summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorAleksey Midenkov <midenok@gmail.com>2021-08-05 23:48:02 +0300
committerAleksey Midenkov <midenok@gmail.com>2021-08-05 23:48:02 +0300
commit160d97a4aaacbefb7f91a7e30a79b4d7937468a8 (patch)
tree85d605ea7c6676d46482da36d26ba081905d78cc
parentb8deb02859bbb869159134fa20ec14cfd875e11b (diff)
downloadmariadb-git-160d97a4aaacbefb7f91a7e30a79b4d7937468a8.tar.gz
MDEV-18734 ASAN heap-use-after-free upon sorting by blob column from partitioned table
ha_partition stores records in array of m_ordered_rec_buffer and uses it for prio queue in ordered index scan. When the records are restored from the array the blob buffers may be already freed or rewritten. The solution is to take temporary ownership of cached blob buffers via String::swap(). When the record is restored from m_ordered_rec_buffer the ownership is returned to table fields. Cleanups: init_record_priority_queue(): removed needless !m_ordered_rec_buffer check as there is same assertion few lines before. dbug_print_row() for arbitrary row pointer
-rw-r--r--mysql-test/suite/federated/federated_partition.result36
-rw-r--r--mysql-test/suite/federated/federated_partition.test25
-rw-r--r--mysql-test/suite/vcol/r/partition.result73
-rw-r--r--mysql-test/suite/vcol/t/partition.test48
-rw-r--r--sql/field.cc1
-rw-r--r--sql/field.h33
-rw-r--r--sql/filesort.cc9
-rw-r--r--sql/ha_partition.cc204
-rw-r--r--sql/ha_partition.h11
9 files changed, 381 insertions, 59 deletions
diff --git a/mysql-test/suite/federated/federated_partition.result b/mysql-test/suite/federated/federated_partition.result
index a2d5fcffd9b..cce5e3cf3e6 100644
--- a/mysql-test/suite/federated/federated_partition.result
+++ b/mysql-test/suite/federated/federated_partition.result
@@ -46,6 +46,42 @@ connection slave;
drop table federated.t1_1;
drop table federated.t1_2;
End of 5.1 tests
+#
+# MDEV-18734 ASAN heap-use-after-free upon sorting by blob column from partitioned table
+#
+connection slave;
+use federated;
+create table t1_1 (x int, b text, key(x));
+create table t1_2 (x int, b text, key(x));
+connection master;
+create table t1 (x int, b text, key(x)) engine=federated
+partition by range columns (x) (
+partition p1 values less than (40) connection='mysql://root@127.0.0.1:SLAVE_PORT/federated/t1_1',
+partition pn values less than (maxvalue) connection='mysql://root@127.0.0.1:SLAVE_PORT/federated/t1_2'
+);
+insert t1 values (1, 1), (2, 2), (3, 3), (4, 4), (5, 5), (6, 6), (7, 7), (8, 8);
+insert t1 select x + 8, x + 8 from t1;
+insert t1 select x + 16, x + 16 from t1;
+insert t1 select x + 49, repeat(x + 49, 100) from t1;
+flush tables;
+# This produces wrong result before MDEV-17573
+select x, left(b, 10) from t1 where x > 30 and x < 60 order by b;
+x left(b, 10)
+31 31
+32 32
+50 5050505050
+51 5151515151
+52 5252525252
+53 5353535353
+54 5454545454
+55 5555555555
+56 5656565656
+57 5757575757
+58 5858585858
+59 5959595959
+drop table t1;
+connection slave;
+drop table t1_1, t1_2;
connection master;
DROP TABLE IF EXISTS federated.t1;
DROP DATABASE IF EXISTS federated;
diff --git a/mysql-test/suite/federated/federated_partition.test b/mysql-test/suite/federated/federated_partition.test
index ef1e27ec505..c5c8433b4cf 100644
--- a/mysql-test/suite/federated/federated_partition.test
+++ b/mysql-test/suite/federated/federated_partition.test
@@ -50,4 +50,29 @@ drop table federated.t1_2;
--echo End of 5.1 tests
+--echo #
