/**********************************************************************//**
Returns a new table, index, or space id. */
UNIV_INTERN
void
dict_hdr_get_new_id(
/*================*/
table_id_t* table_id, /*!< out: table id
(not assigned if NULL) */
index_id_t* index_id, /*!< out: index id
(not assigned if NULL) */
ulint* space_id) /*!< out: space id
(not assigned if NULL) */
{
dict_hdr_t* dict_hdr;
ib_id_t id;
mtr_t mtr;
mtr_start(&mtr);
dict_hdr = dict_hdr_get(&mtr);
if (table_id) {
id = mach_read_from_8(dict_hdr + DICT_HDR_TABLE_ID);
id++;
mlog_write_ull(dict_hdr + DICT_HDR_TABLE_ID, id, &mtr);
*table_id = id;
}
if (index_id) {
id = mach_read_from_8(dict_hdr + DICT_HDR_INDEX_ID);
id++;
mlog_write_ull(dict_hdr + DICT_HDR_INDEX_ID, id, &mtr);
*index_id = id;
}
if (space_id) {
*space_id = mtr_read_ulint(dict_hdr + DICT_HDR_MAX_SPACE_ID,
MLOG_4BYTES, &mtr);
if (fil_assign_new_space_id(space_id)) {
mlog_write_ulint(dict_hdr + DICT_HDR_MAX_SPACE_ID,
*space_id, MLOG_4BYTES, &mtr);
}
}
mtr_commit(&mtr);
}
inline long long int get_int_value(field_def_t *field, byte *value) {
switch (field->fixed_length) {
case 1: return mach_read_from_1(value) & ~(1<<7);
case 2: return mach_read_from_2(value) & ~(1<<15);
case 3: return mach_read_from_3(value) & 0x3FFFFFUL & ~(1L<<23);
case 4: return mach_read_from_4(value) & ~(1L<<31);
case 8: return make_longlong(mach_read_from_8(value)) & ~(1LL<<63);
}
return 0;
}
inline unsigned long long int get_uint_value(field_def_t *field, byte *value) {
switch (field->fixed_length) {
case 1: return mach_read_from_1(value);
case 2: return mach_read_from_2(value);
case 3: return mach_read_from_3(value) & 0x3FFFFFUL;
case 4: return mach_read_from_4(value);
case 8: return make_ulonglong(mach_read_from_8(value));
}
return 0;
}
void process_ibpage(page_t *page) {
ulint page_id;
rec_t *origin;
ulint offsets[MAX_TABLE_FIELDS + 2];
ulint offset, i;
// Skip tables if filter used
if (use_filter_id) {
dulint index_id = mach_read_from_8(page + PAGE_HEADER + PAGE_INDEX_ID);
if (index_id.low != filter_id.low || index_id.high != filter_id.high) {
if (debug) {
page_id = mach_read_from_4(page + FIL_PAGE_OFFSET);
printf("Skipped using index id filter: %lu!\n", page_id);
}
return;
}
}
// Read page id
page_id = mach_read_from_4(page + FIL_PAGE_OFFSET);
if (debug) printf("Page id: %lu\n", page_id);
// Check requested and actual formats
if (!check_page_format(page)) return;
// Find possible data area start point (at least 5 bytes of utility data)
offset = 100 + record_extra_bytes;
if (debug) printf("Starting offset: %lu. Checking %d table definitions.\n", offset, table_definitions_cnt);
// Walk through all possible positions to the end of page
// (start of directory - extra bytes of the last rec)
while (offset < UNIV_PAGE_SIZE - record_extra_bytes) {
// Get record pointer
origin = page + offset;
if (debug) printf("\nChecking offset: %lu: ", offset);
// Check all tables
for (i = 0; i < table_definitions_cnt; i++) {
// Get table info
table_def_t *table = &(table_definitions[i]);
if (debug) printf(" (%s) ", table->name);
// Check if origin points to a valid record
if (check_for_a_record(page, origin, table, offsets) && check_constraints(origin, table, offsets)) {
if (debug) printf("\n---------------------------------------------------\n"
"PAGE%lu: Found a table %s record: %p (offset = %lu)\n", page_id, table->name, origin, offset);
offset += process_ibrec(page, origin, table, offsets);
break;
}
}
// Check from next byte
offset++;
}
}
inline void print_field_value(byte *value, ulint len, field_def_t *field) {
switch (field->type) {
case FT_INTERNAL:
break;
case FT_CHAR:
case FT_TEXT:
print_string((char*)value, len, field);
break;
case FT_UINT:
printf("%llu", get_uint_value(field, value));
break;
case FT_INT:
printf("%lli", get_int_value(field, value));
break;
case FT_FLOAT:
printf("%f", mach_float_read(value));
break;
case FT_DOUBLE:
printf("%lf", mach_double_read(value));
break;
case FT_DATETIME:
print_datetime(make_longlong(mach_read_from_8(value)));
break;
case FT_DATE:
print_date(mach_read_from_3(value));
break;
case FT_TIME:
print_time(mach_read_from_3(value));
break;
case FT_ENUM:
print_enum(mach_read_from_1(value), field);
break;
case FT_DECIMAL:
print_decimal(value, field);
break;
default:
printf("undef(%d)", field->type);
}
}
void process_ibpage(page_t *page) {
static ulint id = 0;
ulint page_id;
dulint index_id;
char tmp[256];
int fn;
// Get page info
page_id = mach_read_from_4(page + FIL_PAGE_OFFSET);
index_id = mach_read_from_8(page + PAGE_HEADER + PAGE_INDEX_ID);
// Skip empty pages
if (ignore_crap && index_id.high == 0 && index_id.low == 0) return;
// Skip tables if filter used
if (use_filter_id && (index_id.low != filter_id.low || index_id.high != filter_id.high)) return;
if (count_pages) {
if (index_id.high >= 1000) {
if (ignore_crap) return;
printf("ERROR: Too high tablespace id! %ld >= 1000!\n", index_id.high);
exit(1);
}
if (index_id.low >= 10000) {
if (ignore_crap) return;
printf("ERROR: Too high index id! %ld >= 10000!\n", index_id.low);
exit(1);
}
page_counters[index_id.high][index_id.low]++;
return;
}
// Create table directory
sprintf(tmp, "pages-%u/%lu-%lu", (unsigned int)timestamp, index_id.high, index_id.low);
mkdir(tmp, 0755);
// Compose page file_name
sprintf(tmp, "pages-%u/%lu-%lu/%lu-%08lu.page", (unsigned int)timestamp, index_id.high, index_id.low, id++, page_id);
printf("Read page #%lu.. saving it to %s\n", page_id, tmp);
// Save page data
fn = open(tmp, O_WRONLY | O_CREAT | O_TRUNC, 0644);
if (!fn) error("Can't open file to save page!");
write(fn, page, UNIV_PAGE_SIZE);
close(fn);
}
static void print_page(uchar *p)
{
int type = mach_read_from_2(p + FIL_PAGE_TYPE);
if (type == FIL_PAGE_TYPE_ALLOCATED) {
return;
}
printf(COLUMN_NAME_FMT " %ld\n", "FIL_PAGE_OFFSET",
mach_read_from_4(p + FIL_PAGE_OFFSET));
printf(COLUMN_NAME_FMT " 0x%08lX\n", "FIL_PAGE_SPACE_OR_CHKSUM",
mach_read_from_4(p + FIL_PAGE_SPACE_OR_CHKSUM));
printf(COLUMN_NAME_FMT " %ld\n", "FIL_PAGE_PREV",
mach_read_from_4(p + FIL_PAGE_PREV));
printf(COLUMN_NAME_FMT " %ld\n", "FIL_PAGE_NEXT",
mach_read_from_4(p + FIL_PAGE_NEXT));
printf(COLUMN_NAME_FMT " %ld\n", "FIL_PAGE_LSN",
mach_read_from_4(p + FIL_PAGE_LSN));
printf(COLUMN_NAME_FMT " %ld\n", "FIL_PAGE_TYPE",
mach_read_from_2(p + FIL_PAGE_TYPE));
dulint flush_lsn_tuple = mach_read_from_6(p + FIL_PAGE_FILE_FLUSH_LSN);
uint64_t flush_lsn = (((uint64_t) flush_lsn_tuple.high) << 32) +
flush_lsn_tuple.low;
printf(COLUMN_NAME_FMT " %" PRIu64 "\n", "FIL_PAGE_FILE_FLUSH_LSN",
flush_lsn);
printf(COLUMN_NAME_FMT " %ld\n", "FIL_PAGE_ARCH_LOG_NO_OR_SPACE_ID",
mach_read_from_4(p + FIL_PAGE_ARCH_LOG_NO_OR_SPACE_ID));
printf(COLUMN_NAME_FMT " 0x%08lX\n", "FIL_PAGE_END_LSN_OLD_CHKSUM",
mach_read_from_4(p + UNIV_PAGE_SIZE - FIL_PAGE_END_LSN_OLD_CHKSUM));
uchar *pd = p + FIL_PAGE_DATA;
if (type == FIL_PAGE_TYPE_FSP_HDR) {
printf(COLUMN_NAME_FMT " %ld\n", "FSEG_HDR_SPACE",
mach_read_from_4(pd + FSEG_HDR_SPACE));
printf(COLUMN_NAME_FMT " %ld\n", "FSEG_HDR_PAGE_NO",
mach_read_from_4(pd + FSEG_HDR_PAGE_NO));
printf(COLUMN_NAME_FMT " %ld\n", "FSEG_HDR_OFFSET",
mach_read_from_4(pd + FSEG_HDR_OFFSET));
} else if (type == FIL_PAGE_INDEX) {
printf(COLUMN_NAME_FMT " 0x%lX\n", "PAGE_N_HEAP",
mach_read_from_2(pd + PAGE_N_HEAP));
printf(COLUMN_NAME_FMT " 0x%lX\n", "PAGE_FREE",
mach_read_from_2(pd + PAGE_FREE));
dulint index_id_tuple = mach_read_from_8(pd + PAGE_INDEX_ID);
uint64_t index_id = (((uint64_t) index_id_tuple.high) << 32) +
index_id_tuple.low;
printf(COLUMN_NAME_FMT " %" PRIu64 "\n", "PAGE_INDEX_ID",
index_id);
printf(COLUMN_NAME_FMT " %ld\n", "PAGE_BTR_SEG_LEAF",
mach_read_from_4(pd + PAGE_BTR_SEG_LEAF + FSEG_HDR_SPACE));
printf(COLUMN_NAME_FMT " %ld\n", "PAGE_BTR_SEG_TOP",
mach_read_from_4(pd + PAGE_BTR_SEG_TOP + FSEG_HDR_SPACE));
#if 0
int i;
for (i = 0; i < 80; i += 4) {
if (i == PAGE_BTR_SEG_LEAF ||
i == PAGE_N_HEAP ||
i == PAGE_INDEX_ID ||
i == PAGE_INDEX_ID + 4) {
continue;
}
char column_name[256];
snprintf(column_name, sizeof(column_name), "FIL_PAGE_DATA + %2d", i);
printf(COLUMN_NAME_FMT " %ld\n", column_name,
mach_read_from_4(p + FIL_PAGE_DATA + i));
}
#endif
}
printf("\n");
}
dict_table_t*
dict_load_table(
/*============*/
/* out: table, NULL if does not exist */
char* name) /* in: table name */
{
dict_table_t* table;
dict_table_t* sys_tables;
btr_pcur_t pcur;
dict_index_t* sys_index;
dtuple_t* tuple;
mem_heap_t* heap;
dfield_t* dfield;
rec_t* rec;
byte* field;
ulint len;
char* buf;
ulint space;
ulint n_cols;
mtr_t mtr;
ut_ad(mutex_own(&(dict_sys->mutex)));
heap = mem_heap_create(1000);
mtr_start(&mtr);
sys_tables = dict_table_get_low("SYS_TABLES");
sys_index = UT_LIST_GET_FIRST(sys_tables->indexes);
tuple = dtuple_create(heap, 1);
dfield = dtuple_get_nth_field(tuple, 0);
dfield_set_data(dfield, name, ut_strlen(name));
dict_index_copy_types(tuple, sys_index, 1);
btr_pcur_open_on_user_rec(sys_index, tuple, PAGE_CUR_GE,
BTR_SEARCH_LEAF, &pcur, &mtr);
rec = btr_pcur_get_rec(&pcur);
if (!btr_pcur_is_on_user_rec(&pcur, &mtr)
|| rec_get_deleted_flag(rec)) {
/* Not found */
btr_pcur_close(&pcur);
mtr_commit(&mtr);
mem_heap_free(heap);
return(NULL);
}
field = rec_get_nth_field(rec, 0, &len);
/* Check if the table name in record is the searched one */
if (len != ut_strlen(name) || ut_memcmp(name, field, len) != 0) {
btr_pcur_close(&pcur);
mtr_commit(&mtr);
mem_heap_free(heap);
return(NULL);
}
ut_a(0 == ut_strcmp("SPACE",
dict_field_get_col(
dict_index_get_nth_field(
dict_table_get_first_index(sys_tables), 9))->name));
field = rec_get_nth_field(rec, 9, &len);
space = mach_read_from_4(field);
ut_a(0 == ut_strcmp("N_COLS",
dict_field_get_col(
dict_index_get_nth_field(
dict_table_get_first_index(sys_tables), 4))->name));
field = rec_get_nth_field(rec, 4, &len);
n_cols = mach_read_from_4(field);
table = dict_mem_table_create(name, space, n_cols);
ut_a(0 == ut_strcmp("ID",
dict_field_get_col(
dict_index_get_nth_field(
dict_table_get_first_index(sys_tables), 3))->name));
field = rec_get_nth_field(rec, 3, &len);
table->id = mach_read_from_8(field);
field = rec_get_nth_field(rec, 5, &len);
table->type = mach_read_from_4(field);
if (table->type == DICT_TABLE_CLUSTER_MEMBER) {
ut_a(0);
field = rec_get_nth_field(rec, 6, &len);
table->mix_id = mach_read_from_8(field);
field = rec_get_nth_field(rec, 8, &len);
buf = mem_heap_alloc(heap, len);
ut_memcpy(buf, field, len);
table->cluster_name = buf;
}
if ((table->type == DICT_TABLE_CLUSTER)
|| (table->type == DICT_TABLE_CLUSTER_MEMBER)) {
field = rec_get_nth_field(rec, 7, &len);
table->mix_len = mach_read_from_4(field);
}
btr_pcur_close(&pcur);
mtr_commit(&mtr);
if (table->type == DICT_TABLE_CLUSTER_MEMBER) {
/* Load the cluster table definition if not yet in
memory cache */
dict_table_get_low(table->cluster_name);
}
dict_load_columns(table, heap);
dict_table_add_to_cache(table);
dict_load_indexes(table, heap);
ut_a(DB_SUCCESS == dict_load_foreigns(table->name));
mem_heap_free(heap);
return(table);
}
/********************************************************************//**
Removes unnecessary history data from a rollback segment. */
static
void
trx_purge_truncate_rseg_history(
/*============================*/
trx_rseg_t* rseg, /*!< in: rollback segment */
trx_id_t limit_trx_no, /*!< in: remove update undo logs whose
trx number is < limit_trx_no */
undo_no_t limit_undo_no) /*!< in: if transaction number is equal
to limit_trx_no, truncate undo records
with undo number < limit_undo_no */
{
fil_addr_t hdr_addr;
fil_addr_t prev_hdr_addr;
trx_rsegf_t* rseg_hdr;
page_t* undo_page;
trx_ulogf_t* log_hdr;
trx_usegf_t* seg_hdr;
ulint n_removed_logs = 0;
mtr_t mtr;
trx_id_t undo_trx_no;
mtr_start(&mtr);
mutex_enter(&(rseg->mutex));
rseg_hdr = trx_rsegf_get(rseg->space, rseg->zip_size,
rseg->page_no, &mtr);
hdr_addr = trx_purge_get_log_from_hist(
flst_get_last(rseg_hdr + TRX_RSEG_HISTORY, &mtr));
loop:
if (hdr_addr.page == FIL_NULL) {
mutex_exit(&(rseg->mutex));
mtr_commit(&mtr);
return;
}
undo_page = trx_undo_page_get(rseg->space, rseg->zip_size,
hdr_addr.page, &mtr);
log_hdr = undo_page + hdr_addr.boffset;
undo_trx_no = mach_read_from_8(log_hdr + TRX_UNDO_TRX_NO);
if (undo_trx_no >= limit_trx_no) {
if (undo_trx_no == limit_trx_no) {
trx_undo_truncate_start(rseg, rseg->space,
hdr_addr.page,
hdr_addr.boffset,
limit_undo_no);
}
mutex_enter(&kernel_mutex);
ut_a(trx_sys->rseg_history_len >= n_removed_logs);
trx_sys->rseg_history_len -= n_removed_logs;
mutex_exit(&kernel_mutex);
flst_truncate_end(rseg_hdr + TRX_RSEG_HISTORY,
log_hdr + TRX_UNDO_HISTORY_NODE,
n_removed_logs, &mtr);
mutex_exit(&(rseg->mutex));
mtr_commit(&mtr);
return;
}
prev_hdr_addr = trx_purge_get_log_from_hist(
flst_get_prev_addr(log_hdr + TRX_UNDO_HISTORY_NODE, &mtr));
n_removed_logs++;
seg_hdr = undo_page + TRX_UNDO_SEG_HDR;
if ((mach_read_from_2(seg_hdr + TRX_UNDO_STATE) == TRX_UNDO_TO_PURGE)
&& (mach_read_from_2(log_hdr + TRX_UNDO_NEXT_LOG) == 0)) {
/* We can free the whole log segment */
mutex_exit(&(rseg->mutex));
mtr_commit(&mtr);
trx_purge_free_segment(rseg, hdr_addr, n_removed_logs);
n_removed_logs = 0;
} else {
mutex_exit(&(rseg->mutex));
mtr_commit(&mtr);
}
mtr_start(&mtr);
mutex_enter(&(rseg->mutex));
rseg_hdr = trx_rsegf_get(rseg->space, rseg->zip_size,
rseg->page_no, &mtr);
hdr_addr = prev_hdr_addr;
goto loop;
}
/***************************************************************************
Updates the last not yet purged history log info in rseg when we have purged
a whole undo log. Advances also purge_sys->purge_trx_no past the purged log. */
static
void
trx_purge_rseg_get_next_history_log(
/*================================*/
trx_rseg_t* rseg) /* in: rollback segment */
{
page_t* undo_page;
trx_ulogf_t* log_hdr;
trx_usegf_t* seg_hdr;
fil_addr_t prev_log_addr;
dulint trx_no;
ibool del_marks;
mtr_t mtr;
ut_ad(mutex_own(&(purge_sys->mutex)));
mutex_enter(&(rseg->mutex));
ut_a(rseg->last_page_no != FIL_NULL);
purge_sys->purge_trx_no = ut_dulint_add(rseg->last_trx_no, 1);
purge_sys->purge_undo_no = ut_dulint_zero;
purge_sys->next_stored = FALSE;
mtr_start(&mtr);
undo_page = trx_undo_page_get_s_latched(rseg->space,
rseg->last_page_no, &mtr);
log_hdr = undo_page + rseg->last_offset;
seg_hdr = undo_page + TRX_UNDO_SEG_HDR;
/* Increase the purge page count by one for every handled log */
purge_sys->n_pages_handled++;
prev_log_addr = trx_purge_get_log_from_hist(
flst_get_prev_addr(log_hdr + TRX_UNDO_HISTORY_NODE, &mtr));
if (prev_log_addr.page == FIL_NULL) {
/* No logs left in the history list */
rseg->last_page_no = FIL_NULL;
mutex_exit(&(rseg->mutex));
mtr_commit(&mtr);
mutex_enter(&kernel_mutex);
/* Add debug code to track history list corruption reported
on the MySQL mailing list on Nov 9, 2004. The fut0lst.c
file-based list was corrupt. The prev node pointer was
FIL_NULL, even though the list length was over 8 million nodes!
We assume that purge truncates the history list in moderate
size pieces, and if we here reach the head of the list, the
list cannot be longer than 20 000 undo logs now. */
if (trx_sys->rseg_history_len > 20000) {
ut_print_timestamp(stderr);
fprintf(stderr,
" InnoDB: Warning: purge reached the"
" head of the history list,\n"
"InnoDB: but its length is still"
" reported as %lu! Make a detailed bug\n"
"InnoDB: report, and submit it"
" to http://bugs.mysql.com\n",
(ulong) trx_sys->rseg_history_len);
}
mutex_exit(&kernel_mutex);
return;
}
mutex_exit(&(rseg->mutex));
mtr_commit(&mtr);
/* Read the trx number and del marks from the previous log header */
mtr_start(&mtr);
log_hdr = trx_undo_page_get_s_latched(rseg->space,
prev_log_addr.page, &mtr)
+ prev_log_addr.boffset;
trx_no = mach_read_from_8(log_hdr + TRX_UNDO_TRX_NO);
del_marks = mach_read_from_2(log_hdr + TRX_UNDO_DEL_MARKS);
mtr_commit(&mtr);
mutex_enter(&(rseg->mutex));
rseg->last_page_no = prev_log_addr.page;
rseg->last_offset = prev_log_addr.boffset;
rseg->last_trx_no = trx_no;
rseg->last_del_marks = del_marks;
mutex_exit(&(rseg->mutex));
}
void process_ibpage(page_t *page) {
ulint page_id;
rec_t *origin;
ulint offsets[MAX_TABLE_FIELDS + 2];
ulint offset, i;
int is_page_valid = 0;
int comp;
unsigned int expected_records = 0;
unsigned int actual_records = 0;
int16_t b, infimum, supremum;
// Skip tables if filter used
if (use_filter_id) {
dulint index_id = mach_read_from_8(page + PAGE_HEADER + PAGE_INDEX_ID);
if (index_id.low != filter_id.low || index_id.high != filter_id.high) {
if (debug) {
page_id = mach_read_from_4(page + FIL_PAGE_OFFSET);
printf("Skipped using index id filter: %lu!\n", page_id);
}
return;
}
}
// Read page id
page_id = mach_read_from_4(page + FIL_PAGE_OFFSET);
if (debug) printf("Page id: %lu\n", page_id);
fprintf(f_result, "-- Page id: %lu", page_id);
// Check requested and actual formats
if (!check_page_format(page)) return;
if(table_definitions_cnt == 0){
fprintf(stderr, "There are no table definitions. Please check include/table_defs.h\n");
exit(EXIT_FAILURE);
}
is_page_valid = check_page(page, &expected_records);
// comp == 1 if page in COMPACT format and 0 if REDUNDANT
comp = page_is_comp(page);
fprintf(f_result, ", Format: %s", (comp ) ? "COMPACT": "REDUNDANT");
infimum = (comp) ? PAGE_NEW_INFIMUM : PAGE_OLD_INFIMUM;
supremum = (comp) ? PAGE_NEW_SUPREMUM : PAGE_OLD_SUPREMUM;
// Find possible data area start point (at least 5 bytes of utility data)
if(is_page_valid){
b = mach_read_from_2(page + infimum - 2);
offset = (comp) ? infimum + b : b;
}
else{
offset = 100 + record_extra_bytes;
}
fprintf(f_result, ", Records list: %s", is_page_valid? "Valid": "Invalid");
fprintf(f_result, ", Expected records: (%u %lu)", expected_records, mach_read_from_2(page + PAGE_HEADER + PAGE_N_RECS));
fprintf(f_result, "\n");
if (debug) printf("Starting offset: %lu (%lX). Checking %d table definitions.\n", offset, offset, table_definitions_cnt);
// Walk through all possible positions to the end of page
// (start of directory - extra bytes of the last rec)
//is_page_valid = 0;
while (offset < UNIV_PAGE_SIZE - record_extra_bytes && ( (offset != supremum ) || !is_page_valid) ) {
// Get record pointer
origin = page + offset;
if (debug) printf("\nChecking offset: 0x%lX: ", offset);
// Check all tables
for (i = 0; i < table_definitions_cnt; i++) {
// Get table info
table_def_t *table = &(table_definitions[i]);
if (debug) printf(" (%s) ", table->name);
// Check if origin points to a valid record
if (check_for_a_record(page, origin, table, offsets) && check_constraints(origin, table, offsets)) {
actual_records++;
if (debug) printf("\n---------------------------------------------------\n"
"PAGE%lu: Found a table %s record: %p (offset = %lu)\n", \
page_id, table->name, origin, offset);
if(is_page_valid){
process_ibrec(page, origin, table, offsets);
b = mach_read_from_2(page + offset - 2);
offset = (comp) ? offset + b : b;
}
else{
offset += process_ibrec(page, origin, table, offsets);
}
if (debug) printf("Next offset: 0x%lX", offset);
break;
}
else{
if(is_page_valid){
b = mach_read_from_2(page + offset - 2);
offset = (comp) ? offset + b : b;
}
else{
offset++;
}
if (debug) printf("\nNext offset: %lX", offset);
}
}
}
fprintf(f_result, "-- Page id: %lu", page_id);
fprintf(f_result, ", Found records: %u", actual_records);
fprintf(f_result, ", Lost records: %s", (actual_records != expected_records) ? "YES": "NO");
fprintf(f_result, ", Leaf page: %s", (mach_read_from_2(page + PAGE_HEADER + PAGE_LEVEL) == 0)? "YES": "NO");
fprintf(f_result, "\n");
}
/************************************************************************
Loads definitions for table indexes. Adds them to the data dictionary cache.
*/
static
void
dict_load_indexes(
/*==============*/
dict_table_t* table, /* in: table */
mem_heap_t* heap) /* in: memory heap for temporary storage */
{
dict_table_t* sys_indexes;
dict_index_t* sys_index;
dict_index_t* index;
btr_pcur_t pcur;
dtuple_t* tuple;
dfield_t* dfield;
rec_t* rec;
byte* field;
ulint len;
ulint name_len;
char* name_buf;
ulint type;
ulint space;
ulint page_no;
ulint n_fields;
byte* buf;
ibool is_sys_table;
dulint id;
mtr_t mtr;
ut_ad(mutex_own(&(dict_sys->mutex)));
if ((ut_dulint_get_high(table->id) == 0)
&& (ut_dulint_get_low(table->id) < DICT_HDR_FIRST_ID)) {
is_sys_table = TRUE;
} else {
is_sys_table = FALSE;
}
mtr_start(&mtr);
sys_indexes = dict_table_get_low("SYS_INDEXES");
sys_index = UT_LIST_GET_FIRST(sys_indexes->indexes);
tuple = dtuple_create(heap, 1);
dfield = dtuple_get_nth_field(tuple, 0);
buf = mem_heap_alloc(heap, 8);
mach_write_to_8(buf, table->id);
dfield_set_data(dfield, buf, 8);
dict_index_copy_types(tuple, sys_index, 1);
btr_pcur_open_on_user_rec(sys_index, tuple, PAGE_CUR_GE,
BTR_SEARCH_LEAF, &pcur, &mtr);
for (;;) {
if (!btr_pcur_is_on_user_rec(&pcur, &mtr)) {
break;
}
rec = btr_pcur_get_rec(&pcur);
field = rec_get_nth_field(rec, 0, &len);
ut_ad(len == 8);
if (ut_memcmp(buf, field, len) != 0) {
break;
}
ut_a(!rec_get_deleted_flag(rec));
field = rec_get_nth_field(rec, 1, &len);
ut_ad(len == 8);
id = mach_read_from_8(field);
ut_a(0 == ut_strcmp("NAME",
dict_field_get_col(
dict_index_get_nth_field(
dict_table_get_first_index(sys_indexes), 4))->name));
field = rec_get_nth_field(rec, 4, &name_len);
name_buf = mem_heap_alloc(heap, name_len + 1);
ut_memcpy(name_buf, field, name_len);
name_buf[name_len] = '\0';
field = rec_get_nth_field(rec, 5, &len);
n_fields = mach_read_from_4(field);
field = rec_get_nth_field(rec, 6, &len);
type = mach_read_from_4(field);
field = rec_get_nth_field(rec, 7, &len);
space = mach_read_from_4(field);
ut_a(0 == ut_strcmp("PAGE_NO",
dict_field_get_col(
dict_index_get_nth_field(
dict_table_get_first_index(sys_indexes), 8))->name));
field = rec_get_nth_field(rec, 8, &len);
page_no = mach_read_from_4(field);
if (is_sys_table
&& ((type & DICT_CLUSTERED)
|| ((table == dict_sys->sys_tables)
&& (name_len == ut_strlen("ID_IND"))
&& (0 == ut_memcmp(name_buf, "ID_IND",
name_len))))) {
/* The index was created in memory already in
booting */
} else {
index = dict_mem_index_create(table->name, name_buf,
space, type, n_fields);
index->page_no = page_no;
index->id = id;
dict_load_fields(table, index, heap);
dict_index_add_to_cache(table, index);
}
btr_pcur_move_to_next_user_rec(&pcur, &mtr);
}
btr_pcur_close(&pcur);
mtr_commit(&mtr);
}
/************************************************************************
Loads definitions for table columns. */
static
void
dict_load_columns(
/*==============*/
dict_table_t* table, /* in: table */
mem_heap_t* heap) /* in: memory heap for temporary storage */
{
dict_table_t* sys_columns;
dict_index_t* sys_index;
btr_pcur_t pcur;
dtuple_t* tuple;
dfield_t* dfield;
rec_t* rec;
byte* field;
ulint len;
byte* buf;
char* name_buf;
char* name;
ulint mtype;
ulint prtype;
ulint col_len;
ulint prec;
ulint i;
mtr_t mtr;
ut_ad(mutex_own(&(dict_sys->mutex)));
mtr_start(&mtr);
sys_columns = dict_table_get_low("SYS_COLUMNS");
sys_index = UT_LIST_GET_FIRST(sys_columns->indexes);
tuple = dtuple_create(heap, 1);
dfield = dtuple_get_nth_field(tuple, 0);
buf = mem_heap_alloc(heap, 8);
mach_write_to_8(buf, table->id);
dfield_set_data(dfield, buf, 8);
dict_index_copy_types(tuple, sys_index, 1);
btr_pcur_open_on_user_rec(sys_index, tuple, PAGE_CUR_GE,
BTR_SEARCH_LEAF, &pcur, &mtr);
for (i = 0; i < table->n_cols - DATA_N_SYS_COLS; i++) {
rec = btr_pcur_get_rec(&pcur);
ut_a(btr_pcur_is_on_user_rec(&pcur, &mtr));
ut_a(!rec_get_deleted_flag(rec));
field = rec_get_nth_field(rec, 0, &len);
ut_ad(len == 8);
ut_a(ut_dulint_cmp(table->id, mach_read_from_8(field)) == 0);
field = rec_get_nth_field(rec, 1, &len);
ut_ad(len == 4);
ut_a(i == mach_read_from_4(field));
ut_a(0 == ut_strcmp("NAME",
dict_field_get_col(
dict_index_get_nth_field(
dict_table_get_first_index(sys_columns), 4))->name));
field = rec_get_nth_field(rec, 4, &len);
name_buf = mem_heap_alloc(heap, len + 1);
ut_memcpy(name_buf, field, len);
name_buf[len] = '\0';
name = name_buf;
field = rec_get_nth_field(rec, 5, &len);
mtype = mach_read_from_4(field);
field = rec_get_nth_field(rec, 6, &len);
prtype = mach_read_from_4(field);
field = rec_get_nth_field(rec, 7, &len);
col_len = mach_read_from_4(field);
ut_a(0 == ut_strcmp("PREC",
dict_field_get_col(
dict_index_get_nth_field(
dict_table_get_first_index(sys_columns), 8))->name));
field = rec_get_nth_field(rec, 8, &len);
prec = mach_read_from_4(field);
dict_mem_table_add_col(table, name, mtype, prtype, col_len,
prec);
btr_pcur_move_to_next_user_rec(&pcur, &mtr);
}
btr_pcur_close(&pcur);
mtr_commit(&mtr);
}
dict_table_t*
dict_load_table_on_id(
/*==================*/
/* out: table; NULL if table does not exist */
dulint table_id) /* in: table id */
{
byte id_buf[8];
btr_pcur_t pcur;
mem_heap_t* heap;
dtuple_t* tuple;
dfield_t* dfield;
dict_index_t* sys_table_ids;
dict_table_t* sys_tables;
rec_t* rec;
byte* field;
ulint len;
dict_table_t* table;
char* name;
mtr_t mtr;
ut_ad(mutex_own(&(dict_sys->mutex)));
/* NOTE that the operation of this function is protected by
the dictionary mutex, and therefore no deadlocks can occur
with other dictionary operations. */
mtr_start(&mtr);
/*---------------------------------------------------*/
/* Get the secondary index based on ID for table SYS_TABLES */
sys_tables = dict_sys->sys_tables;
sys_table_ids = dict_table_get_next_index(
dict_table_get_first_index(sys_tables));
heap = mem_heap_create(256);
tuple = dtuple_create(heap, 1);
dfield = dtuple_get_nth_field(tuple, 0);
/* Write the table id in byte format to id_buf */
mach_write_to_8(id_buf, table_id);
dfield_set_data(dfield, id_buf, 8);
dict_index_copy_types(tuple, sys_table_ids, 1);
btr_pcur_open_on_user_rec(sys_table_ids, tuple, PAGE_CUR_GE,
BTR_SEARCH_LEAF, &pcur, &mtr);
rec = btr_pcur_get_rec(&pcur);
if (!btr_pcur_is_on_user_rec(&pcur, &mtr)
|| rec_get_deleted_flag(rec)) {
/* Not found */
btr_pcur_close(&pcur);
mtr_commit(&mtr);
mem_heap_free(heap);
return(NULL);
}
/*---------------------------------------------------*/
/* Now we have the record in the secondary index containing the
table ID and NAME */
rec = btr_pcur_get_rec(&pcur);
field = rec_get_nth_field(rec, 0, &len);
ut_ad(len == 8);
/* Check if the table id in record is the one searched for */
if (ut_dulint_cmp(table_id, mach_read_from_8(field)) != 0) {
btr_pcur_close(&pcur);
mtr_commit(&mtr);
mem_heap_free(heap);
return(NULL);
}
/* Now we get the table name from the record */
field = rec_get_nth_field(rec, 1, &len);
name = mem_heap_alloc(heap, len + 1);
ut_memcpy(name, field, len);
name[len] = '\0';
/* Load the table definition to memory */
table = dict_load_table(name);
ut_a(table);
btr_pcur_close(&pcur);
mtr_commit(&mtr);
mem_heap_free(heap);
return(table);
}
/*************************************************************//**
Pretty prints a dfield value according to its data type. Also the hex string
is printed if a string contains non-printable characters. */
UNIV_INTERN
void
dfield_print_also_hex(
/*==================*/
const dfield_t* dfield) /*!< in: dfield */
{
const byte* data;
ulint len;
ulint prtype;
ulint i;
ibool print_also_hex;
len = dfield_get_len(dfield);
data = dfield_get_data(dfield);
if (dfield_is_null(dfield)) {
fputs("NULL", stderr);
return;
}
prtype = dtype_get_prtype(dfield_get_type(dfield));
switch (dtype_get_mtype(dfield_get_type(dfield))) {
dulint id;
case DATA_INT:
switch (len) {
ulint val;
case 1:
val = mach_read_from_1(data);
if (!(prtype & DATA_UNSIGNED)) {
val &= ~0x80;
fprintf(stderr, "%ld", (long) val);
} else {
fprintf(stderr, "%lu", (ulong) val);
}
break;
case 2:
val = mach_read_from_2(data);
if (!(prtype & DATA_UNSIGNED)) {
val &= ~0x8000;
fprintf(stderr, "%ld", (long) val);
} else {
fprintf(stderr, "%lu", (ulong) val);
}
break;
case 3:
val = mach_read_from_3(data);
if (!(prtype & DATA_UNSIGNED)) {
val &= ~0x800000;
fprintf(stderr, "%ld", (long) val);
} else {
fprintf(stderr, "%lu", (ulong) val);
}
break;
case 4:
val = mach_read_from_4(data);
if (!(prtype & DATA_UNSIGNED)) {
val &= ~0x80000000;
fprintf(stderr, "%ld", (long) val);
} else {
fprintf(stderr, "%lu", (ulong) val);
}
break;
case 6:
id = mach_read_from_6(data);
fprintf(stderr, "{%lu %lu}",
ut_dulint_get_high(id),
ut_dulint_get_low(id));
break;
case 7:
id = mach_read_from_7(data);
fprintf(stderr, "{%lu %lu}",
ut_dulint_get_high(id),
ut_dulint_get_low(id));
break;
case 8:
id = mach_read_from_8(data);
fprintf(stderr, "{%lu %lu}",
ut_dulint_get_high(id),
ut_dulint_get_low(id));
break;
default:
goto print_hex;
}
break;
case DATA_SYS:
switch (prtype & DATA_SYS_PRTYPE_MASK) {
case DATA_TRX_ID:
id = mach_read_from_6(data);
fprintf(stderr, "trx_id " TRX_ID_FMT,
TRX_ID_PREP_PRINTF(id));
break;
case DATA_ROLL_PTR:
id = mach_read_from_7(data);
fprintf(stderr, "roll_ptr {%lu %lu}",
ut_dulint_get_high(id), ut_dulint_get_low(id));
break;
case DATA_ROW_ID:
id = mach_read_from_6(data);
fprintf(stderr, "row_id {%lu %lu}",
ut_dulint_get_high(id), ut_dulint_get_low(id));
break;
default:
id = mach_dulint_read_compressed(data);
fprintf(stderr, "mix_id {%lu %lu}",
ut_dulint_get_high(id), ut_dulint_get_low(id));
}
break;
case DATA_CHAR:
case DATA_VARCHAR:
print_also_hex = FALSE;
for (i = 0; i < len; i++) {
int c = *data++;
if (!isprint(c)) {
print_also_hex = TRUE;
fprintf(stderr, "\\x%02x", (unsigned char) c);
} else {
putc(c, stderr);
}
}
if (dfield_is_ext(dfield)) {
fputs("(external)", stderr);
}
if (!print_also_hex) {
break;
}
data = dfield_get_data(dfield);
/* fall through */
case DATA_BINARY:
default:
print_hex:
fputs(" Hex: ",stderr);
for (i = 0; i < len; i++) {
fprintf(stderr, "%02lx", (ulint) *data++);
}
if (dfield_is_ext(dfield)) {
fputs("(external)", stderr);
}
}
}
ulint
dict_truncate_index_tree(
/*=====================*/
/* out: new root page number, or
FIL_NULL on failure */
dict_table_t* table, /* in: the table the index belongs to */
btr_pcur_t* pcur, /* in/out: persistent cursor pointing to
record in the clustered index of
SYS_INDEXES table. The cursor may be
repositioned in this call. */
mtr_t* mtr) /* in: mtr having the latch
on the record page. The mtr may be
committed and restarted in this call. */
{
ulint root_page_no;
ulint space;
ulint type;
dulint index_id;
rec_t* rec;
byte* ptr;
ulint len;
ulint comp;
dict_index_t* index;
ut_ad(mutex_own(&(dict_sys->mutex)));
ut_a(!dict_table_is_comp(dict_sys->sys_indexes));
rec = btr_pcur_get_rec(pcur);
ptr = rec_get_nth_field_old(rec, DICT_SYS_INDEXES_PAGE_NO_FIELD, &len);
ut_ad(len == 4);
root_page_no = mtr_read_ulint(ptr, MLOG_4BYTES, mtr);
if (root_page_no == FIL_NULL) {
/* The tree has been freed. */
ut_print_timestamp(stderr);
fprintf(stderr, " InnoDB: Trying to TRUNCATE"
" a missing index of table %s!\n", table->name);
return(FIL_NULL);
}
ptr = rec_get_nth_field_old(rec,
DICT_SYS_INDEXES_SPACE_NO_FIELD, &len);
ut_ad(len == 4);
space = mtr_read_ulint(ptr, MLOG_4BYTES, mtr);
if (!fil_tablespace_exists_in_mem(space)) {
/* It is a single table tablespace and the .ibd file is
missing: do nothing */
ut_print_timestamp(stderr);
fprintf(stderr, " InnoDB: Trying to TRUNCATE"
" a missing .ibd file of table %s!\n", table->name);
return(FIL_NULL);
}
ptr = rec_get_nth_field_old(rec,
DICT_SYS_INDEXES_TYPE_FIELD, &len);
ut_ad(len == 4);
type = mach_read_from_4(ptr);
ptr = rec_get_nth_field_old(rec, 1, &len);
ut_ad(len == 8);
index_id = mach_read_from_8(ptr);
/* We free all the pages but the root page first; this operation
may span several mini-transactions */
btr_free_but_not_root(space, root_page_no);
/* Then we free the root page in the same mini-transaction where
we create the b-tree and write its new root page number to the
appropriate field in the SYS_INDEXES record: this mini-transaction
marks the B-tree totally truncated */
comp = page_is_comp(btr_page_get(space, root_page_no, RW_X_LATCH,
mtr));
btr_free_root(space, root_page_no, mtr);
/* We will temporarily write FIL_NULL to the PAGE_NO field
in SYS_INDEXES, so that the database will not get into an
inconsistent state in case it crashes between the mtr_commit()
below and the following mtr_commit() call. */
page_rec_write_index_page_no(rec, DICT_SYS_INDEXES_PAGE_NO_FIELD,
FIL_NULL, mtr);
/* We will need to commit the mini-transaction in order to avoid
deadlocks in the btr_create() call, because otherwise we would
be freeing and allocating pages in the same mini-transaction. */
btr_pcur_store_position(pcur, mtr);
mtr_commit(mtr);
mtr_start(mtr);
btr_pcur_restore_position(BTR_MODIFY_LEAF, pcur, mtr);
/* Find the index corresponding to this SYS_INDEXES record. */
for (index = UT_LIST_GET_FIRST(table->indexes);
index;
index = UT_LIST_GET_NEXT(indexes, index)) {
if (!ut_dulint_cmp(index->id, index_id)) {
break;
}
}
root_page_no = btr_create(type, space, index_id, comp, mtr);
if (index) {
index->page = (unsigned int) root_page_no;
} else {
ut_print_timestamp(stderr);
fprintf(stderr,
" InnoDB: Index %lu %lu of table %s is missing\n"
"InnoDB: from the data dictionary during TRUNCATE!\n",
ut_dulint_get_high(index_id),
ut_dulint_get_low(index_id),
table->name);
}
return(root_page_no);
}
/*******************************************************************//**
Truncates the index tree associated with a row in SYS_INDEXES table.
@return new root page number, or FIL_NULL on failure */
UNIV_INTERN
ulint
dict_truncate_index_tree(
/*=====================*/
dict_table_t* table, /*!< in: the table the index belongs to */
ulint space, /*!< in: 0=truncate,
nonzero=create the index tree in the
given tablespace */
btr_pcur_t* pcur, /*!< in/out: persistent cursor pointing to
record in the clustered index of
SYS_INDEXES table. The cursor may be
repositioned in this call. */
mtr_t* mtr) /*!< in: mtr having the latch
on the record page. The mtr may be
committed and restarted in this call. */
{
ulint root_page_no;
ibool drop = !space;
ulint zip_size;
ulint type;
index_id_t index_id;
rec_t* rec;
const byte* ptr;
ulint len;
dict_index_t* index;
ut_ad(mutex_own(&(dict_sys->mutex)));
ut_a(!dict_table_is_comp(dict_sys->sys_indexes));
rec = btr_pcur_get_rec(pcur);
ptr = rec_get_nth_field_old(rec, DICT_SYS_INDEXES_PAGE_NO_FIELD, &len);
ut_ad(len == 4);
root_page_no = mtr_read_ulint(ptr, MLOG_4BYTES, mtr);
if (drop && root_page_no == FIL_NULL) {
/* The tree has been freed. */
ut_print_timestamp(stderr);
fprintf(stderr, " InnoDB: Trying to TRUNCATE"
" a missing index of table %s!\n", table->name);
drop = FALSE;
}
ptr = rec_get_nth_field_old(rec,
DICT_SYS_INDEXES_SPACE_NO_FIELD, &len);
ut_ad(len == 4);
if (drop) {
space = mtr_read_ulint(ptr, MLOG_4BYTES, mtr);
}
zip_size = fil_space_get_zip_size(space);
if (UNIV_UNLIKELY(zip_size == ULINT_UNDEFINED)) {
/* It is a single table tablespace and the .ibd file is
missing: do nothing */
ut_print_timestamp(stderr);
fprintf(stderr, " InnoDB: Trying to TRUNCATE"
" a missing .ibd file of table %s!\n", table->name);
return(FIL_NULL);
}
ptr = rec_get_nth_field_old(rec,
DICT_SYS_INDEXES_TYPE_FIELD, &len);
ut_ad(len == 4);
type = mach_read_from_4(ptr);
ptr = rec_get_nth_field_old(rec, 1, &len);
ut_ad(len == 8);
index_id = mach_read_from_8(ptr);
if (!drop) {
goto create;
}
/* We free all the pages but the root page first; this operation
may span several mini-transactions */
btr_free_but_not_root(space, zip_size, root_page_no);
/* Then we free the root page in the same mini-transaction where
we create the b-tree and write its new root page number to the
appropriate field in the SYS_INDEXES record: this mini-transaction
marks the B-tree totally truncated */
btr_block_get(space, zip_size, root_page_no, RW_X_LATCH, NULL, mtr);
btr_free_root(space, zip_size, root_page_no, mtr);
create:
/* We will temporarily write FIL_NULL to the PAGE_NO field
in SYS_INDEXES, so that the database will not get into an
inconsistent state in case it crashes between the mtr_commit()
below and the following mtr_commit() call. */
page_rec_write_field(rec, DICT_SYS_INDEXES_PAGE_NO_FIELD,
FIL_NULL, mtr);
/* We will need to commit the mini-transaction in order to avoid
deadlocks in the btr_create() call, because otherwise we would
be freeing and allocating pages in the same mini-transaction. */
btr_pcur_store_position(pcur, mtr);
mtr_commit(mtr);
mtr_start(mtr);
btr_pcur_restore_position(BTR_MODIFY_LEAF, pcur, mtr);
/* Find the index corresponding to this SYS_INDEXES record. */
for (index = UT_LIST_GET_FIRST(table->indexes);
index;
index = UT_LIST_GET_NEXT(indexes, index)) {
if (index->id == index_id) {
root_page_no = btr_create(type, space, zip_size,
index_id, index, mtr);
index->page = (unsigned int) root_page_no;
return(root_page_no);
}
}
ut_print_timestamp(stderr);
fprintf(stderr,
" InnoDB: Index %llu of table %s is missing\n"
"InnoDB: from the data dictionary during TRUNCATE!\n",
(ullint) index_id,
table->name);
return(FIL_NULL);
}
/***********************************************************************//**
Updates the last not yet purged history log info in rseg when we have purged
a whole undo log. Advances also purge_sys->purge_trx_no past the purged log. */
static
void
trx_purge_rseg_get_next_history_log(
/*================================*/
trx_rseg_t* rseg) /*!< in: rollback segment */
{
page_t* undo_page;
trx_ulogf_t* log_hdr;
fil_addr_t prev_log_addr;
trx_id_t trx_no;
ibool del_marks;
mtr_t mtr;
rseg_queue_t rseg_queue;
const void* ptr;
mutex_enter(&(rseg->mutex));
ut_a(rseg->last_page_no != FIL_NULL);
purge_sys->purge_trx_no = rseg->last_trx_no + 1;
purge_sys->purge_undo_no = 0;
purge_sys->next_stored = FALSE;
mtr_start(&mtr);
undo_page = trx_undo_page_get_s_latched(
rseg->space, rseg->zip_size, rseg->last_page_no, &mtr);
log_hdr = undo_page + rseg->last_offset;
/* Increase the purge page count by one for every handled log */
purge_sys->n_pages_handled++;
prev_log_addr = trx_purge_get_log_from_hist(
flst_get_prev_addr(log_hdr + TRX_UNDO_HISTORY_NODE, &mtr));
if (prev_log_addr.page == FIL_NULL) {
/* No logs left in the history list */
rseg->last_page_no = FIL_NULL;
mutex_exit(&(rseg->mutex));
mtr_commit(&mtr);
mutex_enter(&kernel_mutex);
/* Add debug code to track history list corruption reported
on the MySQL mailing list on Nov 9, 2004. The fut0lst.c
file-based list was corrupt. The prev node pointer was
FIL_NULL, even though the list length was over 8 million nodes!
We assume that purge truncates the history list in large
size pieces, and if we here reach the head of the list, the
list cannot be longer than 2000 000 undo logs now. */
if (trx_sys->rseg_history_len > 2000000) {
ut_print_timestamp(stderr);
fprintf(stderr,
" InnoDB: Warning: purge reached the"
" head of the history list,\n"
"InnoDB: but its length is still"
" reported as %lu! Make a detailed bug\n"
"InnoDB: report, and submit it"
" to http://bugs.mysql.com\n",
(ulong) trx_sys->rseg_history_len);
}
mutex_exit(&kernel_mutex);
return;
}
mutex_exit(&(rseg->mutex));
mtr_commit(&mtr);
/* Read the trx number and del marks from the previous log header */
mtr_start(&mtr);
log_hdr = trx_undo_page_get_s_latched(rseg->space, rseg->zip_size,
prev_log_addr.page, &mtr)
+ prev_log_addr.boffset;
trx_no = mach_read_from_8(log_hdr + TRX_UNDO_TRX_NO);
del_marks = mach_read_from_2(log_hdr + TRX_UNDO_DEL_MARKS);
mtr_commit(&mtr);
mutex_enter(&(rseg->mutex));
rseg->last_page_no = prev_log_addr.page;
rseg->last_offset = prev_log_addr.boffset;
rseg->last_trx_no = trx_no;
rseg->last_del_marks = del_marks;
rseg_queue.rseg = rseg;
rseg_queue.trx_no = rseg->last_trx_no;
/* Purge can also produce events, however these are already ordered
in the rollback segment and any user generated event will be greater
than the events that Purge produces. ie. Purge can never produce
events from an empty rollback segment. */
mutex_enter(&purge_sys->bh_mutex);
ptr = ib_bh_push(purge_sys->ib_bh, &rseg_queue);
ut_a(ptr != NULL);
mutex_exit(&purge_sys->bh_mutex);
mutex_exit(&(rseg->mutex));
}
/*****************************************************************//**
Initializes the data dictionary memory structures when the database is
started. This function is also called when the data dictionary is created. */
UNIV_INTERN
void
dict_boot(void)
/*===========*/
{
dict_table_t* table;
dict_index_t* index;
dict_hdr_t* dict_hdr;
mem_heap_t* heap;
mtr_t mtr;
ulint error;
mtr_start(&mtr);
/* Create the hash tables etc. */
dict_init();
heap = mem_heap_create(450);
mutex_enter(&(dict_sys->mutex));
/* Get the dictionary header */
dict_hdr = dict_hdr_get(&mtr);
/* Because we only write new row ids to disk-based data structure
(dictionary header) when it is divisible by
DICT_HDR_ROW_ID_WRITE_MARGIN, in recovery we will not recover
the latest value of the row id counter. Therefore we advance
the counter at the database startup to avoid overlapping values.
Note that when a user after database startup first time asks for
a new row id, then because the counter is now divisible by
..._MARGIN, it will immediately be updated to the disk-based
header. */
dict_sys->row_id = DICT_HDR_ROW_ID_WRITE_MARGIN
+ ut_uint64_align_up(mach_read_from_8(dict_hdr + DICT_HDR_ROW_ID),
DICT_HDR_ROW_ID_WRITE_MARGIN);
/* Insert into the dictionary cache the descriptions of the basic
system tables */
/*-------------------------*/
table = dict_mem_table_create("SYS_TABLES", DICT_HDR_SPACE, 8, 0);
dict_mem_table_add_col(table, heap, "NAME", DATA_BINARY, 0, 0);
dict_mem_table_add_col(table, heap, "ID", DATA_BINARY, 0, 0);
/* ROW_FORMAT = (N_COLS >> 31) ? COMPACT : REDUNDANT */
dict_mem_table_add_col(table, heap, "N_COLS", DATA_INT, 0, 4);
/* TYPE is either DICT_TABLE_ORDINARY, or (TYPE & DICT_TF_COMPACT)
and (TYPE & DICT_TF_FORMAT_MASK) are nonzero and TYPE = table->flags */
dict_mem_table_add_col(table, heap, "TYPE", DATA_INT, 0, 4);
dict_mem_table_add_col(table, heap, "MIX_ID", DATA_BINARY, 0, 0);
/* MIX_LEN may contain additional table flags when
ROW_FORMAT!=REDUNDANT. Currently, these flags include
DICT_TF2_TEMPORARY. */
dict_mem_table_add_col(table, heap, "MIX_LEN", DATA_INT, 0, 4);
dict_mem_table_add_col(table, heap, "CLUSTER_NAME", DATA_BINARY, 0, 0);
dict_mem_table_add_col(table, heap, "SPACE", DATA_INT, 0, 4);
table->id = DICT_TABLES_ID;
dict_table_add_to_cache(table, heap);
dict_sys->sys_tables = table;
mem_heap_empty(heap);
index = dict_mem_index_create("SYS_TABLES", "CLUST_IND",
DICT_HDR_SPACE,
DICT_UNIQUE | DICT_CLUSTERED, 1);
dict_mem_index_add_field(index, "NAME", 0);
index->id = DICT_TABLES_ID;
error = dict_index_add_to_cache(table, index,
mtr_read_ulint(dict_hdr
+ DICT_HDR_TABLES,
MLOG_4BYTES, &mtr),
FALSE);
ut_a(error == DB_SUCCESS);
/*-------------------------*/
index = dict_mem_index_create("SYS_TABLES", "ID_IND",
DICT_HDR_SPACE, DICT_UNIQUE, 1);
dict_mem_index_add_field(index, "ID", 0);
index->id = DICT_TABLE_IDS_ID;
error = dict_index_add_to_cache(table, index,
mtr_read_ulint(dict_hdr
+ DICT_HDR_TABLE_IDS,
MLOG_4BYTES, &mtr),
FALSE);
ut_a(error == DB_SUCCESS);
/*-------------------------*/
table = dict_mem_table_create("SYS_COLUMNS", DICT_HDR_SPACE, 7, 0);
dict_mem_table_add_col(table, heap, "TABLE_ID", DATA_BINARY, 0, 0);
dict_mem_table_add_col(table, heap, "POS", DATA_INT, 0, 4);
dict_mem_table_add_col(table, heap, "NAME", DATA_BINARY, 0, 0);
dict_mem_table_add_col(table, heap, "MTYPE", DATA_INT, 0, 4);
dict_mem_table_add_col(table, heap, "PRTYPE", DATA_INT, 0, 4);
dict_mem_table_add_col(table, heap, "LEN", DATA_INT, 0, 4);
dict_mem_table_add_col(table, heap, "PREC", DATA_INT, 0, 4);
table->id = DICT_COLUMNS_ID;
dict_table_add_to_cache(table, heap);
dict_sys->sys_columns = table;
mem_heap_empty(heap);
index = dict_mem_index_create("SYS_COLUMNS", "CLUST_IND",
DICT_HDR_SPACE,
DICT_UNIQUE | DICT_CLUSTERED, 2);
dict_mem_index_add_field(index, "TABLE_ID", 0);
dict_mem_index_add_field(index, "POS", 0);
index->id = DICT_COLUMNS_ID;
error = dict_index_add_to_cache(table, index,
mtr_read_ulint(dict_hdr
+ DICT_HDR_COLUMNS,
MLOG_4BYTES, &mtr),
FALSE);
ut_a(error == DB_SUCCESS);
/*-------------------------*/
table = dict_mem_table_create("SYS_INDEXES", DICT_HDR_SPACE, 7, 0);
dict_mem_table_add_col(table, heap, "TABLE_ID", DATA_BINARY, 0, 0);
dict_mem_table_add_col(table, heap, "ID", DATA_BINARY, 0, 0);
dict_mem_table_add_col(table, heap, "NAME", DATA_BINARY, 0, 0);
dict_mem_table_add_col(table, heap, "N_FIELDS", DATA_INT, 0, 4);
dict_mem_table_add_col(table, heap, "TYPE", DATA_INT, 0, 4);
dict_mem_table_add_col(table, heap, "SPACE", DATA_INT, 0, 4);
dict_mem_table_add_col(table, heap, "PAGE_NO", DATA_INT, 0, 4);
/* The '+ 2' below comes from the fields DB_TRX_ID, DB_ROLL_PTR */
#if DICT_SYS_INDEXES_PAGE_NO_FIELD != 6 + 2
#error "DICT_SYS_INDEXES_PAGE_NO_FIELD != 6 + 2"
#endif
#if DICT_SYS_INDEXES_SPACE_NO_FIELD != 5 + 2
#error "DICT_SYS_INDEXES_SPACE_NO_FIELD != 5 + 2"
#endif
#if DICT_SYS_INDEXES_TYPE_FIELD != 4 + 2
#error "DICT_SYS_INDEXES_TYPE_FIELD != 4 + 2"
#endif
#if DICT_SYS_INDEXES_NAME_FIELD != 2 + 2
#error "DICT_SYS_INDEXES_NAME_FIELD != 2 + 2"
#endif
table->id = DICT_INDEXES_ID;
dict_table_add_to_cache(table, heap);
dict_sys->sys_indexes = table;
mem_heap_empty(heap);
index = dict_mem_index_create("SYS_INDEXES", "CLUST_IND",
DICT_HDR_SPACE,
DICT_UNIQUE | DICT_CLUSTERED, 2);
dict_mem_index_add_field(index, "TABLE_ID", 0);
dict_mem_index_add_field(index, "ID", 0);
index->id = DICT_INDEXES_ID;
error = dict_index_add_to_cache(table, index,
mtr_read_ulint(dict_hdr
+ DICT_HDR_INDEXES,
MLOG_4BYTES, &mtr),
FALSE);
ut_a(error == DB_SUCCESS);
/*-------------------------*/
table = dict_mem_table_create("SYS_FIELDS", DICT_HDR_SPACE, 3, 0);
dict_mem_table_add_col(table, heap, "INDEX_ID", DATA_BINARY, 0, 0);
dict_mem_table_add_col(table, heap, "POS", DATA_INT, 0, 4);
dict_mem_table_add_col(table, heap, "COL_NAME", DATA_BINARY, 0, 0);
table->id = DICT_FIELDS_ID;
dict_table_add_to_cache(table, heap);
dict_sys->sys_fields = table;
mem_heap_free(heap);
index = dict_mem_index_create("SYS_FIELDS", "CLUST_IND",
DICT_HDR_SPACE,
DICT_UNIQUE | DICT_CLUSTERED, 2);
dict_mem_index_add_field(index, "INDEX_ID", 0);
dict_mem_index_add_field(index, "POS", 0);
index->id = DICT_FIELDS_ID;
error = dict_index_add_to_cache(table, index,
mtr_read_ulint(dict_hdr
+ DICT_HDR_FIELDS,
MLOG_4BYTES, &mtr),
FALSE);
ut_a(error == DB_SUCCESS);
mtr_commit(&mtr);
/*-------------------------*/
/* Initialize the insert buffer table and index for each tablespace */
ibuf_init_at_db_start();
/* Load definitions of other indexes on system tables */
dict_load_sys_table(dict_sys->sys_tables);
dict_load_sys_table(dict_sys->sys_columns);
dict_load_sys_table(dict_sys->sys_indexes);
dict_load_sys_table(dict_sys->sys_fields);
mutex_exit(&(dict_sys->mutex));
}
