mariadb/mysys/hash.c

717 lines
18 KiB
C
Raw Normal View History

/* Copyright 2000-2008 MySQL AB, 2008 Sun Microsystems, Inc.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; version 2 of the License.
This program is distributed in the hope that it will be useful,
2000-07-31 21:29:14 +02:00
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program; if not, write to the Free Software
Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */
2000-07-31 21:29:14 +02:00
/* The hash functions used for saveing keys */
/* One of key_length or key_length_offset must be given */
/* Key length of 0 isn't allowed */
#include "mysys_priv.h"
#include <m_string.h>
#include <m_ctype.h>
#include "hash.h"
#define NO_RECORD ((uint) -1)
#define LOWFIND 1
#define LOWUSED 2
#define HIGHFIND 4
#define HIGHUSED 8
typedef struct st_hash_info {
uint next; /* index to next key */
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 12:59:39 +03:00
uchar *data; /* data for current entry */
} HASH_LINK;
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
static uint my_hash_mask(uint hashnr, uint buffmax, uint maxlength);
2000-07-31 21:29:14 +02:00
static void movelink(HASH_LINK *array,uint pos,uint next_link,uint newlink);
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 12:59:39 +03:00
static int hashcmp(const HASH *hash, HASH_LINK *pos, const uchar *key,
size_t length);
2000-07-31 21:29:14 +02:00
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 12:59:39 +03:00
static uint calc_hash(const HASH *hash, const uchar *key, size_t length)
{
ulong nr1=1, nr2=4;
hash->charset->coll->hash_sort(hash->charset,(uchar*) key,length,&nr1,&nr2);
return nr1;
}
my_bool
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
_my_hash_init(HASH *hash, uint growth_size, CHARSET_INFO *charset,
ulong size, size_t key_offset, size_t key_length,
my_hash_get_key get_key,
void (*free_element)(void*), uint flags CALLER_INFO_PROTO)
2000-07-31 21:29:14 +02:00
{
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
DBUG_ENTER("my_hash_init");
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 12:59:39 +03:00
DBUG_PRINT("enter",("hash: 0x%lx size: %u", (long) hash, (uint) size));
2000-07-31 21:29:14 +02:00
hash->records=0;
if (my_init_dynamic_array_ci(&hash->array, sizeof(HASH_LINK), size,
growth_size))
2000-07-31 21:29:14 +02:00
{
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
hash->free=0; /* Allow call to my_hash_free */
DBUG_RETURN(1);
2000-07-31 21:29:14 +02:00
}
hash->key_offset=key_offset;
hash->key_length=key_length;
hash->blength=1;
hash->get_key=get_key;
hash->free=free_element;
hash->flags=flags;
2002-03-14 21:44:42 +04:00
hash->charset=charset;
2000-07-31 21:29:14 +02:00
DBUG_RETURN(0);
}
/*
Call hash->free on all elements in hash.
SYNOPSIS
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
my_hash_free_elements()
hash hash table
NOTES:
Sets records to 0
*/
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
static inline void my_hash_free_elements(HASH *hash)
2000-07-31 21:29:14 +02:00
{
if (hash->free)
{
HASH_LINK *data=dynamic_element(&hash->array,0,HASH_LINK*);
HASH_LINK *end= data + hash->records;
while (data < end)
(*hash->free)((data++)->data);
2000-07-31 21:29:14 +02:00
}
hash->records=0;
}
/*
Free memory used by hash.
SYNOPSIS
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
my_hash_free()
hash the hash to delete elements of
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
NOTES: Hash can't be reused without calling my_hash_init again.
*/
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
void my_hash_free(HASH *hash)
{
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
DBUG_ENTER("my_hash_free");
DBUG_PRINT("enter",("hash: 0x%lx", (long) hash));
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
my_hash_free_elements(hash);
hash->free= 0;
delete_dynamic(&hash->array);
2000-07-31 21:29:14 +02:00
DBUG_VOID_RETURN;
}
/*
Delete all elements from the hash (the hash itself is to be reused).
