after getrusage call */ bzero((char*) &rus, sizeof(rus)); #endif === modified file 'mysys/my_rnd.c' --- a/mysys/my_rnd.c 2007-12-14 08:47:00 +0000 +++ b/mysys/my_rnd.c 2009-04-27 18:01:20 +0000 @@ -26,7 +26,7 @@ void my_rnd_init(struct my_rnd_struct *rand_st, ulong seed1, ulong seed2) { -#ifdef HAVE_purify +#ifdef HAVE_valgrind bzero((char*) rand_st,sizeof(*rand_st)); /* Avoid UMC varnings */ #endif rand_st->max_value= 0x3FFFFFFFL; === modified file 'mysys/safemalloc.c' --- a/mysys/safemalloc.c 2008-10-10 15:28:41 +0000 +++ b/mysys/safemalloc.c 2009-04-27 18:01:20 +0000 @@ -304,7 +304,7 @@ void _myfree(void *ptr, const char *file sf_malloc_count--; pthread_mutex_unlock(&THR_LOCK_malloc); -#ifndef HAVE_purify +#ifndef HAVE_valgrind /* Mark this data as free'ed */ if (!sf_malloc_quick) bfill(ptr, irem->datasize, (pchar) FREE_VAL); === modified file 'scripts/mysql_config.pl.in' --- a/scripts/mysql_config.pl.in 2008-11-10 21:12:15 +0000 +++ b/scripts/mysql_config.pl.in 2009-04-27 18:01:20 +0000 @@ -44,7 +44,7 @@ use strict; my @exclude_cflags = qw/DDBUG_OFF DSAFEMALLOC USAFEMALLOC DSAFE_MUTEX DPEDANTIC_SAFEMALLOC DUNIV_MUST_NOT_INLINE DFORCE_INIT_OF_VARS - DEXTRA_DEBUG DHAVE_purify O O[0-9] xO[0-9] W[-A-Za-z]* + DEXTRA_DEBUG DHAVE_valgrind O O[0-9] xO[0-9] W[-A-Za-z]* Xa xstrconst xc99=none unroll2 ip mp restrict/; === modified file 'scripts/mysql_config.sh' --- a/scripts/mysql_config.sh 2008-11-10 21:12:15 +0000 +++ b/scripts/mysql_config.sh 2009-04-27 18:01:20 +0000 @@ -129,7 +129,7 @@ include="-I$pkgincludedir" # to make --cflags usable for HP C++ (aCC) for remove in DDBUG_OFF DSAFEMALLOC USAFEMALLOC DSAFE_MUTEX \ DPEDANTIC_SAFEMALLOC DUNIV_MUST_NOT_INLINE DFORCE_INIT_OF_VARS \ - DEXTRA_DEBUG DHAVE_purify O 'O[0-9]' 'xO[0-9]' 'W[-A-Za-z]*' \ + DEXTRA_DEBUG DHAVE_valgrind O 'O[0-9]' 'xO[0-9]' 'W[-A-Za-z]*' \ 'mtune=[-A-Za-z0-9]*' 'mcpu=[-A-Za-z0-9]*' 'march=[-A-Za-z0-9]*' \ Xa xstrconst "xc99=none" AC99 \ unroll2 ip mp restrict === modified file 'sql/field_conv.cc' --- a/sql/field_conv.cc 2009-01-31 21:22:44 +0000 +++ b/sql/field_conv.cc 2009-04-27 18:01:20 +0000 @@ -773,7 +773,7 @@ int field_conv(Field *to,Field *from) ((Field_varstring*)from)->length_bytes == ((Field_varstring*)to)->length_bytes)) { // Identical fields -#ifdef HAVE_purify +#ifdef HAVE_valgrind /* This may happen if one does 'UPDATE ... SET x=x' */ if (to->ptr != from->ptr) #endif === modified file 'sql/filesort.cc' --- a/sql/filesort.cc 2009-02-19 09:01:25 +0000 +++ b/sql/filesort.cc 2009-04-27 18:01:20 +0000 @@ -953,13 +953,13 @@ static void make_sortkey(register SORTPA if (addonf->null_bit && field->is_null()) { nulls[addonf->null_offset]|= addonf->null_bit; -#ifdef HAVE_purify +#ifdef HAVE_valgrind bzero(to, addonf->length); #endif } else { -#ifdef HAVE_purify +#ifdef HAVE_valgrind uchar *end= field->pack(to, field->ptr); uint length= (uint) ((to + addonf->length) - end); DBUG_ASSERT((int) length >= 0); === modified file 'sql/ha_partition.cc' --- a/sql/ha_partition.cc 2009-02-19 09:01:25 +0000 +++ b/sql/ha_partition.cc 2009-04-27 18:01:20 +0000 @@ -3548,10 +3548,10 @@ void ha_partition::position(const uchar (ref_length - PARTITION_BYTES_IN_POS)); #ifdef SUPPORTING_PARTITION_OVER_DIFFERENT_ENGINES -#ifdef HAVE_purify +#ifdef HAVE_valgrind bzero(ref + PARTITION_BYTES_IN_POS + ref_length, max_ref_length-ref_length); -#endif /* HAVE_purify */ +#endif /* HAVE_valgrind */ #endif DBUG_VOID_RETURN; } === modified file 'sql/hostname.cc' --- a/sql/hostname.cc 2007-10-11 17:29:09 +0000 +++ b/sql/hostname.cc 2009-04-27 18:01:20 +0000 @@ -175,7 +175,7 @@ char * ip_to_hostname(struct in_addr *in char buff[GETHOSTBYADDR_BUFF_SIZE],buff2[GETHOSTBYNAME_BUFF_SIZE]; int tmp_errno; struct hostent tmp_hostent, tmp_hostent2; -#ifdef HAVE_purify +#ifdef HAVE_valgrind bzero(buff,sizeof(buff)); // Bug in purify #endif if (!(hp=gethostbyaddr_r((char*) in,sizeof(*in), === modified file 'sql/item_timefunc.cc' --- a/sql/item_timefunc.cc 2009-02-15 10:58:34 +0000 +++ b/sql/item_timefunc.cc 2009-04-27 18:01:20 +0000 @@ -290,8 +290,8 @@ static bool extract_date_time(DATE_TIME_ LINT_INIT(strict_week_number); /* Remove valgrind varnings when using gcc 3.3 and -O1 */ - PURIFY_OR_LINT_INIT(strict_week_number_year_type); - PURIFY_OR_LINT_INIT(sunday_first_n_first_week_non_iso); + VALGRIND_OR_LINT_INIT(strict_week_number_year_type); + VALGRIND_OR_LINT_INIT(sunday_first_n_first_week_non_iso); if (!sub_pattern_end) bzero((char*) l_time, sizeof(*l_time)); === modified file 'sql/log_event.cc' --- a/sql/log_event.cc 2009-02-19 09:01:25 +0000 +++ b/sql/log_event.cc 2009-04-27 18:01:20 +0000 @@ -6990,7 +6990,7 @@ int Rows_log_event::do_add_row_data(ucha Don't print debug messages when running valgrind since they can trigger false warnings. */ -#ifndef HAVE_purify +#ifndef HAVE_valgrind DBUG_DUMP("row_data", row_data, min(length, 32)); #endif @@ -7771,7 +7771,7 @@ Table_map_log_event::Table_map_log_event Don't print debug messages when running valgrind since they can trigger false warnings. */ -#ifndef HAVE_purify +#ifndef HAVE_valgrind DBUG_DUMP("event buffer", (uchar*) buf, event_len); #endif @@ -8674,7 +8674,7 @@ int Rows_log_event::find_row(const Relay Don't print debug messages when running valgrind since they can trigger false warnings. */ -#ifndef HAVE_purify +#ifndef HAVE_valgrind DBUG_DUMP("key data", m_key, table->key_info->key_length); #endif @@ -8704,7 +8704,7 @@ int Rows_log_event::find_row(const Relay Don't print debug messages when running valgrind since they can trigger false warnings. */ -#ifndef HAVE_purify +#ifndef HAVE_valgrind DBUG_PRINT("info",("found first matching record")); DBUG_DUMP("record[0]", table->record[0], table->s->reclength); #endif @@ -9067,7 +9067,7 @@ Update_rows_log_event::do_exec_row(const Now we have the right row to update. The old row (the one we're looking for) is in record[1] and the new row is in record[0]. */ -#ifndef HAVE_purify +#ifndef HAVE_valgrind /* Don't print debug messages when running valgrind since they can trigger false warnings. === modified file 'sql/log_event_old.cc' --- a/sql/log_event_old.cc 2009-01-31 21:22:44 +0000 +++ b/sql/log_event_old.cc 2009-04-27 18:01:20 +0000 @@ -724,7 +724,7 @@ static int find_and_fetch_row(TABLE *tab Don't print debug messages when running valgrind since they can trigger false warnings. */ -#ifndef HAVE_purify +#ifndef HAVE_valgrind DBUG_DUMP("table->record[0]", table->record[0], table->s->reclength); DBUG_DUMP("table->record[1]", table->record[1], table->s->reclength); #endif @@ -750,7 +750,7 @@ static int find_and_fetch_row(TABLE *tab Don't print debug messages when running valgrind since they can trigger false warnings. */ -#ifndef HAVE_purify +#ifndef HAVE_valgrind DBUG_DUMP("table->record[0]", table->record[0], table->s->reclength); DBUG_DUMP("table->record[1]", table->record[1], table->s->reclength); #endif @@ -1404,7 +1404,7 @@ int Old_rows_log_event::do_add_row_data( Don't print debug messages when running valgrind since they can trigger false warnings. */ -#ifndef HAVE_purify +#ifndef HAVE_valgrind DBUG_DUMP("row_data", row_data, min(length, 32)); #endif @@ -2353,7 +2353,7 @@ int Old_rows_log_event::find_row(const R Don't print debug messages when running valgrind since they can trigger false warnings. */ -#ifndef HAVE_purify +#ifndef HAVE_valgrind DBUG_DUMP("key data", m_key, table->key_info->key_length); #endif @@ -2383,7 +2383,7 @@ int Old_rows_log_event::find_row(const R Don't print debug messages when running valgrind since they can trigger false warnings. */ -#ifndef HAVE_purify +#ifndef HAVE_valgrind DBUG_PRINT("info",("found first matching record")); DBUG_DUMP("record[0]", table->record[0], table->s->reclength); #endif @@ -2878,7 +2878,7 @@ Update_rows_log_event_old::do_exec_row(c Now we have the right row to update. The old row (the one we're looking for) is in record[1] and the new row is in record[0]. */ -#ifndef HAVE_purify +#ifndef HAVE_valgrind /* Don't print debug messages when running valgrind since they can trigger false warnings. === modified file 'sql/my_decimal.h' --- a/sql/my_decimal.h 2008-05-20 07:38:17 +0000 +++ b/sql/my_decimal.h 2009-04-27 18:01:20 +0000 @@ -99,7 +99,7 @@ public: { len= DECIMAL_BUFF_LENGTH; buf= buffer; -#if !defined (HAVE_purify) && !defined(DBUG_OFF) +#if !defined (HAVE_valgrind) && !defined(DBUG_OFF) /* Set buffer to 'random' value to find wrong buffer usage */ for (uint i= 0; i < DECIMAL_BUFF_LENGTH; i++) buffer[i]= i; === modified file 'sql/mysqld.cc' --- a/sql/mysqld.cc 2009-04-08 21:16:10 +0000 +++ b/sql/mysqld.cc 2009-04-27 18:01:20 +0000 @@ -57,12 +57,6 @@ #define mysqld_charset &my_charset_latin1 -#ifdef HAVE_purify -#define IF_PURIFY(A,B) (A) -#else -#define IF_PURIFY(A,B) (B) -#endif - #if SIZEOF_CHARP == 4 #define MAX_MEM_TABLE_SIZE ~(ulong) 0 #else @@ -76,9 +70,9 @@ char pstack_file_name[80]; #endif /* __linux__ */ -/* We have HAVE_purify below as this speeds up the shutdown of MySQL */ +/* We have HAVE_valgrind below as this speeds up the shutdown of MySQL */ -#if defined(HAVE_DEC_3_2_THREADS) || defined(SIGNALS_DONT_BREAK_READ) || defined(HAVE_purify) && defined(__linux__) +#if defined(HAVE_DEC_3_2_THREADS) || defined(SIGNALS_DONT_BREAK_READ) || defined(HAVE_valgrind) && defined(__linux__) #define HAVE_CLOSE_SERVER_SOCK 1 #endif @@ -6520,7 +6514,7 @@ log and this option does nothing anymore purify. These are not suppressed: instead we disable symlinks option if compiled with valgrind support. */ - IF_PURIFY(0,1), 0, 0, 0, 0, 0}, + IF_VALGRIND(0,1), 0, 0, 0, 0, 0}, {"sysdate-is-now", OPT_SYSDATE_IS_NOW, "Non-default option to alias SYSDATE() to NOW() to make it safe-replicable. Since 5.0, SYSDATE() returns a `dynamic' value different for different invocations, even within the same statement.", (uchar**) &global_system_variables.sysdate_is_now, @@ -6561,7 +6555,7 @@ log and this option does nothing anymore 0, 0, 0, 0, 0}, {"use-symbolic-links", 's', "Enable symbolic link support. Deprecated option; use --symbolic-links instead.", (uchar**) &my_use_symdir, (uchar**) &my_use_symdir, 0, GET_BOOL, NO_ARG, - IF_PURIFY(0,1), 0, 0, 0, 0, 0}, + IF_VALGRIND(0,1), 0, 0, 0, 0, 0}, {"user", 'u', "Run mysqld daemon as user.", 0, 0, 0, GET_STR, REQUIRED_ARG, 0, 0, 0, 0, 0, 0}, {"verbose", 'v', "Used with --help option for detailed help", @@ -7661,7 +7655,7 @@ static int mysql_init_variables(void) /* Things reset to zero */ opt_skip_slave_start= opt_reckless_slave = 0; mysql_home[0]= pidfile_name[0]= log_error_file[0]= 0; -#if defined(HAVE_REALPATH) && !defined(HAVE_purify) && !defined(HAVE_BROKEN_REALPATH) +#if defined(HAVE_REALPATH) && !defined(HAVE_valgrind) && !defined(HAVE_BROKEN_REALPATH) /* We can only test for sub paths if my_symlink.c is using realpath */ myisam_test_invalid_symlink= test_if_data_home_dir; #endif === modified file 'sql/opt_range.cc' --- a/sql/opt_range.cc 2009-02-19 09:01:25 +0000 +++ b/sql/opt_range.cc 2009-04-27 18:01:20 +0000 @@ -8347,7 +8347,7 @@ int QUICK_RANGE_SELECT::reset() multi_range_buff->buffer= mrange_buff; multi_range_buff->buffer_end= mrange_buff + mrange_bufsiz; multi_range_buff->end_of_used_area= mrange_buff; -#ifdef HAVE_purify +#ifdef HAVE_valgrind /* We need this until ndb will use the buffer efficiently (Now ndb stores complete row in here, instead of only the used fields === modified file 'sql/opt_range.h' --- a/sql/opt_range.h 2008-08-25 17:18:22 +0000 +++ b/sql/opt_range.h 2009-04-27 18:01:20 +0000 @@ -44,7 +44,7 @@ class QUICK_RANGE :public Sql_alloc { uint16 min_length,max_length,flag; key_part_map min_keypart_map, // bitmap of