+--echo # MDEV-18734 ASAN heap-use-after-free upon sorting by blob column from partitioned table
+--echo #
+connection slave;
+use federated;
+create table t1_1 (x int, b text, key(x));
+create table t1_2 (x int, b text, key(x));
+connection master;
+--replace_result $SLAVE_MYPORT SLAVE_PORT
+eval create table t1 (x int, b text, key(x)) engine=federated
+ partition by range columns (x) (
+ partition p1 values less than (40) connection='mysql://root@127.0.0.1:$SLAVE_MYPORT/federated/t1_1',
+ partition pn values less than (maxvalue) connection='mysql://root@127.0.0.1:$SLAVE_MYPORT/federated/t1_2'
+);
+insert t1 values (1, 1), (2, 2), (3, 3), (4, 4), (5, 5), (6, 6), (7, 7), (8, 8);
+insert t1 select x + 8, x + 8 from t1;
+insert t1 select x + 16, x + 16 from t1;
+insert t1 select x + 49, repeat(x + 49, 100) from t1;
+flush tables;
+--echo # This produces wrong result before MDEV-17573
+select x, left(b, 10) from t1 where x > 30 and x < 60 order by b;
+drop table t1;
+connection slave;
+drop table t1_1, t1_2;
+
source include/federated_cleanup.inc;
diff --git a/mysql-test/suite/vcol/r/partition.result b/mysql-test/suite/vcol/r/partition.result
index bd1353fa145..d7c5052b72a 100644
--- a/mysql-test/suite/vcol/r/partition.result
+++ b/mysql-test/suite/vcol/r/partition.result
@@ -28,3 +28,76 @@ set statement sql_mode= '' for update t1 set i= 1, v= 2;
Warnings:
Warning 1906 The value specified for generated column 'v' in table 't1' has been ignored
drop table t1;
+#
+# MDEV-18734 ASAN heap-use-after-free in my_strnxfrm_simple_internal upon update on versioned partitioned table
+#
+# Cover queue_fix() in ha_partition::handle_ordered_index_scan()
+create or replace table t1 (
+x int auto_increment primary key,
+b text, v mediumtext as (b) virtual,
+index (v(10))
+) partition by range columns (x) (
+partition p1 values less than (3),
+partition p2 values less than (6),
+partition p3 values less than (9),
+partition p4 values less than (12),
+partition p5 values less than (15),
+partition p6 values less than (17),
+partition p7 values less than (19),
+partition p8 values less than (21),
+partition p9 values less than (23),
+partition p10 values less than (25),
+partition p11 values less than (27),
+partition p12 values less than (29),
+partition p13 values less than (31),
+partition p14 values less than (33),
+partition p15 values less than (35),
+partition pn values less than (maxvalue));
+insert into t1 (b) values
+(repeat('q', 8192)), (repeat('z', 8192)), (repeat('a', 8192)), (repeat('b', 8192)),
+(repeat('x', 8192)), (repeat('y', 8192));
+insert t1 (b) select b from t1;
+insert t1 (b) select b from t1;
+insert t1 (b) select b from t1;
+insert t1 (b) select b from t1;
+select x, left(b, 10), left(v, 10) from t1 where x > 30 and x < 60 order by v;
+x left(b, 10) left(v, 10)
+33 aaaaaaaaaa aaaaaaaaaa
+39 aaaaaaaaaa aaaaaaaaaa
+45 aaaaaaaaaa aaaaaaaaaa
+51 aaaaaaaaaa aaaaaaaaaa
+57 aaaaaaaaaa aaaaaaaaaa
+34 bbbbbbbbbb bbbbbbbbbb
+40 bbbbbbbbbb bbbbbbbbbb
+46 bbbbbbbbbb bbbbbbbbbb
+52 bbbbbbbbbb bbbbbbbbbb
+58 bbbbbbbbbb bbbbbbbbbb
+31 qqqqqqqqqq qqqqqqqqqq
+37 qqqqqqqqqq qqqqqqqqqq
+43 qqqqqqqqqq qqqqqqqqqq
+49 qqqqqqqqqq qqqqqqqqqq
+55 qqqqqqqqqq qqqqqqqqqq
+35 xxxxxxxxxx xxxxxxxxxx
+41 xxxxxxxxxx xxxxxxxxxx
+47 xxxxxxxxxx xxxxxxxxxx
+53 xxxxxxxxxx xxxxxxxxxx
+59 xxxxxxxxxx xxxxxxxxxx
+36 yyyyyyyyyy yyyyyyyyyy
+42 yyyyyyyyyy yyyyyyyyyy
+48 yyyyyyyyyy yyyyyyyyyy
+54 yyyyyyyyyy yyyyyyyyyy
+32 zzzzzzzzzz zzzzzzzzzz
+38 zzzzzzzzzz zzzzzzzzzz
+44 zzzzzzzzzz zzzzzzzzzz