SYNOPSIS
my_hash_reset()
hash the hash to delete elements of
*/
void my_hash_reset(HASH *hash)
{
DBUG_ENTER("my_hash_reset");
DBUG_PRINT("enter",("hash: 0x%lxd", (long) hash));
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
my_hash_free_elements(hash);
reset_dynamic(&hash->array);
/* Set row pointers so that the hash can be reused at once */
hash->blength= 1;
DBUG_VOID_RETURN;
}
/* some helper functions */
2000-07-31 21:29:14 +02:00
/*
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 12:59:39 +03:00
This function is char* instead of uchar* as HPUX11 compiler can't
handle inline functions that are not defined as native types
*/
static inline char*
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
my_hash_key(const HASH *hash, const uchar *record, size_t *length,
my_bool first)
2000-07-31 21:29:14 +02:00
{
if (hash->get_key)
return (char*) (*hash->get_key)(record,length,first);
2000-07-31 21:29:14 +02:00
*length=hash->key_length;
return (char*) record+hash->key_offset;
2000-07-31 21:29:14 +02:00
}
/* Calculate pos according to keys */
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
static uint my_hash_mask(uint hashnr, uint buffmax, uint maxlength)
2000-07-31 21:29:14 +02:00
{
if ((hashnr & (buffmax-1)) < maxlength) return (hashnr & (buffmax-1));
return (hashnr & ((buffmax >> 1) -1));
}
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
static uint my_hash_rec_mask(const HASH *hash, HASH_LINK *pos,
uint buffmax, uint maxlength)
2000-07-31 21:29:14 +02:00
{
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 12:59:39 +03:00
size_t length;
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
uchar *key= (uchar*) my_hash_key(hash, pos->data, &length, 0);
return my_hash_mask(calc_hash(hash, key, length), buffmax, maxlength);
2000-07-31 21:29:14 +02:00
}
2000-12-29 16:06:10 +02:00
2000-07-31 21:29:14 +02:00
2004-06-22 17:54:38 +02:00
/* for compilers which can not handle inline */
static
#if !defined(__USLC__) && !defined(__sgi)
inline
#endif
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 12:59:39 +03:00
unsigned int rec_hashnr(HASH *hash,const uchar *record)
2000-07-31 21:29:14 +02:00
{
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 12:59:39 +03:00
size_t length;
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
uchar *key= (uchar*) my_hash_key(hash, record, &length, 0);
return calc_hash(hash,key,length);
2000-07-31 21:29:14 +02:00
}
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
uchar* my_hash_search(const HASH *hash, const uchar *key, size_t length)
{
HASH_SEARCH_STATE state;
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
return my_hash_first(hash, key, length, &state);
}
/*
Search after a record based on a key
NOTE
Assigns the number of the found record to HASH_SEARCH_STATE state
*/
2000-07-31 21:29:14 +02:00
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
uchar* my_hash_first(const HASH *hash, const uchar *key, size_t length,
HASH_SEARCH_STATE *current_record)
2000-07-31 21:29:14 +02:00
{
HASH_LINK *pos;
uint flag,idx;
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
DBUG_ENTER("my_hash_first");
2000-07-31 21:29:14 +02:00
flag=1;
if (hash->records)
{
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
idx= my_hash_mask(calc_hash(hash, key, length ? length : hash->key_length),
hash->blength, hash->records);
2000-07-31 21:29:14 +02:00
do
{
pos= dynamic_element(&hash->array,idx,HASH_LINK*);
if (!hashcmp(hash,pos,key,length))
{
DBUG_PRINT("exit",("found key at %d",idx));
*current_record= idx;
2000-07-31 21:29:14 +02:00
DBUG_RETURN (pos->data);
}
if (flag)
{
flag=0; /* Reset flag */
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
if (my_hash_rec_mask(hash, pos, hash->blength, hash->records) != idx)
2000-07-31 21:29:14 +02:00
break; /* Wrong link */
}
}
while ((idx=pos->next) != NO_RECORD);
}
*current_record= NO_RECORD;
2000-07-31 21:29:14 +02:00
DBUG_RETURN(0);
}
/* Get next record with identical key */
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
/* Can only be called if previous calls was my_hash_search */
2000-07-31 21:29:14 +02:00
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
uchar* my_hash_next(const HASH *hash, const uchar *key, size_t length,
HASH_SEARCH_STATE *current_record)
2000-07-31 21:29:14 +02:00
{
HASH_LINK *pos;
uint idx;
if (*current_record != NO_RECORD)
2000-07-31 21:29:14 +02:00
{
HASH_LINK *data=dynamic_element(&hash->array,0,HASH_LINK*);
for (idx=data[*current_record].next; idx != NO_RECORD ; idx=pos->next)
2000-07-31 21:29:14 +02:00
{
pos=data+idx;
if (!hashcmp(hash,pos,key,length))
{
*current_record= idx;
2000-07-31 21:29:14 +02:00
return pos->data;
}
}
*current_record= NO_RECORD;
2000-07-31 21:29:14 +02:00
}
return 0;
}
/* Change link from pos to new_link */
static void movelink(HASH_LINK *array,uint find,uint next_link,uint newlink)
{
HASH_LINK *old_link;
do
{
old_link=array+next_link;
}
while ((next_link=old_link->next) != find);
old_link->next= newlink;
return;
}
/*
Compare a key in a record to a whole key. Return 0 if identical
SYNOPSIS
hashcmp()
hash hash table
pos position of hash record to use in comparison
key key for comparison
length length of key
NOTES:
If length is 0, comparison is done using the length of the
record being compared against.
RETURN
= 0 key of record == key
!= 0 key of record != key
*/
2000-07-31 21:29:14 +02:00
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 12:59:39 +03:00
static int hashcmp(const HASH *hash, HASH_LINK *pos, const uchar *key,
size_t length)
2000-07-31 21:29:14 +02:00
{
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 12:59:39 +03:00
size_t rec_keylength;
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
uchar *rec_key= (uchar*) my_hash_key(hash, pos->data, &rec_keylength, 1);
2004-03-04 08:50:37 +02:00
return ((length && length != rec_keylength) ||
my_strnncoll(hash->charset, (uchar*) rec_key, rec_keylength,
(uchar*) key, rec_keylength));
2000-07-31 21:29:14 +02:00
}
/* Write a hash-key to the hash-index */
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
my_bool my_hash_insert(HASH *info, const uchar *record)
2000-07-31 21:29:14 +02:00
{
int flag;
size_t idx;
uint halfbuff,hash_nr,first_index;
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 12:59:39 +03:00
uchar *ptr_to_rec,*ptr_to_rec2;
2000-07-31 21:29:14 +02:00
HASH_LINK *data,*empty,*gpos,*gpos2,*pos;
LINT_INIT(gpos);
LINT_INIT(gpos2);
LINT_INIT(ptr_to_rec);
LINT_INIT(ptr_to_rec2);
2000-07-31 21:29:14 +02:00
if (HASH_UNIQUE & info->flags)
{
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
uchar *key= (uchar*) my_hash_key(info, record, &idx, 1);
if (my_hash_search(info, key, idx))
return(TRUE); /* Duplicate entry */
}
2000-07-31 21:29:14 +02:00
flag=0;
if (!(empty=(HASH_LINK*) alloc_dynamic(&info->array)))
return(TRUE); /* No more memory */
data=dynamic_element(&info->array,0,HASH_LINK*);
halfbuff= info->blength >> 1;
idx=first_index=info->records-halfbuff;
if (idx != info->records) /* If some records */
{
do
{
pos=data+idx;
hash_nr=rec_hashnr(info,pos->data);
if (flag == 0) /* First loop; Check if ok */
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
if (my_hash_mask(hash_nr, info->blength, info->records) != first_index)
2000-07-31 21:29:14 +02:00
break;
if (!(hash_nr & halfbuff))
{ /* Key will not move */
if (!(flag & LOWFIND))
{
if (flag & HIGHFIND)
{
flag=LOWFIND | HIGHFIND;
/* key shall be moved to the current empty position */
gpos=empty;
ptr_to_rec=pos->data;
empty=pos; /* This place is now free */
}
else
{
flag=LOWFIND | LOWUSED; /* key isn't changed */
gpos=pos;
ptr_to_rec=pos->data;
}
}
else
{
if (!(flag & LOWUSED))
{
/* Change link of previous LOW-key */
gpos->data=ptr_to_rec;
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 12:59:39 +03:00
gpos->next= (uint) (pos-data);
2000-07-31 21:29:14 +02:00
flag= (flag & HIGHFIND) | (LOWFIND | LOWUSED);
}
gpos=pos;
ptr_to_rec=pos->data;
}
}
else
{ /* key will be moved */
if (!(flag & HIGHFIND))
{
flag= (flag & LOWFIND) | HIGHFIND;
/* key shall be moved to the last (empty) position */
gpos2 = empty; empty=pos;
ptr_to_rec2=pos->data;
}
else
{
if (!(flag & HIGHUSED))
{
/* Change link of previous hash-key and save */
gpos2->data=ptr_to_rec2;
gpos2->next=(uint) (pos-data);
flag= (flag & LOWFIND) | (HIGHFIND | HIGHUSED);
}
gpos2=pos;
ptr_to_rec2=pos->data;
}
}
}
while ((idx=pos->next) != NO_RECORD);
if ((flag & (LOWFIND | LOWUSED)) == LOWFIND)
{
gpos->data=ptr_to_rec;
gpos->next=NO_RECORD;
}
if ((flag & (HIGHFIND | HIGHUSED)) == HIGHFIND)
{
gpos2->data=ptr_to_rec2;
gpos2->next=NO_RECORD;
}
}
/* Check if we are at the empty position */
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
idx= my_hash_mask(rec_hashnr(info, record), info->blength, info->records + 1);
2000-07-31 21:29:14 +02:00
pos=data+idx;
if (pos == empty)
{
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 12:59:39 +03:00
pos->data=(uchar*) record;
2000-07-31 21:29:14 +02:00
pos->next=NO_RECORD;
}
else
{
/* Check if more records in same hash-nr family */
empty[0]=pos[0];
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
gpos= data + my_hash_rec_mask(info, pos, info->blength, info->records + 1);
2000-07-31 21:29:14 +02:00
if (pos == gpos)
{
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 12:59:39 +03:00
pos->data=(uchar*) record;
2000-07-31 21:29:14 +02:00
pos->next=(uint) (empty - data);
}
else
{
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 12:59:39 +03:00
pos->data=(uchar*) record;
2000-07-31 21:29:14 +02:00
pos->next=NO_RECORD;
movelink(data,(uint) (pos-data),(uint) (gpos-data),(uint) (empty-data));
}
}
if (++info->records == info->blength)
info->blength+= info->blength;
return(0);
}
/******************************************************************************
** Remove one record from hash-table. The record with the same record
** ptr is removed.
** if there is a free-function it's called for record if found
******************************************************************************/
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
my_bool my_hash_delete(HASH *hash, uchar *record)
2000-07-31 21:29:14 +02:00
{
uint blength,pos2,pos_hashnr,lastpos_hashnr,idx,empty_index;
HASH_LINK *data,*lastpos,*gpos,*pos,*pos3,*empty;
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
DBUG_ENTER("my_hash_delete");
if (!hash->records)
DBUG_RETURN(1);
2000-07-31 21:29:14 +02:00
blength=hash->blength;
data=dynamic_element(&hash->array,0,HASH_LINK*);
/* Search after record with key */
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
pos= data + my_hash_mask(rec_hashnr(hash, record), blength, hash->records);
2000-07-31 21:29:14 +02:00
gpos = 0;
while (pos->data != record)
{
gpos=pos;
if (pos->next == NO_RECORD)
DBUG_RETURN(1); /* Key not found */
pos=data+pos->next;
}
if ( --(hash->records) < hash->blength >> 1) hash->blength>>=1;
lastpos=data+hash->records;
/* Remove link to record */
empty=pos; empty_index=(uint) (empty-data);
if (gpos)
gpos->next=pos->next; /* unlink current ptr */
else if (pos->next != NO_RECORD)
{
empty=data+(empty_index=pos->next);
pos->data=empty->data;
pos->next=empty->next;
}
if (empty == lastpos) /* last key at wrong pos or no next link */
goto exit;
/* Move the last key (lastpos) */
lastpos_hashnr=rec_hashnr(hash,lastpos->data);
/* pos is where lastpos should be */
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
pos= data + my_hash_mask(lastpos_hashnr, hash->blength, hash->records);
2000-07-31 21:29:14 +02:00
if (pos == empty) /* Move to empty position. */
{
empty[0]=lastpos[0];
goto exit;
}
pos_hashnr=rec_hashnr(hash,pos->data);
/* pos3 is where the pos should be */
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
pos3= data + my_hash_mask(pos_hashnr, hash->blength, hash->records);
2000-07-31 21:29:14 +02:00
if (pos != pos3)
{ /* pos is on wrong posit */
empty[0]=pos[0]; /* Save it here */
pos[0]=lastpos[0]; /* This should be here */
movelink(data,(uint) (pos-data),(uint) (pos3-data),empty_index);
goto exit;
}
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
pos2= my_hash_mask(lastpos_hashnr, blength, hash->records + 1);
if (pos2 == my_hash_mask(pos_hashnr, blength, hash->records + 1))
2000-07-31 21:29:14 +02:00
{ /* Identical key-positions */
if (pos2 != hash->records)
{
empty[0]=lastpos[0];
movelink(data,(uint) (lastpos-data),(uint) (pos-data),empty_index);
goto exit;
}
idx= (uint) (pos-data); /* Link pos->next after lastpos */
}
else idx= NO_RECORD; /* Different positions merge */
empty[0]=lastpos[0];
movelink(data,idx,empty_index,pos->next);
pos->next=empty_index;
exit:
VOID(pop_dynamic(&hash->array));
if (hash->free)
WL#3817: Simplify string / memory area types and make things more consistent (first part) The following type conversions was done: - Changed byte to uchar - Changed gptr to uchar* - Change my_string to char * - Change my_size_t to size_t - Change size_s to size_t Removed declaration of byte, gptr, my_string, my_size_t and size_s. Following function parameter changes was done: - All string functions in mysys/strings was changed to use size_t instead of uint for string lengths. - All read()/write() functions changed to use size_t (including vio). - All protocoll functions changed to use size_t instead of uint - Functions that used a pointer to a string length was changed to use size_t* - Changed malloc(), free() and related functions from using gptr to use void * as this requires fewer casts in the code and is more in line with how the standard functions work. - Added extra length argument to dirname_part() to return the length of the created string. - Changed (at least) following functions to take uchar* as argument: - db_dump() - my_net_write() - net_write_command() - net_store_data() - DBUG_DUMP() - decimal2bin() & bin2decimal() - Changed my_compress() and my_uncompress() to use size_t. Changed one argument to my_uncompress() from a pointer to a value as we only return one value (makes function easier to use). - Changed type of 'pack_data' argument to packfrm() to avoid casts. - Changed in readfrm() and writefrom(), ha_discover and handler::discover() the type for argument 'frmdata' to uchar** to avoid casts. - Changed most Field functions to use uchar* instead of char* (reduced a lot of casts). - Changed field->val_xxx(xxx, new_ptr) to take const pointers. Other changes: - Removed a lot of not needed casts - Added a few new cast required by other changes - Added some cast to my_multi_malloc() arguments for safety (as string lengths needs to be uint, not size_t). - Fixed all calls to hash-get-key functions to use size_t*. (Needed to be done explicitely as this conflict was often hided by casting the function to hash_get_key). - Changed some buffers to memory regions to uchar* to avoid casts. - Changed some string lengths from uint to size_t. - Changed field->ptr to be uchar* instead of char*. This allowed us to get rid of a lot of casts. - Some changes from true -> TRUE, false -> FALSE, unsigned char -> uchar - Include zlib.h in some files as we needed declaration of crc32() - Changed MY_FILE_ERROR to be (size_t) -1. - Changed many variables to hold the result of my_read() / my_write() to be size_t. This was needed to properly detect errors (which are returned as (size_t) -1). - Removed some very old VMS code - Changed packfrm()/unpackfrm() to not be depending on uint size (portability fix) - Removed windows specific code to restore cursor position as this causes slowdown on windows and we should not mix read() and pread() calls anyway as this is not thread safe. Updated function comment to reflect this. Changed function that depended on original behavior of my_pwrite() to itself restore the cursor position (one such case). - Added some missing checking of return value of malloc(). - Changed definition of MOD_PAD_CHAR_TO_FULL_LENGTH to avoid 'long' overflow. - Changed type of table_def::m_size from my_size_t to ulong to reflect that m_size is the number of elements in the array, not a string/memory length. - Moved THD::max_row_length() to table.cc (as it's not depending on THD). Inlined max_row_length_blob() into this function. - More function comments - Fixed some compiler warnings when compiled without partitions. - Removed setting of LEX_STRING() arguments in declaration (portability fix). - Some trivial indentation/variable name changes. - Some trivial code simplifications: - Replaced some calls to alloc_root + memcpy to use strmake_root()/strdup_root(). - Changed some calls from memdup() to strmake() (Safety fix) - Simpler loops in client-simple.c
2007-05-10 12:59:39 +03:00
(*hash->free)((uchar*) record);
2000-07-31 21:29:14 +02:00
DBUG_RETURN(0);
}
/*
Update keys when record has changed.
This is much more efficent than using a delete & insert.
*/
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
my_bool my_hash_update(HASH *hash, uchar *record, uchar *old_key,
size_t old_key_length)
2000-07-31 21:29:14 +02:00
{
uint new_index,new_pos_index,blength,records,empty;
size_t idx;
2000-07-31 21:29:14 +02:00
HASH_LINK org_link,*data,*previous,*pos;
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
DBUG_ENTER("my_hash_update");
if (HASH_UNIQUE & hash->flags)
{
HASH_SEARCH_STATE state;
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
uchar *found, *new_key= (uchar*) my_hash_key(hash, record, &idx, 1);
if ((found= my_hash_first(hash, new_key, idx, &state)))
{
do
{
if (found != record)
DBUG_RETURN(1); /* Duplicate entry */
}
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
while ((found= my_hash_next(hash, new_key, idx, &state)));
}
}
2000-07-31 21:29:14 +02:00
data=dynamic_element(&hash->array,0,HASH_LINK*);
blength=hash->blength; records=hash->records;
/* Search after record with key */
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
idx= my_hash_mask(calc_hash(hash, old_key, (old_key_length ?
old_key_length :
hash->key_length)),
blength, records);
new_index= my_hash_mask(rec_hashnr(hash, record), blength, records);
2000-07-31 21:29:14 +02:00
if (idx == new_index)
DBUG_RETURN(0); /* Nothing to do (No record check) */
previous=0;
for (;;)
{
if ((pos= data+idx)->data == record)
break;
previous=pos;
if ((idx=pos->next) == NO_RECORD)
DBUG_RETURN(1); /* Not found in links */
}
org_link= *pos;
empty=idx;
/* Relink record from current chain */
if (!previous)
{
if (pos->next != NO_RECORD)
{
empty=pos->next;
*pos= data[pos->next];
}
}
else
previous->next=pos->next; /* unlink pos */
/* Move data to correct position */
if (new_index == empty)
{
/*
At this point record is unlinked from the old chain, thus it holds
random position. By the chance this position is equal to position
for the first element in the new chain. That means updated record
is the only record in the new chain.
*/
if (empty != idx)
{
/*
Record was moved while unlinking it from the old chain.
Copy data to a new position.
*/
data[empty]= org_link;
}
data[empty].next= NO_RECORD;
DBUG_RETURN(0);
}
2000-07-31 21:29:14 +02:00
pos=data+new_index;
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
new_pos_index= my_hash_rec_mask(hash, pos, blength, records);
2000-07-31 21:29:14 +02:00
if (new_index != new_pos_index)
{ /* Other record in wrong position */
data[empty] = *pos;
movelink(data,new_index,new_pos_index,empty);
org_link.next=NO_RECORD;
data[new_index]= org_link;
}
else
{ /* Link in chain at right position */
org_link.next=data[new_index].next;
data[empty]=org_link;
data[new_index].next=empty;
}
DBUG_RETURN(0);
}
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
uchar *my_hash_element(HASH *hash, ulong idx)
2000-07-31 21:29:14 +02:00
{
if (idx < hash->records)
return dynamic_element(&hash->array,idx,HASH_LINK*)->data;
return 0;
}
/*
Replace old row with new row. This should only be used when key
isn't changed
*/
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
void my_hash_replace(HASH *hash, HASH_SEARCH_STATE *current_record,
uchar *new_row)
{
if (*current_record != NO_RECORD) /* Safety */
dynamic_element(&hash->array, *current_record, HASH_LINK*)->data= new_row;
}
2000-07-31 21:29:14 +02:00
#ifndef DBUG_OFF
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
my_bool my_hash_check(HASH *hash)
2000-07-31 21:29:14 +02:00
{
int error;
uint i,rec_link,found,max_links,seek,links,idx;
uint records,blength;
HASH_LINK *data,*hash_info;
records=hash->records; blength=hash->blength;
data=dynamic_element(&hash->array,0,HASH_LINK*);
error=0;
for (i=found=max_links=seek=0 ; i < records ; i++)
{
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
if (my_hash_rec_mask(hash, data + i, blength, records) == i)
2000-07-31 21:29:14 +02:00
{
found++; seek++; links=1;
for (idx=data[i].next ;
idx != NO_RECORD && found < records + 1;
idx=hash_info->next)
{
if (idx >= records)
{
DBUG_PRINT("error",
("Found pointer outside array to %d from link starting at %d",
idx,i));
error=1;
}
hash_info=data+idx;
seek+= ++links;
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
if ((rec_link= my_hash_rec_mask(hash, hash_info,
blength, records)) != i)
2000-07-31 21:29:14 +02:00
{
Bug#37958 - test main.plugin crash on Mac OS X when selecting from EXAMPLE engine. This patch contains fixes for two problems: 1. As originally reported, the server crashed on Mac OS X when trying to access an EXAMPLE table after the EXAMPLE plugin was installed. It turned out that the dynamically loaded EXAMPLE plugin called the function hash_earch() from a Mac OS X system library, instead of hash_earch() from MySQL's mysys library. Makefile.am in storage/example does not include libmysys. So the Mac OS X linker arranged the hash_search() function to be linked to the system library when the shared object is loaded. One possible solution would be to include libmysys into the linkage of dynamic plugins. But then we must have a libmysys.so, which must be used by the server too. This could have a minimal performance impact, but foremost the change seems to bee too risky at the current state of MySQL 5.1. The selected solution is to rename MySQL's hash_search() to my_hash_search() like it has been done before with hash_insert() and hash_reset(). Since this is the third time, we need to rename a hash_*() function, I did renamed all hash_*() functions to my_hash_*(). To avoid changing a zillion calls to these functions, and announcing this to hundreds of developers, I added defines that map the old names to the new names. This change is in hash.h and hash.c. 2. The other problem was improper implementation of the handlerton-to-plugin mapping. We use a fixed-size array to hold a plugin reference for each handlerton. On every install of a handler plugin, we allocated a new slot of the array. On uninstall we did not free it. After some uninstall/install cycles the array overflowed. We did not check for overflow. One fix is to check for overflow to stop the crashes. Another fix is to free the array slot at uninstall and search for a free slot at plugin install. This change is in handler.cc.
2008-10-01 12:21:15 +02:00
DBUG_PRINT("error", ("Record in wrong link at %d: Start %d "
"Record: 0x%lx Record-link %d",
idx, i, (long) hash_info->data, rec_link));
2000-07-31 21:29:14 +02:00
error=1;
}
else
found++;
}
if (links > max_links) max_links=links;
}
}
if (found != records)
{
DBUG_PRINT("error",("Found %u of %u records", found, records));
2000-07-31 21:29:14 +02:00
error=1;
}
if (records)
DBUG_PRINT("info",
("records: %u seeks: %d max links: %d hitrate: %.2f",
2000-07-31 21:29:14 +02:00
records,seek,max_links,(float) seek / (float) records));
return error;
}
#endif