linux/fs/f2fs/data.c

1751 lines
40 KiB
C
Raw Normal View History

/*
* fs/f2fs/data.c
*
* Copyright (c) 2012 Samsung Electronics Co., Ltd.
* http://www.samsung.com/
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License version 2 as
* published by the Free Software Foundation.
*/
#include <linux/fs.h>
#include <linux/f2fs_fs.h>
#include <linux/buffer_head.h>
#include <linux/mpage.h>
#include <linux/writeback.h>
#include <linux/backing-dev.h>
#include <linux/pagevec.h>
#include <linux/blkdev.h>
#include <linux/bio.h>
#include <linux/prefetch.h>
#include <linux/uio.h>
#include <linux/cleancache.h>
#include "f2fs.h"
#include "node.h"
#include "segment.h"
#include "trace.h"
#include <trace/events/f2fs.h>
static void f2fs_read_end_io(struct bio *bio)
{
Merge branch 'for-3.14/core' of git://git.kernel.dk/linux-block Pull core block IO changes from Jens Axboe: "The major piece in here is the immutable bio_ve series from Kent, the rest is fairly minor. It was supposed to go in last round, but various issues pushed it to this release instead. The pull request contains: - Various smaller blk-mq fixes from different folks. Nothing major here, just minor fixes and cleanups. - Fix for a memory leak in the error path in the block ioctl code from Christian Engelmayer. - Header export fix from CaiZhiyong. - Finally the immutable biovec changes from Kent Overstreet. This enables some nice future work on making arbitrarily sized bios possible, and splitting more efficient. Related fixes to immutable bio_vecs: - dm-cache immutable fixup from Mike Snitzer. - btrfs immutable fixup from Muthu Kumar. - bio-integrity fix from Nic Bellinger, which is also going to stable" * 'for-3.14/core' of git://git.kernel.dk/linux-block: (44 commits) xtensa: fixup simdisk driver to work with immutable bio_vecs block/blk-mq-cpu.c: use hotcpu_notifier() blk-mq: for_each_* macro correctness block: Fix memory leak in rw_copy_check_uvector() handling bio-integrity: Fix bio_integrity_verify segment start bug block: remove unrelated header files and export symbol blk-mq: uses page->list incorrectly blk-mq: use __smp_call_function_single directly btrfs: fix missing increment of bi_remaining Revert "block: Warn and free bio if bi_end_io is not set" block: Warn and free bio if bi_end_io is not set blk-mq: fix initializing request's start time block: blk-mq: don't export blk_mq_free_queue() block: blk-mq: make blk_sync_queue support mq block: blk-mq: support draining mq queue dm cache: increment bi_remaining when bi_end_io is restored block: fixup for generic bio chaining block: Really silence spurious compiler warnings block: Silence spurious compiler warnings block: Kill bio_pair_split() ...
2014-01-30 19:19:05 +00:00
struct bio_vec *bvec;
int i;
if (f2fs_bio_encrypted(bio)) {
if (bio->bi_error) {
f2fs_release_crypto_ctx(bio->bi_private);
} else {
f2fs_end_io_crypto_work(bio->bi_private, bio);
return;
}
}
bio_for_each_segment_all(bvec, bio, i) {
struct page *page = bvec->bv_page;
if (!bio->bi_error) {
SetPageUptodate(page);
} else {
ClearPageUptodate(page);
SetPageError(page);
}
unlock_page(page);
}
bio_put(bio);
}
static void f2fs_write_end_io(struct bio *bio)
{
struct f2fs_sb_info *sbi = bio->bi_private;
Merge branch 'for-3.14/core' of git://git.kernel.dk/linux-block Pull core block IO changes from Jens Axboe: "The major piece in here is the immutable bio_ve series from Kent, the rest is fairly minor. It was supposed to go in last round, but various issues pushed it to this release instead. The pull request contains: - Various smaller blk-mq fixes from different folks. Nothing major here, just minor fixes and cleanups. - Fix for a memory leak in the error path in the block ioctl code from Christian Engelmayer. - Header export fix from CaiZhiyong. - Finally the immutable biovec changes from Kent Overstreet. This enables some nice future work on making arbitrarily sized bios possible, and splitting more efficient. Related fixes to immutable bio_vecs: - dm-cache immutable fixup from Mike Snitzer. - btrfs immutable fixup from Muthu Kumar. - bio-integrity fix from Nic Bellinger, which is also going to stable" * 'for-3.14/core' of git://git.kernel.dk/linux-block: (44 commits) xtensa: fixup simdisk driver to work with immutable bio_vecs block/blk-mq-cpu.c: use hotcpu_notifier() blk-mq: for_each_* macro correctness block: Fix memory leak in rw_copy_check_uvector() handling bio-integrity: Fix bio_integrity_verify segment start bug block: remove unrelated header files and export symbol blk-mq: uses page->list incorrectly blk-mq: use __smp_call_function_single directly btrfs: fix missing increment of bi_remaining Revert "block: Warn and free bio if bi_end_io is not set" block: Warn and free bio if bi_end_io is not set blk-mq: fix initializing request's start time block: blk-mq: don't export blk_mq_free_queue() block: blk-mq: make blk_sync_queue support mq block: blk-mq: support draining mq queue dm cache: increment bi_remaining when bi_end_io is restored block: fixup for generic bio chaining block: Really silence spurious compiler warnings block: Silence spurious compiler warnings block: Kill bio_pair_split() ...
2014-01-30 19:19:05 +00:00
struct bio_vec *bvec;
int i;
Merge branch 'for-3.14/core' of git://git.kernel.dk/linux-block Pull core block IO changes from Jens Axboe: "The major piece in here is the immutable bio_ve series from Kent, the rest is fairly minor. It was supposed to go in last round, but various issues pushed it to this release instead. The pull request contains: - Various smaller blk-mq fixes from different folks. Nothing major here, just minor fixes and cleanups. - Fix for a memory leak in the error path in the block ioctl code from Christian Engelmayer. - Header export fix from CaiZhiyong. - Finally the immutable biovec changes from Kent Overstreet. This enables some nice future work on making arbitrarily sized bios possible, and splitting more efficient. Related fixes to immutable bio_vecs: - dm-cache immutable fixup from Mike Snitzer. - btrfs immutable fixup from Muthu Kumar. - bio-integrity fix from Nic Bellinger, which is also going to stable" * 'for-3.14/core' of git://git.kernel.dk/linux-block: (44 commits) xtensa: fixup simdisk driver to work with immutable bio_vecs block/blk-mq-cpu.c: use hotcpu_notifier() blk-mq: for_each_* macro correctness block: Fix memory leak in rw_copy_check_uvector() handling bio-integrity: Fix bio_integrity_verify segment start bug block: remove unrelated header files and export symbol blk-mq: uses page->list incorrectly blk-mq: use __smp_call_function_single directly btrfs: fix missing increment of bi_remaining Revert "block: Warn and free bio if bi_end_io is not set" block: Warn and free bio if bi_end_io is not set blk-mq: fix initializing request's start time block: blk-mq: don't export blk_mq_free_queue() block: blk-mq: make blk_sync_queue support mq block: blk-mq: support draining mq queue dm cache: increment bi_remaining when bi_end_io is restored block: fixup for generic bio chaining block: Really silence spurious compiler warnings block: Silence spurious compiler warnings block: Kill bio_pair_split() ...
2014-01-30 19:19:05 +00:00
bio_for_each_segment_all(bvec, bio, i) {
struct page *page = bvec->bv_page;
f2fs_restore_and_release_control_page(&page);
if (unlikely(bio->bi_error)) {
set_page_dirty(page);
set_bit(AS_EIO, &page->mapping->flags);
f2fs_stop_checkpoint(sbi);
}
end_page_writeback(page);
dec_page_count(sbi, F2FS_WRITEBACK);
Merge branch 'for-3.14/core' of git://git.kernel.dk/linux-block Pull core block IO changes from Jens Axboe: "The major piece in here is the immutable bio_ve series from Kent, the rest is fairly minor. It was supposed to go in last round, but various issues pushed it to this release instead. The pull request contains: - Various smaller blk-mq fixes from different folks. Nothing major here, just minor fixes and cleanups. - Fix for a memory leak in the error path in the block ioctl code from Christian Engelmayer. - Header export fix from CaiZhiyong. - Finally the immutable biovec changes from Kent Overstreet. This enables some nice future work on making arbitrarily sized bios possible, and splitting more efficient. Related fixes to immutable bio_vecs: - dm-cache immutable fixup from Mike Snitzer. - btrfs immutable fixup from Muthu Kumar. - bio-integrity fix from Nic Bellinger, which is also going to stable" * 'for-3.14/core' of git://git.kernel.dk/linux-block: (44 commits) xtensa: fixup simdisk driver to work with immutable bio_vecs block/blk-mq-cpu.c: use hotcpu_notifier() blk-mq: for_each_* macro correctness block: Fix memory leak in rw_copy_check_uvector() handling bio-integrity: Fix bio_integrity_verify segment start bug block: remove unrelated header files and export symbol blk-mq: uses page->list incorrectly blk-mq: use __smp_call_function_single directly btrfs: fix missing increment of bi_remaining Revert "block: Warn and free bio if bi_end_io is not set" block: Warn and free bio if bi_end_io is not set blk-mq: fix initializing request's start time block: blk-mq: don't export blk_mq_free_queue() block: blk-mq: make blk_sync_queue support mq block: blk-mq: support draining mq queue dm cache: increment bi_remaining when bi_end_io is restored block: fixup for generic bio chaining block: Really silence spurious compiler warnings block: Silence spurious compiler warnings block: Kill bio_pair_split() ...
2014-01-30 19:19:05 +00:00
}
if (!get_pages(sbi, F2FS_WRITEBACK) &&
!list_empty(&sbi->cp_wait.task_list))
wake_up(&sbi->cp_wait);
bio_put(bio);
}
/*
* Low-level block read/write IO operations.
*/
static struct bio *__bio_alloc(struct f2fs_sb_info *sbi, block_t blk_addr,
int npages, bool is_read)
{
struct bio *bio;
bio = f2fs_bio_alloc(npages);
bio->bi_bdev = sbi->sb->s_bdev;
bio->bi_iter.bi_sector = SECTOR_FROM_BLOCK(blk_addr);
bio->bi_end_io = is_read ? f2fs_read_end_io : f2fs_write_end_io;
bio->bi_private = is_read ? NULL : sbi;
return bio;
}
static void __submit_merged_bio(struct f2fs_bio_info *io)
{
struct f2fs_io_info *fio = &io->fio;
if (!io->bio)
return;
if (is_read_io(fio->rw))
trace_f2fs_submit_read_bio(io->sbi->sb, fio, io->bio);
else
trace_f2fs_submit_write_bio(io->sbi->sb, fio, io->bio);
submit_bio(fio->rw, io->bio);
io->bio = NULL;
}
void f2fs_submit_merged_bio(struct f2fs_sb_info *sbi,
enum page_type type, int rw)
{
enum page_type btype = PAGE_TYPE_OF_BIO(type);
struct f2fs_bio_info *io;
io = is_read_io(rw) ? &sbi->read_io : &sbi->write_io[btype];
down_write(&io->io_rwsem);
/* change META to META_FLUSH in the checkpoint procedure */
if (type >= META_FLUSH) {
io->fio.type = META_FLUSH;
if (test_opt(sbi, NOBARRIER))
io->fio.rw = WRITE_FLUSH | REQ_META | REQ_PRIO;
else
io->fio.rw = WRITE_FLUSH_FUA | REQ_META | REQ_PRIO;
}
__submit_merged_bio(io);
up_write(&io->io_rwsem);
}
/*
* Fill the locked page with data located in the block address.
* Return unlocked page.
*/
int f2fs_submit_page_bio(struct f2fs_io_info *fio)
{
struct bio *bio;
struct page *page = fio->encrypted_page ? fio->encrypted_page : fio->page;
trace_f2fs_submit_page_bio(page, fio);
f2fs_trace_ios(fio, 0);
/* Allocate a new bio */
bio = __bio_alloc(fio->sbi, fio->blk_addr, 1, is_read_io(fio->rw));
if (bio_add_page(bio, page, PAGE_CACHE_SIZE, 0) < PAGE_CACHE_SIZE) {
bio_put(bio);
return -EFAULT;
}
submit_bio(fio->rw, bio);
return 0;
}
void f2fs_submit_page_mbio(struct f2fs_io_info *fio)
{
struct f2fs_sb_info *sbi = fio->sbi;
enum page_type btype = PAGE_TYPE_OF_BIO(fio->type);
struct f2fs_bio_info *io;
bool is_read = is_read_io(fio->rw);
struct page *bio_page;
io = is_read ? &sbi->read_io : &sbi->write_io[btype];
verify_block_addr(sbi, fio->blk_addr);
down_write(&io->io_rwsem);
if (!is_read)
inc_page_count(sbi, F2FS_WRITEBACK);
if (io->bio && (io->last_block_in_bio != fio->blk_addr - 1 ||
io->fio.rw != fio->rw))
__submit_merged_bio(io);
alloc_new:
if (io->bio == NULL) {
int bio_blocks = MAX_BIO_BLOCKS(sbi);
io->bio = __bio_alloc(sbi, fio->blk_addr, bio_blocks, is_read);
io->fio = *fio;
}
bio_page = fio->encrypted_page ? fio->encrypted_page : fio->page;
if (bio_add_page(io->bio, bio_page, PAGE_CACHE_SIZE, 0) <
PAGE_CACHE_SIZE) {
__submit_merged_bio(io);
goto alloc_new;
}
io->last_block_in_bio = fio->blk_addr;
f2fs_trace_ios(fio, 0);
up_write(&io->io_rwsem);
trace_f2fs_submit_page_mbio(fio->page, fio);
}
/*
* Lock ordering for the change of data block address:
* ->data_page
* ->node_page
* update block addresses in the node page
*/
void set_data_blkaddr(struct dnode_of_data *dn)
{
struct f2fs_node *rn;
__le32 *addr_array;
struct page *node_page = dn->node_page;
unsigned int ofs_in_node = dn->ofs_in_node;
f2fs_wait_on_page_writeback(node_page, NODE);
rn = F2FS_NODE(node_page);
/* Get physical address of data block */
addr_array = blkaddr_in_node(rn);
addr_array[ofs_in_node] = cpu_to_le32(dn->data_blkaddr);
set_page_dirty(node_page);
dn->node_changed = true;
}
int reserve_new_block(struct dnode_of_data *dn)
{
struct f2fs_sb_info *sbi = F2FS_I_SB(dn->inode);
if (unlikely(is_inode_flag_set(F2FS_I(dn->inode), FI_NO_ALLOC)))
return -EPERM;
if (unlikely(!inc_valid_block_count(sbi, dn->inode, 1)))
return -ENOSPC;
trace_f2fs_reserve_new_block(dn->inode, dn->nid, dn->ofs_in_node);
dn->data_blkaddr = NEW_ADDR;
set_data_blkaddr(dn);
mark_inode_dirty(dn->inode);
sync_inode_page(dn);
return 0;
}
int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index)
{
bool need_put = dn->inode_page ? false : true;
int err;
err = get_dnode_of_data(dn, index, ALLOC_NODE);
if (err)
return err;
if (dn->data_blkaddr == NULL_ADDR)
err = reserve_new_block(dn);
if (err || need_put)
f2fs_put_dnode(dn);
return err;
}
int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index)
{
struct extent_info ei;
struct inode *inode = dn->inode;
if (f2fs_lookup_extent_cache(inode, index, &ei)) {
dn->data_blkaddr = ei.blk + index - ei.fofs;
return 0;
}
return f2fs_reserve_block(dn, index);
}
struct page *get_read_data_page(struct inode *inode, pgoff_t index,
int rw, bool for_write)
{
struct address_space *mapping = inode->i_mapping;
struct dnode_of_data dn;
struct page *page;
struct extent_info ei;
int err;
struct f2fs_io_info fio = {
.sbi = F2FS_I_SB(inode),
.type = DATA,
.rw = rw,
.encrypted_page = NULL,
};
if (f2fs_encrypted_inode(inode) && S_ISREG(inode->i_mode))
return read_mapping_page(mapping, index, NULL);
page = f2fs_grab_cache_page(mapping, index, for_write);
if (!page)
return ERR_PTR(-ENOMEM);
if (f2fs_lookup_extent_cache(inode, index, &ei)) {
dn.data_blkaddr = ei.blk + index - ei.fofs;
goto got_it;
}
set_new_dnode(&dn, inode, NULL, NULL, 0);
err = get_dnode_of_data(&dn, index, LOOKUP_NODE);
if (err)
goto put_err;
f2fs_put_dnode(&dn);
if (unlikely(dn.data_blkaddr == NULL_ADDR)) {
err = -ENOENT;
goto put_err;
}
got_it:
if (PageUptodate(page)) {
unlock_page(page);
return page;
}
/*
* A new dentry page is allocated but not able to be written, since its
* new inode page couldn't be allocated due to -ENOSPC.