+50 zzzzzzzzzz zzzzzzzzzz
+56 zzzzzzzzzz zzzzzzzzzz
+update t1 set b= 'bar' where v > 'a' limit 20;
+drop table t1;
+# Cover return_top_record() in ha_partition::handle_ordered_index_scan()
+create table t1 (x int primary key, b tinytext, v text as (b) virtual)
+partition by range columns (x) (
+partition p1 values less than (4),
+partition pn values less than (maxvalue));
+insert into t1 (x, b) values (1, ''), (2, ''), (3, 'a'), (4, 'b');
+update t1 set b= 'bar' where x > 0 order by v limit 2;
+drop table t1;
diff --git a/mysql-test/suite/vcol/t/partition.test b/mysql-test/suite/vcol/t/partition.test
index 889724fb1c5..408990b20a6 100644
--- a/mysql-test/suite/vcol/t/partition.test
+++ b/mysql-test/suite/vcol/t/partition.test
@@ -30,3 +30,51 @@ subpartition by hash(v) subpartitions 3 (
insert t1 set i= 0;
set statement sql_mode= '' for update t1 set i= 1, v= 2;
drop table t1;
+
+--echo #
+--echo # MDEV-18734 ASAN heap-use-after-free in my_strnxfrm_simple_internal upon update on versioned partitioned table
+--echo #
+--echo # Cover queue_fix() in ha_partition::handle_ordered_index_scan()
+create or replace table t1 (
+ x int auto_increment primary key,
+ b text, v mediumtext as (b) virtual,
+ index (v(10))
+) partition by range columns (x) (
+ partition p1 values less than (3),
+ partition p2 values less than (6),
+ partition p3 values less than (9),
+ partition p4 values less than (12),
+ partition p5 values less than (15),
+ partition p6 values less than (17),
+ partition p7 values less than (19),
+ partition p8 values less than (21),
+ partition p9 values less than (23),
+ partition p10 values less than (25),
+ partition p11 values less than (27),
+ partition p12 values less than (29),
+ partition p13 values less than (31),
+ partition p14 values less than (33),
+ partition p15 values less than (35),
+ partition pn values less than (maxvalue));
+insert into t1 (b) values
+(repeat('q', 8192)), (repeat('z', 8192)), (repeat('a', 8192)), (repeat('b', 8192)),
+(repeat('x', 8192)), (repeat('y', 8192));
+
+insert t1 (b) select b from t1;
+insert t1 (b) select b from t1;
+insert t1 (b) select b from t1;
+insert t1 (b) select b from t1;
+
+select x, left(b, 10), left(v, 10) from t1 where x > 30 and x < 60 order by v;
+update t1 set b= 'bar' where v > 'a' limit 20;
+
+drop table t1;
+
+--echo # Cover return_top_record() in ha_partition::handle_ordered_index_scan()
+create table t1 (x int primary key, b tinytext, v text as (b) virtual)
+partition by range columns (x) (
+ partition p1 values less than (4),
+ partition pn values less than (maxvalue));
+insert into t1 (x, b) values (1, ''), (2, ''), (3, 'a'), (4, 'b');
+update t1 set b= 'bar' where x > 0 order by v limit 2;
+drop table t1;
diff --git a/sql/field.cc b/sql/field.cc
index 074de35e0cf..4e6bc6b8341 100644
--- a/sql/field.cc
+++ b/sql/field.cc
@@ -8318,6 +8318,7 @@ int Field_blob::store(const char *from,uint length,CHARSET_INFO *cs)
copy_length= copier.well_formed_copy(field_charset,
(char*) value.ptr(), new_length,
cs, from, length);
+ value.length(copy_length);
Field_blob::store_length(copy_length);
bmove(ptr+packlength,(uchar*) &tmp,sizeof(char*));
diff --git a/sql/field.h b/sql/field.h
index f63fb670211..6d8e2aecd6e 100644
--- a/sql/field.h
+++ b/sql/field.h
@@ -3465,6 +3465,12 @@ public:
uchar *new_ptr, uint32 length,
uchar *new_null_ptr, uint new_null_bit);
void sql_type(String &str) const;
+ /**
+ Copy blob buffer into internal storage "value" and update record pointer.
+
+ @retval true Memory allocation error
+ @retval false Success
+ */
inline bool copy()
{
uchar *tmp= get_ptr();
@@ -3477,6 +3483,33 @@ public:
memcpy(ptr+packlength, &tmp, sizeof(char*));
return 0;
}
+ void swap(String &inout, bool set_read_value)
+ {
+ if (set_read_value)
+ read_value.swap(inout);
+ else
+ value.swap(inout);
+ }
+ /**
+ Return pointer to blob cache or NULL if not cached.
+ */
+ String * cached(bool *set_read_value)
+ {
+ char *tmp= (char *) get_ptr();
+ if (!value.is_empty() && tmp == value.ptr())
+ {
+ *set_read_value= false;
+ return &value;
+ }
+
+ if (!read_value.is_empty() && tmp == read_value.ptr())
+ {
+ *set_read_value= true;
+ return &read_value;
+ }
+
+ return NULL;
+ }
/* store value for the duration of the current read record */
inline void swap_value_and_read_value()
{
diff --git a/sql/filesort.cc b/sql/filesort.cc
index d76c39c3bd4..8b019caf8f5 100644
--- a/sql/filesort.cc
+++ b/sql/filesort.cc
@@ -608,6 +608,15 @@ const char* dbug_print_table_row(TABLE *table)
}
+const char* dbug_print_row(TABLE *table, uchar *rec)
+{
+ table->move_fields(table->field, rec, table->record[0]);
+ const char* ret= dbug_print_table_row(table);
+ table->move_fields(table->field, table->record[0], rec);
+ return ret;
+}
+
+
/*
Print a text, SQL-like record representation into dbug trace.
diff --git a/sql/ha_partition.cc b/sql/ha_partition.cc
index 6f50d078dff..4345f9eb9ca 100644
--- a/sql/ha_partition.cc
+++ b/sql/ha_partition.cc
@@ -5103,59 +5103,69 @@ bool ha_partition::init_record_priority_queue()
/*
Initialize the ordered record buffer.
*/
- if (!m_ordered_rec_buffer)
- {
- uint alloc_len;
- uint used_parts= bitmap_bits_set(&m_part_info->read_partitions);
- /* Allocate record buffer for each used partition. */
- m_priority_queue_rec_len= m_rec_length + PARTITION_BYTES_IN_POS;
- if (!m_using_extended_keys)
- m_priority_queue_rec_len += m_file[0]->ref_length;
- alloc_len= used_parts * m_priority_queue_rec_len;
- /* Allocate a key for temporary use when setting up the scan. */
- alloc_len+= table_share->max_key_length;
-
- if (!(m_ordered_rec_buffer= (uchar*)my_malloc(alloc_len, MYF(MY_WME))))
- DBUG_RETURN(true);
+ uint alloc_len;
+ uint used_parts= bitmap_bits_set(&m_part_info->read_partitions);
+ /* Allocate record buffer for each used partition. */
+ m_priority_queue_rec_len= m_rec_length + ORDERED_REC_OFFSET;
+ if (!m_using_extended_keys)
+ m_priority_queue_rec_len += m_file[0]->ref_length;
+ alloc_len= used_parts * m_priority_queue_rec_len;
+ /* Allocate a key for temporary use when setting up the scan. */
+ alloc_len+= table_share->max_key_length;
+ Ordered_blob_storage **blob_storage;
+ Ordered_blob_storage *objs;
+ const size_t n_all= used_parts * table->s->blob_fields;
+
+ if (!my_multi_malloc(MYF(MY_WME), &m_ordered_rec_buffer, alloc_len,
+ &blob_storage, n_all * sizeof(Ordered_blob_storage *),
+ &objs, n_all * sizeof(Ordered_blob_storage), NULL))
+ DBUG_RETURN(true);
- /*
- We set-up one record per partition and each record has 2 bytes in
- front where the partition id is written. This is used by ordered
- index_read.
- We also set-up a reference to the first record for temporary use in
- setting up the scan.
- */
- char *ptr= (char*) m_ordered_rec_buffer;
- uint i;
- for (i= bitmap_get_first_set(&m_part_info->read_partitions);
- i < m_tot_parts;
- i= bitmap_get_next_set(&m_part_info->read_partitions, i))
- {
- DBUG_PRINT("info", ("init rec-buf for part %u", i));
- int2store(ptr, i);
- ptr+= m_priority_queue_rec_len;
- }
- m_start_key.key= (const uchar*)ptr;
-
- /* Initialize priority queue, initialized to reading forward. */
- int (*cmp_func)(void *, uchar *, uchar *);
- void *cmp_arg;
- if (!m_using_extended_keys)
- {
- cmp_func= cmp_key_rowid_part_id;
- cmp_arg= (void*)this;
- }
- else
- {
- cmp_func= cmp_key_part_id;
- cmp_arg= (void*)m_curr_key_info;
- }
- if (init_queue(&m_queue, used_parts, 0, 0, cmp_func, cmp_arg, 0, 0))
+ /*
+ We set-up one record per partition and each record has 2 bytes in
+ front where the partition id is written. This is used by ordered
+ index_read.
+ We also set-up a reference to the first record for temporary use in
+ setting up the scan.
+ */
+ char *ptr= (char*) m_ordered_rec_buffer;
+ uint i;
+ for (i= bitmap_get_first_set(&m_part_info->read_partitions);
+ i < m_tot_parts;
+ i= bitmap_get_next_set(&m_part_info->read_partitions, i))
+ {
+ DBUG_PRINT("info", ("init rec-buf for part %u", i));
+ if (table->s->blob_fields)
{
- my_free(m_ordered_rec_buffer);
- m_ordered_rec_buffer= NULL;
- DBUG_RETURN(true);
+ for (uint j= 0; j < table->s->blob_fields; ++j, ++objs)
+ blob_storage[j]= new (objs) Ordered_blob_storage;
+ *((Ordered_blob_storage ***) ptr)= blob_storage;
+ blob_storage+= table->s->blob_fields;
}
+ int2store(ptr + sizeof(String **), i);
+ ptr+= m_priority_queue_rec_len;
+ }
+ m_start_key.key= (const uchar*)ptr;
+
+ /* Initialize priority queue, initialized to reading forward. */
+ int (*cmp_func)(void *, uchar *, uchar *);
+ void *cmp_arg;
+ if (!m_using_extended_keys)
+ {
+ cmp_func= cmp_key_rowid_part_id;
+ cmp_arg= (void*)this;
+ }
+ else
+ {
+ cmp_func= cmp_key_part_id;
+ cmp_arg= (void*)m_curr_key_info;
+ }
+ if (init_queue(&m_queue, used_parts, ORDERED_PART_NUM_OFFSET,
+ 0, cmp_func, cmp_arg, 0, 0))
+ {
+ my_free(m_ordered_rec_buffer);
+ m_ordered_rec_buffer= NULL;
+ DBUG_RETURN(true);
}
DBUG_RETURN(false);
}
@@ -5170,6 +5180,20 @@ void ha_partition::destroy_record_priority_queue()
DBUG_ENTER("ha_partition::destroy_record_priority_queue");
if (m_ordered_rec_buffer)
{
+ if (table->s->blob_fields)
+ {
+ char *ptr= (char *) m_ordered_rec_buffer;
+ for (uint i= bitmap_get_first_set(&m_part_info->read_partitions);