used keyparts in min_key max_keypart_map; // bitmap of used keyparts in max_key -#ifdef HAVE_purify +#ifdef HAVE_valgrind uint16 dummy; /* Avoid warnings on 'flag' */ #endif QUICK_RANGE(); /* Full range */ @@ -61,7 +61,7 @@ class QUICK_RANGE :public Sql_alloc { min_keypart_map(min_keypart_map_arg), max_keypart_map(max_keypart_map_arg) { -#ifdef HAVE_purify +#ifdef HAVE_valgrind dummy=0; #endif } === modified file 'sql/records.cc' --- a/sql/records.cc 2008-07-17 18:26:55 +0000 +++ b/sql/records.cc 2009-04-27 18:01:20 +0000 @@ -520,7 +520,7 @@ static int init_rr_cache(THD *thd, READ_ info->struct_length+1, MYF(0)))) DBUG_RETURN(1); -#ifdef HAVE_purify +#ifdef HAVE_valgrind // Avoid warnings in qsort bzero(info->cache,rec_cache_size+info->cache_records* info->struct_length+1); #endif === modified file 'sql/rpl_rli.cc' --- a/sql/rpl_rli.cc 2009-02-19 09:01:25 +0000 +++ b/sql/rpl_rli.cc 2009-04-27 18:01:20 +0000 @@ -34,7 +34,7 @@ Relay_log_info::Relay_log_info() no_storage(FALSE), replicate_same_server_id(::replicate_same_server_id), info_fd(-1), cur_log_fd(-1), save_temporary_tables(0), cur_log_old_open_count(0), group_relay_log_pos(0), event_relay_log_pos(0), -#if HAVE_purify +#if HAVE_valgrind is_fake(FALSE), #endif group_master_log_pos(0), log_space_total(0), ignore_log_space_limit(0), === modified file 'sql/rpl_rli.h' --- a/sql/rpl_rli.h 2008-02-27 17:46:06 +0000 +++ b/sql/rpl_rli.h 2009-04-27 18:01:20 +0000 @@ -154,7 +154,7 @@ public: ulonglong event_relay_log_pos; ulonglong future_event_relay_log_pos; -#ifdef HAVE_purify +#ifdef HAVE_valgrind bool is_fake; /* Mark that this is a fake relay log info structure */ #endif === modified file 'sql/slave.cc' --- a/sql/slave.cc 2009-02-19 09:01:25 +0000 +++ b/sql/slave.cc 2009-04-27 18:01:20 +0000 @@ -2014,7 +2014,7 @@ int apply_event_and_update_pos(Log_event if (exec_res == 0) { int error= ev->update_pos(rli); -#ifdef HAVE_purify +#ifdef HAVE_valgrind if (!rli->is_fake) #endif { === modified file 'sql/sql_base.cc' --- a/sql/sql_base.cc 2009-02-19 09:01:25 +0000 +++ b/sql/sql_base.cc 2009-04-27 18:01:20 +0000 @@ -3000,7 +3000,7 @@ TABLE *open_table(THD *thd, TABLE_LIST * table->pos_in_table_list= table_list; table_list->updatable= 1; // It is not derived table nor non-updatable VIEW table->clear_column_bitmaps(); -#if !defined(DBUG_OFF) && !defined(HAVE_purify) +#if !defined(DBUG_OFF) && !defined(HAVE_valgrind) /* Fill record with random values to find bugs where we access fields without first reading them. @@ -7363,7 +7363,7 @@ int setup_wild(THD *thd, TABLE_LIST *tab /* make * substituting permanent */ SELECT_LEX *select_lex= thd->lex->current_select; select_lex->with_wild= 0; -#ifdef HAVE_purify +#ifdef HAVE_valgrind if (&select_lex->item_list != &fields) // Avoid warning #endif select_lex->item_list= fields; === modified file 'sql/sql_binlog.cc' --- a/sql/sql_binlog.cc 2009-01-09 12:49:24 +0000 +++ b/sql/sql_binlog.cc 2009-04-27 18:01:20 +0000 @@ -59,7 +59,7 @@ void mysql_client_binlog_statement(THD* if (!thd->rli_fake) { thd->rli_fake= new Relay_log_info; -#ifdef HAVE_purify +#ifdef HAVE_valgrind thd->rli_fake->is_fake= TRUE; #endif have_fd_event= FALSE; @@ -95,7 +95,7 @@ void mysql_client_binlog_statement(THD* char const *endptr= 0; int bytes_decoded= base64_decode(strptr, coded_len, buf, &endptr); -#ifndef HAVE_purify +#ifndef HAVE_valgrind /* This debug printout should not be used for valgrind builds since it will read from unassigned memory. @@ -188,7 +188,7 @@ void mysql_client_binlog_statement(THD* bufptr += event_len; DBUG_PRINT("info",("ev->get_type_code()=%d", ev->get_type_code())); -#ifndef HAVE_purify +#ifndef HAVE_valgrind /* This debug printout should not be used for valgrind builds since it will read from unassigned memory. === modified file 'sql/sql_class.cc' --- a/sql/sql_class.cc 2009-03-12 22:27:35 +0000 +++ b/sql/sql_class.cc 2009-04-27 18:01:20 +0000 @@ -3480,7 +3480,7 @@ int THD::binlog_update_row(TABLE* table, Don't print debug messages when running valgrind since they can trigger false warnings. */ -#ifndef HAVE_purify +#ifndef HAVE_valgrind DBUG_DUMP("before_record", before_record, table->s->reclength); DBUG_DUMP("after_record", after_record, table->s->reclength); DBUG_DUMP("before_row", before_row, before_size); === modified file 'sql/sql_list.h' --- a/sql/sql_list.h 2009-01-15 18:11:25 +0000 +++ b/sql/sql_list.h 2009-04-27 18:01:20 +0000 @@ -43,7 +43,7 @@ public: static void operator delete[](void *ptr, MEM_ROOT *mem_root) { /* never called */ } static void operator delete[](void *ptr, size_t size) { TRASH(ptr, size); } -#ifdef HAVE_purify +#ifdef HAVE_valgrind bool dummy; inline Sql_alloc() :dummy(0) {} inline ~Sql_alloc() {} === modified file 'sql/sql_load.cc' --- a/sql/sql_load.cc 2009-01-31 21:22:44 +0000 +++ b/sql/sql_load.cc 2009-04-27 18:01:20 +0000 @@ -581,7 +581,7 @@ read_fixed_length(THD *thd, COPY_INFO &i } it.rewind(); uchar *pos=read_info.row_start; -#ifdef HAVE_purify +#ifdef HAVE_valgrind read_info.row_end[0]=0; #endif === modified file 'sql/sql_select.cc' --- a/sql/sql_select.cc 2009-04-08 16:55:26 +0000 +++ b/sql/sql_select.cc 2009-04-27 18:01:20 +0000 @@ -1818,7 +1818,7 @@ JOIN::exec() curr_join->having= curr_join->tmp_having= 0; // Allready done /* Change sum_fields reference to calculated fields in tmp_table */ -#ifdef HAVE_purify +#ifdef HAVE_valgrind if (curr_join != this) #endif curr_join->all_fields= *curr_all_fields; @@ -1840,7 +1840,7 @@ JOIN::exec() fields_list.elements, all_fields)) DBUG_VOID_RETURN; } -#ifdef HAVE_purify +#ifdef HAVE_valgrind if (curr_join != this) #endif { @@ -1994,7 +1994,7 @@ JOIN::exec() tmp_fields_list2, tmp_all_fields2, fields_list.elements, tmp_all_fields1)) DBUG_VOID_RETURN; -#ifdef HAVE_purify +#ifdef HAVE_valgrind /* Some GCCs use memcpy() for struct assignment, even for x=x. GCC bug 19410: http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19410 @@ -2059,7 +2059,7 @@ JOIN::exec() tmp_table_param.save_copy_field= curr_join->tmp_table_param.copy_field; tmp_table_param.save_copy_field_end= curr_join->tmp_table_param.copy_field_end; -#ifdef HAVE_purify +#ifdef HAVE_valgrind if (curr_join != this) #endif { @@ -3848,7 +3848,7 @@ update_ref_and_keys(THD *thd, DYNAMIC_AR continue; } -#ifdef HAVE_purify +#ifdef HAVE_valgrind /* Valgrind complains about overlapped memcpy when save_pos==use. */ if (save_pos != use) #endif @@ -15138,7 +15138,7 @@ setup_copy_fields(THD *thd, TMP_TABLE_PA DBUG_ASSERT (param->field_count > (uint) (copy - copy_start)); copy->set(tmp, item->result_field); item->result_field->move_field(copy->to_ptr,copy->to_null_ptr,1); -#ifdef HAVE_purify +#ifdef HAVE_valgrind copy->to_ptr[copy->from_length]= 0; #endif copy++; === modified file 'sql/table.cc' --- a/sql/table.cc 2009-02-19 09:01:25 +0000 +++ b/sql/table.cc 2009-04-27 18:01:20 +0000 @@ -1694,7 +1694,7 @@ int open_table_from_share(THD *thd, TABL outparam->record[1]= outparam->record[0]; // Safety } -#ifdef HAVE_purify +#ifdef HAVE_valgrind /* We need this because when we read var-length rows, we are not updating bytes after end of varchar @@ -1916,7 +1916,7 @@ partititon_err: } } -#if defined(HAVE_purify) && !defined(DBUG_OFF) +#if defined(HAVE_valgrind) && !defined(DBUG_OFF) bzero((char*) bitmaps, bitmap_size*3); #endif === modified file 'storage/archive/azio.c' --- a/storage/archive/azio.c 2009-04-08 16:55:26 +0000 +++ b/storage/archive/azio.c 2009-04-27 18:01:20 +0000 @@ -63,7 +63,7 @@ void read_header(azio_stream *s, unsigne */ static void *az_allocator(void *dummy, uInt items, uInt size) { - return my_malloc((size_t)items*(size_t)size, IF_PURIFY(MY_ZEROFILL, MYF(0))); + return my_malloc((size_t)items*(size_t)size, IF_VALGRIND(MY_ZEROFILL, MYF(0))); } static void az_free(void *dummy, void *address) === modified file 'storage/innobase/buf/buf0buf.c' --- a/storage/innobase/buf/buf0buf.c 2008-08-20 00:37:41 +0000 +++ b/storage/innobase/buf/buf0buf.c 2009-04-27 18:01:20 +0000 @@ -756,7 +756,7 @@ buf_pool_init( /* Wipe contents of frame to eliminate a Purify warning */ -#ifdef HAVE_purify +#ifdef HAVE_valgrind memset(block->frame, '\0', UNIV_PAGE_SIZE); #endif if (srv_use_awe) { === modified file 'storage/innobase/include/univ.i' --- a/storage/innobase/include/univ.i 2008-03-27 01:40:45 +0000 +++ b/storage/innobase/include/univ.i 2009-04-27 18:01:20 +0000 @@ -107,7 +107,7 @@ operations (very slow); also UNIV_DEBUG #define UNIV_BTR_DEBUG /* check B-tree links */ #define UNIV_LIGHT_MEM_DEBUG /* light memory debugging */ -#ifdef HAVE_purify +#ifdef HAVE_valgrind /* The following sets all new allocated memory to zero before use: this can be used to eliminate unnecessary Purify warnings, but note that it also masks many bugs Purify could detect. For detailed Purify analysis it === modified file 'storage/innobase/srv/srv0start.c' --- a/storage/innobase/srv/srv0start.c 2008-12-14 19:21:24 +0000 +++ b/storage/innobase/srv/srv0start.c 2009-04-27 18:01:20 +0000 @@ -104,7 +104,7 @@ static char* srv_monitor_file_name; /* Avoid warnings when using purify */ -#ifdef HAVE_purify +#ifdef HAVE_valgrind static int inno_bcmp(register const char *s1, register const char *s2, register uint len) { === modified file 'storage/maria/ha_maria.cc' --- a/storage/maria/ha_maria.cc 2009-02-19 09:01:25 +0000 +++ b/storage/maria/ha_maria.cc 2009-04-27 18:01:20 +0000 @@ -3166,7 +3166,7 @@ static int ha_maria_init(void *p) ma_checkpoint_init(checkpoint_interval); maria_multi_threaded= maria_in_ha_maria= TRUE; -#if defined(HAVE_REALPATH) && !defined(HAVE_purify) && !defined(HAVE_BROKEN_REALPATH) +#if defined(HAVE_REALPATH) && !defined(HAVE_valgrind) && !defined(HAVE_BROKEN_REALPATH) /* We can only test for sub paths if my_symlink.c is using realpath */ maria_test_invalid_symlink= test_if_data_home_dir; #endif === modified file 'storage/maria/ma_blockrec.c' --- a/storage/maria/ma_blockrec.c 2009-04-08 08:47:30 +0000 +++ b/storage/maria/ma_blockrec.c 2009-04-27 18:01:20 +0000 @@ -1594,7 +1594,7 @@ static void make_empty_page(MARIA_HA *in bzero(buff, PAGE_HEADER_SIZE); -#if !defined(DONT_ZERO_PAGE_BLOCKS) || defined(HAVE_purify) +#if !defined(DONT_ZERO_PAGE_BLOCKS) || defined(HAVE_valgrind) /* We zero the rest of the block to avoid getting old memory information to disk and to allow the file to be compressed better if archived. === modified file 'storage/maria/ma_check.c' --- a/storage/maria/ma_check.c 2009-02-19 09:01:25 +0000 +++ b/storage/maria/ma_check.c 2009-04-27 18:01:20 +0000 @@ -4522,7 +4522,7 @@ static int sort_key_read(MARIA_SORT_PARA sort_param->current_filepos, 0); sort_param->real_key_length= int_key.data_length + int_key.ref_length; -#ifdef HAVE_purify +#ifdef HAVE_valgrind bzero(key+sort_param->real_key_length, (sort_param->key_length-sort_param->real_key_length)); #endif @@ -4567,7 +4567,7 @@ static int sort_maria_ft_key_read(MARIA_ sort_param->current_filepos); sort_param->real_key_length= int_key.data_length + int_key.ref_length; -#ifdef HAVE_purify +#ifdef HAVE_valgrind if (sort_param->key_length > sort_param->real_key_length) bzero(key+sort_param->real_key_length, (sort_param->key_length-sort_param->real_key_length)); @@ -5104,7 +5104,7 @@ static int sort_get_next_record(MARIA_SO llstr(sort_param->pos,llbuff)); continue; } -#ifdef HAVE_purify +#ifdef HAVE_valgrind bzero(sort_param->rec_buff + block_info.rec_len, share->base.extra_rec_buff_size); #endif === modified file 'storage/maria/ma_loghandler.c' --- a/storage/maria/ma_loghandler.c 2009-03-17 19:05:01 +0000 +++ b/storage/maria/ma_loghandler.c 2009-04-27 18:01:20 +0000 @@ -6194,12 +6194,12 @@ my_bool translog_write_record(LSN *lsn, { uint i; uint len= 0; -#ifdef HAVE_purify +#ifdef HAVE_valgrind ha_checksum checksum= 0; #endif for (i= TRANSLOG_INTERNAL_PARTS; i < part_no; i++) { -#ifdef HAVE_purify +#ifdef HAVE_valgrind /* Find unitialized bytes early */ checksum+= my_checksum(checksum, parts_data[i].str, parts_data[i].length); === modified file 'storage/maria/ma_packrec.c' --- a/storage/maria/ma_packrec.c 2009-01-09 04:23:25 +0000 +++ b/storage/maria/ma_packrec.c 2009-04-27 18:01:20 +0000 @@ -256,7 +256,7 @@ static my_bool _ma_read_pack_info(MARIA_ (uint) (share->pack.header_length-sizeof(header)), MYF(MY_NABP))) goto err2; -#ifdef HAVE_purify +#ifdef HAVE_valgrind /* Zero bytes accessed by fill_buffer */ bzero(disk_cache + (share->pack.header_length-sizeof(header)), share->base.extra_rec_buff_size); === modified file 'storage/maria/ma_page.c' --- a/storage/maria/ma_page.c 2008-09-01 17:31:40 +0000 +++ b/storage/maria/ma_page.c 2009-04-27 18:01:20 +0000 @@ -200,7 +200,7 @@ my_bool _ma_write_keypage(MARIA_PAGE *pa /* Verify that keynr is correct */ DBUG_ASSERT(_ma_get_keynr(share, buff) == page->keyinfo->key_nr); -#if defined(EXTRA_DEBUG) && defined(HAVE_purify) && defined(NOT_ANYMORE) +#if defined(EXTRA_DEBUG) && defined(HAVE_valgrind) && defined(NOT_ANYMORE) { /* This is here to catch uninitialized bytes */ uint length= page->size; === modified file 'storage/maria/ma_pagecrc.c' --- a/storage/maria/ma_pagecrc.c 2008-03-04 11:58:21 +0000 +++ b/storage/maria/ma_pagecrc.c 2009-04-27 18:01:20 +0000 @@ -318,7 +318,7 @@ my_bool maria_page_filler_set_none(uchar __attribute__((unused)), uchar *data_ptr __attribute__((unused))) { -#ifdef HAVE_purify +#ifdef HAVE_valgrind int4store_aligned(page + ((MARIA_SHARE *)data_ptr)->block_size - CRC_SIZE, 0); #endif === modified file 'storage/maria/ma_search.c' --- a/storage/maria/ma_search.c 2009-01-12 11:12:00 +0000 +++ b/storage/maria/ma_search.c 2009-04-27 18:01:20 +0000 @@ -2172,7 +2172,7 @@ int _ma_calc_bin_pack_key_length(const M s_temp->totlength= key_length= (int_key->data_length + int_key->ref_length+ nod_flag); -#ifdef HAVE_purify +#ifdef HAVE_valgrind s_temp->n_length= s_temp->n_ref_length=0; /* For valgrind */ #endif s_temp->key=key; === modified file 'storage/myisam/mi_check.c' --- a/storage/myisam/mi_check.c 2009-02-19 09:01:25 +0000 +++ b/storage/myisam/mi_check.c 2009-04-27 18:01:20 +0000 @@ -3147,7 +3147,7 @@ static int sort_key_read(MI_SORT_PARAM * (info->s->rec_reflength+ _mi_make_key(info, sort_param->key, (uchar*) key, sort_param->record, sort_param->filepos)); -#ifdef HAVE_purify +#ifdef HAVE_valgrind bzero(key+sort_param->real_key_length, (sort_param->key_length-sort_param->real_key_length)); #endif @@ -3187,7 +3187,7 @@ static int sort_ft_key_read(MI_SORT_PARA sort_param->real_key_length=(info->s->rec_reflength+ _ft_make_key(info, sort_param->key, key, wptr++, sort_param->filepos)); -#ifdef HAVE_purify +#ifdef HAVE_valgrind if (sort_param->key_length > sort_param->real_key_length) bzero(key+sort_param->real_key_length, (sort_param->key_length-sort_param->real_key_length)); === modified file 'storage/myisam/mi_page.c' --- a/storage/myisam/mi_page.c 2008-08-28 12:43:44 +0000 +++ b/storage/myisam/mi_page.c 2009-04-27 18:01:20 +0000 @@ -86,7 +86,7 @@ int _mi_write_keypage(register MI_INFO * if ((length=keyinfo->block_length) > IO_SIZE*2 && info->state->key_file_length != page+length) length= ((mi_getint(buff)+IO_SIZE-1) & (uint) ~(IO_SIZE-1)); -#ifdef HAVE_purify +#ifdef HAVE_valgrind { length=mi_getint(buff); bzero((uchar*) buff+length,keyinfo->block_length-length); === modified file 'storage/myisam/mi_search.c' --- a/storage/myisam/mi_search.c 2008-08-28 18:52:23 +0000 +++ b/storage/myisam/mi_search.c 2009-04-27 18:01:20 +0000 @@ -1749,7 +1749,7 @@ _mi_calc_bin_pack_key_length(MI_KEYDEF * uint length,key_length,ref_length; s_temp->totlength=key_length=_mi_keylength(keyinfo,key)+nod_flag; -#ifdef HAVE_purify +#ifdef HAVE_valgrind s_temp->n_length= s_temp->n_ref_length=0; /* For valgrind */ #endif s_temp->key=key; === modified file 'storage/myisammrg/ha_myisammrg.cc' --- a/storage/myisammrg/ha_myisammrg.cc 2008-12-04 00:36:55 +0000 +++ b/storage/myisammrg/ha_myisammrg.cc 2009-04-27 18:01:20 +0000 @@ -837,7 +837,7 @@ int ha_myisammrg::info(uint flag) { if (table->s->key_parts && mrg_info.rec_per_key) { -#ifdef HAVE_purify +#ifdef HAVE_valgrind /* valgrind may be unhappy about it, because optimizer may access values between file->keys and table->key_parts, that will be uninitialized. === modified file 'strings/bcmp.c' --- a/strings/bcmp.c 2007-05-10 09:59:39 +0000 +++ b/strings/bcmp.c 2009-04-27 18:01:20 +0000 @@ -24,7 +24,7 @@ #include <my_global.h> #include "m_string.h" -#ifdef HAVE_purify +#ifdef HAVE_valgrind #undef bcmp #undef HAVE_BCMP #endif @@ -50,7 +50,7 @@ uint len; /* 0 <= len <= 65535 */ #else -#ifndef HAVE_purify +#ifndef HAVE_valgrind size_t bcmp(register const uchar *s1,register const uchar *s2, register size_t len) #else === modified file 'strings/decimal.c' --- a/strings/decimal.c 2008-08-15 19:49:43 +0000 +++ b/strings/decimal.c 2009-04-27 18:01:20 +0000 @@ -144,7 +144,7 @@ static double scaler1[]= { 1.0, 10.0, 1e2, 1e3, 1e4, 1e5, 1e6, 1e7, 1e8, 1e9 }; -#ifdef HAVE_purify +#ifdef HAVE_valgrind #define sanity(d) DBUG_ASSERT((d)->len > 0) #else #define sanity(d) DBUG_ASSERT((d)->len >0 && ((d)->buf[0] | \ === modified file 'strings/strmake.c' --- a/strings/strmake.c 2008-04-28 16:24:05 +0000 +++ b/strings/strmake.c 2009-04-27 18:01:20 +0000 @@ -51,11 +51,11 @@ char *strmake(register char *dst, regist */ if (length) { -#ifdef HAVE_purify +#ifdef HAVE_valgrind dst[length-1]= 'Z'; #else bfill(dst, length-1, (int) 'Z'); -#endif /* HAVE_purify */ +#endif /* HAVE_valgrind */ } #endif /* EXTRA_DEBUG */ return dst-1;

2 1

[Maria-developers] bzr commit into MariaDB 5.1, with Maria 1.5:maria branch (knielsen:2698)
by knielsen＠knielsen-hq.org 06 May '09

06 May '09

#At lp:maria 2698 knielsen(a)knielsen-hq.org 2009-04-27 Misc. small fixes: - Fixes for pool-of-threads patch. - Fixes for push of PBXT storage engine. - mysql-test-run.pl fix. modified: BUILD/compile-dist mysql-test/Makefile.am mysql-test/lib/mtr_unique.pm mysql-test/t/pool_of_threads.cnf sql/set_var.cc per-file messages: BUILD/compile-dist Fix that PBXT was missing in source tarball after `BUILD/compile-dist && make dist` mysql-test/Makefile.am Fix that PBXT test suite was missing from `make dist` source tarball. mysql-test/lib/mtr_unique.pm Better fix to avoid races when chmod'ing the semaphore file. (Though using chmod 666 shared files in /tmp/ is still not a very good solution). mysql-test/t/pool_of_threads.cnf Fix that test case pool_of_threads fails if run on mysqld with no --with-libevent support. sql/set_var.cc Fix missing static declaration on pool_of_threads. === modified file 'BUILD/compile-dist' --- a/BUILD/compile-dist 2009-02-19 09:01:25 +0000 +++ b/BUILD/compile-dist 2009-04-27 13:17:47 +0000 @@ -60,8 +60,6 @@ fi # Make sure to enable all features that affect "make dist" # Remember that configure restricts the man pages to the configured features ! ./configure \ - --with-maria-storage-engine \ - --with-embedded-server \ - --with-ndbcluster -$gmake - + --with-plugins=max-no-ndb \ + --with-embedded-server +$gmake -j4 === modified file 'mysql-test/Makefile.am' --- a/mysql-test/Makefile.am 2009-02-13 12:25:41 +0000 +++ b/mysql-test/Makefile.am 2009-04-27 13:17:47 +0000 @@ -98,7 +98,8 @@ TEST_DIRS = t r include std_data std_dat suite/stress/include suite/stress/t suite/stress/r \ suite/ndb suite/ndb/t suite/ndb/r \ suite/rpl_ndb suite/rpl_ndb/t suite/rpl_ndb/r \ - suite/parts suite/parts/t suite/parts/r suite/parts/inc + suite/parts suite/parts/t suite/parts/r suite/parts/inc \ + suite/pbxt/t suite/pbxt/r # Used by dist-hook and install-data-local to copy all # test files into either dist or install directory === modified file 'mysql-test/lib/mtr_unique.pm' --- a/mysql-test/lib/mtr_unique.pm 2009-04-08 16:55:26 +0000 +++ b/mysql-test/lib/mtr_unique.pm 2009-04-27 13:17:47 +0000 @@ -62,14 +62,14 @@ sub mtr_get_unique_id($$) { die 'lock file is a symbolic link'; } + my $save_umask= umask(0); open SEM, ">", "$file.sem" or die "can't write to $file.sem"; - chmod 0777, "$file.sem"; flock SEM, LOCK_EX or die "can't lock $file.sem"; if(! -e $file) { open FILE, ">", $file or die "can't create $file"; close FILE; } - chmod 0777, $file; + umask($save_umask); msg("HAVE THE LOCK"); @@ -135,8 +135,8 @@ sub mtr_release_unique_id($) { die 'lock file is a symbolic link'; } + my $save_umask= umask(0); open SEM, ">", "$file.sem" or die "can't write to $file.sem"; - chmod 0777, "$file.sem"; flock SEM, LOCK_EX or die "can't lock $file.sem"; msg("HAVE THE LOCK"); @@ -149,7 +149,7 @@ sub mtr_release_unique_id($) { open FILE, ">", $file or die "can't create $file"; close FILE; } - chmod 0777, "$file.sem"; + umask($save_umask); open FILE, "+<", $file or die "can't open $file"; #select undef,undef,undef,0.2; seek FILE, 0, 0; === modified file 'mysql-test/t/pool_of_threads.cnf' --- a/mysql-test/t/pool_of_threads.cnf 2009-03-18 15:46:32 +0000 +++ b/mysql-test/t/pool_of_threads.cnf 2009-04-27 13:17:47 +0000 @@ -2,8 +2,8 @@ [mysqld.1] test-ignore-wrong-options -thread-handling= pool-of-threads -thread_pool_size= 2 +loose-thread-handling= pool-of-threads +loose-thread_pool_size= 2 extra-port= @mysqld.1.#extra-port extra-max-connections=1 === modified file 'sql/set_var.cc' --- a/sql/set_var.cc 2009-04-08 16:55:26 +0000 +++ b/sql/set_var.cc 2009-04-27 13:17:47 +0000 @@ -642,7 +642,7 @@ static sys_var_long_ptr sys_table_lock_w static sys_var_long_ptr sys_thread_cache_size(&vars, "thread_cache_size", &thread_cache_size); #if HAVE_POOL_OF_THREADS == 1 -sys_var_long_ptr sys_thread_pool_size(&vars, "thread_pool_size", +static sys_var_long_ptr sys_thread_pool_size(&vars, "thread_pool_size", &thread_pool_size); #endif static sys_var_thd_enum sys_tx_isolation(&vars, "tx_isolation",

2 1

[Maria-developers] bzr commit into MariaDB 5.1, with Maria 1.5:maria branch (knielsen:2698)
by knielsen＠knielsen-hq.org 06 May '09

06 May '09

#At lp:maria 2698 knielsen(a)knielsen-hq.org 2009-05-06 Fix a build problem with BUILD/compile-pentium64-max on CentOS/Fedora Core 10 amd64. On these systems, there is libz.so but no libz.a. Finding libz.so, ./configure decides to use system zlib. But since BUILD/compile-pentium64-max builds a fully static binary with -all-static, the link of mysqld fails due to missing libz.a. Fix by using bundled zlib in the build script. modified: BUILD/compile-pentium64-max === modified file 'BUILD/compile-pentium64-max' --- a/BUILD/compile-pentium64-max 2007-04-11 12:12:00 +0000 +++ b/BUILD/compile-pentium64-max 2009-05-06 09:51:46 +0000 @@ -4,7 +4,10 @@ path=`dirname $0` . "$path/SETUP.sh" extra_flags="$pentium64_cflags $fast_cflags" -extra_configs="$pentium_configs $max_configs $static_link" +# On CentOS/Fedora Core 10 amd64, there is system libz.so but not +# libz.a, so need to use bundled zlib when building static +# binary. Hence we use --with-zlib-dir=bundled +extra_configs="$pentium_configs $max_configs $static_link --with-zlib-dir=bundled" CC="$CC --pipe" strip=yes

2 1

[Maria-developers] bzr commit into MariaDB 5.1, with Maria 1.5:maria branch (knielsen:2702)
by knielsen＠knielsen-hq.org 05 May '09

05 May '09

#At lp:maria 2702 knielsen(a)knielsen-hq.org 2009-05-05 [merge] test merge 2

1 0

Re: [Maria-developers] main.pool_of_threads fail - Solaris 10
by Adam M. Dutko 27 Apr '09

27 Apr '09

Toby, You might also want to check this page out: http://askmonty.org/wiki/index.php/MariaDB_versus_MySQL -Adam

1 0

[Maria-developers] main.pool_of_threads fail - Solaris 10
by Toby Thain 27 Apr '09

27 Apr '09

Hi list, I am setting up a buildbot for Solaris 10/amd64 (later SPARC) and I see this test failure on Sun cc (only). gcc builds pass all tests. (Although a patch is required.) Any clues? --Toby bzr branch lp:maria ... BUILD/compile-solaris-amd64-forte && cd mysql-test/ && ./mysql-test-run ... main.connect [ pass ] 16399 worker[1] mysql-test-run: WARNING: Process [mysqld.1 - pid: 495, winpid: 495, exit: 1792] died main.pool_of_threads [ fail ] Test ended at 2009-04-26 15:18:41 090426 22:17:50 [Note] /home/buildbot/maria/sql/mysqld: Normal shutdown 090426 22:17:50 [Note] Event Scheduler: Purging the queue. 0 events 090426 22:17:53 InnoDB: Starting shutdown... 090426 22:17:53 InnoDB: Shutdown completed; log sequence number 0 1943684 090426 22:17:53 [Note] /home/buildbot/maria/sql/mysqld: Shutdown complete 090426 22:17:54 InnoDB: Started; log sequence number 0 1943684 090426 22:17:54 [Warning] /home/buildbot/maria/sql/mysqld: unknown variable 'loose-debug-sync-timeout=300' 090426 22:17:54 [Note] Event Scheduler: Loaded 0 events 090426 22:17:54 [Note] /home/buildbot/maria/sql/mysqld: ready for connections. Version: '5.1.32-maria-beta2-log' socket: '/home/buildbot/maria/ mysql-test/var/tmp/mysqld.1.sock' port: 12500 Source distribution CURRENT_TEST: rpl.rpl_truncate_3innodb 090426 22:18:01 [Note] /home/buildbot/maria/sql/mysqld: Normal shutdown 090426 22:18:01 [Note] Event Scheduler: Purging the queue. 0 events 090426 22:18:03 InnoDB: Starting shutdown... 090426 22:18:05 InnoDB: Shutdown completed; log sequence number 0 1968339 090426 22:18:05 [Note] /home/buildbot/maria/sql/mysqld: Shutdown complete CURRENT_TEST: federated.federated 090426 22:18:05 [Note] Plugin 'InnoDB' disabled by command line option 090426 22:18:05 [Warning] /home/buildbot/maria/sql/mysqld: unknown variable 'loose-debug-sync-timeout=300' 090426 22:18:05 [Note] Event Scheduler: Loaded 0 events 090426 22:18:05 [Note] /home/buildbot/maria/sql/mysqld: ready for connections. Version: '5.1.32-maria-beta2-log' socket: '/home/buildbot/maria/ mysql-test/var/tmp/mysqld.1.sock' port: 12500 Source distribution CURRENT_TEST: federated.federated_archive CURRENT_TEST: federated.federated_bug_13118 CURRENT_TEST: federated.federated_bug_25714 CURRENT_TEST: federated.federated_server 090426 22:18:17 [Note] /home/buildbot/maria/sql/mysqld: Normal shutdown 090426 22:18:17 [Note] Event Scheduler: Purging the queue. 0 events 090426 22:18:17 [Note] /home/buildbot/maria/sql/mysqld: Shutdown complete CURRENT_TEST: federated.federated_innodb InnoDB: The first specified data file ./ibdata1 did not exist: InnoDB: a new database to be created! 090426 22:18:18 InnoDB: Setting file ./ibdata1 size to 10 MB InnoDB: Database physically writes the file full: wait... 090426 22:18:18 InnoDB: Log file ./ib_logfile0 did not exist: new to be created InnoDB: Setting log file ./ib_logfile0 size to 5 MB InnoDB: Database physically writes the file full: wait... 090426 22:18:18 InnoDB: Log file ./ib_logfile1 did not exist: new to be created InnoDB: Setting log file ./ib_logfile1 size to 5 MB InnoDB: Database physically writes the file full: wait... InnoDB: Doublewrite buffer not found: creating new InnoDB: Doublewrite buffer created InnoDB: Creating foreign key constraint system tables InnoDB: Foreign key constraint system tables created 090426 22:18:20 InnoDB: Started; log sequence number 0 0 090426 22:18:20 [Warning] /home/buildbot/maria/sql/mysqld: unknown variable 'loose-debug-sync-timeout=300' 090426 22:18:20 [Note] Event Scheduler: Loaded 0 events 090426 22:18:20 [Note] /home/buildbot/maria/sql/mysqld: ready for connections. Version: '5.1.32-maria-beta2-log' socket: '/home/buildbot/maria/ mysql-test/var/tmp/mysqld.1.sock' port: 12500 Source distribution 090426 22:18:20 [Note] /home/buildbot/maria/sql/mysqld: Normal shutdown 090426 22:18:20 [Note] Event Scheduler: Purging the queue. 0 events 090426 22:18:21 InnoDB: Starting shutdown... 090426 22:18:23 InnoDB: Shutdown completed; log sequence number 0 46409 090426 22:18:23 [Note] /home/buildbot/maria/sql/mysqld: Shutdown complete CURRENT_TEST: main.connect 090426 22:18:23 [Note] Plugin 'InnoDB' disabled by command line option 090426 22:18:23 [Warning] /home/buildbot/maria/sql/mysqld: unknown variable 'loose-debug-sync-timeout=300' 090426 22:18:23 [Note] Event Scheduler: Loaded 0 events 090426 22:18:23 [Note] /home/buildbot/maria/sql/mysqld: ready for connections. Version: '5.1.32-maria-beta2-log' socket: '/home/buildbot/maria/ mysql-test/var/tmp/mysqld.1.sock' port: 12500 Source distribution 090426 22:18:29 [Note] Event Scheduler: scheduler thread started with id 21 090426 22:18:40 [Note] Event Scheduler: Killing the scheduler thread, thread id 21 090426 22:18:40 [Note] Event Scheduler: Waiting for the scheduler thread to reply 090426 22:18:40 [Note] Event Scheduler: Stopped 090426 22:18:40 [Note] Event Scheduler: scheduler thread started with id 77 090426 22:18:40 [Note] Event Scheduler: Killing the scheduler thread, thread id 77 090426 22:18:40 [Note] Event Scheduler: Waiting for the scheduler thread to reply 090426 22:18:40 [Note] Event Scheduler: Stopped 090426 22:18:40 [Note] /home/buildbot/maria/sql/mysqld: Normal shutdown 090426 22:18:40 [Note] Event Scheduler: Purging the queue. 0 events 090426 22:18:40 [Note] /home/buildbot/maria/sql/mysqld: Shutdown complete CURRENT_TEST: main.pool_of_threads Unknown option to thread_handling: pool-of-threads Alternatives are: 'one-thread-per-connection','no-threads' 090426 22:18:40 [Note] Plugin 'InnoDB' disabled by command line option 090426 22:18:40 [ERROR] /home/buildbot/maria/sql/mysqld: unknown variable 'thread_pool_size=2' 090426 22:18:40 [ERROR] Aborting 090426 22:18:40 [Note] /home/buildbot/maria/sql/mysqld: Shutdown complete Failed to start mysqld.1 - saving '/home/buildbot/maria/mysql-test/var/log/ main.pool_of_threads/' to '/home/buildbot/maria/mysql-test/var/log/ main.pool_of_threads/' Only 863 of 866 completed. mysql-test-run: *** ERROR: Not all tests completed *** Error code 1 make: Fatal error: Command failed for target `test-ns'

2 1

[Maria-developers] order by x limit y optimization
by Tonu Samuel 26 Apr '09

26 Apr '09

Hi! Just was looking at processlist and thinking "why he cannot optimize such a easy case". I think there is a possible BOOST place (this is common query in many apps): mysql> select count(*) from url; +----------+ | count(*) | +----------+ | 17945916 | +----------+ 1 row in set (0.00 sec) mysql> mysql> show full processlist; +------+------+-----------+-----------+---------+------+----------------+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+ | Id | User | Host | db | Command | Time | State | Info | +------+------+-----------+-----------+---------+------+----------------+-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+ | 78 | root | localhost | eeindexer | Query | 141 | Sorting result | INSERT INTO udm_url_tmp SELECT url.rec_id FROM url WHERE next_index_time<=1240742878 ORDER BY hops LIMIT 256 ..... So, it takes HUGE time to find out 256 rows with least hops. How I would make it manually (MySQL engine has internally better ways to do it): mysql> select min(hops) from url; +-----------+ | min(hops) | +-----------+ | 0 | +-----------+ 1 row in set (0.12 sec) mysql> select count(hops) from url where hops=0; +-------------+ | count(hops) | +-------------+ | 1 | +-------------+ 1 row in set (0.00 sec) mysql> select count(hops) from url where hops=1; +-------------+ | count(hops) | +-------------+ | 92 | +-------------+ 1 row in set (0.00 sec) mysql> select count(hops) from url where hops=2; +-------------+ | count(hops) | +-------------+ | 9790 | +-------------+ 1 row in set (0.00 sec) mysql> mysql> SELECT url.rec_id FROM url WHERE hops in (0,1,2,3) ORDER BY hops LIMIT 256; +--------+ | rec_id | +--------+ | 1 | | 82 | | 121 | | 55 | | 63 | | 39 | | 65 | | 100 | | 129 | | 85 | | 25 | | 11 | | 12 | | 54 | | 3 | | 101 | | 7 | | 50 | | 76 | | 77 | | 10 | | 24 | | 14 | | 2 | | 57 | | 46 | | 44 | | 17 | | 122 | | 135 | | 8 | | 51 | | 104 | | 124 | | 35 | | 56 | | 116 | | 62 | | 41 | | 78 | | 126 | | 15 | | 26 | | 40 | | 90 | | 80 | | 13 | | 36 | | 45 | | 70 | | 60 | | 117 | | 47 | | 114 | | 132 | | 19 | | 96 | | 118 | | 74 | | 9 | | 120 | | 95 | | 23 | | 136 | | 71 | | 130 | | 83 | | 33 | | 111 | | 125 | | 81 | | 43 | | 97 | | 28 | | 73 | | 109 | | 98 | | 105 | | 133 | | 128 | | 106 | | 4 | | 102 | | 127 | | 5 | | 30 | | 115 | | 123 | | 27 | | 6 | | 113 | | 93 | | 48 | | 10047 | | 55273 | | 7120 | | 21448 | | 16872 | | 17985 | | 9862 | | 20979 | | 45717 | | 491 | | 16943 | | 18183 | | 1753 | | 21176 | | 16486 | | 7578 | | 2329 | | 9824 | | 1239 | | 400 | | 6470 | | 8052 | | 2752 | | 1258 | | 16039 | | 593 | | 49887 | | 5884 | | 7119 | | 3011 | | 7819 | | 14710 | | 244 | | 6881 | | 16518 | | 240 | | 16445 | | 49988 | | 62181 | | 61227 | | 61651 | | 18118 | | 9951 | | 54964 | | 61197 | | 16361 | | 21452 | | 15361 | | 2840 | | 16479 | | 17006 | | 2768 | | 7917 | | 61198 | | 21556 | | 2650 | | 2525 | | 55022 | | 54911 | | 55672 | | 15326 | | 14763 | | 15493 | | 55156 | | 21435 | | 49959 | | 18242 | | 1887 | | 6437 | | 19407 | | 7593 | | 2355 | | 16908 | | 16102 | | 1275 | | 57868 | | 15943 | | 7977 | | 16079 | | 14712 | | 49874 | | 17269 | | 55851 | | 57783 | | 3442 | | 18085 | | 50913 | | 4067 | | 19402 | | 2785 | | 7920 | | 16456 | | 19143 | | 15141 | | 54895 | | 3578 | | 2905 | | 16623 | | 16802 | | 6472 | | 5894 | | 7004 | | 14711 | | 2545 | | 19184 | | 2994 | | 3297 | | 7993 | | 10034 | | 51738 | | 16854 | | 7007 | | 7583 | | 15967 | | 3720 | | 54940 | | 15299 | | 2343 | | 9896 | | 7016 | | 19668 | | 16128 | | 21624 | | 7529 | | 3521 | | 16602 | | 16089 | | 54821 | | 51589 | | 61828 | | 19690 | | 3608 | | 17946 | | 61146 | | 2265 | | 9809 | | 2509 | | 54729 | | 2426 | | 21617 | | 7543 | | 1841 | | 3700 | | 21391 | | 21442 | | 55155 | | 7532 | | 3490 | | 7821 | | 1891 | | 9876 | | 3677 | | 10019 | | 2862 | | 16186 | | 16299 | | 16528 | | 50641 | | 5926 | | 55135 | | 16800 | | 54734 | | 19375 | +--------+ 256 rows in set (0.09 sec) mysql> So why not to look on index cardinality, limit size (let say, less than 1% of table), min or max value of field ordered by and then exclude big part of table to avoid unnecessary work? Tõnu

1 0