* In such the case, its blkaddr can be remained as NEW_ADDR.
* see, f2fs_add_link -> get_new_data_page -> init_inode_metadata.
*/
if (dn.data_blkaddr == NEW_ADDR) {
zero_user_segment(page, 0, PAGE_CACHE_SIZE);
SetPageUptodate(page);
unlock_page(page);
return page;
}
fio.blk_addr = dn.data_blkaddr;
fio.page = page;
err = f2fs_submit_page_bio(&fio);
if (err)
goto put_err;
return page;
put_err:
f2fs_put_page(page, 1);
return ERR_PTR(err);
}
struct page *find_data_page(struct inode *inode, pgoff_t index)
{
struct address_space *mapping = inode->i_mapping;
struct page *page;
page = find_get_page(mapping, index);
if (page && PageUptodate(page))
return page;
f2fs_put_page(page, 0);
page = get_read_data_page(inode, index, READ_SYNC, false);
if (IS_ERR(page))
return page;
if (PageUptodate(page))
return page;
wait_on_page_locked(page);
if (unlikely(!PageUptodate(page))) {
f2fs_put_page(page, 0);
return ERR_PTR(-EIO);
}
return page;
}
/*
* If it tries to access a hole, return an error.
* Because, the callers, functions in dir.c and GC, should be able to know
* whether this page exists or not.
*/
struct page *get_lock_data_page(struct inode *inode, pgoff_t index,
bool for_write)
{
struct address_space *mapping = inode->i_mapping;
struct page *page;
repeat:
page = get_read_data_page(inode, index, READ_SYNC, for_write);
if (IS_ERR(page))
return page;
/* wait for read completion */
lock_page(page);
if (unlikely(!PageUptodate(page))) {
f2fs_put_page(page, 1);
return ERR_PTR(-EIO);
}
if (unlikely(page->mapping != mapping)) {
f2fs_put_page(page, 1);
goto repeat;
}
return page;
}
/*
* Caller ensures that this data page is never allocated.
* A new zero-filled data page is allocated in the page cache.
f2fs: introduce a new global lock scheme In the previous version, f2fs uses global locks according to the usage types, such as directory operations, block allocation, block write, and so on. Reference the following lock types in f2fs.h. enum lock_type { RENAME, /* for renaming operations */ DENTRY_OPS, /* for directory operations */ DATA_WRITE, /* for data write */ DATA_NEW, /* for data allocation */ DATA_TRUNC, /* for data truncate */ NODE_NEW, /* for node allocation */ NODE_TRUNC, /* for node truncate */ NODE_WRITE, /* for node write */ NR_LOCK_TYPE, }; In that case, we lose the performance under the multi-threading environment, since every types of operations must be conducted one at a time. In order to address the problem, let's share the locks globally with a mutex array regardless of any types. So, let users grab a mutex and perform their jobs in parallel as much as possbile. For this, I propose a new global lock scheme as follows. 0. Data structure - f2fs_sb_info -> mutex_lock[NR_GLOBAL_LOCKS] - f2fs_sb_info -> node_write 1. mutex_lock_op(sbi) - try to get an avaiable lock from the array. - returns the index of the gottern lock variable. 2. mutex_unlock_op(sbi, index of the lock) - unlock the given index of the lock. 3. mutex_lock_all(sbi) - grab all the locks in the array before the checkpoint. 4. mutex_unlock_all(sbi) - release all the locks in the array after checkpoint. 5. block_operations() - call mutex_lock_all() - sync_dirty_dir_inodes() - grab node_write - sync_node_pages() Note that, the pairs of mutex_lock_op()/mutex_unlock_op() and mutex_lock_all()/mutex_unlock_all() should be used together. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-22 07:21:29 +00:00
*
* Also, caller should grab and release a rwsem by calling f2fs_lock_op() and
* f2fs_unlock_op().
* Note that, ipage is set only by make_empty_dir, and if any error occur,
* ipage should be released by this function.
*/
struct page *get_new_data_page(struct inode *inode,
struct page *ipage, pgoff_t index, bool new_i_size)
{
struct address_space *mapping = inode->i_mapping;
struct page *page;
struct dnode_of_data dn;
int err;
repeat:
page = f2fs_grab_cache_page(mapping, index, true);
if (!page) {
/*
* before exiting, we should make sure ipage will be released
* if any error occur.
*/
f2fs_put_page(ipage, 1);
return ERR_PTR(-ENOMEM);
}
set_new_dnode(&dn, inode, ipage, NULL, 0);
err = f2fs_reserve_block(&dn, index);
if (err) {
f2fs_put_page(page, 1);
return ERR_PTR(err);
}
if (!ipage)
f2fs_put_dnode(&dn);
if (PageUptodate(page))
goto got_it;
if (dn.data_blkaddr == NEW_ADDR) {
zero_user_segment(page, 0, PAGE_CACHE_SIZE);
SetPageUptodate(page);
} else {
f2fs_put_page(page, 1);
page = get_read_data_page(inode, index, READ_SYNC, true);
if (IS_ERR(page))
goto repeat;
/* wait for read completion */
lock_page(page);
}
got_it:
if (new_i_size && i_size_read(inode) <
((loff_t)(index + 1) << PAGE_CACHE_SHIFT)) {
i_size_write(inode, ((loff_t)(index + 1) << PAGE_CACHE_SHIFT));
/* Only the directory inode sets new_i_size */
set_inode_flag(F2FS_I(inode), FI_UPDATE_DIR);
}
return page;
}
static int __allocate_data_block(struct dnode_of_data *dn)
{
struct f2fs_sb_info *sbi = F2FS_I_SB(dn->inode);
struct f2fs_inode_info *fi = F2FS_I(dn->inode);
struct f2fs_summary sum;
struct node_info ni;
int seg = CURSEG_WARM_DATA;
pgoff_t fofs;
if (unlikely(is_inode_flag_set(F2FS_I(dn->inode), FI_NO_ALLOC)))
return -EPERM;
dn->data_blkaddr = datablock_addr(dn->node_page, dn->ofs_in_node);
if (dn->data_blkaddr == NEW_ADDR)
goto alloc;
if (unlikely(!inc_valid_block_count(sbi, dn->inode, 1)))
return -ENOSPC;
alloc:
get_node_info(sbi, dn->nid, &ni);
set_summary(&sum, dn->nid, dn->ofs_in_node, ni.version);
if (dn->ofs_in_node == 0 && dn->inode_page == dn->node_page)
seg = CURSEG_DIRECT_IO;
allocate_data_block(sbi, NULL, dn->data_blkaddr, &dn->data_blkaddr,
&sum, seg);
set_data_blkaddr(dn);
/* update i_size */
fofs = start_bidx_of_node(ofs_of_node(dn->node_page), fi) +
dn->ofs_in_node;
if (i_size_read(dn->inode) < ((loff_t)(fofs + 1) << PAGE_CACHE_SHIFT))
i_size_write(dn->inode,
((loff_t)(fofs + 1) << PAGE_CACHE_SHIFT));
return 0;
}
static int __allocate_data_blocks(struct inode *inode, loff_t offset,
size_t count)
{
struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
struct dnode_of_data dn;
u64 start = F2FS_BYTES_TO_BLK(offset);
u64 len = F2FS_BYTES_TO_BLK(count);
bool allocated;
u64 end_offset;
int err = 0;
while (len) {
f2fs_lock_op(sbi);
/* When reading holes, we need its node page */
set_new_dnode(&dn, inode, NULL, NULL, 0);
err = get_dnode_of_data(&dn, start, ALLOC_NODE);
if (err)
goto out;
allocated = false;
end_offset = ADDRS_PER_PAGE(dn.node_page, F2FS_I(inode));
while (dn.ofs_in_node < end_offset && len) {
block_t blkaddr;
if (unlikely(f2fs_cp_error(sbi))) {
err = -EIO;
f2fs: fix to handle io error in ->direct_IO Here is a oops reported as following message when testing generic/019 of xfstest: ------------[ cut here ]------------ kernel BUG at /home/yuchao/git/f2fs-dev/segment.c:882! invalid opcode: 0000 [#1] SMP Modules linked in: zram lz4_compress lz4_decompress f2fs(O) ip6table_filter ip6_tables ebtable_nat ebtables nf_conntrack_ipv4 nf_def CPU: 2 PID: 25441 Comm: fio Tainted: G O 4.3.0-rc1+ #6 Hardware name: Hewlett-Packard HP Z220 CMT Workstation/1790, BIOS K51 v01.61 05/16/2013 task: ffff8803f4e85580 ti: ffff8803fd61c000 task.ti: ffff8803fd61c000 RIP: 0010:[<ffffffffa0784981>] [<ffffffffa0784981>] new_curseg+0x321/0x330 [f2fs] RSP: 0018:ffff8803fd61f918 EFLAGS: 00010246 RAX: 00000000000007ed RBX: 0000000000000224 RCX: 000000000000001f RDX: 0000000000000800 RSI: ffffffffffffffff RDI: ffff8803f56f4300 RBP: ffff8803fd61f978 R08: 0000000000000000 R09: 0000000000000000 R10: 0000000000000024 R11: ffff8800d23bbd78 R12: ffff8800d0ef0000 R13: 0000000000000224 R14: 0000000000000000 R15: 0000000000000001 FS: 00007f827ff85700(0000) GS:ffff88041ea80000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: ffffffffff600000 CR3: 00000003fef17000 CR4: 00000000001406e0 Stack: 000007ea00000002 0000000100000001 ffff8803f6456248 000007ed0000002b 0000000000000224 ffff880404d1aa20 ffff8803fd61f9c8 ffff8800d0ef0000 ffff8803f6456248 0000000000000001 00000000ffffffff ffffffffa078f358 Call Trace: [<ffffffffa0785b87>] allocate_segment_by_default+0x1a7/0x1f0 [f2fs] [<ffffffffa078322c>] allocate_data_block+0x17c/0x360 [f2fs] [<ffffffffa0779521>] __allocate_data_block+0x131/0x1d0 [f2fs] [<ffffffffa077a995>] f2fs_direct_IO+0x4b5/0x580 [f2fs] [<ffffffff811510ae>] generic_file_direct_write+0xae/0x160 [<ffffffff811518f5>] __generic_file_write_iter+0xd5/0x1f0 [<ffffffff81151e07>] generic_file_write_iter+0xf7/0x200 [<ffffffff81319e38>] ? apparmor_file_permission+0x18/0x20 [<ffffffffa0768480>] ? f2fs_fallocate+0x1190/0x1190 [f2fs] [<ffffffffa07684c6>] f2fs_file_write_iter+0x46/0x90 [f2fs] [<ffffffff8120b4fe>] aio_run_iocb+0x1ee/0x290 [<ffffffff81700f7e>] ? mutex_lock+0x1e/0x50 [<ffffffff8120a1d7>] ? aio_read_events+0x207/0x2b0 [<ffffffff8120b913>] do_io_submit+0x373/0x630 [<ffffffff8120a4f6>] ? SyS_io_getevents+0x56/0xb0 [<ffffffff8120bbe0>] SyS_io_submit+0x10/0x20 [<ffffffff81703857>] entry_SYSCALL_64_fastpath+0x12/0x6a Code: 45 c8 48 8b 78 10 e8 9f 23 bf e0 41 8b 8c 24 cc 03 00 00 89 c7 31 d2 89 c6 89 d8 29 df f7 f1 29 d1 39 cf 0f 83 be fd ff ff eb RIP [<ffffffffa0784981>] new_curseg+0x321/0x330 [f2fs] RSP <ffff8803fd61f918> ---[ end trace 2e577d7f711ddb86 ]--- The reason is that: in the test of generic/019, we will trigger a manmade IO error in block layer through debugfs, after that, prefree segment will no longer be freed, because we always skip doing gc or checkpoint when there occurs an IO error. Meanwhile fio with aio engine generated a large number of direct IOs, which continue allocating spaces in free segment until we run out of them, eventually, results in panic in new_curseg as no more free segment was found. So, this patch changes to return EIO in direct_IO for this condition. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2015-09-21 12:17:52 +00:00
goto sync_out;
}
f2fs: fix to handle io error in ->direct_IO Here is a oops reported as following message when testing generic/019 of xfstest: ------------[ cut here ]------------ kernel BUG at /home/yuchao/git/f2fs-dev/segment.c:882! invalid opcode: 0000 [#1] SMP Modules linked in: zram lz4_compress lz4_decompress f2fs(O) ip6table_filter ip6_tables ebtable_nat ebtables nf_conntrack_ipv4 nf_def CPU: 2 PID: 25441 Comm: fio Tainted: G O 4.3.0-rc1+ #6 Hardware name: Hewlett-Packard HP Z220 CMT Workstation/1790, BIOS K51 v01.61 05/16/2013 task: ffff8803f4e85580 ti: ffff8803fd61c000 task.ti: ffff8803fd61c000 RIP: 0010:[<ffffffffa0784981>] [<ffffffffa0784981>] new_curseg+0x321/0x330 [f2fs] RSP: 0018:ffff8803fd61f918 EFLAGS: 00010246 RAX: 00000000000007ed RBX: 0000000000000224 RCX: 000000000000001f RDX: 0000000000000800 RSI: ffffffffffffffff RDI: ffff8803f56f4300 RBP: ffff8803fd61f978 R08: 0000000000000000 R09: 0000000000000000 R10: 0000000000000024 R11: ffff8800d23bbd78 R12: ffff8800d0ef0000 R13: 0000000000000224 R14: 0000000000000000 R15: 0000000000000001 FS: 00007f827ff85700(0000) GS:ffff88041ea80000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: ffffffffff600000 CR3: 00000003fef17000 CR4: 00000000001406e0 Stack: 000007ea00000002 0000000100000001 ffff8803f6456248 000007ed0000002b 0000000000000224 ffff880404d1aa20 ffff8803fd61f9c8 ffff8800d0ef0000 ffff8803f6456248 0000000000000001 00000000ffffffff ffffffffa078f358 Call Trace: [<ffffffffa0785b87>] allocate_segment_by_default+0x1a7/0x1f0 [f2fs] [<ffffffffa078322c>] allocate_data_block+0x17c/0x360 [f2fs] [<ffffffffa0779521>] __allocate_data_block+0x131/0x1d0 [f2fs] [<ffffffffa077a995>] f2fs_direct_IO+0x4b5/0x580 [f2fs] [<ffffffff811510ae>] generic_file_direct_write+0xae/0x160 [<ffffffff811518f5>] __generic_file_write_iter+0xd5/0x1f0 [<ffffffff81151e07>] generic_file_write_iter+0xf7/0x200 [<ffffffff81319e38>] ? apparmor_file_permission+0x18/0x20 [<ffffffffa0768480>] ? f2fs_fallocate+0x1190/0x1190 [f2fs] [<ffffffffa07684c6>] f2fs_file_write_iter+0x46/0x90 [f2fs] [<ffffffff8120b4fe>] aio_run_iocb+0x1ee/0x290 [<ffffffff81700f7e>] ? mutex_lock+0x1e/0x50 [<ffffffff8120a1d7>] ? aio_read_events+0x207/0x2b0 [<ffffffff8120b913>] do_io_submit+0x373/0x630 [<ffffffff8120a4f6>] ? SyS_io_getevents+0x56/0xb0 [<ffffffff8120bbe0>] SyS_io_submit+0x10/0x20 [<ffffffff81703857>] entry_SYSCALL_64_fastpath+0x12/0x6a Code: 45 c8 48 8b 78 10 e8 9f 23 bf e0 41 8b 8c 24 cc 03 00 00 89 c7 31 d2 89 c6 89 d8 29 df f7 f1 29 d1 39 cf 0f 83 be fd ff ff eb RIP [<ffffffffa0784981>] new_curseg+0x321/0x330 [f2fs] RSP <ffff8803fd61f918> ---[ end trace 2e577d7f711ddb86 ]--- The reason is that: in the test of generic/019, we will trigger a manmade IO error in block layer through debugfs, after that, prefree segment will no longer be freed, because we always skip doing gc or checkpoint when there occurs an IO error. Meanwhile fio with aio engine generated a large number of direct IOs, which continue allocating spaces in free segment until we run out of them, eventually, results in panic in new_curseg as no more free segment was found. So, this patch changes to return EIO in direct_IO for this condition. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2015-09-21 12:17:52 +00:00
blkaddr = datablock_addr(dn.node_page, dn.ofs_in_node);
if (blkaddr == NULL_ADDR || blkaddr == NEW_ADDR) {
err = __allocate_data_block(&dn);
if (err)
goto sync_out;
allocated = true;
}
len--;
start++;
dn.ofs_in_node++;
}
if (allocated)
sync_inode_page(&dn);
f2fs_put_dnode(&dn);
f2fs_unlock_op(sbi);
if (dn.node_changed)
f2fs_balance_fs(sbi);
}
return err;
sync_out:
if (allocated)
sync_inode_page(&dn);
f2fs_put_dnode(&dn);
out:
f2fs_unlock_op(sbi);
if (dn.node_changed)
f2fs_balance_fs(sbi);
return err;
}
/*
* f2fs_map_blocks() now supported readahead/bmap/rw direct_IO with
* f2fs_map_blocks structure.
* If original data blocks are allocated, then give them to blockdev.
* Otherwise,
* a. preallocate requested block addresses
* b. do not use extent cache for better performance
* c. give the block addresses to blockdev
*/
int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map,
int create, int flag)
{
unsigned int maxblocks = map->m_len;
struct dnode_of_data dn;
f2fs: fix to handle io error in ->direct_IO Here is a oops reported as following message when testing generic/019 of xfstest: ------------[ cut here ]------------ kernel BUG at /home/yuchao/git/f2fs-dev/segment.c:882! invalid opcode: 0000 [#1] SMP Modules linked in: zram lz4_compress lz4_decompress f2fs(O) ip6table_filter ip6_tables ebtable_nat ebtables nf_conntrack_ipv4 nf_def CPU: 2 PID: 25441 Comm: fio Tainted: G O 4.3.0-rc1+ #6 Hardware name: Hewlett-Packard HP Z220 CMT Workstation/1790, BIOS K51 v01.61 05/16/2013 task: ffff8803f4e85580 ti: ffff8803fd61c000 task.ti: ffff8803fd61c000 RIP: 0010:[<ffffffffa0784981>] [<ffffffffa0784981>] new_curseg+0x321/0x330 [f2fs] RSP: 0018:ffff8803fd61f918 EFLAGS: 00010246 RAX: 00000000000007ed RBX: 0000000000000224 RCX: 000000000000001f RDX: 0000000000000800 RSI: ffffffffffffffff RDI: ffff8803f56f4300 RBP: ffff8803fd61f978 R08: 0000000000000000 R09: 0000000000000000 R10: 0000000000000024 R11: ffff8800d23bbd78 R12: ffff8800d0ef0000 R13: 0000000000000224 R14: 0000000000000000 R15: 0000000000000001 FS: 00007f827ff85700(0000) GS:ffff88041ea80000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: ffffffffff600000 CR3: 00000003fef17000 CR4: 00000000001406e0 Stack: 000007ea00000002 0000000100000001 ffff8803f6456248 000007ed0000002b 0000000000000224 ffff880404d1aa20 ffff8803fd61f9c8 ffff8800d0ef0000 ffff8803f6456248 0000000000000001 00000000ffffffff ffffffffa078f358 Call Trace: [<ffffffffa0785b87>] allocate_segment_by_default+0x1a7/0x1f0 [f2fs] [<ffffffffa078322c>] allocate_data_block+0x17c/0x360 [f2fs] [<ffffffffa0779521>] __allocate_data_block+0x131/0x1d0 [f2fs] [<ffffffffa077a995>] f2fs_direct_IO+0x4b5/0x580 [f2fs] [<ffffffff811510ae>] generic_file_direct_write+0xae/0x160 [<ffffffff811518f5>] __generic_file_write_iter+0xd5/0x1f0 [<ffffffff81151e07>] generic_file_write_iter+0xf7/0x200 [<ffffffff81319e38>] ? apparmor_file_permission+0x18/0x20 [<ffffffffa0768480>] ? f2fs_fallocate+0x1190/0x1190 [f2fs] [<ffffffffa07684c6>] f2fs_file_write_iter+0x46/0x90 [f2fs] [<ffffffff8120b4fe>] aio_run_iocb+0x1ee/0x290 [<ffffffff81700f7e>] ? mutex_lock+0x1e/0x50 [<ffffffff8120a1d7>] ? aio_read_events+0x207/0x2b0 [<ffffffff8120b913>] do_io_submit+0x373/0x630 [<ffffffff8120a4f6>] ? SyS_io_getevents+0x56/0xb0 [<ffffffff8120bbe0>] SyS_io_submit+0x10/0x20 [<ffffffff81703857>] entry_SYSCALL_64_fastpath+0x12/0x6a Code: 45 c8 48 8b 78 10 e8 9f 23 bf e0 41 8b 8c 24 cc 03 00 00 89 c7 31 d2 89 c6 89 d8 29 df f7 f1 29 d1 39 cf 0f 83 be fd ff ff eb RIP [<ffffffffa0784981>] new_curseg+0x321/0x330 [f2fs] RSP <ffff8803fd61f918> ---[ end trace 2e577d7f711ddb86 ]--- The reason is that: in the test of generic/019, we will trigger a manmade IO error in block layer through debugfs, after that, prefree segment will no longer be freed, because we always skip doing gc or checkpoint when there occurs an IO error. Meanwhile fio with aio engine generated a large number of direct IOs, which continue allocating spaces in free segment until we run out of them, eventually, results in panic in new_curseg as no more free segment was found. So, this patch changes to return EIO in direct_IO for this condition. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2015-09-21 12:17:52 +00:00
struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
int mode = create ? ALLOC_NODE : LOOKUP_NODE_RA;
pgoff_t pgofs, end_offset;
int err = 0, ofs = 1;
struct extent_info ei;
bool allocated = false;
block_t blkaddr;
map->m_len = 0;
map->m_flags = 0;
/* it only supports block size == page size */
pgofs = (pgoff_t)map->m_lblk;
if (f2fs_lookup_extent_cache(inode, pgofs, &ei)) {
map->m_pblk = ei.blk + pgofs - ei.fofs;
map->m_len = min((pgoff_t)maxblocks, ei.fofs + ei.len - pgofs);
map->m_flags = F2FS_MAP_MAPPED;
goto out;
}
if (create)
f2fs_lock_op(sbi);
/* When reading holes, we need its node page */
set_new_dnode(&dn, inode, NULL, NULL, 0);
err = get_dnode_of_data(&dn, pgofs, mode);
if (err) {
if (err == -ENOENT)
err = 0;
goto unlock_out;
}
if (dn.data_blkaddr == NEW_ADDR || dn.data_blkaddr == NULL_ADDR) {
if (create) {
f2fs: fix to handle io error in ->direct_IO Here is a oops reported as following message when testing generic/019 of xfstest: ------------[ cut here ]------------ kernel BUG at /home/yuchao/git/f2fs-dev/segment.c:882! invalid opcode: 0000 [#1] SMP Modules linked in: zram lz4_compress lz4_decompress f2fs(O) ip6table_filter ip6_tables ebtable_nat ebtables nf_conntrack_ipv4 nf_def CPU: 2 PID: 25441 Comm: fio Tainted: G O 4.3.0-rc1+ #6 Hardware name: Hewlett-Packard HP Z220 CMT Workstation/1790, BIOS K51 v01.61 05/16/2013 task: ffff8803f4e85580 ti: ffff8803fd61c000 task.ti: ffff8803fd61c000 RIP: 0010:[<ffffffffa0784981>] [<ffffffffa0784981>] new_curseg+0x321/0x330 [f2fs] RSP: 0018:ffff8803fd61f918 EFLAGS: 00010246 RAX: 00000000000007ed RBX: 0000000000000224 RCX: 000000000000001f RDX: 0000000000000800 RSI: ffffffffffffffff RDI: ffff8803f56f4300 RBP: ffff8803fd61f978 R08: 0000000000000000 R09: 0000000000000000 R10: 0000000000000024 R11: ffff8800d23bbd78 R12: ffff8800d0ef0000 R13: 0000000000000224 R14: 0000000000000000 R15: 0000000000000001 FS: 00007f827ff85700(0000) GS:ffff88041ea80000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: ffffffffff600000 CR3: 00000003fef17000 CR4: 00000000001406e0 Stack: 000007ea00000002 0000000100000001 ffff8803f6456248 000007ed0000002b 0000000000000224 ffff880404d1aa20 ffff8803fd61f9c8 ffff8800d0ef0000 ffff8803f6456248 0000000000000001 00000000ffffffff ffffffffa078f358 Call Trace: [<ffffffffa0785b87>] allocate_segment_by_default+0x1a7/0x1f0 [f2fs] [<ffffffffa078322c>] allocate_data_block+0x17c/0x360 [f2fs] [<ffffffffa0779521>] __allocate_data_block+0x131/0x1d0 [f2fs] [<ffffffffa077a995>] f2fs_direct_IO+0x4b5/0x580 [f2fs] [<ffffffff811510ae>] generic_file_direct_write+0xae/0x160 [<ffffffff811518f5>] __generic_file_write_iter+0xd5/0x1f0 [<ffffffff81151e07>] generic_file_write_iter+0xf7/0x200 [<ffffffff81319e38>] ? apparmor_file_permission+0x18/0x20 [<ffffffffa0768480>] ? f2fs_fallocate+0x1190/0x1190 [f2fs] [<ffffffffa07684c6>] f2fs_file_write_iter+0x46/0x90 [f2fs] [<ffffffff8120b4fe>] aio_run_iocb+0x1ee/0x290 [<ffffffff81700f7e>] ? mutex_lock+0x1e/0x50 [<ffffffff8120a1d7>] ? aio_read_events+0x207/0x2b0 [<ffffffff8120b913>] do_io_submit+0x373/0x630 [<ffffffff8120a4f6>] ? SyS_io_getevents+0x56/0xb0 [<ffffffff8120bbe0>] SyS_io_submit+0x10/0x20 [<ffffffff81703857>] entry_SYSCALL_64_fastpath+0x12/0x6a Code: 45 c8 48 8b 78 10 e8 9f 23 bf e0 41 8b 8c 24 cc 03 00 00 89 c7 31 d2 89 c6 89 d8 29 df f7 f1 29 d1 39 cf 0f 83 be fd ff ff eb RIP [<ffffffffa0784981>] new_curseg+0x321/0x330 [f2fs] RSP <ffff8803fd61f918> ---[ end trace 2e577d7f711ddb86 ]--- The reason is that: in the test of generic/019, we will trigger a manmade IO error in block layer through debugfs, after that, prefree segment will no longer be freed, because we always skip doing gc or checkpoint when there occurs an IO error. Meanwhile fio with aio engine generated a large number of direct IOs, which continue allocating spaces in free segment until we run out of them, eventually, results in panic in new_curseg as no more free segment was found. So, this patch changes to return EIO in direct_IO for this condition. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2015-09-21 12:17:52 +00:00
if (unlikely(f2fs_cp_error(sbi))) {
err = -EIO;
goto put_out;
}
err = __allocate_data_block(&dn);
if (err)
goto put_out;
allocated = true;
map->m_flags = F2FS_MAP_NEW;
} else {
if (flag != F2FS_GET_BLOCK_FIEMAP ||
dn.data_blkaddr != NEW_ADDR) {
if (flag == F2FS_GET_BLOCK_BMAP)
err = -ENOENT;
goto put_out;
}
/*
* preallocated unwritten block should be mapped
* for fiemap.
*/
if (dn.data_blkaddr == NEW_ADDR)
map->m_flags = F2FS_MAP_UNWRITTEN;
}
}
map->m_flags |= F2FS_MAP_MAPPED;
map->m_pblk = dn.data_blkaddr;
map->m_len = 1;
end_offset = ADDRS_PER_PAGE(dn.node_page, F2FS_I(inode));
dn.ofs_in_node++;
pgofs++;
get_next:
if (map->m_len >= maxblocks)
goto sync_out;
if (dn.ofs_in_node >= end_offset) {
if (allocated)
sync_inode_page(&dn);
allocated = false;
f2fs_put_dnode(&dn);
if (create) {
f2fs_unlock_op(sbi);
if (dn.node_changed)
f2fs_balance_fs(sbi);
f2fs_lock_op(sbi);
}
set_new_dnode(&dn, inode, NULL, NULL, 0);
err = get_dnode_of_data(&dn, pgofs, mode);
if (err) {
if (err == -ENOENT)
err = 0;
goto unlock_out;
}
end_offset = ADDRS_PER_PAGE(dn.node_page, F2FS_I(inode));
}
blkaddr = datablock_addr(dn.node_page, dn.ofs_in_node);
if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR) {
if (create) {
if (unlikely(f2fs_cp_error(sbi))) {
err = -EIO;
goto sync_out;
}
err = __allocate_data_block(&dn);
if (err)
goto sync_out;
allocated = true;
map->m_flags |= F2FS_MAP_NEW;
blkaddr = dn.data_blkaddr;
} else {
/*
* we only merge preallocated unwritten blocks
* for fiemap.
*/
if (flag != F2FS_GET_BLOCK_FIEMAP ||
blkaddr != NEW_ADDR)
goto sync_out;
}
}
/* Give more consecutive addresses for the readahead */
if ((map->m_pblk != NEW_ADDR &&
blkaddr == (map->m_pblk + ofs)) ||
(map->m_pblk == NEW_ADDR &&
blkaddr == NEW_ADDR)) {
ofs++;
dn.ofs_in_node++;
pgofs++;
map->m_len++;
goto get_next;
}
sync_out:
if (allocated)
sync_inode_page(&dn);
put_out:
f2fs_put_dnode(&dn);
unlock_out:
if (create) {
f2fs_unlock_op(sbi);
if (dn.node_changed)
f2fs_balance_fs(sbi);
}
out:
trace_f2fs_map_blocks(inode, map, err);
return err;
}
static int __get_data_block(struct inode *inode, sector_t iblock,
struct buffer_head *bh, int create, int flag)
{
struct f2fs_map_blocks map;
int ret;
map.m_lblk = iblock;
map.m_len = bh->b_size >> inode->i_blkbits;
ret = f2fs_map_blocks(inode, &map, create, flag);
if (!ret) {
map_bh(bh, inode->i_sb, map.m_pblk);
bh->b_state = (bh->b_state & ~F2FS_MAP_FLAGS) | map.m_flags;
bh->b_size = map.m_len << inode->i_blkbits;
}
return ret;
}
static int get_data_block(struct inode *inode, sector_t iblock,
struct buffer_head *bh_result, int create, int flag)
{
return __get_data_block(inode, iblock, bh_result, create, flag);
}
static int get_data_block_dio(struct inode *inode, sector_t iblock,
struct buffer_head *bh_result, int create)
{
return __get_data_block(inode, iblock, bh_result, create,
F2FS_GET_BLOCK_DIO);
}
static int get_data_block_bmap(struct inode *inode, sector_t iblock,
struct buffer_head *bh_result, int create)
{
/* Block number less than F2FS MAX BLOCKS */
if (unlikely(iblock >= F2FS_I_SB(inode)->max_file_blocks))
return -EFBIG;
return __get_data_block(inode, iblock, bh_result, create,
F2FS_GET_BLOCK_BMAP);
}
static inline sector_t logical_to_blk(struct inode *inode, loff_t offset)
{
return (offset >> inode->i_blkbits);
}
static inline loff_t blk_to_logical(struct inode *inode, sector_t blk)
{
return (blk << inode->i_blkbits);
}
int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
u64 start, u64 len)
{
struct buffer_head map_bh;
sector_t start_blk, last_blk;
loff_t isize = i_size_read(inode);
u64 logical = 0, phys = 0, size = 0;
u32 flags = 0;
int ret = 0;
ret = fiemap_check_flags(fieinfo, FIEMAP_FLAG_SYNC);
if (ret)
return ret;
if (f2fs_has_inline_data(inode)) {
ret = f2fs_inline_data_fiemap(inode, fieinfo, start, len);
if (ret != -EAGAIN)
return ret;
}
mutex_lock(&inode->i_mutex);
if (start >= isize)
goto out;
if (start + len > isize)
len = isize - start;
if (logical_to_blk(inode, len) == 0)
len = blk_to_logical(inode, 1);
start_blk = logical_to_blk(inode, start);
last_blk = logical_to_blk(inode, start + len - 1);
next:
memset(&map_bh, 0, sizeof(struct buffer_head));
map_bh.b_size = len;
ret = get_data_block(inode, start_blk, &map_bh, 0,
F2FS_GET_BLOCK_FIEMAP);
if (ret)
goto out;
/* HOLE */
if (!buffer_mapped(&map_bh)) {
/* Go through holes util pass the EOF */
if (blk_to_logical(inode, start_blk++) < isize)
goto prep_next;
/* Found a hole beyond isize means no more extents.
* Note that the premise is that filesystems don't
* punch holes beyond isize and keep size unchanged.
*/
flags |= FIEMAP_EXTENT_LAST;
}
if (size)
ret = fiemap_fill_next_extent(fieinfo, logical,
phys, size, flags);
if (start_blk > last_blk || ret)
goto out;
logical = blk_to_logical(inode, start_blk);
phys = blk_to_logical(inode, map_bh.b_blocknr);
size = map_bh.b_size;
flags = 0;
if (buffer_unwritten(&map_bh))
flags = FIEMAP_EXTENT_UNWRITTEN;
start_blk += logical_to_blk(inode, size);
prep_next:
cond_resched();
if (fatal_signal_pending(current))
ret = -EINTR;
else
goto next;
out:
if (ret == 1)
ret = 0;
mutex_unlock(&inode->i_mutex);
return ret;
}
/*
* This function was originally taken from fs/mpage.c, and customized for f2fs.
* Major change was from block_size == page_size in f2fs by default.
*/
static int f2fs_mpage_readpages(struct address_space *mapping,
struct list_head *pages, struct page *page,
unsigned nr_pages)
{
struct bio *bio = NULL;
unsigned page_idx;
sector_t last_block_in_bio = 0;
struct inode *inode = mapping->host;
const unsigned blkbits = inode->i_blkbits;
const unsigned blocksize = 1 << blkbits;
sector_t block_in_file;
sector_t last_block;
sector_t last_block_in_file;
sector_t block_nr;
struct block_device *bdev = inode->i_sb->s_bdev;
struct f2fs_map_blocks map;
map.m_pblk = 0;
map.m_lblk = 0;
map.m_len = 0;
map.m_flags = 0;
for (page_idx = 0; nr_pages; page_idx++, nr_pages--) {
prefetchw(&page->flags);
if (pages) {
page = list_entry(pages->prev, struct page, lru);
list_del(&page->lru);
if (add_to_page_cache_lru(page, mapping,
page->index, GFP_KERNEL))
goto next_page;
}
block_in_file = (sector_t)page->index;
last_block = block_in_file + nr_pages;
last_block_in_file = (i_size_read(inode) + blocksize - 1) >>
blkbits;
if (last_block > last_block_in_file)
last_block = last_block_in_file;
/*
* Map blocks using the previous result first.
*/
if ((map.m_flags & F2FS_MAP_MAPPED) &&
block_in_file > map.m_lblk &&
block_in_file < (map.m_lblk + map.m_len))
goto got_it;
/*
* Then do more f2fs_map_blocks() calls until we are
* done with this page.
*/
map.m_flags = 0;
if (block_in_file < last_block) {
map.m_lblk = block_in_file;
map.m_len = last_block - block_in_file;
if (f2fs_map_blocks(inode, &map, 0,
F2FS_GET_BLOCK_READ))
goto set_error_page;
}
got_it:
if ((map.m_flags & F2FS_MAP_MAPPED)) {
block_nr = map.m_pblk + block_in_file - map.m_lblk;
SetPageMappedToDisk(page);
if (!PageUptodate(page) && !cleancache_get_page(page)) {
SetPageUptodate(page);
goto confused;
}
} else {
zero_user_segment(page, 0, PAGE_CACHE_SIZE);
SetPageUptodate(page);
unlock_page(page);
goto next_page;
}
/*
* This page will go to BIO. Do we need to send this
* BIO off first?
*/
if (bio && (last_block_in_bio != block_nr - 1)) {
submit_and_realloc:
submit_bio(READ, bio);
bio = NULL;
}
if (bio == NULL) {
struct f2fs_crypto_ctx *ctx = NULL;
if (f2fs_encrypted_inode(inode) &&
S_ISREG(inode->i_mode)) {
ctx = f2fs_get_crypto_ctx(inode);
if (IS_ERR(ctx))
goto set_error_page;
/* wait the page to be moved by cleaning */
f2fs crypto: fix racing of accessing encrypted page among different competitors Since we use different page cache (normally inode's page cache for R/W and meta inode's page cache for GC) to cache the same physical block which is belong to an encrypted inode. Writeback of these two page cache should be exclusive, but now we didn't handle writeback state well, so there may be potential racing problem: a) kworker: f2fs_gc: - f2fs_write_data_pages - f2fs_write_data_page - do_write_data_page - write_data_page - f2fs_submit_page_mbio (page#1 in inode's page cache was queued in f2fs bio cache, and be ready to write to new blkaddr) - gc_data_segment - move_encrypted_block - pagecache_get_page (page#2 in meta inode's page cache was cached with the invalid datas of physical block located in new blkaddr) - f2fs_submit_page_mbio (page#1 was submitted, later, page#2 with invalid data will be submitted) b) f2fs_gc: - gc_data_segment - move_encrypted_block - f2fs_submit_page_mbio (page#1 in meta inode's page cache was queued in f2fs bio cache, and be ready to write to new blkaddr) user thread: - f2fs_write_begin - f2fs_submit_page_bio (we submit the request to block layer to update page#2 in inode's page cache with physical block located in new blkaddr, so here we may read gabbage data from new blkaddr since GC hasn't writebacked the page#1 yet) This patch fixes above potential racing problem for encrypted inode. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2015-10-08 05:27:34 +00:00
f2fs_wait_on_encrypted_page_writeback(
F2FS_I_SB(inode), block_nr);
}
bio = bio_alloc(GFP_KERNEL,
min_t(int, nr_pages, BIO_MAX_PAGES));
if (!bio) {
if (ctx)
f2fs_release_crypto_ctx(ctx);
goto set_error_page;
}
bio->bi_bdev = bdev;
bio->bi_iter.bi_sector = SECTOR_FROM_BLOCK(block_nr);
bio->bi_end_io = f2fs_read_end_io;
bio->bi_private = ctx;
}
if (bio_add_page(bio, page, blocksize, 0) < blocksize)
goto submit_and_realloc;
last_block_in_bio = block_nr;
goto next_page;
set_error_page:
SetPageError(page);
zero_user_segment(page, 0, PAGE_CACHE_SIZE);
unlock_page(page);
goto next_page;
confused:
if (bio) {
submit_bio(READ, bio);
bio = NULL;
}
unlock_page(page);
next_page:
if (pages)
page_cache_release(page);
}
BUG_ON(pages && !list_empty(pages));
if (bio)
submit_bio(READ, bio);
return 0;
}
static int f2fs_read_data_page(struct file *file, struct page *page)
{
struct inode *inode = page->mapping->host;
int ret = -EAGAIN;
trace_f2fs_readpage(page, DATA);
/* If the file has inline data, try to read it directly */
if (f2fs_has_inline_data(inode))
ret = f2fs_read_inline_data(inode, page);
if (ret == -EAGAIN)
ret = f2fs_mpage_readpages(page->mapping, NULL, page, 1);
return ret;
}
static int f2fs_read_data_pages(struct file *file,
struct address_space *mapping,
struct list_head *pages, unsigned nr_pages)
{
struct inode *inode = file->f_mapping->host;
struct page *page = list_entry(pages->prev, struct page, lru);
trace_f2fs_readpages(inode, page, nr_pages);
/* If the file has inline data, skip readpages */
if (f2fs_has_inline_data(inode))
return 0;
return f2fs_mpage_readpages(mapping, pages, NULL, nr_pages);
}
int do_write_data_page(struct f2fs_io_info *fio)
{
struct page *page = fio->page;
struct inode *inode = page->mapping->host;
struct dnode_of_data dn;
int err = 0;
set_new_dnode(&dn, inode, NULL, NULL, 0);
err = get_dnode_of_data(&dn, page->index, LOOKUP_NODE);
if (err)
return err;
fio->blk_addr = dn.data_blkaddr;
/* This page is already truncated */
if (fio->blk_addr == NULL_ADDR) {
ClearPageUptodate(page);
goto out_writepage;
}
if (f2fs_encrypted_inode(inode) && S_ISREG(inode->i_mode)) {
f2fs crypto: fix racing of accessing encrypted page among different competitors Since we use different page cache (normally inode's page cache for R/W and meta inode's page cache for GC) to cache the same physical block which is belong to an encrypted inode. Writeback of these two page cache should be exclusive, but now we didn't handle writeback state well, so there may be potential racing problem: a) kworker: f2fs_gc: - f2fs_write_data_pages - f2fs_write_data_page - do_write_data_page - write_data_page - f2fs_submit_page_mbio (page#1 in inode's page cache was queued in f2fs bio cache, and be ready to write to new blkaddr) - gc_data_segment - move_encrypted_block - pagecache_get_page (page#2 in meta inode's page cache was cached with the invalid datas of physical block located in new blkaddr) - f2fs_submit_page_mbio (page#1 was submitted, later, page#2 with invalid data will be submitted) b) f2fs_gc: - gc_data_segment - move_encrypted_block - f2fs_submit_page_mbio (page#1 in meta inode's page cache was queued in f2fs bio cache, and be ready to write to new blkaddr) user thread: - f2fs_write_begin - f2fs_submit_page_bio (we submit the request to block layer to update page#2 in inode's page cache with physical block located in new blkaddr, so here we may read gabbage data from new blkaddr since GC hasn't writebacked the page#1 yet) This patch fixes above potential racing problem for encrypted inode. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2015-10-08 05:27:34 +00:00
/* wait for GCed encrypted page writeback */
f2fs_wait_on_encrypted_page_writeback(F2FS_I_SB(inode),
fio->blk_addr);
fio->encrypted_page = f2fs_encrypt(inode, fio->page);
if (IS_ERR(fio->encrypted_page)) {
err = PTR_ERR(fio->encrypted_page);
goto out_writepage;
}
}
set_page_writeback(page);
/*
* If current allocation needs SSR,
* it had better in-place writes for updated data.
*/
if (unlikely(fio->blk_addr != NEW_ADDR &&
!is_cold_data(page) &&
!IS_ATOMIC_WRITTEN_PAGE(page) &&
need_inplace_update(inode))) {
rewrite_data_page(fio);
set_inode_flag(F2FS_I(inode), FI_UPDATE_WRITE);
trace_f2fs_do_write_data_page(page, IPU);
} else {
write_data_page(&dn, fio);
set_data_blkaddr(&dn);
f2fs_update_extent_cache(&dn);
trace_f2fs_do_write_data_page(page, OPU);
set_inode_flag(F2FS_I(inode), FI_APPEND_WRITE);
if (page->index == 0)
set_inode_flag(F2FS_I(inode), FI_FIRST_BLOCK_WRITTEN);
}
out_writepage:
f2fs_put_dnode(&dn);
return err;
}
static int f2fs_write_data_page(struct page *page,
struct writeback_control *wbc)
{
struct inode *inode = page->mapping->host;
struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
loff_t i_size = i_size_read(inode);
const pgoff_t end_index = ((unsigned long long) i_size)
>> PAGE_CACHE_SHIFT;
unsigned offset = 0;
f2fs: introduce a new global lock scheme In the previous version, f2fs uses global locks according to the usage types, such as directory operations, block allocation, block write, and so on. Reference the following lock types in f2fs.h. enum lock_type { RENAME, /* for renaming operations */ DENTRY_OPS, /* for directory operations */ DATA_WRITE, /* for data write */ DATA_NEW, /* for data allocation */ DATA_TRUNC, /* for data truncate */ NODE_NEW, /* for node allocation */ NODE_TRUNC, /* for node truncate */ NODE_WRITE, /* for node write */ NR_LOCK_TYPE, }; In that case, we lose the performance under the multi-threading environment, since every types of operations must be conducted one at a time. In order to address the problem, let's share the locks globally with a mutex array regardless of any types. So, let users grab a mutex and perform their jobs in parallel as much as possbile. For this, I propose a new global lock scheme as follows. 0. Data structure - f2fs_sb_info -> mutex_lock[NR_GLOBAL_LOCKS] - f2fs_sb_info -> node_write 1. mutex_lock_op(sbi) - try to get an avaiable lock from the array. - returns the index of the gottern lock variable. 2. mutex_unlock_op(sbi, index of the lock) - unlock the given index of the lock. 3. mutex_lock_all(sbi) - grab all the locks in the array before the checkpoint. 4. mutex_unlock_all(sbi) - release all the locks in the array after checkpoint. 5. block_operations() - call mutex_lock_all() - sync_dirty_dir_inodes() - grab node_write - sync_node_pages() Note that, the pairs of mutex_lock_op()/mutex_unlock_op() and mutex_lock_all()/mutex_unlock_all() should be used together. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-22 07:21:29 +00:00
bool need_balance_fs = false;
int err = 0;
struct f2fs_io_info fio = {
.sbi = sbi,
.type = DATA,
.rw = (wbc->sync_mode == WB_SYNC_ALL) ? WRITE_SYNC : WRITE,
.page = page,
.encrypted_page = NULL,
};
trace_f2fs_writepage(page, DATA);
if (page->index < end_index)
f2fs: introduce a new global lock scheme In the previous version, f2fs uses global locks according to the usage types, such as directory operations, block allocation, block write, and so on. Reference the following lock types in f2fs.h. enum lock_type { RENAME, /* for renaming operations */ DENTRY_OPS, /* for directory operations */ DATA_WRITE, /* for data write */ DATA_NEW, /* for data allocation */ DATA_TRUNC, /* for data truncate */ NODE_NEW, /* for node allocation */ NODE_TRUNC, /* for node truncate */ NODE_WRITE, /* for node write */ NR_LOCK_TYPE, }; In that case, we lose the performance under the multi-threading environment, since every types of operations must be conducted one at a time. In order to address the problem, let's share the locks globally with a mutex array regardless of any types. So, let users grab a mutex and perform their jobs in parallel as much as possbile. For this, I propose a new global lock scheme as follows. 0. Data structure - f2fs_sb_info -> mutex_lock[NR_GLOBAL_LOCKS] - f2fs_sb_info -> node_write 1. mutex_lock_op(sbi) - try to get an avaiable lock from the array. - returns the index of the gottern lock variable. 2. mutex_unlock_op(sbi, index of the lock) - unlock the given index of the lock. 3. mutex_lock_all(sbi) - grab all the locks in the array before the checkpoint. 4. mutex_unlock_all(sbi) - release all the locks in the array after checkpoint. 5. block_operations() - call mutex_lock_all() - sync_dirty_dir_inodes() - grab node_write - sync_node_pages() Note that, the pairs of mutex_lock_op()/mutex_unlock_op() and mutex_lock_all()/mutex_unlock_all() should be used together. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-22 07:21:29 +00:00
goto write;
/*
* If the offset is out-of-range of file size,
* this page does not have to be written to disk.
*/
offset = i_size & (PAGE_CACHE_SIZE - 1);
if ((page->index >= end_index + 1) || !offset)
f2fs: introduce a new global lock scheme In the previous version, f2fs uses global locks according to the usage types, such as directory operations, block allocation, block write, and so on. Reference the following lock types in f2fs.h. enum lock_type { RENAME, /* for renaming operations */ DENTRY_OPS, /* for directory operations */ DATA_WRITE, /* for data write */ DATA_NEW, /* for data allocation */ DATA_TRUNC, /* for data truncate */ NODE_NEW, /* for node allocation */ NODE_TRUNC, /* for node truncate */ NODE_WRITE, /* for node write */ NR_LOCK_TYPE, }; In that case, we lose the performance under the multi-threading environment, since every types of operations must be conducted one at a time. In order to address the problem, let's share the locks globally with a mutex array regardless of any types. So, let users grab a mutex and perform their jobs in parallel as much as possbile. For this, I propose a new global lock scheme as follows. 0. Data structure - f2fs_sb_info -> mutex_lock[NR_GLOBAL_LOCKS] - f2fs_sb_info -> node_write 1. mutex_lock_op(sbi) - try to get an avaiable lock from the array. - returns the index of the gottern lock variable. 2. mutex_unlock_op(sbi, index of the lock) - unlock the given index of the lock. 3. mutex_lock_all(sbi) - grab all the locks in the array before the checkpoint. 4. mutex_unlock_all(sbi) - release all the locks in the array after checkpoint. 5. block_operations() - call mutex_lock_all() - sync_dirty_dir_inodes() - grab node_write - sync_node_pages() Note that, the pairs of mutex_lock_op()/mutex_unlock_op() and mutex_lock_all()/mutex_unlock_all() should be used together. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-22 07:21:29 +00:00
goto out;
zero_user_segment(page, offset, PAGE_CACHE_SIZE);
f2fs: introduce a new global lock scheme In the previous version, f2fs uses global locks according to the usage types, such as directory operations, block allocation, block write, and so on. Reference the following lock types in f2fs.h. enum lock_type { RENAME, /* for renaming operations */ DENTRY_OPS, /* for directory operations */ DATA_WRITE, /* for data write */ DATA_NEW, /* for data allocation */ DATA_TRUNC, /* for data truncate */ NODE_NEW, /* for node allocation */ NODE_TRUNC, /* for node truncate */ NODE_WRITE, /* for node write */ NR_LOCK_TYPE, }; In that case, we lose the performance under the multi-threading environment, since every types of operations must be conducted one at a time. In order to address the problem, let's share the locks globally with a mutex array regardless of any types. So, let users grab a mutex and perform their jobs in parallel as much as possbile. For this, I propose a new global lock scheme as follows. 0. Data structure - f2fs_sb_info -> mutex_lock[NR_GLOBAL_LOCKS] - f2fs_sb_info -> node_write 1. mutex_lock_op(sbi) - try to get an avaiable lock from the array. - returns the index of the gottern lock variable. 2. mutex_unlock_op(sbi, index of the lock) - unlock the given index of the lock. 3. mutex_lock_all(sbi) - grab all the locks in the array before the checkpoint. 4. mutex_unlock_all(sbi) - release all the locks in the array after checkpoint. 5. block_operations() - call mutex_lock_all() - sync_dirty_dir_inodes() - grab node_write - sync_node_pages() Note that, the pairs of mutex_lock_op()/mutex_unlock_op() and mutex_lock_all()/mutex_unlock_all() should be used together. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-22 07:21:29 +00:00
write:
if (unlikely(is_sbi_flag_set(sbi, SBI_POR_DOING)))
goto redirty_out;
if (f2fs_is_drop_cache(inode))
goto out;
if (f2fs_is_volatile_file(inode) && !wbc->for_reclaim &&
available_free_memory(sbi, BASE_CHECK))
goto redirty_out;
f2fs: introduce a new global lock scheme In the previous version, f2fs uses global locks according to the usage types, such as directory operations, block allocation, block write, and so on. Reference the following lock types in f2fs.h. enum lock_type { RENAME, /* for renaming operations */ DENTRY_OPS, /* for directory operations */ DATA_WRITE, /* for data write */ DATA_NEW, /* for data allocation */ DATA_TRUNC, /* for data truncate */ NODE_NEW, /* for node allocation */ NODE_TRUNC, /* for node truncate */ NODE_WRITE, /* for node write */ NR_LOCK_TYPE, }; In that case, we lose the performance under the multi-threading environment, since every types of operations must be conducted one at a time. In order to address the problem, let's share the locks globally with a mutex array regardless of any types. So, let users grab a mutex and perform their jobs in parallel as much as possbile. For this, I propose a new global lock scheme as follows. 0. Data structure - f2fs_sb_info -> mutex_lock[NR_GLOBAL_LOCKS] - f2fs_sb_info -> node_write 1. mutex_lock_op(sbi) - try to get an avaiable lock from the array. - returns the index of the gottern lock variable. 2. mutex_unlock_op(sbi, index of the lock) - unlock the given index of the lock. 3. mutex_lock_all(sbi) - grab all the locks in the array before the checkpoint. 4. mutex_unlock_all(sbi) - release all the locks in the array after checkpoint. 5. block_operations() - call mutex_lock_all() - sync_dirty_dir_inodes() - grab node_write - sync_node_pages() Note that, the pairs of mutex_lock_op()/mutex_unlock_op() and mutex_lock_all()/mutex_unlock_all() should be used together. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-22 07:21:29 +00:00
/* Dentry blocks are controlled by checkpoint */
if (S_ISDIR(inode->i_mode)) {
if (unlikely(f2fs_cp_error(sbi)))
goto redirty_out;
err = do_write_data_page(&fio);
goto done;
}
/* we should bypass data pages to proceed the kworkder jobs */
if (unlikely(f2fs_cp_error(sbi))) {
SetPageError(page);
goto out;
}
if (!wbc->for_reclaim)
f2fs: introduce a new global lock scheme In the previous version, f2fs uses global locks according to the usage types, such as directory operations, block allocation, block write, and so on. Reference the following lock types in f2fs.h. enum lock_type { RENAME, /* for renaming operations */ DENTRY_OPS, /* for directory operations */ DATA_WRITE, /* for data write */ DATA_NEW, /* for data allocation */ DATA_TRUNC, /* for data truncate */ NODE_NEW, /* for node allocation */ NODE_TRUNC, /* for node truncate */ NODE_WRITE, /* for node write */ NR_LOCK_TYPE, }; In that case, we lose the performance under the multi-threading environment, since every types of operations must be conducted one at a time. In order to address the problem, let's share the locks globally with a mutex array regardless of any types. So, let users grab a mutex and perform their jobs in parallel as much as possbile. For this, I propose a new global lock scheme as follows. 0. Data structure - f2fs_sb_info -> mutex_lock[NR_GLOBAL_LOCKS] - f2fs_sb_info -> node_write 1. mutex_lock_op(sbi) - try to get an avaiable lock from the array. - returns the index of the gottern lock variable. 2. mutex_unlock_op(sbi, index of the lock) - unlock the given index of the lock. 3. mutex_lock_all(sbi) - grab all the locks in the array before the checkpoint. 4. mutex_unlock_all(sbi) - release all the locks in the array after checkpoint. 5. block_operations() - call mutex_lock_all() - sync_dirty_dir_inodes() - grab node_write - sync_node_pages() Note that, the pairs of mutex_lock_op()/mutex_unlock_op() and mutex_lock_all()/mutex_unlock_all() should be used together. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-22 07:21:29 +00:00
need_balance_fs = true;
else if (has_not_enough_free_secs(sbi, 0))
f2fs: introduce a new global lock scheme In the previous version, f2fs uses global locks according to the usage types, such as directory operations, block allocation, block write, and so on. Reference the following lock types in f2fs.h. enum lock_type { RENAME, /* for renaming operations */ DENTRY_OPS, /* for directory operations */ DATA_WRITE, /* for data write */ DATA_NEW, /* for data allocation */ DATA_TRUNC, /* for data truncate */ NODE_NEW, /* for node allocation */ NODE_TRUNC, /* for node truncate */ NODE_WRITE, /* for node write */ NR_LOCK_TYPE, }; In that case, we lose the performance under the multi-threading environment, since every types of operations must be conducted one at a time. In order to address the problem, let's share the locks globally with a mutex array regardless of any types. So, let users grab a mutex and perform their jobs in parallel as much as possbile. For this, I propose a new global lock scheme as follows. 0. Data structure - f2fs_sb_info -> mutex_lock[NR_GLOBAL_LOCKS] - f2fs_sb_info -> node_write 1. mutex_lock_op(sbi) - try to get an avaiable lock from the array. - returns the index of the gottern lock variable. 2. mutex_unlock_op(sbi, index of the lock) - unlock the given index of the lock. 3. mutex_lock_all(sbi) - grab all the locks in the array before the checkpoint. 4. mutex_unlock_all(sbi) - release all the locks in the array after checkpoint. 5. block_operations() - call mutex_lock_all() - sync_dirty_dir_inodes() - grab node_write - sync_node_pages() Note that, the pairs of mutex_lock_op()/mutex_unlock_op() and mutex_lock_all()/mutex_unlock_all() should be used together. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-22 07:21:29 +00:00
goto redirty_out;
err = -EAGAIN;
f2fs_lock_op(sbi);
if (f2fs_has_inline_data(inode))
err = f2fs_write_inline_data(inode, page);
if (err == -EAGAIN)
err = do_write_data_page(&fio);
f2fs_unlock_op(sbi);
done:
if (err && err != -ENOENT)
goto redirty_out;
clear_cold_data(page);
f2fs: introduce a new global lock scheme In the previous version, f2fs uses global locks according to the usage types, such as directory operations, block allocation, block write, and so on. Reference the following lock types in f2fs.h. enum lock_type { RENAME, /* for renaming operations */ DENTRY_OPS, /* for directory operations */ DATA_WRITE, /* for data write */ DATA_NEW, /* for data allocation */ DATA_TRUNC, /* for data truncate */ NODE_NEW, /* for node allocation */ NODE_TRUNC, /* for node truncate */ NODE_WRITE, /* for node write */ NR_LOCK_TYPE, }; In that case, we lose the performance under the multi-threading environment, since every types of operations must be conducted one at a time. In order to address the problem, let's share the locks globally with a mutex array regardless of any types. So, let users grab a mutex and perform their jobs in parallel as much as possbile. For this, I propose a new global lock scheme as follows. 0. Data structure - f2fs_sb_info -> mutex_lock[NR_GLOBAL_LOCKS] - f2fs_sb_info -> node_write 1. mutex_lock_op(sbi) - try to get an avaiable lock from the array. - returns the index of the gottern lock variable. 2. mutex_unlock_op(sbi, index of the lock) - unlock the given index of the lock. 3. mutex_lock_all(sbi) - grab all the locks in the array before the checkpoint. 4. mutex_unlock_all(sbi) - release all the locks in the array after checkpoint. 5. block_operations() - call mutex_lock_all() - sync_dirty_dir_inodes() - grab node_write - sync_node_pages() Note that, the pairs of mutex_lock_op()/mutex_unlock_op() and mutex_lock_all()/mutex_unlock_all() should be used together. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-22 07:21:29 +00:00
out:
inode_dec_dirty_pages(inode);
if (err)
ClearPageUptodate(page);
unlock_page(page);
f2fs: introduce a new global lock scheme In the previous version, f2fs uses global locks according to the usage types, such as directory operations, block allocation, block write, and so on. Reference the following lock types in f2fs.h. enum lock_type { RENAME, /* for renaming operations */ DENTRY_OPS, /* for directory operations */ DATA_WRITE, /* for data write */ DATA_NEW, /* for data allocation */ DATA_TRUNC, /* for data truncate */ NODE_NEW, /* for node allocation */ NODE_TRUNC, /* for node truncate */ NODE_WRITE, /* for node write */ NR_LOCK_TYPE, }; In that case, we lose the performance under the multi-threading environment, since every types of operations must be conducted one at a time. In order to address the problem, let's share the locks globally with a mutex array regardless of any types. So, let users grab a mutex and perform their jobs in parallel as much as possbile. For this, I propose a new global lock scheme as follows. 0. Data structure - f2fs_sb_info -> mutex_lock[NR_GLOBAL_LOCKS] - f2fs_sb_info -> node_write 1. mutex_lock_op(sbi) - try to get an avaiable lock from the array. - returns the index of the gottern lock variable. 2. mutex_unlock_op(sbi, index of the lock) - unlock the given index of the lock. 3. mutex_lock_all(sbi) - grab all the locks in the array before the checkpoint. 4. mutex_unlock_all(sbi) - release all the locks in the array after checkpoint. 5. block_operations() - call mutex_lock_all() - sync_dirty_dir_inodes() - grab node_write - sync_node_pages() Note that, the pairs of mutex_lock_op()/mutex_unlock_op() and mutex_lock_all()/mutex_unlock_all() should be used together. Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
2012-11-22 07:21:29 +00:00
if (need_balance_fs)
f2fs_balance_fs(sbi);
f2fs: write pending bios when cp_error is set When testing ioc_shutdown, put_super is able to be hanged by waiting for writebacking pages as follows. INFO: task umount:2723 blocked for more than 120 seconds. Tainted: G O 4.4.0-rc3+ #8 "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. umount D ffff88000859f9d8 0 2723 2110 0x00000000 ffff88000859f9d8 0000000000000000 0000000000000000 ffffffff81e11540 ffff880078c225c0 ffff8800085a0000 ffff88007fc17440 7fffffffffffffff ffffffff818239f0 ffff88000859fb48 ffff88000859f9f0 ffffffff8182310c Call Trace: [<ffffffff818239f0>] ? bit_wait+0x50/0x50 [<ffffffff8182310c>] schedule+0x3c/0x90 [<ffffffff81827fb9>] schedule_timeout+0x2d9/0x430 [<ffffffff810e0f8f>] ? mark_held_locks+0x6f/0xa0 [<ffffffff8111614d>] ? ktime_get+0x7d/0x140 [<ffffffff818239f0>] ? bit_wait+0x50/0x50 [<ffffffff8106a655>] ? kvm_clock_get_cycles+0x25/0x30 [<ffffffff8111617c>] ? ktime_get+0xac/0x140 [<ffffffff818239f0>] ? bit_wait+0x50/0x50 [<ffffffff81822564>] io_schedule_timeout+0xa4/0x110 [<ffffffff81823a25>] bit_wait_io+0x35/0x50 [<ffffffff818235bd>] __wait_on_bit+0x5d/0x90 [<ffffffff811b9e8b>] wait_on_page_bit+0xcb/0xf0 [<ffffffff810d5f90>] ? autoremove_wake_function+0x40/0x40 [<ffffffff811cf84c>] truncate_inode_pages_range+0x4bc/0x840 [<ffffffff811cfc3d>] truncate_inode_pages_final+0x4d/0x60 [<ffffffffc023ced5>] f2fs_evict_inode+0x75/0x400 [f2fs] [<ffffffff812639bc>] evict+0xbc/0x190 [<ffffffff81263d19>] iput+0x229/0x2c0 [<ffffffffc0241885>] f2fs_put_super+0x105/0x1a0 [f2fs] [<ffffffff8124756a>] generic_shutdown_super+0x6a/0xf0 [<ffffffff812478f7>] kill_block_super+0x27/0x70 [<ffffffffc0241290>] kill_f2fs_super+0x20/0x30 [f2fs] [<ffffffff81247b03>] deactivate_locked_super+0x43/0x70 [<ffffffff81247f4c>] deactivate_super+0x5c/0x60 [<ffffffff81268d2f>] cleanup_mnt+0x3f/0x90 [<ffffffff81268dc2>] __cleanup_mnt+0x12/0x20 [<ffffffff810ac463>] task_work_run+0x73/0xa0 [<ffffffff810032ac>] exit_to_usermode_loop+0xcc/0xd0 [<ffffffff81003e7c>] syscall_return_slowpath+0xcc/0xe0 [<ffffffff81829ea2>] int_ret_from_sys_call+0x25/0x9f Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2015-12-31 21:08:02 +00:00
if (wbc->for_reclaim || unlikely(f2fs_cp_error(sbi))) {
f2fs_submit_merged_bio(sbi, DATA, WRITE);
remove_dirty_inode(inode);
}
return 0;
redirty_out:
redirty_page_for_writepage(wbc, page);
return AOP_WRITEPAGE_ACTIVATE;
}
static int __f2fs_writepage(struct page *page, struct writeback_control *wbc,
void *data)
{
struct address_space *mapping = data;
int ret = mapping->a_ops->writepage(page, wbc);
mapping_set_error(mapping, ret);
return ret;
}
/*
* This function was copied from write_cche_pages from mm/page-writeback.c.
* The major change is making write step of cold data page separately from
* warm/hot data page.
*/
static int f2fs_write_cache_pages(struct address_space *mapping,
struct writeback_control *wbc, writepage_t writepage,
void *data)
{
int ret = 0;
int done = 0;
struct pagevec pvec;
int nr_pages;
pgoff_t uninitialized_var(writeback_index);
pgoff_t index;
pgoff_t end; /* Inclusive */
pgoff_t done_index;
int cycled;
int range_whole = 0;
int tag;
int step = 0;
pagevec_init(&pvec, 0);
next:
if (wbc->range_cyclic) {
writeback_index = mapping->writeback_index; /* prev offset */
index = writeback_index;
if (index == 0)
cycled = 1;
else
cycled = 0;
end = -1;
} else {
index = wbc->range_start >> PAGE_CACHE_SHIFT;
end = wbc->range_end >> PAGE_CACHE_SHIFT;
if (wbc->range_start == 0 && wbc->range_end == LLONG_MAX)
range_whole = 1;
cycled = 1; /* ignore range_cyclic tests */
}
if (wbc->sync_mode == WB_SYNC_ALL || wbc->tagged_writepages)
tag = PAGECACHE_TAG_TOWRITE;
else
tag = PAGECACHE_TAG_DIRTY;
retry:
if (wbc->sync_mode == WB_SYNC_ALL || wbc->tagged_writepages)
tag_pages_for_writeback(mapping, index, end);
done_index = index;
while (!done && (index <= end)) {
int i;
nr_pages = pagevec_lookup_tag(&pvec, mapping, &index, tag,
min(end - index, (pgoff_t)PAGEVEC_SIZE - 1) + 1);
if (nr_pages == 0)
break;
for (i = 0; i < nr_pages; i++) {
struct page *page = pvec.pages[i];
if (page->index > end) {
done = 1;
break;
}
done_index = page->index;
lock_page(page);
if (unlikely(page->mapping != mapping)) {
continue_unlock:
unlock_page(page);
continue;
}
if (!PageDirty(page)) {
/* someone wrote it for us */
goto continue_unlock;
}
if (step == is_cold_data(page))
goto continue_unlock;
if (PageWriteback(page)) {
if (wbc->sync_mode != WB_SYNC_NONE)
f2fs_wait_on_page_writeback(page, DATA);
else
goto continue_unlock;
}
BUG_ON(PageWriteback(page));
if (!clear_page_dirty_for_io(page))
goto continue_unlock;
ret = (*writepage)(page, wbc, data);
if (unlikely(ret)) {
if (ret == AOP_WRITEPAGE_ACTIVATE) {
unlock_page(page);
ret = 0;
} else {
done_index = page->index + 1;
done = 1;
break;
}
}
if (--wbc->nr_to_write <= 0 &&
wbc->sync_mode == WB_SYNC_NONE) {
done = 1;
break;
}
}
pagevec_release(&pvec);
cond_resched();
}
if (step < 1) {
step++;
goto next;
}
if (!cycled && !done) {
cycled = 1;
index = 0;
end = writeback_index - 1;
goto retry;
}
if (wbc->range_cyclic || (range_whole && wbc->nr_to_write > 0))
mapping->writeback_index = done_index;
return ret;
}
static int f2fs_write_data_pages(struct address_space *mapping,
struct writeback_control *wbc)
{
struct inode *inode = mapping->host;
struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
bool locked = false;
int ret;
long diff;
trace_f2fs_writepages(mapping->host, wbc, DATA);
/* deal with chardevs and other special file */
if (!mapping->a_ops->writepage)
return 0;
/* skip writing if there is no dirty page in this inode */
if (!get_dirty_pages(inode) && wbc->sync_mode == WB_SYNC_NONE)
return 0;
if (S_ISDIR(inode->i_mode) && wbc->sync_mode == WB_SYNC_NONE &&
get_dirty_pages(inode) < nr_pages_to_skip(sbi, DATA) &&
available_free_memory(sbi, DIRTY_DENTS))
goto skip_write;
/* skip writing during file defragment */
if (is_inode_flag_set(F2FS_I(inode), FI_DO_DEFRAG))
goto skip_write;
f2fs: avoid to trigger writepage during POR This patch doesn't make any effect on previous behavior, since f2fs_write_data_page bypasses writing the page during POR. But, the difference is that this patch avoids holding writepages mutex. This is to avoid the following false warning, since this can happen only when mount and shutdown are triggered at the same time. ====================================================== [ INFO: possible circular locking dependency detected ] 4.0.0-rc1+ #3 Tainted: G O ------------------------------------------------------- kworker/u8:0/2270 is trying to acquire lock: (&sbi->gc_mutex){+.+.+.}, at: [<ffffffffa02bdd33>] f2fs_balance_fs+0x73/0x90 [f2fs] but task is already holding lock: (&sbi->writepages){+.+...}, at: [<ffffffffa02b261b>] f2fs_write_data_pages+0xcb/0x3a0 [f2fs] which lock already depends on the new lock. the existing dependency chain (in reverse order) is: -> #2 (&sbi->writepages){+.+...}: [<ffffffff810e2b11>] lock_acquire+0xe1/0x2f0 [<ffffffff8185e1b3>] mutex_lock_nested+0x63/0x530 [<ffffffffa02b261b>] f2fs_write_data_pages+0xcb/0x3a0 [f2fs] [<ffffffff811c38c1>] do_writepages+0x21/0x50 [<ffffffff8126c5a6>] __writeback_single_inode+0x76/0xbf0 [<ffffffff8126e23a>] writeback_single_inode+0xea/0x1c0 [<ffffffff8126e425>] write_inode_now+0x95/0xa0 [<ffffffff81259dab>] iput+0x20b/0x3f0 [<ffffffffa02c1c8b>] recover_data.constprop.14+0x26b/0xa80 [f2fs] [<ffffffffa02c2776>] recover_fsync_data+0x2b6/0x5e0 [f2fs] [<ffffffffa02a9744>] f2fs_fill_super+0xb24/0xb90 [f2fs] [<ffffffff8123d7f4>] mount_bdev+0x1a4/0x1e0 [<ffffffffa02a3c85>] f2fs_mount+0x15/0x20 [f2fs] [<ffffffff8123e159>] mount_fs+0x39/0x180 [<ffffffff8125e51b>] vfs_kern_mount+0x6b/0x160 [<ffffffff81261554>] do_mount+0x204/0xbe0 [<ffffffff8126223b>] SyS_mount+0x8b/0xe0 [<ffffffff81863e6d>] system_call_fastpath+0x16/0x1b -> #1 (&sbi->cp_mutex){+.+...}: [<ffffffff810e2b11>] lock_acquire+0xe1/0x2f0 [<ffffffff8185e1b3>] mutex_lock_nested+0x63/0x530 [<ffffffffa02acbf2>] write_checkpoint+0x42/0x1230 [f2fs] [<ffffffffa02a847d>] f2fs_sync_fs+0x9d/0x2a0 [f2fs] [<ffffffff81272f82>] sync_filesystem+0x82/0xb0 [<ffffffff8123c214>] generic_shutdown_super+0x34/0x100 [<ffffffff8123c5f7>] kill_block_super+0x27/0x70 [<ffffffffa02a3c60>] kill_f2fs_super+0x20/0x30 [f2fs] [<ffffffff8123ca49>] deactivate_locked_super+0x49/0x80 [<ffffffff8123d05e>] deactivate_super+0x4e/0x70 [<ffffffff8125df63>] cleanup_mnt+0x43/0x90 [<ffffffff8125e002>] __cleanup_mnt+0x12/0x20 [<ffffffff810a82e4>] task_work_run+0xc4/0xf0 [<ffffffff8101f0bd>] do_notify_resume+0x8d/0xa0 [<ffffffff81864141>] int_signal+0x12/0x17 -> #0 (&sbi->gc_mutex){+.+.+.}: [<ffffffff810e2866>] __lock_acquire+0x1ac6/0x1c90 [<ffffffff810e2b11>] lock_acquire+0xe1/0x2f0 [<ffffffff8185e1b3>] mutex_lock_nested+0x63/0x530 [<ffffffffa02bdd33>] f2fs_balance_fs+0x73/0x90 [f2fs] [<ffffffffa02b5938>] f2fs_write_data_page+0x348/0x5b0 [f2fs] [<ffffffffa02af9da>] __f2fs_writepage+0x1a/0x50 [f2fs] [<ffffffff811c1b54>] write_cache_pages+0x274/0x6f0 [<ffffffffa02b2630>] f2fs_write_data_pages+0xe0/0x3a0 [f2fs] [<ffffffff811c38c1>] do_writepages+0x21/0x50 [<ffffffff8126c5a6>] __writeback_single_inode+0x76/0xbf0 [<ffffffff8126d44a>] writeback_sb_inodes+0x32a/0x710 [<ffffffff8126d8cf>] __writeback_inodes_wb+0x9f/0xd0 [<ffffffff8126dcdb>] wb_writeback+0x3db/0x850 [<ffffffff8126e848>] bdi_writeback_workfn+0x148/0x980 [<ffffffff810a3782>] process_one_work+0x1e2/0x840 [<ffffffff810a3f01>] worker_thread+0x121/0x460 [<ffffffff810a9dc8>] kthread+0xf8/0x110 [<ffffffff81863dbc>] ret_from_fork+0x7c/0xb0 Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2015-02-27 21:37:39 +00:00
/* during POR, we don't need to trigger writepage at all. */
if (unlikely(is_sbi_flag_set(sbi, SBI_POR_DOING)))
goto skip_write;
diff = nr_pages_to_write(sbi, DATA, wbc);
if (!S_ISDIR(inode->i_mode)) {
mutex_lock(&sbi->writepages);
locked = true;
}
ret = f2fs_write_cache_pages(mapping, wbc, __f2fs_writepage, mapping);
f2fs: enhance multithread performance In ->writepages, we use writepages mutex lock to serialize all block address allocation and page submitting pairs from different inodes. This method makes our delayed dirty pages of one inode being written continously as many as possible. But there is one problem that we did not submit current cached bio in protection region of writepages mutex lock, so there is a small chance that we submit the one of other thread's as below, resulting in splitting more bios. thread 1 thread 2 ->writepages lock(writepages) ->write_cache_pages unlock(writepages) lock(writepages) ->write_cache_pages ->f2fs_submit_merged_bio ->writepage unlock(writepages) fs_mark-6535 [002] .... 2242.270230: f2fs_submit_write_bio: dev = (1,0), WRITE_SYNC, DATA, sector = 5766152, size = 524288 fs_mark-6536 [000] .... 2242.270361: f2fs_submit_write_bio: dev = (1,0), WRITE_SYNC, DATA, sector = 5767176, size = 4096 fs_mark-6536 [000] .... 2242.270370: f2fs_submit_write_bio: dev = (1,0), WRITE_SYNC, NODE, sector = 8138112, size = 4096 fs_mark-6535 [002] .... 2242.270776: f2fs_submit_write_bio: dev = (1,0), WRITE_SYNC, DATA, sector = 5767184, size = 516096 This may really increase time of block layer works, and may cause larger IO lantency. This patch moves the submitting operation into region of writepages mutex lock to avoid bio splits when concurrently writebacking is intensive. my test environment: virtual machine, intel cpu i5 2500, 8GB size memory, 4GB size ramdisk time fs_mark -t 16 -L 1 -s 524288 -S 1 -d /mnt/f2fs/ before: real 0m4.244s user 0m0.088s sys 0m12.336s after: real 0m3.822s user 0m0.072s sys 0m10.760s Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2015-07-06 12:31:49 +00:00
f2fs_submit_merged_bio(sbi, DATA, WRITE);
if (locked)
mutex_unlock(&sbi->writepages);
remove_dirty_inode(inode);
wbc->nr_to_write = max((long)0, wbc->nr_to_write - diff);
return ret;
skip_write:
wbc->pages_skipped += get_dirty_pages(inode);
return 0;
}
static void f2fs_write_failed(struct address_space *mapping, loff_t to)
{
struct inode *inode = mapping->host;
loff_t i_size = i_size_read(inode);
if (to > i_size) {
truncate_pagecache(inode, i_size);
truncate_blocks(inode, i_size, true);
}
}
static int prepare_write_begin(struct f2fs_sb_info *sbi,
struct page *page, loff_t pos, unsigned len,
block_t *blk_addr, bool *node_changed)
{
struct inode *inode = page->mapping->host;
pgoff_t index = page->index;
struct dnode_of_data dn;
struct page *ipage;
bool locked = false;
struct extent_info ei;
int err = 0;
if (f2fs_has_inline_data(inode) ||
(pos & PAGE_CACHE_MASK) >= i_size_read(inode)) {
f2fs_lock_op(sbi);
locked = true;
}
restart:
/* check inline_data */
ipage = get_node_page(sbi, inode->i_ino);
if (IS_ERR(ipage)) {
err = PTR_ERR(ipage);
goto unlock_out;
}
set_new_dnode(&dn, inode, ipage, ipage, 0);
if (f2fs_has_inline_data(inode)) {
if (pos + len <= MAX_INLINE_DATA) {
read_inline_data(page, ipage);
set_inode_flag(F2FS_I(inode), FI_DATA_EXIST);
sync_inode_page(&dn);
} else {
err = f2fs_convert_inline_page(&dn, page);
if (err)
goto out;
if (dn.data_blkaddr == NULL_ADDR)
err = f2fs_get_block(&dn, index);
}
} else if (locked) {
err = f2fs_get_block(&dn, index);
} else {
if (f2fs_lookup_extent_cache(inode, index, &ei)) {
dn.data_blkaddr = ei.blk + index - ei.fofs;
} else {
bool restart = false;
/* hole case */
err = get_dnode_of_data(&dn, index, LOOKUP_NODE);
if (err || (!err && dn.data_blkaddr == NULL_ADDR))
restart = true;
if (restart) {
f2fs_put_dnode(&dn);
f2fs_lock_op(sbi);
locked = true;
goto restart;
}
}
}
/* convert_inline_page can make node_changed */
*blk_addr = dn.data_blkaddr;
*node_changed = dn.node_changed;
out:
f2fs_put_dnode(&dn);
unlock_out:
if (locked)
f2fs_unlock_op(sbi);
return err;
}
static int f2fs_write_begin(struct file *file, struct address_space *mapping,
loff_t pos, unsigned len, unsigned flags,
struct page **pagep, void **fsdata)
{
struct inode *inode = mapping->host;
struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
struct page *page = NULL;
pgoff_t index = ((unsigned long long) pos) >> PAGE_CACHE_SHIFT;
bool need_balance = false;
block_t blkaddr = NULL_ADDR;
int err = 0;
trace_f2fs_write_begin(inode, pos, len, flags);
/*
* We should check this at this moment to avoid deadlock on inode page
* and #0 page. The locking rule for inline_data conversion should be:
* lock_page(page #0) -> lock_page(inode_page)
*/
if (index != 0) {
err = f2fs_convert_inline_inode(inode);
if (err)
goto fail;
}
repeat:
page = grab_cache_page_write_begin(mapping, index, flags);
if (!page) {
err = -ENOMEM;
goto fail;
}
*pagep = page;
err = prepare_write_begin(sbi, page, pos, len,
&blkaddr, &need_balance);
if (err)
goto fail;
if (need_balance && has_not_enough_free_secs(sbi, 0)) {
unlock_page(page);
f2fs_balance_fs(sbi);
lock_page(page);
if (page->mapping != mapping) {
/* The page got truncated from under us */
f2fs_put_page(page, 1);
goto repeat;
}
}
f2fs_wait_on_page_writeback(page, DATA);
f2fs crypto: fix racing of accessing encrypted page among different competitors Since we use different page cache (normally inode's page cache for R/W and meta inode's page cache for GC) to cache the same physical block which is belong to an encrypted inode. Writeback of these two page cache should be exclusive, but now we didn't handle writeback state well, so there may be potential racing problem: a) kworker: f2fs_gc: - f2fs_write_data_pages - f2fs_write_data_page - do_write_data_page - write_data_page - f2fs_submit_page_mbio (page#1 in inode's page cache was queued in f2fs bio cache, and be ready to write to new blkaddr) - gc_data_segment - move_encrypted_block - pagecache_get_page (page#2 in meta inode's page cache was cached with the invalid datas of physical block located in new blkaddr) - f2fs_submit_page_mbio (page#1 was submitted, later, page#2 with invalid data will be submitted) b) f2fs_gc: - gc_data_segment - move_encrypted_block - f2fs_submit_page_mbio (page#1 in meta inode's page cache was queued in f2fs bio cache, and be ready to write to new blkaddr) user thread: - f2fs_write_begin - f2fs_submit_page_bio (we submit the request to block layer to update page#2 in inode's page cache with physical block located in new blkaddr, so here we may read gabbage data from new blkaddr since GC hasn't writebacked the page#1 yet) This patch fixes above potential racing problem for encrypted inode. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2015-10-08 05:27:34 +00:00
/* wait for GCed encrypted page writeback */
if (f2fs_encrypted_inode(inode) && S_ISREG(inode->i_mode))
f2fs_wait_on_encrypted_page_writeback(sbi, blkaddr);
f2fs crypto: fix racing of accessing encrypted page among different competitors Since we use different page cache (normally inode's page cache for R/W and meta inode's page cache for GC) to cache the same physical block which is belong to an encrypted inode. Writeback of these two page cache should be exclusive, but now we didn't handle writeback state well, so there may be potential racing problem: a) kworker: f2fs_gc: - f2fs_write_data_pages - f2fs_write_data_page - do_write_data_page - write_data_page - f2fs_submit_page_mbio (page#1 in inode's page cache was queued in f2fs bio cache, and be ready to write to new blkaddr) - gc_data_segment - move_encrypted_block - pagecache_get_page (page#2 in meta inode's page cache was cached with the invalid datas of physical block located in new blkaddr) - f2fs_submit_page_mbio (page#1 was submitted, later, page#2 with invalid data will be submitted) b) f2fs_gc: - gc_data_segment - move_encrypted_block - f2fs_submit_page_mbio (page#1 in meta inode's page cache was queued in f2fs bio cache, and be ready to write to new blkaddr) user thread: - f2fs_write_begin - f2fs_submit_page_bio (we submit the request to block layer to update page#2 in inode's page cache with physical block located in new blkaddr, so here we may read gabbage data from new blkaddr since GC hasn't writebacked the page#1 yet) This patch fixes above potential racing problem for encrypted inode. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2015-10-08 05:27:34 +00:00
if (len == PAGE_CACHE_SIZE)
goto out_update;
if (PageUptodate(page))
goto out_clear;
if ((pos & PAGE_CACHE_MASK) >= i_size_read(inode)) {
unsigned start = pos & (PAGE_CACHE_SIZE - 1);
unsigned end = start + len;
/* Reading beyond i_size is simple: memset to zero */
zero_user_segments(page, 0, start, end, PAGE_CACHE_SIZE);
goto out_update;
}
if (blkaddr == NEW_ADDR) {
zero_user_segment(page, 0, PAGE_CACHE_SIZE);
} else {
struct f2fs_io_info fio = {
.sbi = sbi,
.type = DATA,
.rw = READ_SYNC,
.blk_addr = blkaddr,
.page = page,
.encrypted_page = NULL,
};
err = f2fs_submit_page_bio(&fio);
if (err)
goto fail;
lock_page(page);
if (unlikely(!PageUptodate(page))) {
err = -EIO;
goto fail;
}
if (unlikely(page->mapping != mapping)) {
f2fs_put_page(page, 1);
goto repeat;
}
/* avoid symlink page */
if (f2fs_encrypted_inode(inode) && S_ISREG(inode->i_mode)) {
err = f2fs_decrypt_one(inode, page);
if (err)
goto fail;
}
}
out_update:
SetPageUptodate(page);
out_clear:
clear_cold_data(page);
return 0;
fail:
f2fs_put_page(page, 1);
f2fs_write_failed(mapping, pos + len);
return err;
}
static int f2fs_write_end(struct file *file,
struct address_space *mapping,
loff_t pos, unsigned len, unsigned copied,
struct page *page, void *fsdata)
{
struct inode *inode = page->mapping->host;
trace_f2fs_write_end(inode, pos, len, copied);
set_page_dirty(page);
if (pos + copied > i_size_read(inode)) {
i_size_write(inode, pos + copied);
mark_inode_dirty(inode);
update_inode_page(inode);
}
f2fs_put_page(page, 1);
return copied;
}
static int check_direct_IO(struct inode *inode, struct iov_iter *iter,
loff_t offset)
{
unsigned blocksize_mask = inode->i_sb->s_blocksize - 1;
if (offset & blocksize_mask)
return -EINVAL;
if (iov_iter_alignment(iter) & blocksize_mask)
return -EINVAL;
return 0;
}
static ssize_t f2fs_direct_IO(struct kiocb *iocb, struct iov_iter *iter,
loff_t offset)
{
struct file *file = iocb->ki_filp;
struct address_space *mapping = file->f_mapping;
struct inode *inode = mapping->host;
size_t count = iov_iter_count(iter);
int err;
/* we don't need to use inline_data strictly */
err = f2fs_convert_inline_inode(inode);
if (err)
return err;
if (f2fs_encrypted_inode(inode) && S_ISREG(inode->i_mode))
return 0;
f2fs: report EINVAL for unalignment direct IO We run ltp testcase with f2fs and obtain a TFAIL in diotest4, the result in detail is as fallow: dio04 <<<test_start>>> tag=dio04 stime=1432278894 cmdline="diotest4" contacts="" analysis=exit <<<test_output>>> diotest4 1 TPASS : Negative Offset diotest4 2 TPASS : removed diotest4 3 TFAIL : diotest4.c:129: write allows odd count.returns 1: Success diotest4 4 TFAIL : diotest4.c:183: Odd count of read and write diotest4 5 TPASS : Read beyond the file size ...... the result of ext4 with same environment: dio04 <<<test_start>>> tag=dio04 stime=1432259643 cmdline="diotest4" contacts="" analysis=exit <<<test_output>>> diotest4 1 TPASS : Negative Offset diotest4 2 TPASS : removed diotest4 3 TPASS : Odd count of read and write diotest4 4 TPASS : Read beyond the file size ...... The reason is that when triggering DIO in f2fs, we will return zero value in ->direct_IO if writer's buffer offset, file offset and transfer size is not alignment to block size of filesystem, resulting in falling back into buffered write instead of returning -EINVAL. This patch fixes that problem by returning correct error number for above case, and removing the judgement condition in check_direct_IO to make sure the verification will be enabled for direct reader too. Besides, Jaegeuk Kim pointed out that there is expectional cases we should always make direct-io falling back into buffered write, such as dio in encrypted file. Signed-off-by: Yunlei He <heyunlei@huawei.com> [Chao Yu make small change and add detail description in commit message] Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2015-08-07 10:39:32 +00:00
err = check_direct_IO(inode, iter, offset);
if (err)
return err;
trace_f2fs_direct_IO_enter(inode, offset, count, iov_iter_rw(iter));
f2fs: fix to handle io error in ->direct_IO Here is a oops reported as following message when testing generic/019 of xfstest: ------------[ cut here ]------------ kernel BUG at /home/yuchao/git/f2fs-dev/segment.c:882! invalid opcode: 0000 [#1] SMP Modules linked in: zram lz4_compress lz4_decompress f2fs(O) ip6table_filter ip6_tables ebtable_nat ebtables nf_conntrack_ipv4 nf_def CPU: 2 PID: 25441 Comm: fio Tainted: G O 4.3.0-rc1+ #6 Hardware name: Hewlett-Packard HP Z220 CMT Workstation/1790, BIOS K51 v01.61 05/16/2013 task: ffff8803f4e85580 ti: ffff8803fd61c000 task.ti: ffff8803fd61c000 RIP: 0010:[<ffffffffa0784981>] [<ffffffffa0784981>] new_curseg+0x321/0x330 [f2fs] RSP: 0018:ffff8803fd61f918 EFLAGS: 00010246 RAX: 00000000000007ed RBX: 0000000000000224 RCX: 000000000000001f RDX: 0000000000000800 RSI: ffffffffffffffff RDI: ffff8803f56f4300 RBP: ffff8803fd61f978 R08: 0000000000000000 R09: 0000000000000000 R10: 0000000000000024 R11: ffff8800d23bbd78 R12: ffff8800d0ef0000 R13: 0000000000000224 R14: 0000000000000000 R15: 0000000000000001 FS: 00007f827ff85700(0000) GS:ffff88041ea80000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: ffffffffff600000 CR3: 00000003fef17000 CR4: 00000000001406e0 Stack: 000007ea00000002 0000000100000001 ffff8803f6456248 000007ed0000002b 0000000000000224 ffff880404d1aa20 ffff8803fd61f9c8 ffff8800d0ef0000 ffff8803f6456248 0000000000000001 00000000ffffffff ffffffffa078f358 Call Trace: [<ffffffffa0785b87>] allocate_segment_by_default+0x1a7/0x1f0 [f2fs] [<ffffffffa078322c>] allocate_data_block+0x17c/0x360 [f2fs] [<ffffffffa0779521>] __allocate_data_block+0x131/0x1d0 [f2fs] [<ffffffffa077a995>] f2fs_direct_IO+0x4b5/0x580 [f2fs] [<ffffffff811510ae>] generic_file_direct_write+0xae/0x160 [<ffffffff811518f5>] __generic_file_write_iter+0xd5/0x1f0 [<ffffffff81151e07>] generic_file_write_iter+0xf7/0x200 [<ffffffff81319e38>] ? apparmor_file_permission+0x18/0x20 [<ffffffffa0768480>] ? f2fs_fallocate+0x1190/0x1190 [f2fs] [<ffffffffa07684c6>] f2fs_file_write_iter+0x46/0x90 [f2fs] [<ffffffff8120b4fe>] aio_run_iocb+0x1ee/0x290 [<ffffffff81700f7e>] ? mutex_lock+0x1e/0x50 [<ffffffff8120a1d7>] ? aio_read_events+0x207/0x2b0 [<ffffffff8120b913>] do_io_submit+0x373/0x630 [<ffffffff8120a4f6>] ? SyS_io_getevents+0x56/0xb0 [<ffffffff8120bbe0>] SyS_io_submit+0x10/0x20 [<ffffffff81703857>] entry_SYSCALL_64_fastpath+0x12/0x6a Code: 45 c8 48 8b 78 10 e8 9f 23 bf e0 41 8b 8c 24 cc 03 00 00 89 c7 31 d2 89 c6 89 d8 29 df f7 f1 29 d1 39 cf 0f 83 be fd ff ff eb RIP [<ffffffffa0784981>] new_curseg+0x321/0x330 [f2fs] RSP <ffff8803fd61f918> ---[ end trace 2e577d7f711ddb86 ]--- The reason is that: in the test of generic/019, we will trigger a manmade IO error in block layer through debugfs, after that, prefree segment will no longer be freed, because we always skip doing gc or checkpoint when there occurs an IO error. Meanwhile fio with aio engine generated a large number of direct IOs, which continue allocating spaces in free segment until we run out of them, eventually, results in panic in new_curseg as no more free segment was found. So, this patch changes to return EIO in direct_IO for this condition. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2015-09-21 12:17:52 +00:00
if (iov_iter_rw(iter) == WRITE) {
err = __allocate_data_blocks(inode, offset, count);
if (err)
f2fs: fix to handle io error in ->direct_IO Here is a oops reported as following message when testing generic/019 of xfstest: ------------[ cut here ]------------ kernel BUG at /home/yuchao/git/f2fs-dev/segment.c:882! invalid opcode: 0000 [#1] SMP Modules linked in: zram lz4_compress lz4_decompress f2fs(O) ip6table_filter ip6_tables ebtable_nat ebtables nf_conntrack_ipv4 nf_def CPU: 2 PID: 25441 Comm: fio Tainted: G O 4.3.0-rc1+ #6 Hardware name: Hewlett-Packard HP Z220 CMT Workstation/1790, BIOS K51 v01.61 05/16/2013 task: ffff8803f4e85580 ti: ffff8803fd61c000 task.ti: ffff8803fd61c000 RIP: 0010:[<ffffffffa0784981>] [<ffffffffa0784981>] new_curseg+0x321/0x330 [f2fs] RSP: 0018:ffff8803fd61f918 EFLAGS: 00010246 RAX: 00000000000007ed RBX: 0000000000000224 RCX: 000000000000001f RDX: 0000000000000800 RSI: ffffffffffffffff RDI: ffff8803f56f4300 RBP: ffff8803fd61f978 R08: 0000000000000000 R09: 0000000000000000 R10: 0000000000000024 R11: ffff8800d23bbd78 R12: ffff8800d0ef0000 R13: 0000000000000224 R14: 0000000000000000 R15: 0000000000000001 FS: 00007f827ff85700(0000) GS:ffff88041ea80000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: ffffffffff600000 CR3: 00000003fef17000 CR4: 00000000001406e0 Stack: 000007ea00000002 0000000100000001 ffff8803f6456248 000007ed0000002b 0000000000000224 ffff880404d1aa20 ffff8803fd61f9c8 ffff8800d0ef0000 ffff8803f6456248 0000000000000001 00000000ffffffff ffffffffa078f358 Call Trace: [<ffffffffa0785b87>] allocate_segment_by_default+0x1a7/0x1f0 [f2fs] [<ffffffffa078322c>] allocate_data_block+0x17c/0x360 [f2fs] [<ffffffffa0779521>] __allocate_data_block+0x131/0x1d0 [f2fs] [<ffffffffa077a995>] f2fs_direct_IO+0x4b5/0x580 [f2fs] [<ffffffff811510ae>] generic_file_direct_write+0xae/0x160 [<ffffffff811518f5>] __generic_file_write_iter+0xd5/0x1f0 [<ffffffff81151e07>] generic_file_write_iter+0xf7/0x200 [<ffffffff81319e38>] ? apparmor_file_permission+0x18/0x20 [<ffffffffa0768480>] ? f2fs_fallocate+0x1190/0x1190 [f2fs] [<ffffffffa07684c6>] f2fs_file_write_iter+0x46/0x90 [f2fs] [<ffffffff8120b4fe>] aio_run_iocb+0x1ee/0x290 [<ffffffff81700f7e>] ? mutex_lock+0x1e/0x50 [<ffffffff8120a1d7>] ? aio_read_events+0x207/0x2b0 [<ffffffff8120b913>] do_io_submit+0x373/0x630 [<ffffffff8120a4f6>] ? SyS_io_getevents+0x56/0xb0 [<ffffffff8120bbe0>] SyS_io_submit+0x10/0x20 [<ffffffff81703857>] entry_SYSCALL_64_fastpath+0x12/0x6a Code: 45 c8 48 8b 78 10 e8 9f 23 bf e0 41 8b 8c 24 cc 03 00 00 89 c7 31 d2 89 c6 89 d8 29 df f7 f1 29 d1 39 cf 0f 83 be fd ff ff eb RIP [<ffffffffa0784981>] new_curseg+0x321/0x330 [f2fs] RSP <ffff8803fd61f918> ---[ end trace 2e577d7f711ddb86 ]--- The reason is that: in the test of generic/019, we will trigger a manmade IO error in block layer through debugfs, after that, prefree segment will no longer be freed, because we always skip doing gc or checkpoint when there occurs an IO error. Meanwhile fio with aio engine generated a large number of direct IOs, which continue allocating spaces in free segment until we run out of them, eventually, results in panic in new_curseg as no more free segment was found. So, this patch changes to return EIO in direct_IO for this condition. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2015-09-21 12:17:52 +00:00
goto out;
}
err = blockdev_direct_IO(iocb, inode, iter, offset, get_data_block_dio);
f2fs: fix to handle io error in ->direct_IO Here is a oops reported as following message when testing generic/019 of xfstest: ------------[ cut here ]------------ kernel BUG at /home/yuchao/git/f2fs-dev/segment.c:882! invalid opcode: 0000 [#1] SMP Modules linked in: zram lz4_compress lz4_decompress f2fs(O) ip6table_filter ip6_tables ebtable_nat ebtables nf_conntrack_ipv4 nf_def CPU: 2 PID: 25441 Comm: fio Tainted: G O 4.3.0-rc1+ #6 Hardware name: Hewlett-Packard HP Z220 CMT Workstation/1790, BIOS K51 v01.61 05/16/2013 task: ffff8803f4e85580 ti: ffff8803fd61c000 task.ti: ffff8803fd61c000 RIP: 0010:[<ffffffffa0784981>] [<ffffffffa0784981>] new_curseg+0x321/0x330 [f2fs] RSP: 0018:ffff8803fd61f918 EFLAGS: 00010246 RAX: 00000000000007ed RBX: 0000000000000224 RCX: 000000000000001f RDX: 0000000000000800 RSI: ffffffffffffffff RDI: ffff8803f56f4300 RBP: ffff8803fd61f978 R08: 0000000000000000 R09: 0000000000000000 R10: 0000000000000024 R11: ffff8800d23bbd78 R12: ffff8800d0ef0000 R13: 0000000000000224 R14: 0000000000000000 R15: 0000000000000001 FS: 00007f827ff85700(0000) GS:ffff88041ea80000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: ffffffffff600000 CR3: 00000003fef17000 CR4: 00000000001406e0 Stack: 000007ea00000002 0000000100000001 ffff8803f6456248 000007ed0000002b 0000000000000224 ffff880404d1aa20 ffff8803fd61f9c8 ffff8800d0ef0000 ffff8803f6456248 0000000000000001 00000000ffffffff ffffffffa078f358 Call Trace: [<ffffffffa0785b87>] allocate_segment_by_default+0x1a7/0x1f0 [f2fs] [<ffffffffa078322c>] allocate_data_block+0x17c/0x360 [f2fs] [<ffffffffa0779521>] __allocate_data_block+0x131/0x1d0 [f2fs] [<ffffffffa077a995>] f2fs_direct_IO+0x4b5/0x580 [f2fs] [<ffffffff811510ae>] generic_file_direct_write+0xae/0x160 [<ffffffff811518f5>] __generic_file_write_iter+0xd5/0x1f0 [<ffffffff81151e07>] generic_file_write_iter+0xf7/0x200 [<ffffffff81319e38>] ? apparmor_file_permission+0x18/0x20 [<ffffffffa0768480>] ? f2fs_fallocate+0x1190/0x1190 [f2fs] [<ffffffffa07684c6>] f2fs_file_write_iter+0x46/0x90 [f2fs] [<ffffffff8120b4fe>] aio_run_iocb+0x1ee/0x290 [<ffffffff81700f7e>] ? mutex_lock+0x1e/0x50 [<ffffffff8120a1d7>] ? aio_read_events+0x207/0x2b0 [<ffffffff8120b913>] do_io_submit+0x373/0x630 [<ffffffff8120a4f6>] ? SyS_io_getevents+0x56/0xb0 [<ffffffff8120bbe0>] SyS_io_submit+0x10/0x20 [<ffffffff81703857>] entry_SYSCALL_64_fastpath+0x12/0x6a Code: 45 c8 48 8b 78 10 e8 9f 23 bf e0 41 8b 8c 24 cc 03 00 00 89 c7 31 d2 89 c6 89 d8 29 df f7 f1 29 d1 39 cf 0f 83 be fd ff ff eb RIP [<ffffffffa0784981>] new_curseg+0x321/0x330 [f2fs] RSP <ffff8803fd61f918> ---[ end trace 2e577d7f711ddb86 ]--- The reason is that: in the test of generic/019, we will trigger a manmade IO error in block layer through debugfs, after that, prefree segment will no longer be freed, because we always skip doing gc or checkpoint when there occurs an IO error. Meanwhile fio with aio engine generated a large number of direct IOs, which continue allocating spaces in free segment until we run out of them, eventually, results in panic in new_curseg as no more free segment was found. So, this patch changes to return EIO in direct_IO for this condition. Signed-off-by: Chao Yu <chao2.yu@samsung.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
2015-09-21 12:17:52 +00:00
out:
if (err < 0 && iov_iter_rw(iter) == WRITE)
f2fs_write_failed(mapping, offset + count);
trace_f2fs_direct_IO_exit(inode, offset, count, iov_iter_rw(iter), err);
return err;
}
void f2fs_invalidate_page(struct page *page, unsigned int offset,
unsigned int length)
{
struct inode *inode = page->mapping->host;
struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
if (inode->i_ino >= F2FS_ROOT_INO(sbi) &&
(offset % PAGE_CACHE_SIZE || length != PAGE_CACHE_SIZE))
return;
if (PageDirty(page)) {
if (inode->i_ino == F2FS_META_INO(sbi))
dec_page_count(sbi, F2FS_DIRTY_META);
else if (inode->i_ino == F2FS_NODE_INO(sbi))
dec_page_count(sbi, F2FS_DIRTY_NODES);
else
inode_dec_dirty_pages(inode);
}
/* This is atomic written page, keep Private */
if (IS_ATOMIC_WRITTEN_PAGE(page))
return;
ClearPagePrivate(page);
}
int f2fs_release_page(struct page *page, gfp_t wait)
{
/* If this is dirty page, keep PagePrivate */
if (PageDirty(page))
return 0;
/* This is atomic written page, keep Private */
if (IS_ATOMIC_WRITTEN_PAGE(page))
return 0;
ClearPagePrivate(page);
return 1;
}
static int f2fs_set_data_page_dirty(struct page *page)
{
struct address_space *mapping = page->mapping;
struct inode *inode = mapping->host;
trace_f2fs_set_page_dirty(page, DATA);
SetPageUptodate(page);
if (f2fs_is_atomic_file(inode)) {
if (!IS_ATOMIC_WRITTEN_PAGE(page)) {
register_inmem_page(inode, page);
return 1;
}
/*
* Previously, this page has been registered, we just
* return here.
*/
return 0;
}
if (!PageDirty(page)) {
__set_page_dirty_nobuffers(page);
update_dirty_page(inode, page);
return 1;
}
return 0;
}
static sector_t f2fs_bmap(struct address_space *mapping, sector_t block)
{
struct inode *inode = mapping->host;
if (f2fs_has_inline_data(inode))
return 0;
/* make sure allocating whole blocks */
if (mapping_tagged(mapping, PAGECACHE_TAG_DIRTY))
filemap_write_and_wait(mapping);
return generic_block_bmap(mapping, block, get_data_block_bmap);
}
const struct address_space_operations f2fs_dblock_aops = {
.readpage = f2fs_read_data_page,
.readpages = f2fs_read_data_pages,
.writepage = f2fs_write_data_page,
.writepages = f2fs_write_data_pages,
.write_begin = f2fs_write_begin,
.write_end = f2fs_write_end,
.set_page_dirty = f2fs_set_data_page_dirty,
.invalidatepage = f2fs_invalidate_page,
.releasepage = f2fs_release_page,
.direct_IO = f2fs_direct_IO,
.bmap = f2fs_bmap,
};