+ i < m_tot_parts;
+ i= bitmap_get_next_set(&m_part_info->read_partitions, i))
+ {
+ Ordered_blob_storage **blob_storage= *((Ordered_blob_storage ***) ptr);
+ for (uint b= 0; b < table->s->blob_fields; ++b)
+ blob_storage[b]->blob.free();
+ ptr+= m_priority_queue_rec_len;
+ }
+ }
+
delete_queue(&m_queue);
my_free(m_ordered_rec_buffer);
m_ordered_rec_buffer= NULL;
@@ -5394,7 +5418,7 @@ static int cmp_part_ids(uchar *ref1, uchar *ref2)
extern "C" int cmp_key_part_id(void *key_p, uchar *ref1, uchar *ref2)
{
int res;
- if ((res= key_rec_cmp(key_p, ref1 + PARTITION_BYTES_IN_POS,
+ if ((res= key_rec_cmp(key_p, ref1 + PARTITION_BYTES_IN_POS,
ref2 + PARTITION_BYTES_IN_POS)))
{
return res;
@@ -6133,8 +6157,8 @@ int ha_partition::handle_ordered_index_scan(uchar *buf, bool reverse_order)
{
DBUG_PRINT("info", ("reading from part %u (scan_type: %u)",
i, m_index_scan_type));
- DBUG_ASSERT(i == uint2korr(part_rec_buf_ptr));
- uchar *rec_buf_ptr= part_rec_buf_ptr + PARTITION_BYTES_IN_POS;
+ DBUG_ASSERT(i == uint2korr(part_rec_buf_ptr + ORDERED_PART_NUM_OFFSET));
+ uchar *rec_buf_ptr= part_rec_buf_ptr + ORDERED_REC_OFFSET;
int error;
handler *file= m_file[i];
@@ -6162,6 +6186,7 @@ int ha_partition::handle_ordered_index_scan(uchar *buf, bool reverse_order)
error= file->read_range_first(m_start_key.key? &m_start_key: NULL,
end_range, eq_range, TRUE);
memcpy(rec_buf_ptr, table->record[0], m_rec_length);
+
reverse_order= FALSE;
break;
}
@@ -6181,6 +6206,11 @@ int ha_partition::handle_ordered_index_scan(uchar *buf, bool reverse_order)
Initialize queue without order first, simply insert
*/
queue_element(&m_queue, j++)= part_rec_buf_ptr;
+ if (table->s->blob_fields)
+ {
+ Ordered_blob_storage **storage= *((Ordered_blob_storage ***) part_rec_buf_ptr);
+ swap_blobs(rec_buf_ptr, storage, false);
+ }
}
else if (error != HA_ERR_KEY_NOT_FOUND && error != HA_ERR_END_OF_FILE)
{
@@ -6229,10 +6259,15 @@ void ha_partition::return_top_record(uchar *buf)
{
uint part_id;
uchar *key_buffer= queue_top(&m_queue);
- uchar *rec_buffer= key_buffer + PARTITION_BYTES_IN_POS;
+ uchar *rec_buffer= key_buffer + ORDERED_REC_OFFSET;
- part_id= uint2korr(key_buffer);
+ part_id= uint2korr(key_buffer + ORDERED_PART_NUM_OFFSET);
memcpy(buf, rec_buffer, m_rec_length);
+ if (table->s->blob_fields)
+ {
+ Ordered_blob_storage **storage= *((Ordered_blob_storage ***) key_buffer);
+ swap_blobs(buf, storage, true);
+ }
m_last_part= part_id;
m_top_entry= part_id;
}
@@ -6268,7 +6303,7 @@ int ha_partition::handle_ordered_index_scan_key_not_found()
This partition is used and did return HA_ERR_KEY_NOT_FOUND
in index_read_map.
*/
- curr_rec_buf= part_buf + PARTITION_BYTES_IN_POS;
+ curr_rec_buf= part_buf + ORDERED_REC_OFFSET;
error= m_file[i]->ha_index_next(curr_rec_buf);
/* HA_ERR_KEY_NOT_FOUND is not allowed from index_next! */
DBUG_ASSERT(error != HA_ERR_KEY_NOT_FOUND);
@@ -6293,6 +6328,48 @@ int ha_partition::handle_ordered_index_scan_key_not_found()
}
+void ha_partition::swap_blobs(uchar * rec_buf, Ordered_blob_storage ** storage, bool restore)
+{
+ uint *ptr, *end;
+ uint blob_n= 0;
+ table->move_fields(table->field, rec_buf, table->record[0]);
+ for (ptr= table->s->blob_field, end= ptr + table->s->blob_fields;
+ ptr != end; ++ptr, ++blob_n)
+ {
+ DBUG_ASSERT(*ptr < table->s->fields);
+ Field_blob *blob= (Field_blob*) table->field[*ptr];
+ DBUG_ASSERT(blob->flags & BLOB_FLAG);
+ DBUG_ASSERT(blob->field_index == *ptr);
+ if (!bitmap_is_set(table->read_set, *ptr) || blob->is_null())
+ continue;
+
+ Ordered_blob_storage &s= *storage[blob_n];
+
+ if (restore)
+ {
+ /*
+ We protect only blob cache (value or read_value). If the cache was
+ empty that doesn't mean the blob was empty. Blobs allocated by a
+ storage engine should work just fine.
+ */
+ if (!s.blob.is_empty())
+ blob->swap(s.blob, s.set_read_value);
+ }
+ else
+ {
+ bool set_read_value;
+ String *cached= blob->cached(&set_read_value);
+ if (cached)
+ {
+ cached->swap(s.blob);
+ s.set_read_value= set_read_value;
+ }
+ }
+ }
+ table->move_fields(table->field, table->record[0], rec_buf);
+}
+
+
/*
Common routine to handle index_next with ordered results
@@ -6311,7 +6388,8 @@ int ha_partition::handle_ordered_next(uchar *buf, bool is_next_same)
{
int error;
uint part_id= m_top_entry;
- uchar *rec_buf= queue_top(&m_queue) + PARTITION_BYTES_IN_POS;
+ uchar *part_rec_buf_ptr= queue_top(&m_queue);
+ uchar *rec_buf= part_rec_buf_ptr + ORDERED_REC_OFFSET;
handler *file;
DBUG_ENTER("ha_partition::handle_ordered_next");
@@ -6354,7 +6432,15 @@ int ha_partition::handle_ordered_next(uchar *buf, bool is_next_same)
if (m_index_scan_type == partition_read_range)
{
error= file->read_range_next();
- memcpy(rec_buf, table->record[0], m_rec_length);
+ if (!error)
+ {
+ memcpy(rec_buf, table->record[0], m_rec_length);
+ if (table->s->blob_fields)
+ {
+ Ordered_blob_storage **storage= *((Ordered_blob_storage ***) part_rec_buf_ptr);
+ swap_blobs(rec_buf, storage, false);
+ }
+ }
}
else if (!is_next_same)
error= file->ha_index_next(rec_buf);
@@ -6410,7 +6496,7 @@ int ha_partition::handle_ordered_prev(uchar *buf)
{
int error;
uint part_id= m_top_entry;
- uchar *rec_buf= queue_top(&m_queue) + PARTITION_BYTES_IN_POS;
+ uchar *rec_buf= queue_top(&m_queue) + ORDERED_REC_OFFSET;
handler *file= m_file[part_id];
DBUG_ENTER("ha_partition::handle_ordered_prev");
diff --git a/sql/ha_partition.h b/sql/ha_partition.h
index 34174089bf6..2fccc37ec1f 100644
--- a/sql/ha_partition.h
+++ b/sql/ha_partition.h
@@ -21,7 +21,17 @@
#include "sql_partition.h" /* part_id_range, partition_element */
#include "queues.h" /* QUEUE */
+struct Ordered_blob_storage
+{
+ String blob;
+ bool set_read_value;
+ Ordered_blob_storage() : set_read_value(false)
+ {}
+};
+
#define PARTITION_BYTES_IN_POS 2
+#define ORDERED_PART_NUM_OFFSET sizeof(Ordered_blob_storage **)
+#define ORDERED_REC_OFFSET (ORDERED_PART_NUM_OFFSET + PARTITION_BYTES_IN_POS)
/** Struct used for partition_name_hash */
@@ -630,6 +640,7 @@ private:
int handle_ordered_next(uchar * buf, bool next_same);
int handle_ordered_prev(uchar * buf);
void return_top_record(uchar * buf);
+ void swap_blobs(uchar* rec_buf, Ordered_blob_storage ** storage, bool restore);
public:
/*
-------------------------------------------------------------------------