Commit d09426f9 authored by Marko Mäkelä's avatar Marko Mäkelä

MDEV-26537 InnoDB corrupts files due to incorrect st_blksize calculation

The st_blksize returned by fstat(2) is not documented to be
a power of 2, like we assumed in
commit 58252fff (MDEV-26040).
While on Linux, the st_blksize appears to report the file system
block size (which hopefully is not smaller than the sector size
of the underlying block device), on FreeBSD we observed
st_blksize values that might have been something similar to st_size.

Also IBM AIX was affected by this. A simple test case would
lead to a crash when using the minimum innodb_buffer_pool_size=5m
on both FreeBSD and AIX:

seq -f 'create table t%g engine=innodb select * from seq_1_to_200000;' \
1 100|mysql test&
seq -f 'create table u%g engine=innodb select * from seq_1_to_200000;' \
1 100|mysql test&

We will fix this by not trusting st_blksize at all, and assuming that
the smallest allowed write size (for O_DIRECT) is 4096 bytes. We hope
that no storage systems with larger block size exist. Anything larger
than 4096 bytes should be unlikely, given that it is the minimum
virtual memory page size of many contemporary processors.

MariaDB Server on Microsoft Windows was not affected by this.

While the 512-byte sector size of the venerable Seagate ST-225 is still
in widespread use, the minimum innodb_page_size is 4096 bytes, and
innodb_log_file_size can be set in integer multiples of 65536 bytes.

The only occasion where InnoDB uses smaller data file block sizes than
4096 bytes is with ROW_FORMAT=COMPRESSED tables with KEY_BLOCK_SIZE=1
or KEY_BLOCK_SIZE=2 (or innodb_page_size=4096). For such tables,
we will from now on preallocate space in integer multiples of 4096 bytes
and let regular writes extend the file by 1024, 2048, or 3072 bytes.

The view INFORMATION_SCHEMA.INNODB_SYS_TABLESPACES.FS_BLOCK_SIZE
should report the raw st_blksize.

For page_compressed tables, the function fil_space_get_block_size()
will map to 512 any st_blksize value that is larger than 4096.

os_file_set_size(): Assume that the file system block size is 4096 bytes,
and only support extending files to integer multiples of 4096 bytes.

fil_space_extend_must_retry(): Round down the preallocation size to
an integer multiple of 4096 bytes.
parent 1c378f1b
...@@ -46,6 +46,12 @@ perl; ...@@ -46,6 +46,12 @@ perl;
print "# bytes: ", (-s "$ENV{MYSQLD_DATADIR}/test/t1.ibd"), "\n"; print "# bytes: ", (-s "$ENV{MYSQLD_DATADIR}/test/t1.ibd"), "\n";
EOF EOF
INSERT INTO t1 SELECT seq,REPEAT('a',30000) FROM seq_1_to_20; INSERT INTO t1 SELECT seq,REPEAT('a',30000) FROM seq_1_to_20;
# Ensure that the file will be extended with the last 1024-byte page
# after the file was pre-extended in 4096-byte increments.
--disable_query_log
FLUSH TABLE t1 FOR EXPORT;
UNLOCK TABLES;
--enable_query_log
perl; perl;
print "# bytes: ", (-s "$ENV{MYSQLD_DATADIR}/test/t1.ibd"), "\n"; print "# bytes: ", (-s "$ENV{MYSQLD_DATADIR}/test/t1.ibd"), "\n";
EOF EOF
......
...@@ -980,11 +980,16 @@ fil_space_extend_must_retry( ...@@ -980,11 +980,16 @@ fil_space_extend_must_retry(
const page_size_t pageSize(space->flags); const page_size_t pageSize(space->flags);
const ulint page_size = pageSize.physical(); const ulint page_size = pageSize.physical();
/* fil_read_first_page() expects UNIV_PAGE_SIZE bytes. /* fil_read_first_page() expects innodb_page_size bytes.
fil_node_open_file() expects at least 4 * UNIV_PAGE_SIZE bytes.*/ fil_node_open_file() expects at least 4 * innodb_page_size bytes.
os_file_set_size() expects multiples of 4096 bytes.
For ROW_FORMAT=COMPRESSED tables using 1024-byte or 2048-byte
pages, we will preallocate up to an integer multiple of 4096 bytes,
and let normal writes append 1024, 2048, or 3072 bytes to the file. */
os_offset_t new_size = std::max( os_offset_t new_size = std::max(
os_offset_t(size - file_start_page_no) * page_size, (os_offset_t(size - file_start_page_no) * page_size)
os_offset_t(FIL_IBD_FILE_INITIAL_SIZE * UNIV_PAGE_SIZE)); & ~os_offset_t(4095),
os_offset_t(FIL_IBD_FILE_INITIAL_SIZE << srv_page_size_shift));
*success = os_file_set_size(node->name, node->handle, new_size, *success = os_file_set_size(node->name, node->handle, new_size,
FSP_FLAGS_HAS_PAGE_COMPRESSION(space->flags)); FSP_FLAGS_HAS_PAGE_COMPRESSION(space->flags));
......
...@@ -5405,6 +5405,8 @@ os_file_set_size( ...@@ -5405,6 +5405,8 @@ os_file_set_size(
os_offset_t size, os_offset_t size,
bool is_sparse) bool is_sparse)
{ {
ut_ad(!(size & 4095));
#ifdef _WIN32 #ifdef _WIN32
/* On Windows, changing file size works well and as expected for both /* On Windows, changing file size works well and as expected for both
sparse and normal files. sparse and normal files.
...@@ -5446,7 +5448,7 @@ os_file_set_size( ...@@ -5446,7 +5448,7 @@ os_file_set_size(
if (current_size >= size) { if (current_size >= size) {
return true; return true;
} }
current_size &= ~os_offset_t(statbuf.st_blksize - 1); current_size &= ~4095ULL;
err = posix_fallocate(file, current_size, err = posix_fallocate(file, current_size,
size - current_size); size - current_size);
} }
...@@ -5486,8 +5488,7 @@ os_file_set_size( ...@@ -5486,8 +5488,7 @@ os_file_set_size(
if (fstat(file, &statbuf)) { if (fstat(file, &statbuf)) {
return false; return false;
} }
os_offset_t current_size = statbuf.st_size os_offset_t current_size = statbuf.st_size & ~4095ULL;
& ~os_offset_t(statbuf.st_blksize - 1);
#endif #endif
if (current_size >= size) { if (current_size >= size) {
return true; return true;
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment