linux-stable/fs/btrfs/file.c

3134 lines
83 KiB
C
Raw Normal View History

/*
* Copyright (C) 2007 Oracle. All rights reserved.
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public
* License v2 as published by the Free Software Foundation.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* General Public License for more details.
*
* You should have received a copy of the GNU General Public
* License along with this program; if not, write to the
* Free Software Foundation, Inc., 59 Temple Place - Suite 330,
* Boston, MA 021110-1307, USA.
*/
#include <linux/fs.h>
#include <linux/pagemap.h>
#include <linux/highmem.h>
#include <linux/time.h>
#include <linux/init.h>
#include <linux/string.h>
#include <linux/backing-dev.h>
#include <linux/mpage.h>
#include <linux/falloc.h>
#include <linux/swap.h>
#include <linux/writeback.h>
#include <linux/compat.h>
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h percpu.h is included by sched.h and module.h and thus ends up being included when building most .c files. percpu.h includes slab.h which in turn includes gfp.h making everything defined by the two files universally available and complicating inclusion dependencies. percpu.h -> slab.h dependency is about to be removed. Prepare for this change by updating users of gfp and slab facilities include those headers directly instead of assuming availability. As this conversion needs to touch large number of source files, the following script is used as the basis of conversion. http://userweb.kernel.org/~tj/misc/slabh-sweep.py The script does the followings. * Scan files for gfp and slab usages and update includes such that only the necessary includes are there. ie. if only gfp is used, gfp.h, if slab is used, slab.h. * When the script inserts a new include, it looks at the include blocks and try to put the new include such that its order conforms to its surrounding. It's put in the include block which contains core kernel includes, in the same order that the rest are ordered - alphabetical, Christmas tree, rev-Xmas-tree or at the end if there doesn't seem to be any matching order. * If the script can't find a place to put a new include (mostly because the file doesn't have fitting include block), it prints out an error message indicating which .h file needs to be added to the file. The conversion was done in the following steps. 1. The initial automatic conversion of all .c files updated slightly over 4000 files, deleting around 700 includes and adding ~480 gfp.h and ~3000 slab.h inclusions. The script emitted errors for ~400 files. 2. Each error was manually checked. Some didn't need the inclusion, some needed manual addition while adding it to implementation .h or embedding .c file was more appropriate for others. This step added inclusions to around 150 files. 3. The script was run again and the output was compared to the edits from #2 to make sure no file was left behind. 4. Several build tests were done and a couple of problems were fixed. e.g. lib/decompress_*.c used malloc/free() wrappers around slab APIs requiring slab.h to be added manually. 5. The script was run on all .h files but without automatically editing them as sprinkling gfp.h and slab.h inclusions around .h files could easily lead to inclusion dependency hell. Most gfp.h inclusion directives were ignored as stuff from gfp.h was usually wildly available and often used in preprocessor macros. Each slab.h inclusion directive was examined and added manually as necessary. 6. percpu.h was updated not to include slab.h. 7. Build test were done on the following configurations and failures were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my distributed build env didn't work with gcov compiles) and a few more options had to be turned off depending on archs to make things build (like ipr on powerpc/64 which failed due to missing writeq). * x86 and x86_64 UP and SMP allmodconfig and a custom test config. * powerpc and powerpc64 SMP allmodconfig * sparc and sparc64 SMP allmodconfig * ia64 SMP allmodconfig * s390 SMP allmodconfig * alpha SMP allmodconfig * um on x86_64 SMP allmodconfig 8. percpu.h modifications were reverted so that it could be applied as a separate patch and serve as bisection point. Given the fact that I had only a couple of failures from tests on step 6, I'm fairly confident about the coverage of this conversion patch. If there is a breakage, it's likely to be something in one of the arch headers which should be easily discoverable easily on most builds of the specific arch. Signed-off-by: Tejun Heo <tj@kernel.org> Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org> Cc: Ingo Molnar <mingo@redhat.com> Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 08:04:11 +00:00
#include <linux/slab.h>
#include <linux/btrfs.h>
#include <linux/uio.h>
#include "ctree.h"
#include "disk-io.h"
#include "transaction.h"
#include "btrfs_inode.h"
#include "print-tree.h"
#include "tree-log.h"
#include "locking.h"
#include "volumes.h"
Btrfs: rework qgroup accounting Currently qgroups account for space by intercepting delayed ref updates to fs trees. It does this by adding sequence numbers to delayed ref updates so that it can figure out how the tree looked before the update so we can adjust the counters properly. The problem with this is that it does not allow delayed refs to be merged, so if you say are defragging an extent with 5k snapshots pointing to it we will thrash the delayed ref lock because we need to go back and manually merge these things together. Instead we want to process quota changes when we know they are going to happen, like when we first allocate an extent, we free a reference for an extent, we add new references etc. This patch accomplishes this by only adding qgroup operations for real ref changes. We only modify the sequence number when we need to lookup roots for bytenrs, this reduces the amount of churn on the sequence number and allows us to merge delayed refs as we add them most of the time. This patch encompasses a bunch of architectural changes 1) qgroup ref operations: instead of tracking qgroup operations through the delayed refs we simply add new ref operations whenever we notice that we need to when we've modified the refs themselves. 2) tree mod seq: we no longer have this separation of major/minor counters. this makes the sequence number stuff much more sane and we can remove some locking that was needed to protect the counter. 3) delayed ref seq: we now read the tree mod seq number and use that as our sequence. This means each new delayed ref doesn't have it's own unique sequence number, rather whenever we go to lookup backrefs we inc the sequence number so we can make sure to keep any new operations from screwing up our world view at that given point. This allows us to merge delayed refs during runtime. With all of these changes the delayed ref stuff is a little saner and the qgroup accounting stuff no longer goes negative in some cases like it was before. Thanks, Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-05-14 00:30:47 +00:00
#include "qgroup.h"
#include "compression.h"
static struct kmem_cache *btrfs_inode_defrag_cachep;
/*
* when auto defrag is enabled we
* queue up these defrag structs to remember which
* inodes need defragging passes
*/
struct inode_defrag {
struct rb_node rb_node;
/* objectid */
u64 ino;
/*
* transid where the defrag was added, we search for
* extents newer than this
*/
u64 transid;
/* root objectid */
u64 root;
/* last offset we were able to defrag */
u64 last_offset;
/* if we've wrapped around back to zero once already */
int cycled;
};
static int __compare_inode_defrag(struct inode_defrag *defrag1,
struct inode_defrag *defrag2)
{
if (defrag1->root > defrag2->root)
return 1;
else if (defrag1->root < defrag2->root)
return -1;
else if (defrag1->ino > defrag2->ino)
return 1;
else if (defrag1->ino < defrag2->ino)
return -1;
else
return 0;
}
/* pop a record for an inode into the defrag tree. The lock
* must be held already
*
* If you're inserting a record for an older transid than an
* existing record, the transid already in the tree is lowered
*
* If an existing record is found the defrag item you
* pass in is freed
*/
static int __btrfs_add_inode_defrag(struct btrfs_inode *inode,
struct inode_defrag *defrag)
{
struct btrfs_fs_info *fs_info = btrfs_sb(inode->vfs_inode.i_sb);
struct inode_defrag *entry;
struct rb_node **p;
struct rb_node *parent = NULL;
int ret;
p = &fs_info->defrag_inodes.rb_node;
while (*p) {
parent = *p;
entry = rb_entry(parent, struct inode_defrag, rb_node);
ret = __compare_inode_defrag(defrag, entry);
if (ret < 0)
p = &parent->rb_left;
else if (ret > 0)
p = &parent->rb_right;
else {
/* if we're reinserting an entry for
* an old defrag run, make sure to
* lower the transid of our existing record
*/
if (defrag->transid < entry->transid)
entry->transid = defrag->transid;
if (defrag->last_offset > entry->last_offset)
entry->last_offset = defrag->last_offset;
return -EEXIST;
}
}
set_bit(BTRFS_INODE_IN_DEFRAG, &inode->runtime_flags);
rb_link_node(&defrag->rb_node, parent, p);
rb_insert_color(&defrag->rb_node, &fs_info->defrag_inodes);
return 0;
}
static inline int __need_auto_defrag(struct btrfs_fs_info *fs_info)
{
if (!btrfs_test_opt(fs_info, AUTO_DEFRAG))
return 0;
if (btrfs_fs_closing(fs_info))
return 0;
return 1;
}
/*
* insert a defrag record for this inode if auto defrag is
* enabled
*/
int btrfs_add_inode_defrag(struct btrfs_trans_handle *trans,
struct btrfs_inode *inode)
{
struct btrfs_fs_info *fs_info = btrfs_sb(inode->vfs_inode.i_sb);
struct btrfs_root *root = inode->root;
struct inode_defrag *defrag;
u64 transid;
int ret;
if (!__need_auto_defrag(fs_info))
return 0;
if (test_bit(BTRFS_INODE_IN_DEFRAG, &inode->runtime_flags))
return 0;
if (trans)
transid = trans->transid;
else
transid = inode->root->last_trans;
defrag = kmem_cache_zalloc(btrfs_inode_defrag_cachep, GFP_NOFS);
if (!defrag)
return -ENOMEM;
defrag->ino = btrfs_ino(inode);
defrag->transid = transid;
defrag->root = root->root_key.objectid;
spin_lock(&fs_info->defrag_inodes_lock);
if (!test_bit(BTRFS_INODE_IN_DEFRAG, &inode->runtime_flags)) {
/*
* If we set IN_DEFRAG flag and evict the inode from memory,
* and then re-read this inode, this new inode doesn't have
* IN_DEFRAG flag. At the case, we may find the existed defrag.
*/
ret = __btrfs_add_inode_defrag(inode, defrag);
if (ret)
kmem_cache_free(btrfs_inode_defrag_cachep, defrag);
} else {
kmem_cache_free(btrfs_inode_defrag_cachep, defrag);
}
spin_unlock(&fs_info->defrag_inodes_lock);
return 0;
}
/*
* Requeue the defrag object. If there is a defrag object that points to
* the same inode in the tree, we will merge them together (by
* __btrfs_add_inode_defrag()) and free the one that we want to requeue.
*/
static void btrfs_requeue_inode_defrag(struct btrfs_inode *inode,
struct inode_defrag *defrag)
{
struct btrfs_fs_info *fs_info = btrfs_sb(inode->vfs_inode.i_sb);
int ret;
if (!__need_auto_defrag(fs_info))
goto out;
/*
* Here we don't check the IN_DEFRAG flag, because we need merge
* them together.
*/
spin_lock(&fs_info->defrag_inodes_lock);
ret = __btrfs_add_inode_defrag(inode, defrag);
spin_unlock(&fs_info->defrag_inodes_lock);
if (ret)
goto out;
return;
out:
kmem_cache_free(btrfs_inode_defrag_cachep, defrag);
}
/*
* pick the defragable inode that we want, if it doesn't exist, we will get
* the next one.
*/
static struct inode_defrag *
btrfs_pick_defrag_inode(struct btrfs_fs_info *fs_info, u64 root, u64 ino)
{
struct inode_defrag *entry = NULL;
struct inode_defrag tmp;
struct rb_node *p;
struct rb_node *parent = NULL;
int ret;
tmp.ino = ino;
tmp.root = root;
spin_lock(&fs_info->defrag_inodes_lock);
p = fs_info->defrag_inodes.rb_node;
while (p) {
parent = p;
entry = rb_entry(parent, struct inode_defrag, rb_node);
ret = __compare_inode_defrag(&tmp, entry);
if (ret < 0)
p = parent->rb_left;
else if (ret > 0)
p = parent->rb_right;
else
goto out;
}
if (parent && __compare_inode_defrag(&tmp, entry) > 0) {
parent = rb_next(parent);
if (parent)
entry = rb_entry(parent, struct inode_defrag, rb_node);
else
entry = NULL;
}
out:
if (entry)
rb_erase(parent, &fs_info->defrag_inodes);
spin_unlock(&fs_info->defrag_inodes_lock);
return entry;
}
void btrfs_cleanup_defrag_inodes(struct btrfs_fs_info *fs_info)
{
struct inode_defrag *defrag;
struct rb_node *node;
spin_lock(&fs_info->defrag_inodes_lock);
node = rb_first(&fs_info->defrag_inodes);
while (node) {
rb_erase(node, &fs_info->defrag_inodes);
defrag = rb_entry(node, struct inode_defrag, rb_node);
kmem_cache_free(btrfs_inode_defrag_cachep, defrag);
cond_resched_lock(&fs_info->defrag_inodes_lock);
node = rb_first(&fs_info->defrag_inodes);
}
spin_unlock(&fs_info->defrag_inodes_lock);
}
#define BTRFS_DEFRAG_BATCH 1024
static int __btrfs_run_defrag_inode(struct btrfs_fs_info *fs_info,
struct inode_defrag *defrag)
{
struct btrfs_root *inode_root;
struct inode *inode;
struct btrfs_key key;
struct btrfs_ioctl_defrag_range_args range;
int num_defrag;
int index;
int ret;
/* get the inode */
key.objectid = defrag->root;
key.type = BTRFS_ROOT_ITEM_KEY;
key.offset = (u64)-1;
index = srcu_read_lock(&fs_info->subvol_srcu);
inode_root = btrfs_read_fs_root_no_name(fs_info, &key);
if (IS_ERR(inode_root)) {
ret = PTR_ERR(inode_root);
goto cleanup;
}
key.objectid = defrag->ino;
key.type = BTRFS_INODE_ITEM_KEY;
key.offset = 0;
inode = btrfs_iget(fs_info->sb, &key, inode_root, NULL);
if (IS_ERR(inode)) {
ret = PTR_ERR(inode);
goto cleanup;
}
srcu_read_unlock(&fs_info->subvol_srcu, index);
/* do a chunk of defrag */
clear_bit(BTRFS_INODE_IN_DEFRAG, &BTRFS_I(inode)->runtime_flags);
memset(&range, 0, sizeof(range));
range.len = (u64)-1;
range.start = defrag->last_offset;
sb_start_write(fs_info->sb);
num_defrag = btrfs_defrag_file(inode, NULL, &range, defrag->transid,
BTRFS_DEFRAG_BATCH);
sb_end_write(fs_info->sb);
/*
* if we filled the whole defrag batch, there
* must be more work to do. Queue this defrag
* again
*/
if (num_defrag == BTRFS_DEFRAG_BATCH) {
defrag->last_offset = range.start;
btrfs_requeue_inode_defrag(BTRFS_I(inode), defrag);
} else if (defrag->last_offset && !defrag->cycled) {
/*
* we didn't fill our defrag batch, but
* we didn't start at zero. Make sure we loop
* around to the start of the file.
*/
defrag->last_offset = 0;
defrag->cycled = 1;
btrfs_requeue_inode_defrag(BTRFS_I(inode), defrag);
} else {
kmem_cache_free(btrfs_inode_defrag_cachep, defrag);
}
iput(inode);
return 0;
cleanup:
srcu_read_unlock(&fs_info->subvol_srcu, index);
kmem_cache_free(btrfs_inode_defrag_cachep, defrag);
return ret;
}
/*
* run through the list of inodes in the FS that need
* defragging
*/
int btrfs_run_defrag_inodes(struct btrfs_fs_info *fs_info)
{
struct inode_defrag *defrag;
u64 first_ino = 0;
u64 root_objectid = 0;
atomic_inc(&fs_info->defrag_running);
while (1) {
/* Pause the auto defragger. */
if (test_bit(BTRFS_FS_STATE_REMOUNTING,
&fs_info->fs_state))
break;
if (!__need_auto_defrag(fs_info))
break;
/* find an inode to defrag */
defrag = btrfs_pick_defrag_inode(fs_info, root_objectid,
first_ino);
if (!defrag) {
if (root_objectid || first_ino) {
root_objectid = 0;
first_ino = 0;
continue;
} else {
break;
}
}
first_ino = defrag->ino + 1;
root_objectid = defrag->root;
__btrfs_run_defrag_inode(fs_info, defrag);
}
atomic_dec(&fs_info->defrag_running);
/*
* during unmount, we use the transaction_wait queue to
* wait for the defragger to stop
*/
wake_up(&fs_info->transaction_wait);
return 0;
}
/* simple helper to fault in pages and copy. This should go away
* and be replaced with calls into generic code.
*/
static noinline int btrfs_copy_from_user(loff_t pos, size_t write_bytes,
struct page **prepared_pages,
struct iov_iter *i)
{
size_t copied = 0;
size_t total_copied = 0;
int pg = 0;
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
int offset = pos & (PAGE_SIZE - 1);
while (write_bytes > 0) {
size_t count = min_t(size_t,
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
PAGE_SIZE - offset, write_bytes);
struct page *page = prepared_pages[pg];
/*
* Copy data from userspace to the current page
*/
copied = iov_iter_copy_from_user_atomic(page, i, offset, count);
/* Flush processor's dcache for this page */
flush_dcache_page(page);
/*
* if we get a partial write, we can end up with
* partially up to date pages. These add
* a lot of complexity, so make sure they don't
* happen by forcing this copy to be retried.
*
* The rest of the btrfs_file_write code will fall
* back to page at a time copies after we return 0.
*/
if (!PageUptodate(page) && copied < count)
copied = 0;
iov_iter_advance(i, copied);
write_bytes -= copied;
total_copied += copied;
/* Return to btrfs_file_write_iter to fault page */
if (unlikely(copied == 0))
break;
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
if (copied < PAGE_SIZE - offset) {
offset += copied;
} else {
pg++;
offset = 0;
}
}
return total_copied;
}
/*
* unlocks pages after btrfs_file_write is done with them
*/
static void btrfs_drop_pages(struct page **pages, size_t num_pages)
{
size_t i;
for (i = 0; i < num_pages; i++) {
/* page checked is some magic around finding pages that
* have been modified without going through btrfs_set_page_dirty
mm: non-atomically mark page accessed during page cache allocation where possible aops->write_begin may allocate a new page and make it visible only to have mark_page_accessed called almost immediately after. Once the page is visible the atomic operations are necessary which is noticable overhead when writing to an in-memory filesystem like tmpfs but should also be noticable with fast storage. The objective of the patch is to initialse the accessed information with non-atomic operations before the page is visible. The bulk of filesystems directly or indirectly use grab_cache_page_write_begin or find_or_create_page for the initial allocation of a page cache page. This patch adds an init_page_accessed() helper which behaves like the first call to mark_page_accessed() but may called before the page is visible and can be done non-atomically. The primary APIs of concern in this care are the following and are used by most filesystems. find_get_page find_lock_page find_or_create_page grab_cache_page_nowait grab_cache_page_write_begin All of them are very similar in detail to the patch creates a core helper pagecache_get_page() which takes a flags parameter that affects its behavior such as whether the page should be marked accessed or not. Then old API is preserved but is basically a thin wrapper around this core function. Each of the filesystems are then updated to avoid calling mark_page_accessed when it is known that the VM interfaces have already done the job. There is a slight snag in that the timing of the mark_page_accessed() has now changed so in rare cases it's possible a page gets to the end of the LRU as PageReferenced where as previously it might have been repromoted. This is expected to be rare but it's worth the filesystem people thinking about it in case they see a problem with the timing change. It is also the case that some filesystems may be marking pages accessed that previously did not but it makes sense that filesystems have consistent behaviour in this regard. The test case used to evaulate this is a simple dd of a large file done multiple times with the file deleted on each iterations. The size of the file is 1/10th physical memory to avoid dirty page balancing. In the async case it will be possible that the workload completes without even hitting the disk and will have variable results but highlight the impact of mark_page_accessed for async IO. The sync results are expected to be more stable. The exception is tmpfs where the normal case is for the "IO" to not hit the disk. The test machine was single socket and UMA to avoid any scheduling or NUMA artifacts. Throughput and wall times are presented for sync IO, only wall times are shown for async as the granularity reported by dd and the variability is unsuitable for comparison. As async results were variable do to writback timings, I'm only reporting the maximum figures. The sync results were stable enough to make the mean and stddev uninteresting. The performance results are reported based on a run with no profiling. Profile data is based on a separate run with oprofile running. async dd 3.15.0-rc3 3.15.0-rc3 vanilla accessed-v2 ext3 Max elapsed 13.9900 ( 0.00%) 11.5900 ( 17.16%) tmpfs Max elapsed 0.5100 ( 0.00%) 0.4900 ( 3.92%) btrfs Max elapsed 12.8100 ( 0.00%) 12.7800 ( 0.23%) ext4 Max elapsed 18.6000 ( 0.00%) 13.3400 ( 28.28%) xfs Max elapsed 12.5600 ( 0.00%) 2.0900 ( 83.36%) The XFS figure is a bit strange as it managed to avoid a worst case by sheer luck but the average figures looked reasonable. samples percentage ext3 86107 0.9783 vmlinux-3.15.0-rc4-vanilla mark_page_accessed ext3 23833 0.2710 vmlinux-3.15.0-rc4-accessed-v3r25 mark_page_accessed ext3 5036 0.0573 vmlinux-3.15.0-rc4-accessed-v3r25 init_page_accessed ext4 64566 0.8961 vmlinux-3.15.0-rc4-vanilla mark_page_accessed ext4 5322 0.0713 vmlinux-3.15.0-rc4-accessed-v3r25 mark_page_accessed ext4 2869 0.0384 vmlinux-3.15.0-rc4-accessed-v3r25 init_page_accessed xfs 62126 1.7675 vmlinux-3.15.0-rc4-vanilla mark_page_accessed xfs 1904 0.0554 vmlinux-3.15.0-rc4-accessed-v3r25 init_page_accessed xfs 103 0.0030 vmlinux-3.15.0-rc4-accessed-v3r25 mark_page_accessed btrfs 10655 0.1338 vmlinux-3.15.0-rc4-vanilla mark_page_accessed btrfs 2020 0.0273 vmlinux-3.15.0-rc4-accessed-v3r25 init_page_accessed btrfs 587 0.0079 vmlinux-3.15.0-rc4-accessed-v3r25 mark_page_accessed tmpfs 59562 3.2628 vmlinux-3.15.0-rc4-vanilla mark_page_accessed tmpfs 1210 0.0696 vmlinux-3.15.0-rc4-accessed-v3r25 init_page_accessed tmpfs 94 0.0054 vmlinux-3.15.0-rc4-accessed-v3r25 mark_page_accessed [akpm@linux-foundation.org: don't run init_page_accessed() against an uninitialised pointer] Signed-off-by: Mel Gorman <mgorman@suse.de> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Jan Kara <jack@suse.cz> Cc: Michal Hocko <mhocko@suse.cz> Cc: Hugh Dickins <hughd@google.com> Cc: Dave Hansen <dave.hansen@intel.com> Cc: Theodore Ts'o <tytso@mit.edu> Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com> Cc: Oleg Nesterov <oleg@redhat.com> Cc: Rik van Riel <riel@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Tested-by: Prabhakar Lad <prabhakar.csengg@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-06-04 23:10:31 +00:00
* clear it here. There should be no need to mark the pages
* accessed as prepare_pages should have marked them accessed
* in prepare_pages via find_or_create_page()
*/
ClearPageChecked(pages[i]);
unlock_page(pages[i]);
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
put_page(pages[i]);
}
}
/*
* after copy_from_user, pages need to be dirtied and we need to make
* sure holes are created between the current EOF and the start of
* any next extents (if required).
*
* this also makes the decision about creating an inline extent vs
* doing real data extents, marking pages dirty and delalloc as required.
*/
int btrfs_dirty_pages(struct inode *inode, struct page **pages,
size_t num_pages, loff_t pos, size_t write_bytes,
struct extent_state **cached)
{
struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
int err = 0;
int i;
u64 num_bytes;
u64 start_pos;
u64 end_of_last_block;
u64 end_pos = pos + write_bytes;
loff_t isize = i_size_read(inode);
start_pos = pos & ~((u64) fs_info->sectorsize - 1);
num_bytes = round_up(write_bytes + pos - start_pos,
fs_info->sectorsize);
end_of_last_block = start_pos + num_bytes - 1;
err = btrfs_set_extent_delalloc(inode, start_pos, end_of_last_block,
cached, 0);
if (err)
return err;
Btrfs: proper -ENOSPC handling At the start of a transaction we do a btrfs_reserve_metadata_space() and specify how many items we plan on modifying. Then once we've done our modifications and such, just call btrfs_unreserve_metadata_space() for the same number of items we reserved. For keeping track of metadata needed for data I've had to add an extent_io op for when we merge extents. This lets us track space properly when we are doing sequential writes, so we don't end up reserving way more metadata space than what we need. The only place where the metadata space accounting is not done is in the relocation code. This is because Yan is going to be reworking that code in the near future, so running btrfs-vol -b could still possibly result in a ENOSPC related panic. This patch also turns off the metadata_ratio stuff in order to allow users to more efficiently use their disk space. This patch makes it so we track how much metadata we need for an inode's delayed allocation extents by tracking how many extents are currently waiting for allocation. It introduces two new callbacks for the extent_io tree's, merge_extent_hook and split_extent_hook. These help us keep track of when we merge delalloc extents together and split them up. Reservations are handled prior to any actually dirty'ing occurs, and then we unreserve after we dirty. btrfs_unreserve_metadata_for_delalloc() will make the appropriate unreservations as needed based on the number of reservations we currently have and the number of extents we currently have. Doing the reservation outside of doing any of the actual dirty'ing lets us do things like filemap_flush() the inode to try and force delalloc to happen, or as a last resort actually start allocation on all delalloc inodes in the fs. This has survived dbench, fs_mark and an fsx torture test. Signed-off-by: Josef Bacik <jbacik@redhat.com> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2009-09-11 20:12:44 +00:00
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
for (i = 0; i < num_pages; i++) {
struct page *p = pages[i];
SetPageUptodate(p);
ClearPageChecked(p);
set_page_dirty(p);
}
/*
* we've only changed i_size in ram, and we haven't updated
* the disk i_size. There is no need to log the inode
* at this time.
*/
if (end_pos > isize)
i_size_write(inode, end_pos);
return 0;
}
/*
* this drops all the extents in the cache that intersect the range
* [start, end]. Existing extents are split as required.
*/
void btrfs_drop_extent_cache(struct btrfs_inode *inode, u64 start, u64 end,
int skip_pinned)
{
struct extent_map *em;
struct extent_map *split = NULL;
struct extent_map *split2 = NULL;
struct extent_map_tree *em_tree = &inode->extent_tree;
u64 len = end - start + 1;
Btrfs: turbo charge fsync At least for the vm workload. Currently on fsync we will 1) Truncate all items in the log tree for the given inode if they exist and 2) Copy all items for a given inode into the log The problem with this is that for things like VMs you can have lots of extents from the fragmented writing behavior, and worst yet you may have only modified a few extents, not the entire thing. This patch fixes this problem by tracking which transid modified our extent, and then when we do the tree logging we find all of the extents we've modified in our current transaction, sort them and commit them. We also only truncate up to the xattrs of the inode and copy that stuff in normally, and then just drop any extents in the range we have that exist in the log already. Here are some numbers of a 50 meg fio job that does random writes and fsync()s after every write Original Patched SATA drive 82KB/s 140KB/s Fusion drive 431KB/s 2532KB/s So around 2-6 times faster depending on your hardware. There are a few corner cases, for example if you truncate at all we have to do it the old way since there is no way to be sure what is in the log is ok. This probably could be done smarter, but if you write-fsync-truncate-write-fsync you deserve what you get. All this work is in RAM of course so if your inode gets evicted from cache and you read it in and fsync it we'll do it the slow way if we are still in the same transaction that we last modified the inode in. The biggest cool part of this is that it requires no changes to the recovery code, so if you fsync with this patch and crash and load an old kernel, it will run the recovery and be a-ok. I have tested this pretty thoroughly with an fsync tester and everything comes back fine, as well as xfstests. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2012-08-17 17:14:17 +00:00
u64 gen;
int ret;
int testend = 1;
unsigned long flags;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
int compressed = 0;
2013-04-05 20:51:15 +00:00
bool modified;
WARN_ON(end < start);
if (end == (u64)-1) {
len = (u64)-1;
testend = 0;
}
while (1) {
int no_splits = 0;
2013-04-05 20:51:15 +00:00
modified = false;
if (!split)
split = alloc_extent_map();
if (!split2)
split2 = alloc_extent_map();
if (!split || !split2)
no_splits = 1;
write_lock(&em_tree->lock);
em = lookup_extent_mapping(em_tree, start, len);
if (!em) {
write_unlock(&em_tree->lock);
break;
}
flags = em->flags;
Btrfs: turbo charge fsync At least for the vm workload. Currently on fsync we will 1) Truncate all items in the log tree for the given inode if they exist and 2) Copy all items for a given inode into the log The problem with this is that for things like VMs you can have lots of extents from the fragmented writing behavior, and worst yet you may have only modified a few extents, not the entire thing. This patch fixes this problem by tracking which transid modified our extent, and then when we do the tree logging we find all of the extents we've modified in our current transaction, sort them and commit them. We also only truncate up to the xattrs of the inode and copy that stuff in normally, and then just drop any extents in the range we have that exist in the log already. Here are some numbers of a 50 meg fio job that does random writes and fsync()s after every write Original Patched SATA drive 82KB/s 140KB/s Fusion drive 431KB/s 2532KB/s So around 2-6 times faster depending on your hardware. There are a few corner cases, for example if you truncate at all we have to do it the old way since there is no way to be sure what is in the log is ok. This probably could be done smarter, but if you write-fsync-truncate-write-fsync you deserve what you get. All this work is in RAM of course so if your inode gets evicted from cache and you read it in and fsync it we'll do it the slow way if we are still in the same transaction that we last modified the inode in. The biggest cool part of this is that it requires no changes to the recovery code, so if you fsync with this patch and crash and load an old kernel, it will run the recovery and be a-ok. I have tested this pretty thoroughly with an fsync tester and everything comes back fine, as well as xfstests. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2012-08-17 17:14:17 +00:00
gen = em->generation;
if (skip_pinned && test_bit(EXTENT_FLAG_PINNED, &em->flags)) {
if (testend && em->start + em->len >= start + len) {
free_extent_map(em);
write_unlock(&em_tree->lock);
break;
}
start = em->start + em->len;
if (testend)
len = start + len - (em->start + em->len);
free_extent_map(em);
write_unlock(&em_tree->lock);
continue;
}
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
compressed = test_bit(EXTENT_FLAG_COMPRESSED, &em->flags);
clear_bit(EXTENT_FLAG_PINNED, &em->flags);
clear_bit(EXTENT_FLAG_LOGGING, &flags);
2013-04-05 20:51:15 +00:00
modified = !list_empty(&em->list);
if (no_splits)
goto next;
if (em->start < start) {
split->start = em->start;
split->len = start - em->start;
if (em->block_start < EXTENT_MAP_LAST_BYTE) {
split->orig_start = em->orig_start;
split->block_start = em->block_start;
if (compressed)
split->block_len = em->block_len;
else
split->block_len = split->len;
split->orig_block_len = max(split->block_len,
em->orig_block_len);
split->ram_bytes = em->ram_bytes;
} else {
split->orig_start = split->start;
split->block_len = 0;
split->block_start = em->block_start;
split->orig_block_len = 0;
split->ram_bytes = split->len;
}
Btrfs: turbo charge fsync At least for the vm workload. Currently on fsync we will 1) Truncate all items in the log tree for the given inode if they exist and 2) Copy all items for a given inode into the log The problem with this is that for things like VMs you can have lots of extents from the fragmented writing behavior, and worst yet you may have only modified a few extents, not the entire thing. This patch fixes this problem by tracking which transid modified our extent, and then when we do the tree logging we find all of the extents we've modified in our current transaction, sort them and commit them. We also only truncate up to the xattrs of the inode and copy that stuff in normally, and then just drop any extents in the range we have that exist in the log already. Here are some numbers of a 50 meg fio job that does random writes and fsync()s after every write Original Patched SATA drive 82KB/s 140KB/s Fusion drive 431KB/s 2532KB/s So around 2-6 times faster depending on your hardware. There are a few corner cases, for example if you truncate at all we have to do it the old way since there is no way to be sure what is in the log is ok. This probably could be done smarter, but if you write-fsync-truncate-write-fsync you deserve what you get. All this work is in RAM of course so if your inode gets evicted from cache and you read it in and fsync it we'll do it the slow way if we are still in the same transaction that we last modified the inode in. The biggest cool part of this is that it requires no changes to the recovery code, so if you fsync with this patch and crash and load an old kernel, it will run the recovery and be a-ok. I have tested this pretty thoroughly with an fsync tester and everything comes back fine, as well as xfstests. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2012-08-17 17:14:17 +00:00
split->generation = gen;
split->bdev = em->bdev;
split->flags = flags;
split->compress_type = em->compress_type;
Btrfs: more efficient btrfs_drop_extent_cache While droping extent map structures from the extent cache that cover our target range, we would remove each extent map structure from the red black tree and then add either 1 or 2 new extent map structures if the former extent map covered sections outside our target range. This change simply attempts to replace the existing extent map structure with a new one that covers the subsection we're not interested in, instead of doing a red black remove operation followed by an insertion operation. The number of elements in an inode's extent map tree can get very high for large files under random writes. For example, while running the following test: sysbench --test=fileio --file-num=1 --file-total-size=10G \ --file-test-mode=rndrw --num-threads=32 --file-block-size=32768 \ --max-requests=500000 --file-rw-ratio=2 [prepare|run] I captured the following histogram capturing the number of extent_map items in the red black tree while that test was running: Count: 122462 Range: 1.000 - 172231.000; Mean: 96415.831; Median: 101855.000; Stddev: 49700.981 Percentiles: 90th: 160120.000; 95th: 166335.000; 99th: 171070.000 1.000 - 5.231: 452 | 5.231 - 187.392: 87 | 187.392 - 585.911: 206 | 585.911 - 1827.438: 623 | 1827.438 - 5695.245: 1962 # 5695.245 - 17744.861: 6204 #### 17744.861 - 55283.764: 21115 ############ 55283.764 - 172231.000: 91813 ##################################################### Benchmark: sysbench --test=fileio --file-num=1 --file-total-size=10G --file-test-mode=rndwr \ --num-threads=64 --file-block-size=32768 --max-requests=0 --max-time=60 \ --file-io-mode=sync --file-fsync-freq=0 [prepare|run] Before this change: 122.1Mb/sec After this change: 125.07Mb/sec (averages of 5 test runs) Test machine: quad core intel i5-3570K, 32Gb of ram, SSD Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com> Signed-off-by: Josef Bacik <jbacik@fb.com>
2014-02-25 14:15:13 +00:00
replace_extent_mapping(em_tree, em, split, modified);
free_extent_map(split);
split = split2;
split2 = NULL;
}
if (testend && em->start + em->len > start + len) {
u64 diff = start + len - em->start;
split->start = start + len;
split->len = em->start + em->len - (start + len);
split->bdev = em->bdev;
split->flags = flags;
split->compress_type = em->compress_type;
Btrfs: turbo charge fsync At least for the vm workload. Currently on fsync we will 1) Truncate all items in the log tree for the given inode if they exist and 2) Copy all items for a given inode into the log The problem with this is that for things like VMs you can have lots of extents from the fragmented writing behavior, and worst yet you may have only modified a few extents, not the entire thing. This patch fixes this problem by tracking which transid modified our extent, and then when we do the tree logging we find all of the extents we've modified in our current transaction, sort them and commit them. We also only truncate up to the xattrs of the inode and copy that stuff in normally, and then just drop any extents in the range we have that exist in the log already. Here are some numbers of a 50 meg fio job that does random writes and fsync()s after every write Original Patched SATA drive 82KB/s 140KB/s Fusion drive 431KB/s 2532KB/s So around 2-6 times faster depending on your hardware. There are a few corner cases, for example if you truncate at all we have to do it the old way since there is no way to be sure what is in the log is ok. This probably could be done smarter, but if you write-fsync-truncate-write-fsync you deserve what you get. All this work is in RAM of course so if your inode gets evicted from cache and you read it in and fsync it we'll do it the slow way if we are still in the same transaction that we last modified the inode in. The biggest cool part of this is that it requires no changes to the recovery code, so if you fsync with this patch and crash and load an old kernel, it will run the recovery and be a-ok. I have tested this pretty thoroughly with an fsync tester and everything comes back fine, as well as xfstests. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2012-08-17 17:14:17 +00:00
split->generation = gen;
if (em->block_start < EXTENT_MAP_LAST_BYTE) {
split->orig_block_len = max(em->block_len,
em->orig_block_len);
split->ram_bytes = em->ram_bytes;
if (compressed) {
split->block_len = em->block_len;
split->block_start = em->block_start;
split->orig_start = em->orig_start;
} else {
split->block_len = split->len;
split->block_start = em->block_start
+ diff;
split->orig_start = em->orig_start;
}
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
} else {
split->ram_bytes = split->len;
split->orig_start = split->start;
split->block_len = 0;
split->block_start = em->block_start;
split->orig_block_len = 0;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
}
Btrfs: more efficient btrfs_drop_extent_cache While droping extent map structures from the extent cache that cover our target range, we would remove each extent map structure from the red black tree and then add either 1 or 2 new extent map structures if the former extent map covered sections outside our target range. This change simply attempts to replace the existing extent map structure with a new one that covers the subsection we're not interested in, instead of doing a red black remove operation followed by an insertion operation. The number of elements in an inode's extent map tree can get very high for large files under random writes. For example, while running the following test: sysbench --test=fileio --file-num=1 --file-total-size=10G \ --file-test-mode=rndrw --num-threads=32 --file-block-size=32768 \ --max-requests=500000 --file-rw-ratio=2 [prepare|run] I captured the following histogram capturing the number of extent_map items in the red black tree while that test was running: Count: 122462 Range: 1.000 - 172231.000; Mean: 96415.831; Median: 101855.000; Stddev: 49700.981 Percentiles: 90th: 160120.000; 95th: 166335.000; 99th: 171070.000 1.000 - 5.231: 452 | 5.231 - 187.392: 87 | 187.392 - 585.911: 206 | 585.911 - 1827.438: 623 | 1827.438 - 5695.245: 1962 # 5695.245 - 17744.861: 6204 #### 17744.861 - 55283.764: 21115 ############ 55283.764 - 172231.000: 91813 ##################################################### Benchmark: sysbench --test=fileio --file-num=1 --file-total-size=10G --file-test-mode=rndwr \ --num-threads=64 --file-block-size=32768 --max-requests=0 --max-time=60 \ --file-io-mode=sync --file-fsync-freq=0 [prepare|run] Before this change: 122.1Mb/sec After this change: 125.07Mb/sec (averages of 5 test runs) Test machine: quad core intel i5-3570K, 32Gb of ram, SSD Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com> Signed-off-by: Josef Bacik <jbacik@fb.com>
2014-02-25 14:15:13 +00:00
if (extent_map_in_tree(em)) {
replace_extent_mapping(em_tree, em, split,
modified);
} else {
ret = add_extent_mapping(em_tree, split,
modified);
ASSERT(ret == 0); /* Logic error */
}
free_extent_map(split);
split = NULL;
}
next:
Btrfs: more efficient btrfs_drop_extent_cache While droping extent map structures from the extent cache that cover our target range, we would remove each extent map structure from the red black tree and then add either 1 or 2 new extent map structures if the former extent map covered sections outside our target range. This change simply attempts to replace the existing extent map structure with a new one that covers the subsection we're not interested in, instead of doing a red black remove operation followed by an insertion operation. The number of elements in an inode's extent map tree can get very high for large files under random writes. For example, while running the following test: sysbench --test=fileio --file-num=1 --file-total-size=10G \ --file-test-mode=rndrw --num-threads=32 --file-block-size=32768 \ --max-requests=500000 --file-rw-ratio=2 [prepare|run] I captured the following histogram capturing the number of extent_map items in the red black tree while that test was running: Count: 122462 Range: 1.000 - 172231.000; Mean: 96415.831; Median: 101855.000; Stddev: 49700.981 Percentiles: 90th: 160120.000; 95th: 166335.000; 99th: 171070.000 1.000 - 5.231: 452 | 5.231 - 187.392: 87 | 187.392 - 585.911: 206 | 585.911 - 1827.438: 623 | 1827.438 - 5695.245: 1962 # 5695.245 - 17744.861: 6204 #### 17744.861 - 55283.764: 21115 ############ 55283.764 - 172231.000: 91813 ##################################################### Benchmark: sysbench --test=fileio --file-num=1 --file-total-size=10G --file-test-mode=rndwr \ --num-threads=64 --file-block-size=32768 --max-requests=0 --max-time=60 \ --file-io-mode=sync --file-fsync-freq=0 [prepare|run] Before this change: 122.1Mb/sec After this change: 125.07Mb/sec (averages of 5 test runs) Test machine: quad core intel i5-3570K, 32Gb of ram, SSD Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com> Signed-off-by: Josef Bacik <jbacik@fb.com>
2014-02-25 14:15:13 +00:00
if (extent_map_in_tree(em))
remove_extent_mapping(em_tree, em);
write_unlock(&em_tree->lock);
/* once for us */
free_extent_map(em);
/* once for the tree*/
free_extent_map(em);
}
if (split)
free_extent_map(split);
if (split2)
free_extent_map(split2);
}
/*
* this is very complex, but the basic idea is to drop all extents
* in the range start - end. hint_block is filled in with a block number
* that would be a good hint to the block allocator for this file.
*
* If an extent intersects the range but is not entirely inside the range
* it is either truncated or split. Anything entirely inside the range
* is deleted from the tree.
*/
Btrfs: turbo charge fsync At least for the vm workload. Currently on fsync we will 1) Truncate all items in the log tree for the given inode if they exist and 2) Copy all items for a given inode into the log The problem with this is that for things like VMs you can have lots of extents from the fragmented writing behavior, and worst yet you may have only modified a few extents, not the entire thing. This patch fixes this problem by tracking which transid modified our extent, and then when we do the tree logging we find all of the extents we've modified in our current transaction, sort them and commit them. We also only truncate up to the xattrs of the inode and copy that stuff in normally, and then just drop any extents in the range we have that exist in the log already. Here are some numbers of a 50 meg fio job that does random writes and fsync()s after every write Original Patched SATA drive 82KB/s 140KB/s Fusion drive 431KB/s 2532KB/s So around 2-6 times faster depending on your hardware. There are a few corner cases, for example if you truncate at all we have to do it the old way since there is no way to be sure what is in the log is ok. This probably could be done smarter, but if you write-fsync-truncate-write-fsync you deserve what you get. All this work is in RAM of course so if your inode gets evicted from cache and you read it in and fsync it we'll do it the slow way if we are still in the same transaction that we last modified the inode in. The biggest cool part of this is that it requires no changes to the recovery code, so if you fsync with this patch and crash and load an old kernel, it will run the recovery and be a-ok. I have tested this pretty thoroughly with an fsync tester and everything comes back fine, as well as xfstests. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2012-08-17 17:14:17 +00:00
int __btrfs_drop_extents(struct btrfs_trans_handle *trans,
struct btrfs_root *root, struct inode *inode,
struct btrfs_path *path, u64 start, u64 end,
Btrfs: faster file extent item replace operations When writing to a file we drop existing file extent items that cover the write range and then add a new file extent item that represents that write range. Before this change we were doing a tree lookup to remove the file extent items, and then after we did another tree lookup to insert the new file extent item. Most of the time all the file extent items we need to drop are located within a single leaf - this is the leaf where our new file extent item ends up at. Therefore, in this common case just combine these 2 operations into a single one. By avoiding the second btree navigation for insertion of the new file extent item, we reduce btree node/leaf lock acquisitions/releases, btree block/leaf COW operations, CPU time on btree node/leaf key binary searches, etc. Besides for file writes, this is an operation that happens for file fsync's as well. However log btrees are much less likely to big as big as regular fs btrees, therefore the impact of this change is smaller. The following benchmark was performed against an SSD drive and a HDD drive, both for random and sequential writes: sysbench --test=fileio --file-num=4096 --file-total-size=8G \ --file-test-mode=[rndwr|seqwr] --num-threads=512 \ --file-block-size=8192 \ --max-requests=1000000 \ --file-fsync-freq=0 --file-io-mode=sync [prepare|run] All results below are averages of 10 runs of the respective test. ** SSD sequential writes Before this change: 225.88 Mb/sec After this change: 277.26 Mb/sec ** SSD random writes Before this change: 49.91 Mb/sec After this change: 56.39 Mb/sec ** HDD sequential writes Before this change: 68.53 Mb/sec After this change: 69.87 Mb/sec ** HDD random writes Before this change: 13.04 Mb/sec After this change: 14.39 Mb/sec Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-01-07 11:42:27 +00:00
u64 *drop_end, int drop_cache,
int replace_extent,
u32 extent_item_size,
int *key_inserted)
{
struct btrfs_fs_info *fs_info = root->fs_info;
struct extent_buffer *leaf;
struct btrfs_file_extent_item *fi;
struct btrfs_key key;
struct btrfs_key new_key;
u64 ino = btrfs_ino(BTRFS_I(inode));
u64 search_start = start;
u64 disk_bytenr = 0;
u64 num_bytes = 0;
u64 extent_offset = 0;
u64 extent_end = 0;
u64 last_end = start;
int del_nr = 0;
int del_slot = 0;
int extent_type;
int recow;
int ret;
int modify_tree = -1;
int update_refs;
int found = 0;
Btrfs: faster file extent item replace operations When writing to a file we drop existing file extent items that cover the write range and then add a new file extent item that represents that write range. Before this change we were doing a tree lookup to remove the file extent items, and then after we did another tree lookup to insert the new file extent item. Most of the time all the file extent items we need to drop are located within a single leaf - this is the leaf where our new file extent item ends up at. Therefore, in this common case just combine these 2 operations into a single one. By avoiding the second btree navigation for insertion of the new file extent item, we reduce btree node/leaf lock acquisitions/releases, btree block/leaf COW operations, CPU time on btree node/leaf key binary searches, etc. Besides for file writes, this is an operation that happens for file fsync's as well. However log btrees are much less likely to big as big as regular fs btrees, therefore the impact of this change is smaller. The following benchmark was performed against an SSD drive and a HDD drive, both for random and sequential writes: sysbench --test=fileio --file-num=4096 --file-total-size=8G \ --file-test-mode=[rndwr|seqwr] --num-threads=512 \ --file-block-size=8192 \ --max-requests=1000000 \ --file-fsync-freq=0 --file-io-mode=sync [prepare|run] All results below are averages of 10 runs of the respective test. ** SSD sequential writes Before this change: 225.88 Mb/sec After this change: 277.26 Mb/sec ** SSD random writes Before this change: 49.91 Mb/sec After this change: 56.39 Mb/sec ** HDD sequential writes Before this change: 68.53 Mb/sec After this change: 69.87 Mb/sec ** HDD random writes Before this change: 13.04 Mb/sec After this change: 14.39 Mb/sec Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-01-07 11:42:27 +00:00
int leafs_visited = 0;
if (drop_cache)
btrfs_drop_extent_cache(BTRFS_I(inode), start, end - 1, 0);
if (start >= BTRFS_I(inode)->disk_i_size && !replace_extent)
modify_tree = 0;
update_refs = (test_bit(BTRFS_ROOT_REF_COWS, &root->state) ||
root == fs_info->tree_root);
while (1) {
recow = 0;
ret = btrfs_lookup_file_extent(trans, root, path, ino,
search_start, modify_tree);
if (ret < 0)
break;
if (ret > 0 && path->slots[0] > 0 && search_start == start) {
leaf = path->nodes[0];
btrfs_item_key_to_cpu(leaf, &key, path->slots[0] - 1);
if (key.objectid == ino &&
key.type == BTRFS_EXTENT_DATA_KEY)
path->slots[0]--;
}
ret = 0;
Btrfs: faster file extent item replace operations When writing to a file we drop existing file extent items that cover the write range and then add a new file extent item that represents that write range. Before this change we were doing a tree lookup to remove the file extent items, and then after we did another tree lookup to insert the new file extent item. Most of the time all the file extent items we need to drop are located within a single leaf - this is the leaf where our new file extent item ends up at. Therefore, in this common case just combine these 2 operations into a single one. By avoiding the second btree navigation for insertion of the new file extent item, we reduce btree node/leaf lock acquisitions/releases, btree block/leaf COW operations, CPU time on btree node/leaf key binary searches, etc. Besides for file writes, this is an operation that happens for file fsync's as well. However log btrees are much less likely to big as big as regular fs btrees, therefore the impact of this change is smaller. The following benchmark was performed against an SSD drive and a HDD drive, both for random and sequential writes: sysbench --test=fileio --file-num=4096 --file-total-size=8G \ --file-test-mode=[rndwr|seqwr] --num-threads=512 \ --file-block-size=8192 \ --max-requests=1000000 \ --file-fsync-freq=0 --file-io-mode=sync [prepare|run] All results below are averages of 10 runs of the respective test. ** SSD sequential writes Before this change: 225.88 Mb/sec After this change: 277.26 Mb/sec ** SSD random writes Before this change: 49.91 Mb/sec After this change: 56.39 Mb/sec ** HDD sequential writes Before this change: 68.53 Mb/sec After this change: 69.87 Mb/sec ** HDD random writes Before this change: 13.04 Mb/sec After this change: 14.39 Mb/sec Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-01-07 11:42:27 +00:00
leafs_visited++;
next_slot:
leaf = path->nodes[0];
if (path->slots[0] >= btrfs_header_nritems(leaf)) {
BUG_ON(del_nr > 0);
ret = btrfs_next_leaf(root, path);
if (ret < 0)
break;
if (ret > 0) {
ret = 0;
break;
}
Btrfs: faster file extent item replace operations When writing to a file we drop existing file extent items that cover the write range and then add a new file extent item that represents that write range. Before this change we were doing a tree lookup to remove the file extent items, and then after we did another tree lookup to insert the new file extent item. Most of the time all the file extent items we need to drop are located within a single leaf - this is the leaf where our new file extent item ends up at. Therefore, in this common case just combine these 2 operations into a single one. By avoiding the second btree navigation for insertion of the new file extent item, we reduce btree node/leaf lock acquisitions/releases, btree block/leaf COW operations, CPU time on btree node/leaf key binary searches, etc. Besides for file writes, this is an operation that happens for file fsync's as well. However log btrees are much less likely to big as big as regular fs btrees, therefore the impact of this change is smaller. The following benchmark was performed against an SSD drive and a HDD drive, both for random and sequential writes: sysbench --test=fileio --file-num=4096 --file-total-size=8G \ --file-test-mode=[rndwr|seqwr] --num-threads=512 \ --file-block-size=8192 \ --max-requests=1000000 \ --file-fsync-freq=0 --file-io-mode=sync [prepare|run] All results below are averages of 10 runs of the respective test. ** SSD sequential writes Before this change: 225.88 Mb/sec After this change: 277.26 Mb/sec ** SSD random writes Before this change: 49.91 Mb/sec After this change: 56.39 Mb/sec ** HDD sequential writes Before this change: 68.53 Mb/sec After this change: 69.87 Mb/sec ** HDD random writes Before this change: 13.04 Mb/sec After this change: 14.39 Mb/sec Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-01-07 11:42:27 +00:00
leafs_visited++;
leaf = path->nodes[0];
recow = 1;
}
btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
Btrfs: fix race leading to incorrect item deletion when dropping extents While running a stress test I got the following warning triggered: [191627.672810] ------------[ cut here ]------------ [191627.673949] WARNING: CPU: 8 PID: 8447 at fs/btrfs/file.c:779 __btrfs_drop_extents+0x391/0xa50 [btrfs]() (...) [191627.701485] Call Trace: [191627.702037] [<ffffffff8145f077>] dump_stack+0x4f/0x7b [191627.702992] [<ffffffff81095de5>] ? console_unlock+0x356/0x3a2 [191627.704091] [<ffffffff8104b3b0>] warn_slowpath_common+0xa1/0xbb [191627.705380] [<ffffffffa0664499>] ? __btrfs_drop_extents+0x391/0xa50 [btrfs] [191627.706637] [<ffffffff8104b46d>] warn_slowpath_null+0x1a/0x1c [191627.707789] [<ffffffffa0664499>] __btrfs_drop_extents+0x391/0xa50 [btrfs] [191627.709155] [<ffffffff8115663c>] ? cache_alloc_debugcheck_after.isra.32+0x171/0x1d0 [191627.712444] [<ffffffff81155007>] ? kmemleak_alloc_recursive.constprop.40+0x16/0x18 [191627.714162] [<ffffffffa06570c9>] insert_reserved_file_extent.constprop.40+0x83/0x24e [btrfs] [191627.715887] [<ffffffffa065422b>] ? start_transaction+0x3bb/0x610 [btrfs] [191627.717287] [<ffffffffa065b604>] btrfs_finish_ordered_io+0x273/0x4e2 [btrfs] [191627.728865] [<ffffffffa065b888>] finish_ordered_fn+0x15/0x17 [btrfs] [191627.730045] [<ffffffffa067d688>] normal_work_helper+0x14c/0x32c [btrfs] [191627.731256] [<ffffffffa067d96a>] btrfs_endio_write_helper+0x12/0x14 [btrfs] [191627.732661] [<ffffffff81061119>] process_one_work+0x24c/0x4ae [191627.733822] [<ffffffff810615b0>] worker_thread+0x206/0x2c2 [191627.734857] [<ffffffff810613aa>] ? process_scheduled_works+0x2f/0x2f [191627.736052] [<ffffffff810613aa>] ? process_scheduled_works+0x2f/0x2f [191627.737349] [<ffffffff810669a6>] kthread+0xef/0xf7 [191627.738267] [<ffffffff810f3b3a>] ? time_hardirqs_on+0x15/0x28 [191627.739330] [<ffffffff810668b7>] ? __kthread_parkme+0xad/0xad [191627.741976] [<ffffffff81465592>] ret_from_fork+0x42/0x70 [191627.743080] [<ffffffff810668b7>] ? __kthread_parkme+0xad/0xad [191627.744206] ---[ end trace bbfddacb7aaada8d ]--- $ cat -n fs/btrfs/file.c 691 int __btrfs_drop_extents(struct btrfs_trans_handle *trans, (...) 758 btrfs_item_key_to_cpu(leaf, &key, path->slots[0]); 759 if (key.objectid > ino || 760 key.type > BTRFS_EXTENT_DATA_KEY || key.offset >= end) 761 break; 762 763 fi = btrfs_item_ptr(leaf, path->slots[0], 764 struct btrfs_file_extent_item); 765 extent_type = btrfs_file_extent_type(leaf, fi); 766 767 if (extent_type == BTRFS_FILE_EXTENT_REG || 768 extent_type == BTRFS_FILE_EXTENT_PREALLOC) { (...) 774 } else if (extent_type == BTRFS_FILE_EXTENT_INLINE) { (...) 778 } else { 779 WARN_ON(1); 780 extent_end = search_start; 781 } (...) This happened because the item we were processing did not match a file extent item (its key type != BTRFS_EXTENT_DATA_KEY), and even on this case we cast the item to a struct btrfs_file_extent_item pointer and then find a type field value that does not match any of the expected values (BTRFS_FILE_EXTENT_[REG|PREALLOC|INLINE]). This scenario happens due to a tiny time window where a race can happen as exemplified below. For example, consider the following scenario where we're using the NO_HOLES feature and we have the following two neighbour leafs: Leaf X (has N items) Leaf Y [ ... (257 INODE_ITEM 0) (257 INODE_REF 256) ] [ (257 EXTENT_DATA 8192), ... ] slot N - 2 slot N - 1 slot 0 Our inode 257 has an implicit hole in the range [0, 8K[ (implicit rather than explicit because NO_HOLES is enabled). Now if our inode has an ordered extent for the range [4K, 8K[ that is finishing, the following can happen: CPU 1 CPU 2 btrfs_finish_ordered_io() insert_reserved_file_extent() __btrfs_drop_extents() Searches for the key (257 EXTENT_DATA 4096) through btrfs_lookup_file_extent() Key not found and we get a path where path->nodes[0] == leaf X and path->slots[0] == N Because path->slots[0] is >= btrfs_header_nritems(leaf X), we call btrfs_next_leaf() btrfs_next_leaf() releases the path inserts key (257 INODE_REF 4096) at the end of leaf X, leaf X now has N + 1 keys, and the new key is at slot N btrfs_next_leaf() searches for key (257 INODE_REF 256), with path->keep_locks set to 1, because it was the last key it saw in leaf X finds it in leaf X again and notices it's no longer the last key of the leaf, so it returns 0 with path->nodes[0] == leaf X and path->slots[0] == N (which is now < btrfs_header_nritems(leaf X)), pointing to the new key (257 INODE_REF 4096) __btrfs_drop_extents() casts the item at path->nodes[0], slot path->slots[0], to a struct btrfs_file_extent_item - it does not skip keys for the target inode with a type less than BTRFS_EXTENT_DATA_KEY (BTRFS_INODE_REF_KEY < BTRFS_EXTENT_DATA_KEY) sees a bogus value for the type field triggering the WARN_ON in the trace shown above, and sets extent_end = search_start (4096) does the if-then-else logic to fixup 0 length extent items created by a past bug from hole punching: if (extent_end == key.offset && extent_end >= search_start) goto delete_extent_item; that evaluates to true and it ends up deleting the key pointed to by path->slots[0], (257 INODE_REF 4096), from leaf X The same could happen for example for a xattr that ends up having a key with an offset value that matches search_start (very unlikely but not impossible). So fix this by ensuring that keys smaller than BTRFS_EXTENT_DATA_KEY are skipped, never casted to struct btrfs_file_extent_item and never deleted by accident. Also protect against the unexpected case of getting a key for a lower inode number by skipping that key and issuing a warning. Cc: stable@vger.kernel.org Signed-off-by: Filipe Manana <fdmanana@suse.com>
2015-11-06 13:33:33 +00:00
if (key.objectid > ino)
break;
if (WARN_ON_ONCE(key.objectid < ino) ||
key.type < BTRFS_EXTENT_DATA_KEY) {
ASSERT(del_nr == 0);
path->slots[0]++;
goto next_slot;
}
if (key.type > BTRFS_EXTENT_DATA_KEY || key.offset >= end)
break;
fi = btrfs_item_ptr(leaf, path->slots[0],
struct btrfs_file_extent_item);
extent_type = btrfs_file_extent_type(leaf, fi);
if (extent_type == BTRFS_FILE_EXTENT_REG ||
extent_type == BTRFS_FILE_EXTENT_PREALLOC) {
disk_bytenr = btrfs_file_extent_disk_bytenr(leaf, fi);
num_bytes = btrfs_file_extent_disk_num_bytes(leaf, fi);
extent_offset = btrfs_file_extent_offset(leaf, fi);
extent_end = key.offset +
btrfs_file_extent_num_bytes(leaf, fi);
} else if (extent_type == BTRFS_FILE_EXTENT_INLINE) {
extent_end = key.offset +
btrfs_file_extent_inline_len(leaf,
path->slots[0], fi);
} else {
Btrfs: fix race leading to incorrect item deletion when dropping extents While running a stress test I got the following warning triggered: [191627.672810] ------------[ cut here ]------------ [191627.673949] WARNING: CPU: 8 PID: 8447 at fs/btrfs/file.c:779 __btrfs_drop_extents+0x391/0xa50 [btrfs]() (...) [191627.701485] Call Trace: [191627.702037] [<ffffffff8145f077>] dump_stack+0x4f/0x7b [191627.702992] [<ffffffff81095de5>] ? console_unlock+0x356/0x3a2 [191627.704091] [<ffffffff8104b3b0>] warn_slowpath_common+0xa1/0xbb [191627.705380] [<ffffffffa0664499>] ? __btrfs_drop_extents+0x391/0xa50 [btrfs] [191627.706637] [<ffffffff8104b46d>] warn_slowpath_null+0x1a/0x1c [191627.707789] [<ffffffffa0664499>] __btrfs_drop_extents+0x391/0xa50 [btrfs] [191627.709155] [<ffffffff8115663c>] ? cache_alloc_debugcheck_after.isra.32+0x171/0x1d0 [191627.712444] [<ffffffff81155007>] ? kmemleak_alloc_recursive.constprop.40+0x16/0x18 [191627.714162] [<ffffffffa06570c9>] insert_reserved_file_extent.constprop.40+0x83/0x24e [btrfs] [191627.715887] [<ffffffffa065422b>] ? start_transaction+0x3bb/0x610 [btrfs] [191627.717287] [<ffffffffa065b604>] btrfs_finish_ordered_io+0x273/0x4e2 [btrfs] [191627.728865] [<ffffffffa065b888>] finish_ordered_fn+0x15/0x17 [btrfs] [191627.730045] [<ffffffffa067d688>] normal_work_helper+0x14c/0x32c [btrfs] [191627.731256] [<ffffffffa067d96a>] btrfs_endio_write_helper+0x12/0x14 [btrfs] [191627.732661] [<ffffffff81061119>] process_one_work+0x24c/0x4ae [191627.733822] [<ffffffff810615b0>] worker_thread+0x206/0x2c2 [191627.734857] [<ffffffff810613aa>] ? process_scheduled_works+0x2f/0x2f [191627.736052] [<ffffffff810613aa>] ? process_scheduled_works+0x2f/0x2f [191627.737349] [<ffffffff810669a6>] kthread+0xef/0xf7 [191627.738267] [<ffffffff810f3b3a>] ? time_hardirqs_on+0x15/0x28 [191627.739330] [<ffffffff810668b7>] ? __kthread_parkme+0xad/0xad [191627.741976] [<ffffffff81465592>] ret_from_fork+0x42/0x70 [191627.743080] [<ffffffff810668b7>] ? __kthread_parkme+0xad/0xad [191627.744206] ---[ end trace bbfddacb7aaada8d ]--- $ cat -n fs/btrfs/file.c 691 int __btrfs_drop_extents(struct btrfs_trans_handle *trans, (...) 758 btrfs_item_key_to_cpu(leaf, &key, path->slots[0]); 759 if (key.objectid > ino || 760 key.type > BTRFS_EXTENT_DATA_KEY || key.offset >= end) 761 break; 762 763 fi = btrfs_item_ptr(leaf, path->slots[0], 764 struct btrfs_file_extent_item); 765 extent_type = btrfs_file_extent_type(leaf, fi); 766 767 if (extent_type == BTRFS_FILE_EXTENT_REG || 768 extent_type == BTRFS_FILE_EXTENT_PREALLOC) { (...) 774 } else if (extent_type == BTRFS_FILE_EXTENT_INLINE) { (...) 778 } else { 779 WARN_ON(1); 780 extent_end = search_start; 781 } (...) This happened because the item we were processing did not match a file extent item (its key type != BTRFS_EXTENT_DATA_KEY), and even on this case we cast the item to a struct btrfs_file_extent_item pointer and then find a type field value that does not match any of the expected values (BTRFS_FILE_EXTENT_[REG|PREALLOC|INLINE]). This scenario happens due to a tiny time window where a race can happen as exemplified below. For example, consider the following scenario where we're using the NO_HOLES feature and we have the following two neighbour leafs: Leaf X (has N items) Leaf Y [ ... (257 INODE_ITEM 0) (257 INODE_REF 256) ] [ (257 EXTENT_DATA 8192), ... ] slot N - 2 slot N - 1 slot 0 Our inode 257 has an implicit hole in the range [0, 8K[ (implicit rather than explicit because NO_HOLES is enabled). Now if our inode has an ordered extent for the range [4K, 8K[ that is finishing, the following can happen: CPU 1 CPU 2 btrfs_finish_ordered_io() insert_reserved_file_extent() __btrfs_drop_extents() Searches for the key (257 EXTENT_DATA 4096) through btrfs_lookup_file_extent() Key not found and we get a path where path->nodes[0] == leaf X and path->slots[0] == N Because path->slots[0] is >= btrfs_header_nritems(leaf X), we call btrfs_next_leaf() btrfs_next_leaf() releases the path inserts key (257 INODE_REF 4096) at the end of leaf X, leaf X now has N + 1 keys, and the new key is at slot N btrfs_next_leaf() searches for key (257 INODE_REF 256), with path->keep_locks set to 1, because it was the last key it saw in leaf X finds it in leaf X again and notices it's no longer the last key of the leaf, so it returns 0 with path->nodes[0] == leaf X and path->slots[0] == N (which is now < btrfs_header_nritems(leaf X)), pointing to the new key (257 INODE_REF 4096) __btrfs_drop_extents() casts the item at path->nodes[0], slot path->slots[0], to a struct btrfs_file_extent_item - it does not skip keys for the target inode with a type less than BTRFS_EXTENT_DATA_KEY (BTRFS_INODE_REF_KEY < BTRFS_EXTENT_DATA_KEY) sees a bogus value for the type field triggering the WARN_ON in the trace shown above, and sets extent_end = search_start (4096) does the if-then-else logic to fixup 0 length extent items created by a past bug from hole punching: if (extent_end == key.offset && extent_end >= search_start) goto delete_extent_item; that evaluates to true and it ends up deleting the key pointed to by path->slots[0], (257 INODE_REF 4096), from leaf X The same could happen for example for a xattr that ends up having a key with an offset value that matches search_start (very unlikely but not impossible). So fix this by ensuring that keys smaller than BTRFS_EXTENT_DATA_KEY are skipped, never casted to struct btrfs_file_extent_item and never deleted by accident. Also protect against the unexpected case of getting a key for a lower inode number by skipping that key and issuing a warning. Cc: stable@vger.kernel.org Signed-off-by: Filipe Manana <fdmanana@suse.com>
2015-11-06 13:33:33 +00:00
/* can't happen */
BUG();
}
Btrfs: fix leaf corruption caused by ENOSPC while hole punching While running a stress test with multiple threads writing to the same btrfs file system, I ended up with a situation where a leaf was corrupted in that it had 2 file extent item keys that had the same exact key. I was able to detect this quickly thanks to the following patch which triggers an assertion as soon as a leaf is marked dirty if there are duplicated keys or out of order keys: Btrfs: check if items are ordered when a leaf is marked dirty (https://patchwork.kernel.org/patch/3955431/) Basically while running the test, I got the following in dmesg: [28877.415877] WARNING: CPU: 2 PID: 10706 at fs/btrfs/file.c:553 btrfs_drop_extent_cache+0x435/0x440 [btrfs]() (...) [28877.415917] Call Trace: [28877.415922] [<ffffffff816f1189>] dump_stack+0x4e/0x68 [28877.415926] [<ffffffff8104a32c>] warn_slowpath_common+0x8c/0xc0 [28877.415929] [<ffffffff8104a37a>] warn_slowpath_null+0x1a/0x20 [28877.415944] [<ffffffffa03775a5>] btrfs_drop_extent_cache+0x435/0x440 [btrfs] [28877.415949] [<ffffffff8118e7be>] ? kmem_cache_alloc+0xfe/0x1c0 [28877.415962] [<ffffffffa03777d9>] fill_holes+0x229/0x3e0 [btrfs] [28877.415972] [<ffffffffa0345865>] ? block_rsv_add_bytes+0x55/0x80 [btrfs] [28877.415984] [<ffffffffa03792cb>] btrfs_fallocate+0xb6b/0xc20 [btrfs] (...) [29854.132560] BTRFS critical (device sdc): corrupt leaf, bad key order: block=955232256,root=1, slot=24 [29854.132565] BTRFS info (device sdc): leaf 955232256 total ptrs 40 free space 778 (...) [29854.132637] item 23 key (3486 108 667648) itemoff 2694 itemsize 53 [29854.132638] extent data disk bytenr 14574411776 nr 286720 [29854.132639] extent data offset 0 nr 286720 ram 286720 [29854.132640] item 24 key (3486 108 954368) itemoff 2641 itemsize 53 [29854.132641] extent data disk bytenr 0 nr 0 [29854.132643] extent data offset 0 nr 0 ram 0 [29854.132644] item 25 key (3486 108 954368) itemoff 2588 itemsize 53 [29854.132645] extent data disk bytenr 8699670528 nr 77824 [29854.132646] extent data offset 0 nr 77824 ram 77824 [29854.132647] item 26 key (3486 108 1146880) itemoff 2535 itemsize 53 [29854.132648] extent data disk bytenr 8699670528 nr 77824 [29854.132649] extent data offset 0 nr 77824 ram 77824 (...) [29854.132707] kernel BUG at fs/btrfs/ctree.h:3901! (...) [29854.132771] Call Trace: [29854.132779] [<ffffffffa0342b5c>] setup_items_for_insert+0x2dc/0x400 [btrfs] [29854.132791] [<ffffffffa0378537>] __btrfs_drop_extents+0xba7/0xdd0 [btrfs] [29854.132794] [<ffffffff8109c0d6>] ? trace_hardirqs_on_caller+0x16/0x1d0 [29854.132797] [<ffffffff8109c29d>] ? trace_hardirqs_on+0xd/0x10 [29854.132800] [<ffffffff8118e7be>] ? kmem_cache_alloc+0xfe/0x1c0 [29854.132810] [<ffffffffa036783b>] insert_reserved_file_extent.constprop.66+0xab/0x310 [btrfs] [29854.132820] [<ffffffffa036a6c6>] __btrfs_prealloc_file_range+0x116/0x340 [btrfs] [29854.132830] [<ffffffffa0374d53>] btrfs_prealloc_file_range+0x23/0x30 [btrfs] (...) So this is caused by getting an -ENOSPC error while punching a file hole, more specifically, we get -ENOSPC error from __btrfs_drop_extents in the while loop of file.c:btrfs_punch_hole() when it's unable to modify the btree to delete one or more file extent items due to lack of enough free space. When this happens, in btrfs_punch_hole(), we attempt to reclaim free space by switching our transaction block reservation object to root->fs_info->trans_block_rsv, end our transaction and start a new transaction basically - and, we keep increasing our current offset (cur_offset) as long as it's smaller than the end of the target range (lockend) - this makes use leave the loop with cur_offset == drop_end which in turn makes us call fill_holes() for inserting a file extent item that represents a 0 bytes range hole (and this insertion succeeds, as in the meanwhile more space became available). This 0 bytes file hole extent item is a problem because any subsequent caller of __btrfs_drop_extents (regular file writes, or fallocate calls for e.g.), with a start file offset that is equal to the offset of the hole, will not remove this extent item due to the following conditional in the while loop of __btrfs_drop_extents: if (extent_end <= search_start) { path->slots[0]++; goto next_slot; } This later makes the call to setup_items_for_insert() (at the very end of __btrfs_drop_extents), insert a new file extent item with the same offset as the 0 bytes file hole extent item that follows it. Needless is to say that this causes chaos, either when reading the leaf from disk (btree_readpage_end_io_hook), where we perform leaf sanity checks or in subsequent operations that manipulate file extent items, as in the fallocate call as shown by the dmesg trace above. Without my other patch to perform the leaf sanity checks once a leaf is marked as dirty (if the integrity checker is enabled), it would have been much harder to debug this issue. This change might fix a few similar issues reported by users in the mailing list regarding assertion failures in btrfs_set_item_key_safe calls performed by __btrfs_drop_extents, such as the following report: http://comments.gmane.org/gmane.comp.file-systems.btrfs/32938 Asking fill_holes() to create a 0 bytes wide file hole item also produced the first warning in the trace above, as we passed a range to btrfs_drop_extent_cache that has an end smaller (by -1) than its start. On 3.14 kernels this issue manifests itself through leaf corruption, as we get duplicated file extent item keys in a leaf when calling setup_items_for_insert(), but on older kernels, setup_items_for_insert() isn't called by __btrfs_drop_extents(), instead we have callers of __btrfs_drop_extents(), namely the functions inode.c:insert_inline_extent() and inode.c:insert_reserved_file_extent(), calling btrfs_insert_empty_item() to insert the new file extent item, which would fail with error -EEXIST, instead of inserting a duplicated key - which is still a serious issue as it would make all similar file extent item replace operations keep failing if they target the same file range. Cc: stable@vger.kernel.org Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-04-29 12:18:40 +00:00
/*
* Don't skip extent items representing 0 byte lengths. They
* used to be created (bug) if while punching holes we hit
* -ENOSPC condition. So if we find one here, just ensure we
* delete it, otherwise we would insert a new file extent item
* with the same key (offset) as that 0 bytes length file
* extent item in the call to setup_items_for_insert() later
* in this function.
*/
if (extent_end == key.offset && extent_end >= search_start) {
last_end = extent_end;
Btrfs: fix leaf corruption caused by ENOSPC while hole punching While running a stress test with multiple threads writing to the same btrfs file system, I ended up with a situation where a leaf was corrupted in that it had 2 file extent item keys that had the same exact key. I was able to detect this quickly thanks to the following patch which triggers an assertion as soon as a leaf is marked dirty if there are duplicated keys or out of order keys: Btrfs: check if items are ordered when a leaf is marked dirty (https://patchwork.kernel.org/patch/3955431/) Basically while running the test, I got the following in dmesg: [28877.415877] WARNING: CPU: 2 PID: 10706 at fs/btrfs/file.c:553 btrfs_drop_extent_cache+0x435/0x440 [btrfs]() (...) [28877.415917] Call Trace: [28877.415922] [<ffffffff816f1189>] dump_stack+0x4e/0x68 [28877.415926] [<ffffffff8104a32c>] warn_slowpath_common+0x8c/0xc0 [28877.415929] [<ffffffff8104a37a>] warn_slowpath_null+0x1a/0x20 [28877.415944] [<ffffffffa03775a5>] btrfs_drop_extent_cache+0x435/0x440 [btrfs] [28877.415949] [<ffffffff8118e7be>] ? kmem_cache_alloc+0xfe/0x1c0 [28877.415962] [<ffffffffa03777d9>] fill_holes+0x229/0x3e0 [btrfs] [28877.415972] [<ffffffffa0345865>] ? block_rsv_add_bytes+0x55/0x80 [btrfs] [28877.415984] [<ffffffffa03792cb>] btrfs_fallocate+0xb6b/0xc20 [btrfs] (...) [29854.132560] BTRFS critical (device sdc): corrupt leaf, bad key order: block=955232256,root=1, slot=24 [29854.132565] BTRFS info (device sdc): leaf 955232256 total ptrs 40 free space 778 (...) [29854.132637] item 23 key (3486 108 667648) itemoff 2694 itemsize 53 [29854.132638] extent data disk bytenr 14574411776 nr 286720 [29854.132639] extent data offset 0 nr 286720 ram 286720 [29854.132640] item 24 key (3486 108 954368) itemoff 2641 itemsize 53 [29854.132641] extent data disk bytenr 0 nr 0 [29854.132643] extent data offset 0 nr 0 ram 0 [29854.132644] item 25 key (3486 108 954368) itemoff 2588 itemsize 53 [29854.132645] extent data disk bytenr 8699670528 nr 77824 [29854.132646] extent data offset 0 nr 77824 ram 77824 [29854.132647] item 26 key (3486 108 1146880) itemoff 2535 itemsize 53 [29854.132648] extent data disk bytenr 8699670528 nr 77824 [29854.132649] extent data offset 0 nr 77824 ram 77824 (...) [29854.132707] kernel BUG at fs/btrfs/ctree.h:3901! (...) [29854.132771] Call Trace: [29854.132779] [<ffffffffa0342b5c>] setup_items_for_insert+0x2dc/0x400 [btrfs] [29854.132791] [<ffffffffa0378537>] __btrfs_drop_extents+0xba7/0xdd0 [btrfs] [29854.132794] [<ffffffff8109c0d6>] ? trace_hardirqs_on_caller+0x16/0x1d0 [29854.132797] [<ffffffff8109c29d>] ? trace_hardirqs_on+0xd/0x10 [29854.132800] [<ffffffff8118e7be>] ? kmem_cache_alloc+0xfe/0x1c0 [29854.132810] [<ffffffffa036783b>] insert_reserved_file_extent.constprop.66+0xab/0x310 [btrfs] [29854.132820] [<ffffffffa036a6c6>] __btrfs_prealloc_file_range+0x116/0x340 [btrfs] [29854.132830] [<ffffffffa0374d53>] btrfs_prealloc_file_range+0x23/0x30 [btrfs] (...) So this is caused by getting an -ENOSPC error while punching a file hole, more specifically, we get -ENOSPC error from __btrfs_drop_extents in the while loop of file.c:btrfs_punch_hole() when it's unable to modify the btree to delete one or more file extent items due to lack of enough free space. When this happens, in btrfs_punch_hole(), we attempt to reclaim free space by switching our transaction block reservation object to root->fs_info->trans_block_rsv, end our transaction and start a new transaction basically - and, we keep increasing our current offset (cur_offset) as long as it's smaller than the end of the target range (lockend) - this makes use leave the loop with cur_offset == drop_end which in turn makes us call fill_holes() for inserting a file extent item that represents a 0 bytes range hole (and this insertion succeeds, as in the meanwhile more space became available). This 0 bytes file hole extent item is a problem because any subsequent caller of __btrfs_drop_extents (regular file writes, or fallocate calls for e.g.), with a start file offset that is equal to the offset of the hole, will not remove this extent item due to the following conditional in the while loop of __btrfs_drop_extents: if (extent_end <= search_start) { path->slots[0]++; goto next_slot; } This later makes the call to setup_items_for_insert() (at the very end of __btrfs_drop_extents), insert a new file extent item with the same offset as the 0 bytes file hole extent item that follows it. Needless is to say that this causes chaos, either when reading the leaf from disk (btree_readpage_end_io_hook), where we perform leaf sanity checks or in subsequent operations that manipulate file extent items, as in the fallocate call as shown by the dmesg trace above. Without my other patch to perform the leaf sanity checks once a leaf is marked as dirty (if the integrity checker is enabled), it would have been much harder to debug this issue. This change might fix a few similar issues reported by users in the mailing list regarding assertion failures in btrfs_set_item_key_safe calls performed by __btrfs_drop_extents, such as the following report: http://comments.gmane.org/gmane.comp.file-systems.btrfs/32938 Asking fill_holes() to create a 0 bytes wide file hole item also produced the first warning in the trace above, as we passed a range to btrfs_drop_extent_cache that has an end smaller (by -1) than its start. On 3.14 kernels this issue manifests itself through leaf corruption, as we get duplicated file extent item keys in a leaf when calling setup_items_for_insert(), but on older kernels, setup_items_for_insert() isn't called by __btrfs_drop_extents(), instead we have callers of __btrfs_drop_extents(), namely the functions inode.c:insert_inline_extent() and inode.c:insert_reserved_file_extent(), calling btrfs_insert_empty_item() to insert the new file extent item, which would fail with error -EEXIST, instead of inserting a duplicated key - which is still a serious issue as it would make all similar file extent item replace operations keep failing if they target the same file range. Cc: stable@vger.kernel.org Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-04-29 12:18:40 +00:00
goto delete_extent_item;
}
Btrfs: fix leaf corruption caused by ENOSPC while hole punching While running a stress test with multiple threads writing to the same btrfs file system, I ended up with a situation where a leaf was corrupted in that it had 2 file extent item keys that had the same exact key. I was able to detect this quickly thanks to the following patch which triggers an assertion as soon as a leaf is marked dirty if there are duplicated keys or out of order keys: Btrfs: check if items are ordered when a leaf is marked dirty (https://patchwork.kernel.org/patch/3955431/) Basically while running the test, I got the following in dmesg: [28877.415877] WARNING: CPU: 2 PID: 10706 at fs/btrfs/file.c:553 btrfs_drop_extent_cache+0x435/0x440 [btrfs]() (...) [28877.415917] Call Trace: [28877.415922] [<ffffffff816f1189>] dump_stack+0x4e/0x68 [28877.415926] [<ffffffff8104a32c>] warn_slowpath_common+0x8c/0xc0 [28877.415929] [<ffffffff8104a37a>] warn_slowpath_null+0x1a/0x20 [28877.415944] [<ffffffffa03775a5>] btrfs_drop_extent_cache+0x435/0x440 [btrfs] [28877.415949] [<ffffffff8118e7be>] ? kmem_cache_alloc+0xfe/0x1c0 [28877.415962] [<ffffffffa03777d9>] fill_holes+0x229/0x3e0 [btrfs] [28877.415972] [<ffffffffa0345865>] ? block_rsv_add_bytes+0x55/0x80 [btrfs] [28877.415984] [<ffffffffa03792cb>] btrfs_fallocate+0xb6b/0xc20 [btrfs] (...) [29854.132560] BTRFS critical (device sdc): corrupt leaf, bad key order: block=955232256,root=1, slot=24 [29854.132565] BTRFS info (device sdc): leaf 955232256 total ptrs 40 free space 778 (...) [29854.132637] item 23 key (3486 108 667648) itemoff 2694 itemsize 53 [29854.132638] extent data disk bytenr 14574411776 nr 286720 [29854.132639] extent data offset 0 nr 286720 ram 286720 [29854.132640] item 24 key (3486 108 954368) itemoff 2641 itemsize 53 [29854.132641] extent data disk bytenr 0 nr 0 [29854.132643] extent data offset 0 nr 0 ram 0 [29854.132644] item 25 key (3486 108 954368) itemoff 2588 itemsize 53 [29854.132645] extent data disk bytenr 8699670528 nr 77824 [29854.132646] extent data offset 0 nr 77824 ram 77824 [29854.132647] item 26 key (3486 108 1146880) itemoff 2535 itemsize 53 [29854.132648] extent data disk bytenr 8699670528 nr 77824 [29854.132649] extent data offset 0 nr 77824 ram 77824 (...) [29854.132707] kernel BUG at fs/btrfs/ctree.h:3901! (...) [29854.132771] Call Trace: [29854.132779] [<ffffffffa0342b5c>] setup_items_for_insert+0x2dc/0x400 [btrfs] [29854.132791] [<ffffffffa0378537>] __btrfs_drop_extents+0xba7/0xdd0 [btrfs] [29854.132794] [<ffffffff8109c0d6>] ? trace_hardirqs_on_caller+0x16/0x1d0 [29854.132797] [<ffffffff8109c29d>] ? trace_hardirqs_on+0xd/0x10 [29854.132800] [<ffffffff8118e7be>] ? kmem_cache_alloc+0xfe/0x1c0 [29854.132810] [<ffffffffa036783b>] insert_reserved_file_extent.constprop.66+0xab/0x310 [btrfs] [29854.132820] [<ffffffffa036a6c6>] __btrfs_prealloc_file_range+0x116/0x340 [btrfs] [29854.132830] [<ffffffffa0374d53>] btrfs_prealloc_file_range+0x23/0x30 [btrfs] (...) So this is caused by getting an -ENOSPC error while punching a file hole, more specifically, we get -ENOSPC error from __btrfs_drop_extents in the while loop of file.c:btrfs_punch_hole() when it's unable to modify the btree to delete one or more file extent items due to lack of enough free space. When this happens, in btrfs_punch_hole(), we attempt to reclaim free space by switching our transaction block reservation object to root->fs_info->trans_block_rsv, end our transaction and start a new transaction basically - and, we keep increasing our current offset (cur_offset) as long as it's smaller than the end of the target range (lockend) - this makes use leave the loop with cur_offset == drop_end which in turn makes us call fill_holes() for inserting a file extent item that represents a 0 bytes range hole (and this insertion succeeds, as in the meanwhile more space became available). This 0 bytes file hole extent item is a problem because any subsequent caller of __btrfs_drop_extents (regular file writes, or fallocate calls for e.g.), with a start file offset that is equal to the offset of the hole, will not remove this extent item due to the following conditional in the while loop of __btrfs_drop_extents: if (extent_end <= search_start) { path->slots[0]++; goto next_slot; } This later makes the call to setup_items_for_insert() (at the very end of __btrfs_drop_extents), insert a new file extent item with the same offset as the 0 bytes file hole extent item that follows it. Needless is to say that this causes chaos, either when reading the leaf from disk (btree_readpage_end_io_hook), where we perform leaf sanity checks or in subsequent operations that manipulate file extent items, as in the fallocate call as shown by the dmesg trace above. Without my other patch to perform the leaf sanity checks once a leaf is marked as dirty (if the integrity checker is enabled), it would have been much harder to debug this issue. This change might fix a few similar issues reported by users in the mailing list regarding assertion failures in btrfs_set_item_key_safe calls performed by __btrfs_drop_extents, such as the following report: http://comments.gmane.org/gmane.comp.file-systems.btrfs/32938 Asking fill_holes() to create a 0 bytes wide file hole item also produced the first warning in the trace above, as we passed a range to btrfs_drop_extent_cache that has an end smaller (by -1) than its start. On 3.14 kernels this issue manifests itself through leaf corruption, as we get duplicated file extent item keys in a leaf when calling setup_items_for_insert(), but on older kernels, setup_items_for_insert() isn't called by __btrfs_drop_extents(), instead we have callers of __btrfs_drop_extents(), namely the functions inode.c:insert_inline_extent() and inode.c:insert_reserved_file_extent(), calling btrfs_insert_empty_item() to insert the new file extent item, which would fail with error -EEXIST, instead of inserting a duplicated key - which is still a serious issue as it would make all similar file extent item replace operations keep failing if they target the same file range. Cc: stable@vger.kernel.org Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-04-29 12:18:40 +00:00
if (extent_end <= search_start) {
path->slots[0]++;
goto next_slot;
}
found = 1;
search_start = max(key.offset, start);
if (recow || !modify_tree) {
modify_tree = -1;
btrfs_release_path(path);
continue;
}
/*
* | - range to drop - |
* | -------- extent -------- |
*/
if (start > key.offset && end < extent_end) {
BUG_ON(del_nr > 0);
if (extent_type == BTRFS_FILE_EXTENT_INLINE) {
ret = -EOPNOTSUPP;
break;
}
memcpy(&new_key, &key, sizeof(new_key));
new_key.offset = start;
ret = btrfs_duplicate_item(trans, root, path,
&new_key);
if (ret == -EAGAIN) {
btrfs_release_path(path);
continue;
}
if (ret < 0)
break;
leaf = path->nodes[0];
fi = btrfs_item_ptr(leaf, path->slots[0] - 1,
struct btrfs_file_extent_item);
btrfs_set_file_extent_num_bytes(leaf, fi,
start - key.offset);
fi = btrfs_item_ptr(leaf, path->slots[0],
struct btrfs_file_extent_item);
extent_offset += start - key.offset;
btrfs_set_file_extent_offset(leaf, fi, extent_offset);
btrfs_set_file_extent_num_bytes(leaf, fi,
extent_end - start);
btrfs_mark_buffer_dirty(leaf);
Btrfs: turbo charge fsync At least for the vm workload. Currently on fsync we will 1) Truncate all items in the log tree for the given inode if they exist and 2) Copy all items for a given inode into the log The problem with this is that for things like VMs you can have lots of extents from the fragmented writing behavior, and worst yet you may have only modified a few extents, not the entire thing. This patch fixes this problem by tracking which transid modified our extent, and then when we do the tree logging we find all of the extents we've modified in our current transaction, sort them and commit them. We also only truncate up to the xattrs of the inode and copy that stuff in normally, and then just drop any extents in the range we have that exist in the log already. Here are some numbers of a 50 meg fio job that does random writes and fsync()s after every write Original Patched SATA drive 82KB/s 140KB/s Fusion drive 431KB/s 2532KB/s So around 2-6 times faster depending on your hardware. There are a few corner cases, for example if you truncate at all we have to do it the old way since there is no way to be sure what is in the log is ok. This probably could be done smarter, but if you write-fsync-truncate-write-fsync you deserve what you get. All this work is in RAM of course so if your inode gets evicted from cache and you read it in and fsync it we'll do it the slow way if we are still in the same transaction that we last modified the inode in. The biggest cool part of this is that it requires no changes to the recovery code, so if you fsync with this patch and crash and load an old kernel, it will run the recovery and be a-ok. I have tested this pretty thoroughly with an fsync tester and everything comes back fine, as well as xfstests. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2012-08-17 17:14:17 +00:00
if (update_refs && disk_bytenr > 0) {
ret = btrfs_inc_extent_ref(trans, fs_info,
disk_bytenr, num_bytes, 0,
root->root_key.objectid,
new_key.objectid,
Btrfs: fix regression running delayed references when using qgroups In the kernel 4.2 merge window we had a big changes to the implementation of delayed references and qgroups which made the no_quota field of delayed references not used anymore. More specifically the no_quota field is not used anymore as of: commit 0ed4792af0e8 ("btrfs: qgroup: Switch to new extent-oriented qgroup mechanism.") Leaving the no_quota field actually prevents delayed references from getting merged, which in turn cause the following BUG_ON(), at fs/btrfs/extent-tree.c, to be hit when qgroups are enabled: static int run_delayed_tree_ref(...) { (...) BUG_ON(node->ref_mod != 1); (...) } This happens on a scenario like the following: 1) Ref1 bytenr X, action = BTRFS_ADD_DELAYED_REF, no_quota = 1, added. 2) Ref2 bytenr X, action = BTRFS_DROP_DELAYED_REF, no_quota = 0, added. It's not merged with Ref1 because Ref1->no_quota != Ref2->no_quota. 3) Ref3 bytenr X, action = BTRFS_ADD_DELAYED_REF, no_quota = 1, added. It's not merged with the reference at the tail of the list of refs for bytenr X because the reference at the tail, Ref2 is incompatible due to Ref2->no_quota != Ref3->no_quota. 4) Ref4 bytenr X, action = BTRFS_DROP_DELAYED_REF, no_quota = 0, added. It's not merged with the reference at the tail of the list of refs for bytenr X because the reference at the tail, Ref3 is incompatible due to Ref3->no_quota != Ref4->no_quota. 5) We run delayed references, trigger merging of delayed references, through __btrfs_run_delayed_refs() -> btrfs_merge_delayed_refs(). 6) Ref1 and Ref3 are merged as Ref1->no_quota = Ref3->no_quota and all other conditions are satisfied too. So Ref1 gets a ref_mod value of 2. 7) Ref2 and Ref4 are merged as Ref2->no_quota = Ref4->no_quota and all other conditions are satisfied too. So Ref2 gets a ref_mod value of 2. 8) Ref1 and Ref2 aren't merged, because they have different values for their no_quota field. 9) Delayed reference Ref1 is picked for running (select_delayed_ref() always prefers references with an action == BTRFS_ADD_DELAYED_REF). So run_delayed_tree_ref() is called for Ref1 which triggers the BUG_ON because Ref1->red_mod != 1 (equals 2). So fix this by removing the no_quota field, as it's not used anymore as of commit 0ed4792af0e8 ("btrfs: qgroup: Switch to new extent-oriented qgroup mechanism."). The use of no_quota was also buggy in at least two places: 1) At delayed-refs.c:btrfs_add_delayed_tree_ref() - we were setting no_quota to 0 instead of 1 when the following condition was true: is_fstree(ref_root) || !fs_info->quota_enabled 2) At extent-tree.c:__btrfs_inc_extent_ref() - we were attempting to reset a node's no_quota when the condition "!is_fstree(root_objectid) || !root->fs_info->quota_enabled" was true but we did it only in an unused local stack variable, that is, we never reset the no_quota value in the node itself. This fixes the remainder of problems several people have been having when running delayed references, mostly while a balance is running in parallel, on a 4.2+ kernel. Very special thanks to Stéphane Lesimple for helping debugging this issue and testing this fix on his multi terabyte filesystem (which took more than one day to balance alone, plus fsck, etc). Also, this fixes deadlock issue when using the clone ioctl with qgroups enabled, as reported by Elias Probst in the mailing list. The deadlock happens because after calling btrfs_insert_empty_item we have our path holding a write lock on a leaf of the fs/subvol tree and then before releasing the path we called check_ref() which did backref walking, when qgroups are enabled, and tried to read lock the same leaf. The trace for this case is the following: INFO: task systemd-nspawn:6095 blocked for more than 120 seconds. (...) Call Trace: [<ffffffff86999201>] schedule+0x74/0x83 [<ffffffff863ef64c>] btrfs_tree_read_lock+0xc0/0xea [<ffffffff86137ed7>] ? wait_woken+0x74/0x74 [<ffffffff8639f0a7>] btrfs_search_old_slot+0x51a/0x810 [<ffffffff863a129b>] btrfs_next_old_leaf+0xdf/0x3ce [<ffffffff86413a00>] ? ulist_add_merge+0x1b/0x127 [<ffffffff86411688>] __resolve_indirect_refs+0x62a/0x667 [<ffffffff863ef546>] ? btrfs_clear_lock_blocking_rw+0x78/0xbe [<ffffffff864122d3>] find_parent_nodes+0xaf3/0xfc6 [<ffffffff86412838>] __btrfs_find_all_roots+0x92/0xf0 [<ffffffff864128f2>] btrfs_find_all_roots+0x45/0x65 [<ffffffff8639a75b>] ? btrfs_get_tree_mod_seq+0x2b/0x88 [<ffffffff863e852e>] check_ref+0x64/0xc4 [<ffffffff863e9e01>] btrfs_clone+0x66e/0xb5d [<ffffffff863ea77f>] btrfs_ioctl_clone+0x48f/0x5bb [<ffffffff86048a68>] ? native_sched_clock+0x28/0x77 [<ffffffff863ed9b0>] btrfs_ioctl+0xabc/0x25cb (...) The problem goes away by eleminating check_ref(), which no longer is needed as its purpose was to get a value for the no_quota field of a delayed reference (this patch removes the no_quota field as mentioned earlier). Reported-by: Stéphane Lesimple <stephane_btrfs@lesimple.fr> Tested-by: Stéphane Lesimple <stephane_btrfs@lesimple.fr> Reported-by: Elias Probst <mail@eliasprobst.eu> Reported-by: Peter Becker <floyd.net@gmail.com> Reported-by: Malte Schröder <malte@tnxip.de> Reported-by: Derek Dongray <derek@valedon.co.uk> Reported-by: Erkki Seppala <flux-btrfs@inside.org> Cc: stable@vger.kernel.org # 4.2+ Signed-off-by: Filipe Manana <fdmanana@suse.com> Reviewed-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
2015-10-23 06:52:54 +00:00
start - extent_offset);
BUG_ON(ret); /* -ENOMEM */
}
key.offset = start;
}
/*
* From here on out we will have actually dropped something, so
* last_end can be updated.
*/
last_end = extent_end;
/*
* | ---- range to drop ----- |
* | -------- extent -------- |
*/
if (start <= key.offset && end < extent_end) {
if (extent_type == BTRFS_FILE_EXTENT_INLINE) {
ret = -EOPNOTSUPP;
break;
}
memcpy(&new_key, &key, sizeof(new_key));
new_key.offset = end;
btrfs_set_item_key_safe(fs_info, path, &new_key);
extent_offset += end - key.offset;
btrfs_set_file_extent_offset(leaf, fi, extent_offset);
btrfs_set_file_extent_num_bytes(leaf, fi,
extent_end - end);
btrfs_mark_buffer_dirty(leaf);
if (update_refs && disk_bytenr > 0)
inode_sub_bytes(inode, end - key.offset);
break;
}
search_start = extent_end;
/*
* | ---- range to drop ----- |
* | -------- extent -------- |
*/
if (start > key.offset && end >= extent_end) {
BUG_ON(del_nr > 0);
if (extent_type == BTRFS_FILE_EXTENT_INLINE) {
ret = -EOPNOTSUPP;
break;
}
btrfs_set_file_extent_num_bytes(leaf, fi,
start - key.offset);
btrfs_mark_buffer_dirty(leaf);
if (update_refs && disk_bytenr > 0)
inode_sub_bytes(inode, extent_end - start);
if (end == extent_end)
break;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
path->slots[0]++;
goto next_slot;
}
/*
* | ---- range to drop ----- |
* | ------ extent ------ |
*/
if (start <= key.offset && end >= extent_end) {
Btrfs: fix leaf corruption caused by ENOSPC while hole punching While running a stress test with multiple threads writing to the same btrfs file system, I ended up with a situation where a leaf was corrupted in that it had 2 file extent item keys that had the same exact key. I was able to detect this quickly thanks to the following patch which triggers an assertion as soon as a leaf is marked dirty if there are duplicated keys or out of order keys: Btrfs: check if items are ordered when a leaf is marked dirty (https://patchwork.kernel.org/patch/3955431/) Basically while running the test, I got the following in dmesg: [28877.415877] WARNING: CPU: 2 PID: 10706 at fs/btrfs/file.c:553 btrfs_drop_extent_cache+0x435/0x440 [btrfs]() (...) [28877.415917] Call Trace: [28877.415922] [<ffffffff816f1189>] dump_stack+0x4e/0x68 [28877.415926] [<ffffffff8104a32c>] warn_slowpath_common+0x8c/0xc0 [28877.415929] [<ffffffff8104a37a>] warn_slowpath_null+0x1a/0x20 [28877.415944] [<ffffffffa03775a5>] btrfs_drop_extent_cache+0x435/0x440 [btrfs] [28877.415949] [<ffffffff8118e7be>] ? kmem_cache_alloc+0xfe/0x1c0 [28877.415962] [<ffffffffa03777d9>] fill_holes+0x229/0x3e0 [btrfs] [28877.415972] [<ffffffffa0345865>] ? block_rsv_add_bytes+0x55/0x80 [btrfs] [28877.415984] [<ffffffffa03792cb>] btrfs_fallocate+0xb6b/0xc20 [btrfs] (...) [29854.132560] BTRFS critical (device sdc): corrupt leaf, bad key order: block=955232256,root=1, slot=24 [29854.132565] BTRFS info (device sdc): leaf 955232256 total ptrs 40 free space 778 (...) [29854.132637] item 23 key (3486 108 667648) itemoff 2694 itemsize 53 [29854.132638] extent data disk bytenr 14574411776 nr 286720 [29854.132639] extent data offset 0 nr 286720 ram 286720 [29854.132640] item 24 key (3486 108 954368) itemoff 2641 itemsize 53 [29854.132641] extent data disk bytenr 0 nr 0 [29854.132643] extent data offset 0 nr 0 ram 0 [29854.132644] item 25 key (3486 108 954368) itemoff 2588 itemsize 53 [29854.132645] extent data disk bytenr 8699670528 nr 77824 [29854.132646] extent data offset 0 nr 77824 ram 77824 [29854.132647] item 26 key (3486 108 1146880) itemoff 2535 itemsize 53 [29854.132648] extent data disk bytenr 8699670528 nr 77824 [29854.132649] extent data offset 0 nr 77824 ram 77824 (...) [29854.132707] kernel BUG at fs/btrfs/ctree.h:3901! (...) [29854.132771] Call Trace: [29854.132779] [<ffffffffa0342b5c>] setup_items_for_insert+0x2dc/0x400 [btrfs] [29854.132791] [<ffffffffa0378537>] __btrfs_drop_extents+0xba7/0xdd0 [btrfs] [29854.132794] [<ffffffff8109c0d6>] ? trace_hardirqs_on_caller+0x16/0x1d0 [29854.132797] [<ffffffff8109c29d>] ? trace_hardirqs_on+0xd/0x10 [29854.132800] [<ffffffff8118e7be>] ? kmem_cache_alloc+0xfe/0x1c0 [29854.132810] [<ffffffffa036783b>] insert_reserved_file_extent.constprop.66+0xab/0x310 [btrfs] [29854.132820] [<ffffffffa036a6c6>] __btrfs_prealloc_file_range+0x116/0x340 [btrfs] [29854.132830] [<ffffffffa0374d53>] btrfs_prealloc_file_range+0x23/0x30 [btrfs] (...) So this is caused by getting an -ENOSPC error while punching a file hole, more specifically, we get -ENOSPC error from __btrfs_drop_extents in the while loop of file.c:btrfs_punch_hole() when it's unable to modify the btree to delete one or more file extent items due to lack of enough free space. When this happens, in btrfs_punch_hole(), we attempt to reclaim free space by switching our transaction block reservation object to root->fs_info->trans_block_rsv, end our transaction and start a new transaction basically - and, we keep increasing our current offset (cur_offset) as long as it's smaller than the end of the target range (lockend) - this makes use leave the loop with cur_offset == drop_end which in turn makes us call fill_holes() for inserting a file extent item that represents a 0 bytes range hole (and this insertion succeeds, as in the meanwhile more space became available). This 0 bytes file hole extent item is a problem because any subsequent caller of __btrfs_drop_extents (regular file writes, or fallocate calls for e.g.), with a start file offset that is equal to the offset of the hole, will not remove this extent item due to the following conditional in the while loop of __btrfs_drop_extents: if (extent_end <= search_start) { path->slots[0]++; goto next_slot; } This later makes the call to setup_items_for_insert() (at the very end of __btrfs_drop_extents), insert a new file extent item with the same offset as the 0 bytes file hole extent item that follows it. Needless is to say that this causes chaos, either when reading the leaf from disk (btree_readpage_end_io_hook), where we perform leaf sanity checks or in subsequent operations that manipulate file extent items, as in the fallocate call as shown by the dmesg trace above. Without my other patch to perform the leaf sanity checks once a leaf is marked as dirty (if the integrity checker is enabled), it would have been much harder to debug this issue. This change might fix a few similar issues reported by users in the mailing list regarding assertion failures in btrfs_set_item_key_safe calls performed by __btrfs_drop_extents, such as the following report: http://comments.gmane.org/gmane.comp.file-systems.btrfs/32938 Asking fill_holes() to create a 0 bytes wide file hole item also produced the first warning in the trace above, as we passed a range to btrfs_drop_extent_cache that has an end smaller (by -1) than its start. On 3.14 kernels this issue manifests itself through leaf corruption, as we get duplicated file extent item keys in a leaf when calling setup_items_for_insert(), but on older kernels, setup_items_for_insert() isn't called by __btrfs_drop_extents(), instead we have callers of __btrfs_drop_extents(), namely the functions inode.c:insert_inline_extent() and inode.c:insert_reserved_file_extent(), calling btrfs_insert_empty_item() to insert the new file extent item, which would fail with error -EEXIST, instead of inserting a duplicated key - which is still a serious issue as it would make all similar file extent item replace operations keep failing if they target the same file range. Cc: stable@vger.kernel.org Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-04-29 12:18:40 +00:00
delete_extent_item:
if (del_nr == 0) {
del_slot = path->slots[0];
del_nr = 1;
} else {
BUG_ON(del_slot + del_nr != path->slots[0]);
del_nr++;
}
Btrfs: turbo charge fsync At least for the vm workload. Currently on fsync we will 1) Truncate all items in the log tree for the given inode if they exist and 2) Copy all items for a given inode into the log The problem with this is that for things like VMs you can have lots of extents from the fragmented writing behavior, and worst yet you may have only modified a few extents, not the entire thing. This patch fixes this problem by tracking which transid modified our extent, and then when we do the tree logging we find all of the extents we've modified in our current transaction, sort them and commit them. We also only truncate up to the xattrs of the inode and copy that stuff in normally, and then just drop any extents in the range we have that exist in the log already. Here are some numbers of a 50 meg fio job that does random writes and fsync()s after every write Original Patched SATA drive 82KB/s 140KB/s Fusion drive 431KB/s 2532KB/s So around 2-6 times faster depending on your hardware. There are a few corner cases, for example if you truncate at all we have to do it the old way since there is no way to be sure what is in the log is ok. This probably could be done smarter, but if you write-fsync-truncate-write-fsync you deserve what you get. All this work is in RAM of course so if your inode gets evicted from cache and you read it in and fsync it we'll do it the slow way if we are still in the same transaction that we last modified the inode in. The biggest cool part of this is that it requires no changes to the recovery code, so if you fsync with this patch and crash and load an old kernel, it will run the recovery and be a-ok. I have tested this pretty thoroughly with an fsync tester and everything comes back fine, as well as xfstests. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2012-08-17 17:14:17 +00:00
if (update_refs &&
extent_type == BTRFS_FILE_EXTENT_INLINE) {
inode_sub_bytes(inode,
extent_end - key.offset);
extent_end = ALIGN(extent_end,
fs_info->sectorsize);
Btrfs: turbo charge fsync At least for the vm workload. Currently on fsync we will 1) Truncate all items in the log tree for the given inode if they exist and 2) Copy all items for a given inode into the log The problem with this is that for things like VMs you can have lots of extents from the fragmented writing behavior, and worst yet you may have only modified a few extents, not the entire thing. This patch fixes this problem by tracking which transid modified our extent, and then when we do the tree logging we find all of the extents we've modified in our current transaction, sort them and commit them. We also only truncate up to the xattrs of the inode and copy that stuff in normally, and then just drop any extents in the range we have that exist in the log already. Here are some numbers of a 50 meg fio job that does random writes and fsync()s after every write Original Patched SATA drive 82KB/s 140KB/s Fusion drive 431KB/s 2532KB/s So around 2-6 times faster depending on your hardware. There are a few corner cases, for example if you truncate at all we have to do it the old way since there is no way to be sure what is in the log is ok. This probably could be done smarter, but if you write-fsync-truncate-write-fsync you deserve what you get. All this work is in RAM of course so if your inode gets evicted from cache and you read it in and fsync it we'll do it the slow way if we are still in the same transaction that we last modified the inode in. The biggest cool part of this is that it requires no changes to the recovery code, so if you fsync with this patch and crash and load an old kernel, it will run the recovery and be a-ok. I have tested this pretty thoroughly with an fsync tester and everything comes back fine, as well as xfstests. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2012-08-17 17:14:17 +00:00
} else if (update_refs && disk_bytenr > 0) {
ret = btrfs_free_extent(trans, fs_info,
disk_bytenr, num_bytes, 0,
root->root_key.objectid,
Btrfs: Mixed back reference (FORWARD ROLLING FORMAT CHANGE) This commit introduces a new kind of back reference for btrfs metadata. Once a filesystem has been mounted with this commit, IT WILL NO LONGER BE MOUNTABLE BY OLDER KERNELS. When a tree block in subvolume tree is cow'd, the reference counts of all extents it points to are increased by one. At transaction commit time, the old root of the subvolume is recorded in a "dead root" data structure, and the btree it points to is later walked, dropping reference counts and freeing any blocks where the reference count goes to 0. The increments done during cow and decrements done after commit cancel out, and the walk is a very expensive way to go about freeing the blocks that are no longer referenced by the new btree root. This commit reduces the transaction overhead by avoiding the need for dead root records. When a non-shared tree block is cow'd, we free the old block at once, and the new block inherits old block's references. When a tree block with reference count > 1 is cow'd, we increase the reference counts of all extents the new block points to by one, and decrease the old block's reference count by one. This dead tree avoidance code removes the need to modify the reference counts of lower level extents when a non-shared tree block is cow'd. But we still need to update back ref for all pointers in the block. This is because the location of the block is recorded in the back ref item. We can solve this by introducing a new type of back ref. The new back ref provides information about pointer's key, level and in which tree the pointer lives. This information allow us to find the pointer by searching the tree. The shortcoming of the new back ref is that it only works for pointers in tree blocks referenced by their owner trees. This is mostly a problem for snapshots, where resolving one of these fuzzy back references would be O(number_of_snapshots) and quite slow. The solution used here is to use the fuzzy back references in the common case where a given tree block is only referenced by one root, and use the full back references when multiple roots have a reference on a given block. This commit adds per subvolume red-black tree to keep trace of cached inodes. The red-black tree helps the balancing code to find cached inodes whose inode numbers within a given range. This commit improves the balancing code by introducing several data structures to keep the state of balancing. The most important one is the back ref cache. It caches how the upper level tree blocks are referenced. This greatly reduce the overhead of checking back ref. The improved balancing code scales significantly better with a large number of snapshots. This is a very large commit and was written in a number of pieces. But, they depend heavily on the disk format change and were squashed together to make sure git bisect didn't end up in a bad state wrt space balancing or the format change. Signed-off-by: Yan Zheng <zheng.yan@oracle.com> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2009-06-10 14:45:14 +00:00
key.objectid, key.offset -
Btrfs: fix regression running delayed references when using qgroups In the kernel 4.2 merge window we had a big changes to the implementation of delayed references and qgroups which made the no_quota field of delayed references not used anymore. More specifically the no_quota field is not used anymore as of: commit 0ed4792af0e8 ("btrfs: qgroup: Switch to new extent-oriented qgroup mechanism.") Leaving the no_quota field actually prevents delayed references from getting merged, which in turn cause the following BUG_ON(), at fs/btrfs/extent-tree.c, to be hit when qgroups are enabled: static int run_delayed_tree_ref(...) { (...) BUG_ON(node->ref_mod != 1); (...) } This happens on a scenario like the following: 1) Ref1 bytenr X, action = BTRFS_ADD_DELAYED_REF, no_quota = 1, added. 2) Ref2 bytenr X, action = BTRFS_DROP_DELAYED_REF, no_quota = 0, added. It's not merged with Ref1 because Ref1->no_quota != Ref2->no_quota. 3) Ref3 bytenr X, action = BTRFS_ADD_DELAYED_REF, no_quota = 1, added. It's not merged with the reference at the tail of the list of refs for bytenr X because the reference at the tail, Ref2 is incompatible due to Ref2->no_quota != Ref3->no_quota. 4) Ref4 bytenr X, action = BTRFS_DROP_DELAYED_REF, no_quota = 0, added. It's not merged with the reference at the tail of the list of refs for bytenr X because the reference at the tail, Ref3 is incompatible due to Ref3->no_quota != Ref4->no_quota. 5) We run delayed references, trigger merging of delayed references, through __btrfs_run_delayed_refs() -> btrfs_merge_delayed_refs(). 6) Ref1 and Ref3 are merged as Ref1->no_quota = Ref3->no_quota and all other conditions are satisfied too. So Ref1 gets a ref_mod value of 2. 7) Ref2 and Ref4 are merged as Ref2->no_quota = Ref4->no_quota and all other conditions are satisfied too. So Ref2 gets a ref_mod value of 2. 8) Ref1 and Ref2 aren't merged, because they have different values for their no_quota field. 9) Delayed reference Ref1 is picked for running (select_delayed_ref() always prefers references with an action == BTRFS_ADD_DELAYED_REF). So run_delayed_tree_ref() is called for Ref1 which triggers the BUG_ON because Ref1->red_mod != 1 (equals 2). So fix this by removing the no_quota field, as it's not used anymore as of commit 0ed4792af0e8 ("btrfs: qgroup: Switch to new extent-oriented qgroup mechanism."). The use of no_quota was also buggy in at least two places: 1) At delayed-refs.c:btrfs_add_delayed_tree_ref() - we were setting no_quota to 0 instead of 1 when the following condition was true: is_fstree(ref_root) || !fs_info->quota_enabled 2) At extent-tree.c:__btrfs_inc_extent_ref() - we were attempting to reset a node's no_quota when the condition "!is_fstree(root_objectid) || !root->fs_info->quota_enabled" was true but we did it only in an unused local stack variable, that is, we never reset the no_quota value in the node itself. This fixes the remainder of problems several people have been having when running delayed references, mostly while a balance is running in parallel, on a 4.2+ kernel. Very special thanks to Stéphane Lesimple for helping debugging this issue and testing this fix on his multi terabyte filesystem (which took more than one day to balance alone, plus fsck, etc). Also, this fixes deadlock issue when using the clone ioctl with qgroups enabled, as reported by Elias Probst in the mailing list. The deadlock happens because after calling btrfs_insert_empty_item we have our path holding a write lock on a leaf of the fs/subvol tree and then before releasing the path we called check_ref() which did backref walking, when qgroups are enabled, and tried to read lock the same leaf. The trace for this case is the following: INFO: task systemd-nspawn:6095 blocked for more than 120 seconds. (...) Call Trace: [<ffffffff86999201>] schedule+0x74/0x83 [<ffffffff863ef64c>] btrfs_tree_read_lock+0xc0/0xea [<ffffffff86137ed7>] ? wait_woken+0x74/0x74 [<ffffffff8639f0a7>] btrfs_search_old_slot+0x51a/0x810 [<ffffffff863a129b>] btrfs_next_old_leaf+0xdf/0x3ce [<ffffffff86413a00>] ? ulist_add_merge+0x1b/0x127 [<ffffffff86411688>] __resolve_indirect_refs+0x62a/0x667 [<ffffffff863ef546>] ? btrfs_clear_lock_blocking_rw+0x78/0xbe [<ffffffff864122d3>] find_parent_nodes+0xaf3/0xfc6 [<ffffffff86412838>] __btrfs_find_all_roots+0x92/0xf0 [<ffffffff864128f2>] btrfs_find_all_roots+0x45/0x65 [<ffffffff8639a75b>] ? btrfs_get_tree_mod_seq+0x2b/0x88 [<ffffffff863e852e>] check_ref+0x64/0xc4 [<ffffffff863e9e01>] btrfs_clone+0x66e/0xb5d [<ffffffff863ea77f>] btrfs_ioctl_clone+0x48f/0x5bb [<ffffffff86048a68>] ? native_sched_clock+0x28/0x77 [<ffffffff863ed9b0>] btrfs_ioctl+0xabc/0x25cb (...) The problem goes away by eleminating check_ref(), which no longer is needed as its purpose was to get a value for the no_quota field of a delayed reference (this patch removes the no_quota field as mentioned earlier). Reported-by: Stéphane Lesimple <stephane_btrfs@lesimple.fr> Tested-by: Stéphane Lesimple <stephane_btrfs@lesimple.fr> Reported-by: Elias Probst <mail@eliasprobst.eu> Reported-by: Peter Becker <floyd.net@gmail.com> Reported-by: Malte Schröder <malte@tnxip.de> Reported-by: Derek Dongray <derek@valedon.co.uk> Reported-by: Erkki Seppala <flux-btrfs@inside.org> Cc: stable@vger.kernel.org # 4.2+ Signed-off-by: Filipe Manana <fdmanana@suse.com> Reviewed-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
2015-10-23 06:52:54 +00:00
extent_offset);
BUG_ON(ret); /* -ENOMEM */
inode_sub_bytes(inode,
extent_end - key.offset);
}
if (end == extent_end)
break;
if (path->slots[0] + 1 < btrfs_header_nritems(leaf)) {
path->slots[0]++;
goto next_slot;
}
ret = btrfs_del_items(trans, root, path, del_slot,
del_nr);
if (ret) {
btrfs_abort_transaction(trans, ret);
Btrfs: turbo charge fsync At least for the vm workload. Currently on fsync we will 1) Truncate all items in the log tree for the given inode if they exist and 2) Copy all items for a given inode into the log The problem with this is that for things like VMs you can have lots of extents from the fragmented writing behavior, and worst yet you may have only modified a few extents, not the entire thing. This patch fixes this problem by tracking which transid modified our extent, and then when we do the tree logging we find all of the extents we've modified in our current transaction, sort them and commit them. We also only truncate up to the xattrs of the inode and copy that stuff in normally, and then just drop any extents in the range we have that exist in the log already. Here are some numbers of a 50 meg fio job that does random writes and fsync()s after every write Original Patched SATA drive 82KB/s 140KB/s Fusion drive 431KB/s 2532KB/s So around 2-6 times faster depending on your hardware. There are a few corner cases, for example if you truncate at all we have to do it the old way since there is no way to be sure what is in the log is ok. This probably could be done smarter, but if you write-fsync-truncate-write-fsync you deserve what you get. All this work is in RAM of course so if your inode gets evicted from cache and you read it in and fsync it we'll do it the slow way if we are still in the same transaction that we last modified the inode in. The biggest cool part of this is that it requires no changes to the recovery code, so if you fsync with this patch and crash and load an old kernel, it will run the recovery and be a-ok. I have tested this pretty thoroughly with an fsync tester and everything comes back fine, as well as xfstests. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2012-08-17 17:14:17 +00:00
break;
}
del_nr = 0;
del_slot = 0;
btrfs_release_path(path);
continue;
}
BUG_ON(1);
}
if (!ret && del_nr > 0) {
Btrfs: faster file extent item replace operations When writing to a file we drop existing file extent items that cover the write range and then add a new file extent item that represents that write range. Before this change we were doing a tree lookup to remove the file extent items, and then after we did another tree lookup to insert the new file extent item. Most of the time all the file extent items we need to drop are located within a single leaf - this is the leaf where our new file extent item ends up at. Therefore, in this common case just combine these 2 operations into a single one. By avoiding the second btree navigation for insertion of the new file extent item, we reduce btree node/leaf lock acquisitions/releases, btree block/leaf COW operations, CPU time on btree node/leaf key binary searches, etc. Besides for file writes, this is an operation that happens for file fsync's as well. However log btrees are much less likely to big as big as regular fs btrees, therefore the impact of this change is smaller. The following benchmark was performed against an SSD drive and a HDD drive, both for random and sequential writes: sysbench --test=fileio --file-num=4096 --file-total-size=8G \ --file-test-mode=[rndwr|seqwr] --num-threads=512 \ --file-block-size=8192 \ --max-requests=1000000 \ --file-fsync-freq=0 --file-io-mode=sync [prepare|run] All results below are averages of 10 runs of the respective test. ** SSD sequential writes Before this change: 225.88 Mb/sec After this change: 277.26 Mb/sec ** SSD random writes Before this change: 49.91 Mb/sec After this change: 56.39 Mb/sec ** HDD sequential writes Before this change: 68.53 Mb/sec After this change: 69.87 Mb/sec ** HDD random writes Before this change: 13.04 Mb/sec After this change: 14.39 Mb/sec Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-01-07 11:42:27 +00:00
/*
* Set path->slots[0] to first slot, so that after the delete
* if items are move off from our leaf to its immediate left or
* right neighbor leafs, we end up with a correct and adjusted
* path->slots[0] for our insertion (if replace_extent != 0).
Btrfs: faster file extent item replace operations When writing to a file we drop existing file extent items that cover the write range and then add a new file extent item that represents that write range. Before this change we were doing a tree lookup to remove the file extent items, and then after we did another tree lookup to insert the new file extent item. Most of the time all the file extent items we need to drop are located within a single leaf - this is the leaf where our new file extent item ends up at. Therefore, in this common case just combine these 2 operations into a single one. By avoiding the second btree navigation for insertion of the new file extent item, we reduce btree node/leaf lock acquisitions/releases, btree block/leaf COW operations, CPU time on btree node/leaf key binary searches, etc. Besides for file writes, this is an operation that happens for file fsync's as well. However log btrees are much less likely to big as big as regular fs btrees, therefore the impact of this change is smaller. The following benchmark was performed against an SSD drive and a HDD drive, both for random and sequential writes: sysbench --test=fileio --file-num=4096 --file-total-size=8G \ --file-test-mode=[rndwr|seqwr] --num-threads=512 \ --file-block-size=8192 \ --max-requests=1000000 \ --file-fsync-freq=0 --file-io-mode=sync [prepare|run] All results below are averages of 10 runs of the respective test. ** SSD sequential writes Before this change: 225.88 Mb/sec After this change: 277.26 Mb/sec ** SSD random writes Before this change: 49.91 Mb/sec After this change: 56.39 Mb/sec ** HDD sequential writes Before this change: 68.53 Mb/sec After this change: 69.87 Mb/sec ** HDD random writes Before this change: 13.04 Mb/sec After this change: 14.39 Mb/sec Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-01-07 11:42:27 +00:00
*/
path->slots[0] = del_slot;
ret = btrfs_del_items(trans, root, path, del_slot, del_nr);
if (ret)
btrfs_abort_transaction(trans, ret);
}
Btrfs: faster file extent item replace operations When writing to a file we drop existing file extent items that cover the write range and then add a new file extent item that represents that write range. Before this change we were doing a tree lookup to remove the file extent items, and then after we did another tree lookup to insert the new file extent item. Most of the time all the file extent items we need to drop are located within a single leaf - this is the leaf where our new file extent item ends up at. Therefore, in this common case just combine these 2 operations into a single one. By avoiding the second btree navigation for insertion of the new file extent item, we reduce btree node/leaf lock acquisitions/releases, btree block/leaf COW operations, CPU time on btree node/leaf key binary searches, etc. Besides for file writes, this is an operation that happens for file fsync's as well. However log btrees are much less likely to big as big as regular fs btrees, therefore the impact of this change is smaller. The following benchmark was performed against an SSD drive and a HDD drive, both for random and sequential writes: sysbench --test=fileio --file-num=4096 --file-total-size=8G \ --file-test-mode=[rndwr|seqwr] --num-threads=512 \ --file-block-size=8192 \ --max-requests=1000000 \ --file-fsync-freq=0 --file-io-mode=sync [prepare|run] All results below are averages of 10 runs of the respective test. ** SSD sequential writes Before this change: 225.88 Mb/sec After this change: 277.26 Mb/sec ** SSD random writes Before this change: 49.91 Mb/sec After this change: 56.39 Mb/sec ** HDD sequential writes Before this change: 68.53 Mb/sec After this change: 69.87 Mb/sec ** HDD random writes Before this change: 13.04 Mb/sec After this change: 14.39 Mb/sec Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-01-07 11:42:27 +00:00
leaf = path->nodes[0];
/*
* If btrfs_del_items() was called, it might have deleted a leaf, in
* which case it unlocked our path, so check path->locks[0] matches a
* write lock.
*/
if (!ret && replace_extent && leafs_visited == 1 &&
(path->locks[0] == BTRFS_WRITE_LOCK_BLOCKING ||
path->locks[0] == BTRFS_WRITE_LOCK) &&
btrfs_leaf_free_space(fs_info, leaf) >=
sizeof(struct btrfs_item) + extent_item_size) {
key.objectid = ino;
key.type = BTRFS_EXTENT_DATA_KEY;
key.offset = start;
if (!del_nr && path->slots[0] < btrfs_header_nritems(leaf)) {
struct btrfs_key slot_key;
btrfs_item_key_to_cpu(leaf, &slot_key, path->slots[0]);
if (btrfs_comp_cpu_keys(&key, &slot_key) > 0)
path->slots[0]++;
Btrfs: faster file extent item replace operations When writing to a file we drop existing file extent items that cover the write range and then add a new file extent item that represents that write range. Before this change we were doing a tree lookup to remove the file extent items, and then after we did another tree lookup to insert the new file extent item. Most of the time all the file extent items we need to drop are located within a single leaf - this is the leaf where our new file extent item ends up at. Therefore, in this common case just combine these 2 operations into a single one. By avoiding the second btree navigation for insertion of the new file extent item, we reduce btree node/leaf lock acquisitions/releases, btree block/leaf COW operations, CPU time on btree node/leaf key binary searches, etc. Besides for file writes, this is an operation that happens for file fsync's as well. However log btrees are much less likely to big as big as regular fs btrees, therefore the impact of this change is smaller. The following benchmark was performed against an SSD drive and a HDD drive, both for random and sequential writes: sysbench --test=fileio --file-num=4096 --file-total-size=8G \ --file-test-mode=[rndwr|seqwr] --num-threads=512 \ --file-block-size=8192 \ --max-requests=1000000 \ --file-fsync-freq=0 --file-io-mode=sync [prepare|run] All results below are averages of 10 runs of the respective test. ** SSD sequential writes Before this change: 225.88 Mb/sec After this change: 277.26 Mb/sec ** SSD random writes Before this change: 49.91 Mb/sec After this change: 56.39 Mb/sec ** HDD sequential writes Before this change: 68.53 Mb/sec After this change: 69.87 Mb/sec ** HDD random writes Before this change: 13.04 Mb/sec After this change: 14.39 Mb/sec Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-01-07 11:42:27 +00:00
}
setup_items_for_insert(root, path, &key,
&extent_item_size,
extent_item_size,
sizeof(struct btrfs_item) +
extent_item_size, 1);
*key_inserted = 1;
}
Btrfs: faster file extent item replace operations When writing to a file we drop existing file extent items that cover the write range and then add a new file extent item that represents that write range. Before this change we were doing a tree lookup to remove the file extent items, and then after we did another tree lookup to insert the new file extent item. Most of the time all the file extent items we need to drop are located within a single leaf - this is the leaf where our new file extent item ends up at. Therefore, in this common case just combine these 2 operations into a single one. By avoiding the second btree navigation for insertion of the new file extent item, we reduce btree node/leaf lock acquisitions/releases, btree block/leaf COW operations, CPU time on btree node/leaf key binary searches, etc. Besides for file writes, this is an operation that happens for file fsync's as well. However log btrees are much less likely to big as big as regular fs btrees, therefore the impact of this change is smaller. The following benchmark was performed against an SSD drive and a HDD drive, both for random and sequential writes: sysbench --test=fileio --file-num=4096 --file-total-size=8G \ --file-test-mode=[rndwr|seqwr] --num-threads=512 \ --file-block-size=8192 \ --max-requests=1000000 \ --file-fsync-freq=0 --file-io-mode=sync [prepare|run] All results below are averages of 10 runs of the respective test. ** SSD sequential writes Before this change: 225.88 Mb/sec After this change: 277.26 Mb/sec ** SSD random writes Before this change: 49.91 Mb/sec After this change: 56.39 Mb/sec ** HDD sequential writes Before this change: 68.53 Mb/sec After this change: 69.87 Mb/sec ** HDD random writes Before this change: 13.04 Mb/sec After this change: 14.39 Mb/sec Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-01-07 11:42:27 +00:00
if (!replace_extent || !(*key_inserted))
btrfs_release_path(path);
if (drop_end)
*drop_end = found ? min(end, last_end) : end;
Btrfs: turbo charge fsync At least for the vm workload. Currently on fsync we will 1) Truncate all items in the log tree for the given inode if they exist and 2) Copy all items for a given inode into the log The problem with this is that for things like VMs you can have lots of extents from the fragmented writing behavior, and worst yet you may have only modified a few extents, not the entire thing. This patch fixes this problem by tracking which transid modified our extent, and then when we do the tree logging we find all of the extents we've modified in our current transaction, sort them and commit them. We also only truncate up to the xattrs of the inode and copy that stuff in normally, and then just drop any extents in the range we have that exist in the log already. Here are some numbers of a 50 meg fio job that does random writes and fsync()s after every write Original Patched SATA drive 82KB/s 140KB/s Fusion drive 431KB/s 2532KB/s So around 2-6 times faster depending on your hardware. There are a few corner cases, for example if you truncate at all we have to do it the old way since there is no way to be sure what is in the log is ok. This probably could be done smarter, but if you write-fsync-truncate-write-fsync you deserve what you get. All this work is in RAM of course so if your inode gets evicted from cache and you read it in and fsync it we'll do it the slow way if we are still in the same transaction that we last modified the inode in. The biggest cool part of this is that it requires no changes to the recovery code, so if you fsync with this patch and crash and load an old kernel, it will run the recovery and be a-ok. I have tested this pretty thoroughly with an fsync tester and everything comes back fine, as well as xfstests. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2012-08-17 17:14:17 +00:00
return ret;
}
int btrfs_drop_extents(struct btrfs_trans_handle *trans,
struct btrfs_root *root, struct inode *inode, u64 start,
u64 end, int drop_cache)
Btrfs: turbo charge fsync At least for the vm workload. Currently on fsync we will 1) Truncate all items in the log tree for the given inode if they exist and 2) Copy all items for a given inode into the log The problem with this is that for things like VMs you can have lots of extents from the fragmented writing behavior, and worst yet you may have only modified a few extents, not the entire thing. This patch fixes this problem by tracking which transid modified our extent, and then when we do the tree logging we find all of the extents we've modified in our current transaction, sort them and commit them. We also only truncate up to the xattrs of the inode and copy that stuff in normally, and then just drop any extents in the range we have that exist in the log already. Here are some numbers of a 50 meg fio job that does random writes and fsync()s after every write Original Patched SATA drive 82KB/s 140KB/s Fusion drive 431KB/s 2532KB/s So around 2-6 times faster depending on your hardware. There are a few corner cases, for example if you truncate at all we have to do it the old way since there is no way to be sure what is in the log is ok. This probably could be done smarter, but if you write-fsync-truncate-write-fsync you deserve what you get. All this work is in RAM of course so if your inode gets evicted from cache and you read it in and fsync it we'll do it the slow way if we are still in the same transaction that we last modified the inode in. The biggest cool part of this is that it requires no changes to the recovery code, so if you fsync with this patch and crash and load an old kernel, it will run the recovery and be a-ok. I have tested this pretty thoroughly with an fsync tester and everything comes back fine, as well as xfstests. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2012-08-17 17:14:17 +00:00
{
struct btrfs_path *path;
int ret;
path = btrfs_alloc_path();
if (!path)
return -ENOMEM;
ret = __btrfs_drop_extents(trans, root, inode, path, start, end, NULL,
Btrfs: faster file extent item replace operations When writing to a file we drop existing file extent items that cover the write range and then add a new file extent item that represents that write range. Before this change we were doing a tree lookup to remove the file extent items, and then after we did another tree lookup to insert the new file extent item. Most of the time all the file extent items we need to drop are located within a single leaf - this is the leaf where our new file extent item ends up at. Therefore, in this common case just combine these 2 operations into a single one. By avoiding the second btree navigation for insertion of the new file extent item, we reduce btree node/leaf lock acquisitions/releases, btree block/leaf COW operations, CPU time on btree node/leaf key binary searches, etc. Besides for file writes, this is an operation that happens for file fsync's as well. However log btrees are much less likely to big as big as regular fs btrees, therefore the impact of this change is smaller. The following benchmark was performed against an SSD drive and a HDD drive, both for random and sequential writes: sysbench --test=fileio --file-num=4096 --file-total-size=8G \ --file-test-mode=[rndwr|seqwr] --num-threads=512 \ --file-block-size=8192 \ --max-requests=1000000 \ --file-fsync-freq=0 --file-io-mode=sync [prepare|run] All results below are averages of 10 runs of the respective test. ** SSD sequential writes Before this change: 225.88 Mb/sec After this change: 277.26 Mb/sec ** SSD random writes Before this change: 49.91 Mb/sec After this change: 56.39 Mb/sec ** HDD sequential writes Before this change: 68.53 Mb/sec After this change: 69.87 Mb/sec ** HDD random writes Before this change: 13.04 Mb/sec After this change: 14.39 Mb/sec Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-01-07 11:42:27 +00:00
drop_cache, 0, 0, NULL);
btrfs_free_path(path);
return ret;
}
static int extent_mergeable(struct extent_buffer *leaf, int slot,
u64 objectid, u64 bytenr, u64 orig_offset,
u64 *start, u64 *end)
{
struct btrfs_file_extent_item *fi;
struct btrfs_key key;
u64 extent_end;
if (slot < 0 || slot >= btrfs_header_nritems(leaf))
return 0;
btrfs_item_key_to_cpu(leaf, &key, slot);
if (key.objectid != objectid || key.type != BTRFS_EXTENT_DATA_KEY)
return 0;
fi = btrfs_item_ptr(leaf, slot, struct btrfs_file_extent_item);
if (btrfs_file_extent_type(leaf, fi) != BTRFS_FILE_EXTENT_REG ||
btrfs_file_extent_disk_bytenr(leaf, fi) != bytenr ||
btrfs_file_extent_offset(leaf, fi) != key.offset - orig_offset ||
btrfs_file_extent_compression(leaf, fi) ||
btrfs_file_extent_encryption(leaf, fi) ||
btrfs_file_extent_other_encoding(leaf, fi))
return 0;
extent_end = key.offset + btrfs_file_extent_num_bytes(leaf, fi);
if ((*start && *start != key.offset) || (*end && *end != extent_end))
return 0;
*start = key.offset;
*end = extent_end;
return 1;
}
/*
* Mark extent in the range start - end as written.
*
* This changes extent type from 'pre-allocated' to 'regular'. If only
* part of extent is marked as written, the extent will be split into
* two or three.
*/
int btrfs_mark_extent_written(struct btrfs_trans_handle *trans,
struct btrfs_inode *inode, u64 start, u64 end)
{
struct btrfs_fs_info *fs_info = btrfs_sb(inode->vfs_inode.i_sb);
struct btrfs_root *root = inode->root;
struct extent_buffer *leaf;
struct btrfs_path *path;
struct btrfs_file_extent_item *fi;
struct btrfs_key key;
struct btrfs_key new_key;
u64 bytenr;
u64 num_bytes;
u64 extent_end;
Btrfs: Mixed back reference (FORWARD ROLLING FORMAT CHANGE) This commit introduces a new kind of back reference for btrfs metadata. Once a filesystem has been mounted with this commit, IT WILL NO LONGER BE MOUNTABLE BY OLDER KERNELS. When a tree block in subvolume tree is cow'd, the reference counts of all extents it points to are increased by one. At transaction commit time, the old root of the subvolume is recorded in a "dead root" data structure, and the btree it points to is later walked, dropping reference counts and freeing any blocks where the reference count goes to 0. The increments done during cow and decrements done after commit cancel out, and the walk is a very expensive way to go about freeing the blocks that are no longer referenced by the new btree root. This commit reduces the transaction overhead by avoiding the need for dead root records. When a non-shared tree block is cow'd, we free the old block at once, and the new block inherits old block's references. When a tree block with reference count > 1 is cow'd, we increase the reference counts of all extents the new block points to by one, and decrease the old block's reference count by one. This dead tree avoidance code removes the need to modify the reference counts of lower level extents when a non-shared tree block is cow'd. But we still need to update back ref for all pointers in the block. This is because the location of the block is recorded in the back ref item. We can solve this by introducing a new type of back ref. The new back ref provides information about pointer's key, level and in which tree the pointer lives. This information allow us to find the pointer by searching the tree. The shortcoming of the new back ref is that it only works for pointers in tree blocks referenced by their owner trees. This is mostly a problem for snapshots, where resolving one of these fuzzy back references would be O(number_of_snapshots) and quite slow. The solution used here is to use the fuzzy back references in the common case where a given tree block is only referenced by one root, and use the full back references when multiple roots have a reference on a given block. This commit adds per subvolume red-black tree to keep trace of cached inodes. The red-black tree helps the balancing code to find cached inodes whose inode numbers within a given range. This commit improves the balancing code by introducing several data structures to keep the state of balancing. The most important one is the back ref cache. It caches how the upper level tree blocks are referenced. This greatly reduce the overhead of checking back ref. The improved balancing code scales significantly better with a large number of snapshots. This is a very large commit and was written in a number of pieces. But, they depend heavily on the disk format change and were squashed together to make sure git bisect didn't end up in a bad state wrt space balancing or the format change. Signed-off-by: Yan Zheng <zheng.yan@oracle.com> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2009-06-10 14:45:14 +00:00
u64 orig_offset;
u64 other_start;
u64 other_end;
u64 split;
int del_nr = 0;
int del_slot = 0;
int recow;
int ret;
u64 ino = btrfs_ino(inode);
path = btrfs_alloc_path();
if (!path)
return -ENOMEM;
again:
recow = 0;
split = start;
key.objectid = ino;
key.type = BTRFS_EXTENT_DATA_KEY;
key.offset = split;
ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
if (ret < 0)
goto out;
if (ret > 0 && path->slots[0] > 0)
path->slots[0]--;
leaf = path->nodes[0];
btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
if (key.objectid != ino ||
key.type != BTRFS_EXTENT_DATA_KEY) {
ret = -EINVAL;
btrfs_abort_transaction(trans, ret);
goto out;
}
fi = btrfs_item_ptr(leaf, path->slots[0],
struct btrfs_file_extent_item);
if (btrfs_file_extent_type(leaf, fi) != BTRFS_FILE_EXTENT_PREALLOC) {
ret = -EINVAL;
btrfs_abort_transaction(trans, ret);
goto out;
}
extent_end = key.offset + btrfs_file_extent_num_bytes(leaf, fi);
if (key.offset > start || extent_end < end) {
ret = -EINVAL;
btrfs_abort_transaction(trans, ret);
goto out;
}
bytenr = btrfs_file_extent_disk_bytenr(leaf, fi);
num_bytes = btrfs_file_extent_disk_num_bytes(leaf, fi);
Btrfs: Mixed back reference (FORWARD ROLLING FORMAT CHANGE) This commit introduces a new kind of back reference for btrfs metadata. Once a filesystem has been mounted with this commit, IT WILL NO LONGER BE MOUNTABLE BY OLDER KERNELS. When a tree block in subvolume tree is cow'd, the reference counts of all extents it points to are increased by one. At transaction commit time, the old root of the subvolume is recorded in a "dead root" data structure, and the btree it points to is later walked, dropping reference counts and freeing any blocks where the reference count goes to 0. The increments done during cow and decrements done after commit cancel out, and the walk is a very expensive way to go about freeing the blocks that are no longer referenced by the new btree root. This commit reduces the transaction overhead by avoiding the need for dead root records. When a non-shared tree block is cow'd, we free the old block at once, and the new block inherits old block's references. When a tree block with reference count > 1 is cow'd, we increase the reference counts of all extents the new block points to by one, and decrease the old block's reference count by one. This dead tree avoidance code removes the need to modify the reference counts of lower level extents when a non-shared tree block is cow'd. But we still need to update back ref for all pointers in the block. This is because the location of the block is recorded in the back ref item. We can solve this by introducing a new type of back ref. The new back ref provides information about pointer's key, level and in which tree the pointer lives. This information allow us to find the pointer by searching the tree. The shortcoming of the new back ref is that it only works for pointers in tree blocks referenced by their owner trees. This is mostly a problem for snapshots, where resolving one of these fuzzy back references would be O(number_of_snapshots) and quite slow. The solution used here is to use the fuzzy back references in the common case where a given tree block is only referenced by one root, and use the full back references when multiple roots have a reference on a given block. This commit adds per subvolume red-black tree to keep trace of cached inodes. The red-black tree helps the balancing code to find cached inodes whose inode numbers within a given range. This commit improves the balancing code by introducing several data structures to keep the state of balancing. The most important one is the back ref cache. It caches how the upper level tree blocks are referenced. This greatly reduce the overhead of checking back ref. The improved balancing code scales significantly better with a large number of snapshots. This is a very large commit and was written in a number of pieces. But, they depend heavily on the disk format change and were squashed together to make sure git bisect didn't end up in a bad state wrt space balancing or the format change. Signed-off-by: Yan Zheng <zheng.yan@oracle.com> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2009-06-10 14:45:14 +00:00
orig_offset = key.offset - btrfs_file_extent_offset(leaf, fi);
memcpy(&new_key, &key, sizeof(new_key));
if (start == key.offset && end < extent_end) {
other_start = 0;
other_end = start;
if (extent_mergeable(leaf, path->slots[0] - 1,
ino, bytenr, orig_offset,
&other_start, &other_end)) {
new_key.offset = end;
btrfs_set_item_key_safe(fs_info, path, &new_key);
fi = btrfs_item_ptr(leaf, path->slots[0],
struct btrfs_file_extent_item);
btrfs_set_file_extent_generation(leaf, fi,
trans->transid);
btrfs_set_file_extent_num_bytes(leaf, fi,
extent_end - end);
btrfs_set_file_extent_offset(leaf, fi,
end - orig_offset);
fi = btrfs_item_ptr(leaf, path->slots[0] - 1,
struct btrfs_file_extent_item);
btrfs_set_file_extent_generation(leaf, fi,
trans->transid);
btrfs_set_file_extent_num_bytes(leaf, fi,
end - other_start);
btrfs_mark_buffer_dirty(leaf);
goto out;
}
}
if (start > key.offset && end == extent_end) {
other_start = end;
other_end = 0;
if (extent_mergeable(leaf, path->slots[0] + 1,
ino, bytenr, orig_offset,
&other_start, &other_end)) {
fi = btrfs_item_ptr(leaf, path->slots[0],
struct btrfs_file_extent_item);
btrfs_set_file_extent_num_bytes(leaf, fi,
start - key.offset);
btrfs_set_file_extent_generation(leaf, fi,
trans->transid);
path->slots[0]++;
new_key.offset = start;
btrfs_set_item_key_safe(fs_info, path, &new_key);
fi = btrfs_item_ptr(leaf, path->slots[0],
struct btrfs_file_extent_item);
btrfs_set_file_extent_generation(leaf, fi,
trans->transid);
btrfs_set_file_extent_num_bytes(leaf, fi,
other_end - start);
btrfs_set_file_extent_offset(leaf, fi,
start - orig_offset);
btrfs_mark_buffer_dirty(leaf);
goto out;
}
}
while (start > key.offset || end < extent_end) {
if (key.offset == start)
split = end;
new_key.offset = split;
ret = btrfs_duplicate_item(trans, root, path, &new_key);
if (ret == -EAGAIN) {
btrfs_release_path(path);
goto again;
}
if (ret < 0) {
btrfs_abort_transaction(trans, ret);
goto out;
}
leaf = path->nodes[0];
fi = btrfs_item_ptr(leaf, path->slots[0] - 1,
struct btrfs_file_extent_item);
btrfs_set_file_extent_generation(leaf, fi, trans->transid);
btrfs_set_file_extent_num_bytes(leaf, fi,
split - key.offset);
fi = btrfs_item_ptr(leaf, path->slots[0],
struct btrfs_file_extent_item);
btrfs_set_file_extent_generation(leaf, fi, trans->transid);
btrfs_set_file_extent_offset(leaf, fi, split - orig_offset);
btrfs_set_file_extent_num_bytes(leaf, fi,
extent_end - split);
btrfs_mark_buffer_dirty(leaf);
ret = btrfs_inc_extent_ref(trans, fs_info, bytenr, num_bytes,
0, root->root_key.objectid,
Btrfs: fix regression running delayed references when using qgroups In the kernel 4.2 merge window we had a big changes to the implementation of delayed references and qgroups which made the no_quota field of delayed references not used anymore. More specifically the no_quota field is not used anymore as of: commit 0ed4792af0e8 ("btrfs: qgroup: Switch to new extent-oriented qgroup mechanism.") Leaving the no_quota field actually prevents delayed references from getting merged, which in turn cause the following BUG_ON(), at fs/btrfs/extent-tree.c, to be hit when qgroups are enabled: static int run_delayed_tree_ref(...) { (...) BUG_ON(node->ref_mod != 1); (...) } This happens on a scenario like the following: 1) Ref1 bytenr X, action = BTRFS_ADD_DELAYED_REF, no_quota = 1, added. 2) Ref2 bytenr X, action = BTRFS_DROP_DELAYED_REF, no_quota = 0, added. It's not merged with Ref1 because Ref1->no_quota != Ref2->no_quota. 3) Ref3 bytenr X, action = BTRFS_ADD_DELAYED_REF, no_quota = 1, added. It's not merged with the reference at the tail of the list of refs for bytenr X because the reference at the tail, Ref2 is incompatible due to Ref2->no_quota != Ref3->no_quota. 4) Ref4 bytenr X, action = BTRFS_DROP_DELAYED_REF, no_quota = 0, added. It's not merged with the reference at the tail of the list of refs for bytenr X because the reference at the tail, Ref3 is incompatible due to Ref3->no_quota != Ref4->no_quota. 5) We run delayed references, trigger merging of delayed references, through __btrfs_run_delayed_refs() -> btrfs_merge_delayed_refs(). 6) Ref1 and Ref3 are merged as Ref1->no_quota = Ref3->no_quota and all other conditions are satisfied too. So Ref1 gets a ref_mod value of 2. 7) Ref2 and Ref4 are merged as Ref2->no_quota = Ref4->no_quota and all other conditions are satisfied too. So Ref2 gets a ref_mod value of 2. 8) Ref1 and Ref2 aren't merged, because they have different values for their no_quota field. 9) Delayed reference Ref1 is picked for running (select_delayed_ref() always prefers references with an action == BTRFS_ADD_DELAYED_REF). So run_delayed_tree_ref() is called for Ref1 which triggers the BUG_ON because Ref1->red_mod != 1 (equals 2). So fix this by removing the no_quota field, as it's not used anymore as of commit 0ed4792af0e8 ("btrfs: qgroup: Switch to new extent-oriented qgroup mechanism."). The use of no_quota was also buggy in at least two places: 1) At delayed-refs.c:btrfs_add_delayed_tree_ref() - we were setting no_quota to 0 instead of 1 when the following condition was true: is_fstree(ref_root) || !fs_info->quota_enabled 2) At extent-tree.c:__btrfs_inc_extent_ref() - we were attempting to reset a node's no_quota when the condition "!is_fstree(root_objectid) || !root->fs_info->quota_enabled" was true but we did it only in an unused local stack variable, that is, we never reset the no_quota value in the node itself. This fixes the remainder of problems several people have been having when running delayed references, mostly while a balance is running in parallel, on a 4.2+ kernel. Very special thanks to Stéphane Lesimple for helping debugging this issue and testing this fix on his multi terabyte filesystem (which took more than one day to balance alone, plus fsck, etc). Also, this fixes deadlock issue when using the clone ioctl with qgroups enabled, as reported by Elias Probst in the mailing list. The deadlock happens because after calling btrfs_insert_empty_item we have our path holding a write lock on a leaf of the fs/subvol tree and then before releasing the path we called check_ref() which did backref walking, when qgroups are enabled, and tried to read lock the same leaf. The trace for this case is the following: INFO: task systemd-nspawn:6095 blocked for more than 120 seconds. (...) Call Trace: [<ffffffff86999201>] schedule+0x74/0x83 [<ffffffff863ef64c>] btrfs_tree_read_lock+0xc0/0xea [<ffffffff86137ed7>] ? wait_woken+0x74/0x74 [<ffffffff8639f0a7>] btrfs_search_old_slot+0x51a/0x810 [<ffffffff863a129b>] btrfs_next_old_leaf+0xdf/0x3ce [<ffffffff86413a00>] ? ulist_add_merge+0x1b/0x127 [<ffffffff86411688>] __resolve_indirect_refs+0x62a/0x667 [<ffffffff863ef546>] ? btrfs_clear_lock_blocking_rw+0x78/0xbe [<ffffffff864122d3>] find_parent_nodes+0xaf3/0xfc6 [<ffffffff86412838>] __btrfs_find_all_roots+0x92/0xf0 [<ffffffff864128f2>] btrfs_find_all_roots+0x45/0x65 [<ffffffff8639a75b>] ? btrfs_get_tree_mod_seq+0x2b/0x88 [<ffffffff863e852e>] check_ref+0x64/0xc4 [<ffffffff863e9e01>] btrfs_clone+0x66e/0xb5d [<ffffffff863ea77f>] btrfs_ioctl_clone+0x48f/0x5bb [<ffffffff86048a68>] ? native_sched_clock+0x28/0x77 [<ffffffff863ed9b0>] btrfs_ioctl+0xabc/0x25cb (...) The problem goes away by eleminating check_ref(), which no longer is needed as its purpose was to get a value for the no_quota field of a delayed reference (this patch removes the no_quota field as mentioned earlier). Reported-by: Stéphane Lesimple <stephane_btrfs@lesimple.fr> Tested-by: Stéphane Lesimple <stephane_btrfs@lesimple.fr> Reported-by: Elias Probst <mail@eliasprobst.eu> Reported-by: Peter Becker <floyd.net@gmail.com> Reported-by: Malte Schröder <malte@tnxip.de> Reported-by: Derek Dongray <derek@valedon.co.uk> Reported-by: Erkki Seppala <flux-btrfs@inside.org> Cc: stable@vger.kernel.org # 4.2+ Signed-off-by: Filipe Manana <fdmanana@suse.com> Reviewed-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
2015-10-23 06:52:54 +00:00
ino, orig_offset);
if (ret) {
btrfs_abort_transaction(trans, ret);
goto out;
}
if (split == start) {
key.offset = start;
} else {
if (start != key.offset) {
ret = -EINVAL;
btrfs_abort_transaction(trans, ret);
goto out;
}
path->slots[0]--;
extent_end = end;
}
recow = 1;
}
other_start = end;
other_end = 0;
if (extent_mergeable(leaf, path->slots[0] + 1,
ino, bytenr, orig_offset,
&other_start, &other_end)) {
if (recow) {
btrfs_release_path(path);
goto again;
}
extent_end = other_end;
del_slot = path->slots[0] + 1;
del_nr++;
ret = btrfs_free_extent(trans, fs_info, bytenr, num_bytes,
0, root->root_key.objectid,
Btrfs: fix regression running delayed references when using qgroups In the kernel 4.2 merge window we had a big changes to the implementation of delayed references and qgroups which made the no_quota field of delayed references not used anymore. More specifically the no_quota field is not used anymore as of: commit 0ed4792af0e8 ("btrfs: qgroup: Switch to new extent-oriented qgroup mechanism.") Leaving the no_quota field actually prevents delayed references from getting merged, which in turn cause the following BUG_ON(), at fs/btrfs/extent-tree.c, to be hit when qgroups are enabled: static int run_delayed_tree_ref(...) { (...) BUG_ON(node->ref_mod != 1); (...) } This happens on a scenario like the following: 1) Ref1 bytenr X, action = BTRFS_ADD_DELAYED_REF, no_quota = 1, added. 2) Ref2 bytenr X, action = BTRFS_DROP_DELAYED_REF, no_quota = 0, added. It's not merged with Ref1 because Ref1->no_quota != Ref2->no_quota. 3) Ref3 bytenr X, action = BTRFS_ADD_DELAYED_REF, no_quota = 1, added. It's not merged with the reference at the tail of the list of refs for bytenr X because the reference at the tail, Ref2 is incompatible due to Ref2->no_quota != Ref3->no_quota. 4) Ref4 bytenr X, action = BTRFS_DROP_DELAYED_REF, no_quota = 0, added. It's not merged with the reference at the tail of the list of refs for bytenr X because the reference at the tail, Ref3 is incompatible due to Ref3->no_quota != Ref4->no_quota. 5) We run delayed references, trigger merging of delayed references, through __btrfs_run_delayed_refs() -> btrfs_merge_delayed_refs(). 6) Ref1 and Ref3 are merged as Ref1->no_quota = Ref3->no_quota and all other conditions are satisfied too. So Ref1 gets a ref_mod value of 2. 7) Ref2 and Ref4 are merged as Ref2->no_quota = Ref4->no_quota and all other conditions are satisfied too. So Ref2 gets a ref_mod value of 2. 8) Ref1 and Ref2 aren't merged, because they have different values for their no_quota field. 9) Delayed reference Ref1 is picked for running (select_delayed_ref() always prefers references with an action == BTRFS_ADD_DELAYED_REF). So run_delayed_tree_ref() is called for Ref1 which triggers the BUG_ON because Ref1->red_mod != 1 (equals 2). So fix this by removing the no_quota field, as it's not used anymore as of commit 0ed4792af0e8 ("btrfs: qgroup: Switch to new extent-oriented qgroup mechanism."). The use of no_quota was also buggy in at least two places: 1) At delayed-refs.c:btrfs_add_delayed_tree_ref() - we were setting no_quota to 0 instead of 1 when the following condition was true: is_fstree(ref_root) || !fs_info->quota_enabled 2) At extent-tree.c:__btrfs_inc_extent_ref() - we were attempting to reset a node's no_quota when the condition "!is_fstree(root_objectid) || !root->fs_info->quota_enabled" was true but we did it only in an unused local stack variable, that is, we never reset the no_quota value in the node itself. This fixes the remainder of problems several people have been having when running delayed references, mostly while a balance is running in parallel, on a 4.2+ kernel. Very special thanks to Stéphane Lesimple for helping debugging this issue and testing this fix on his multi terabyte filesystem (which took more than one day to balance alone, plus fsck, etc). Also, this fixes deadlock issue when using the clone ioctl with qgroups enabled, as reported by Elias Probst in the mailing list. The deadlock happens because after calling btrfs_insert_empty_item we have our path holding a write lock on a leaf of the fs/subvol tree and then before releasing the path we called check_ref() which did backref walking, when qgroups are enabled, and tried to read lock the same leaf. The trace for this case is the following: INFO: task systemd-nspawn:6095 blocked for more than 120 seconds. (...) Call Trace: [<ffffffff86999201>] schedule+0x74/0x83 [<ffffffff863ef64c>] btrfs_tree_read_lock+0xc0/0xea [<ffffffff86137ed7>] ? wait_woken+0x74/0x74 [<ffffffff8639f0a7>] btrfs_search_old_slot+0x51a/0x810 [<ffffffff863a129b>] btrfs_next_old_leaf+0xdf/0x3ce [<ffffffff86413a00>] ? ulist_add_merge+0x1b/0x127 [<ffffffff86411688>] __resolve_indirect_refs+0x62a/0x667 [<ffffffff863ef546>] ? btrfs_clear_lock_blocking_rw+0x78/0xbe [<ffffffff864122d3>] find_parent_nodes+0xaf3/0xfc6 [<ffffffff86412838>] __btrfs_find_all_roots+0x92/0xf0 [<ffffffff864128f2>] btrfs_find_all_roots+0x45/0x65 [<ffffffff8639a75b>] ? btrfs_get_tree_mod_seq+0x2b/0x88 [<ffffffff863e852e>] check_ref+0x64/0xc4 [<ffffffff863e9e01>] btrfs_clone+0x66e/0xb5d [<ffffffff863ea77f>] btrfs_ioctl_clone+0x48f/0x5bb [<ffffffff86048a68>] ? native_sched_clock+0x28/0x77 [<ffffffff863ed9b0>] btrfs_ioctl+0xabc/0x25cb (...) The problem goes away by eleminating check_ref(), which no longer is needed as its purpose was to get a value for the no_quota field of a delayed reference (this patch removes the no_quota field as mentioned earlier). Reported-by: Stéphane Lesimple <stephane_btrfs@lesimple.fr> Tested-by: Stéphane Lesimple <stephane_btrfs@lesimple.fr> Reported-by: Elias Probst <mail@eliasprobst.eu> Reported-by: Peter Becker <floyd.net@gmail.com> Reported-by: Malte Schröder <malte@tnxip.de> Reported-by: Derek Dongray <derek@valedon.co.uk> Reported-by: Erkki Seppala <flux-btrfs@inside.org> Cc: stable@vger.kernel.org # 4.2+ Signed-off-by: Filipe Manana <fdmanana@suse.com> Reviewed-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
2015-10-23 06:52:54 +00:00
ino, orig_offset);
if (ret) {
btrfs_abort_transaction(trans, ret);
goto out;
}
}
other_start = 0;
other_end = start;
if (extent_mergeable(leaf, path->slots[0] - 1,
ino, bytenr, orig_offset,
&other_start, &other_end)) {
if (recow) {
btrfs_release_path(path);
goto again;
}
key.offset = other_start;
del_slot = path->slots[0];
del_nr++;
ret = btrfs_free_extent(trans, fs_info, bytenr, num_bytes,
0, root->root_key.objectid,
Btrfs: fix regression running delayed references when using qgroups In the kernel 4.2 merge window we had a big changes to the implementation of delayed references and qgroups which made the no_quota field of delayed references not used anymore. More specifically the no_quota field is not used anymore as of: commit 0ed4792af0e8 ("btrfs: qgroup: Switch to new extent-oriented qgroup mechanism.") Leaving the no_quota field actually prevents delayed references from getting merged, which in turn cause the following BUG_ON(), at fs/btrfs/extent-tree.c, to be hit when qgroups are enabled: static int run_delayed_tree_ref(...) { (...) BUG_ON(node->ref_mod != 1); (...) } This happens on a scenario like the following: 1) Ref1 bytenr X, action = BTRFS_ADD_DELAYED_REF, no_quota = 1, added. 2) Ref2 bytenr X, action = BTRFS_DROP_DELAYED_REF, no_quota = 0, added. It's not merged with Ref1 because Ref1->no_quota != Ref2->no_quota. 3) Ref3 bytenr X, action = BTRFS_ADD_DELAYED_REF, no_quota = 1, added. It's not merged with the reference at the tail of the list of refs for bytenr X because the reference at the tail, Ref2 is incompatible due to Ref2->no_quota != Ref3->no_quota. 4) Ref4 bytenr X, action = BTRFS_DROP_DELAYED_REF, no_quota = 0, added. It's not merged with the reference at the tail of the list of refs for bytenr X because the reference at the tail, Ref3 is incompatible due to Ref3->no_quota != Ref4->no_quota. 5) We run delayed references, trigger merging of delayed references, through __btrfs_run_delayed_refs() -> btrfs_merge_delayed_refs(). 6) Ref1 and Ref3 are merged as Ref1->no_quota = Ref3->no_quota and all other conditions are satisfied too. So Ref1 gets a ref_mod value of 2. 7) Ref2 and Ref4 are merged as Ref2->no_quota = Ref4->no_quota and all other conditions are satisfied too. So Ref2 gets a ref_mod value of 2. 8) Ref1 and Ref2 aren't merged, because they have different values for their no_quota field. 9) Delayed reference Ref1 is picked for running (select_delayed_ref() always prefers references with an action == BTRFS_ADD_DELAYED_REF). So run_delayed_tree_ref() is called for Ref1 which triggers the BUG_ON because Ref1->red_mod != 1 (equals 2). So fix this by removing the no_quota field, as it's not used anymore as of commit 0ed4792af0e8 ("btrfs: qgroup: Switch to new extent-oriented qgroup mechanism."). The use of no_quota was also buggy in at least two places: 1) At delayed-refs.c:btrfs_add_delayed_tree_ref() - we were setting no_quota to 0 instead of 1 when the following condition was true: is_fstree(ref_root) || !fs_info->quota_enabled 2) At extent-tree.c:__btrfs_inc_extent_ref() - we were attempting to reset a node's no_quota when the condition "!is_fstree(root_objectid) || !root->fs_info->quota_enabled" was true but we did it only in an unused local stack variable, that is, we never reset the no_quota value in the node itself. This fixes the remainder of problems several people have been having when running delayed references, mostly while a balance is running in parallel, on a 4.2+ kernel. Very special thanks to Stéphane Lesimple for helping debugging this issue and testing this fix on his multi terabyte filesystem (which took more than one day to balance alone, plus fsck, etc). Also, this fixes deadlock issue when using the clone ioctl with qgroups enabled, as reported by Elias Probst in the mailing list. The deadlock happens because after calling btrfs_insert_empty_item we have our path holding a write lock on a leaf of the fs/subvol tree and then before releasing the path we called check_ref() which did backref walking, when qgroups are enabled, and tried to read lock the same leaf. The trace for this case is the following: INFO: task systemd-nspawn:6095 blocked for more than 120 seconds. (...) Call Trace: [<ffffffff86999201>] schedule+0x74/0x83 [<ffffffff863ef64c>] btrfs_tree_read_lock+0xc0/0xea [<ffffffff86137ed7>] ? wait_woken+0x74/0x74 [<ffffffff8639f0a7>] btrfs_search_old_slot+0x51a/0x810 [<ffffffff863a129b>] btrfs_next_old_leaf+0xdf/0x3ce [<ffffffff86413a00>] ? ulist_add_merge+0x1b/0x127 [<ffffffff86411688>] __resolve_indirect_refs+0x62a/0x667 [<ffffffff863ef546>] ? btrfs_clear_lock_blocking_rw+0x78/0xbe [<ffffffff864122d3>] find_parent_nodes+0xaf3/0xfc6 [<ffffffff86412838>] __btrfs_find_all_roots+0x92/0xf0 [<ffffffff864128f2>] btrfs_find_all_roots+0x45/0x65 [<ffffffff8639a75b>] ? btrfs_get_tree_mod_seq+0x2b/0x88 [<ffffffff863e852e>] check_ref+0x64/0xc4 [<ffffffff863e9e01>] btrfs_clone+0x66e/0xb5d [<ffffffff863ea77f>] btrfs_ioctl_clone+0x48f/0x5bb [<ffffffff86048a68>] ? native_sched_clock+0x28/0x77 [<ffffffff863ed9b0>] btrfs_ioctl+0xabc/0x25cb (...) The problem goes away by eleminating check_ref(), which no longer is needed as its purpose was to get a value for the no_quota field of a delayed reference (this patch removes the no_quota field as mentioned earlier). Reported-by: Stéphane Lesimple <stephane_btrfs@lesimple.fr> Tested-by: Stéphane Lesimple <stephane_btrfs@lesimple.fr> Reported-by: Elias Probst <mail@eliasprobst.eu> Reported-by: Peter Becker <floyd.net@gmail.com> Reported-by: Malte Schröder <malte@tnxip.de> Reported-by: Derek Dongray <derek@valedon.co.uk> Reported-by: Erkki Seppala <flux-btrfs@inside.org> Cc: stable@vger.kernel.org # 4.2+ Signed-off-by: Filipe Manana <fdmanana@suse.com> Reviewed-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
2015-10-23 06:52:54 +00:00
ino, orig_offset);
if (ret) {
btrfs_abort_transaction(trans, ret);
goto out;
}
}
if (del_nr == 0) {
fi = btrfs_item_ptr(leaf, path->slots[0],
struct btrfs_file_extent_item);
btrfs_set_file_extent_type(leaf, fi,
BTRFS_FILE_EXTENT_REG);
btrfs_set_file_extent_generation(leaf, fi, trans->transid);
btrfs_mark_buffer_dirty(leaf);
} else {
fi = btrfs_item_ptr(leaf, del_slot - 1,
struct btrfs_file_extent_item);
btrfs_set_file_extent_type(leaf, fi,
BTRFS_FILE_EXTENT_REG);
btrfs_set_file_extent_generation(leaf, fi, trans->transid);
btrfs_set_file_extent_num_bytes(leaf, fi,
extent_end - key.offset);
btrfs_mark_buffer_dirty(leaf);
ret = btrfs_del_items(trans, root, path, del_slot, del_nr);
if (ret < 0) {
btrfs_abort_transaction(trans, ret);
goto out;
}
}
out:
btrfs_free_path(path);
return 0;
}
Btrfs: fix regressions in copy_from_user handling Commit 914ee295af418e936ec20a08c1663eaabe4cd07a fixed deadlocks in btrfs_file_write where we would catch page faults on pages we had locked. But, there were a few problems: 1) The x86-32 iov_iter_copy_from_user_atomic code always fails to copy data when the amount to copy is more than 4K and the offset to start copying from is not page aligned. The result was btrfs_file_write looping forever retrying the iov_iter_copy_from_user_atomic We deal with this by changing btrfs_file_write to drop down to single page copies when iov_iter_copy_from_user_atomic starts returning failure. 2) The btrfs_file_write code was leaking delalloc reservations when iov_iter_copy_from_user_atomic returned zero. The looping above would result in the entire filesystem running out of delalloc reservations and constantly trying to flush things to disk. 3) btrfs_file_write will lock down page cache pages, make sure any writeback is finished, do the copy_from_user and then release them. Before the loop runs we check the first and last pages in the write to see if they are only being partially modified. If the start or end of the write isn't aligned, we make sure the corresponding pages are up to date so that we don't introduce garbage into the file. With the copy_from_user changes, we're allowing the VM to reclaim the pages after a partial update from copy_from_user, but we're not making sure the page cache page is up to date when we loop around to resume the write. We deal with this by pushing the up to date checks down into the page prep code. This fits better with how the rest of file_write works. Signed-off-by: Chris Mason <chris.mason@oracle.com> Reported-by: Mitch Harder <mitch.harder@sabayonlinux.org> cc: stable@kernel.org
2011-02-28 14:52:08 +00:00
/*
* on error we return an unlocked page and the error value
* on success we return a locked page and 0
*/
static int prepare_uptodate_page(struct inode *inode,
struct page *page, u64 pos,
bool force_uptodate)
Btrfs: fix regressions in copy_from_user handling Commit 914ee295af418e936ec20a08c1663eaabe4cd07a fixed deadlocks in btrfs_file_write where we would catch page faults on pages we had locked. But, there were a few problems: 1) The x86-32 iov_iter_copy_from_user_atomic code always fails to copy data when the amount to copy is more than 4K and the offset to start copying from is not page aligned. The result was btrfs_file_write looping forever retrying the iov_iter_copy_from_user_atomic We deal with this by changing btrfs_file_write to drop down to single page copies when iov_iter_copy_from_user_atomic starts returning failure. 2) The btrfs_file_write code was leaking delalloc reservations when iov_iter_copy_from_user_atomic returned zero. The looping above would result in the entire filesystem running out of delalloc reservations and constantly trying to flush things to disk. 3) btrfs_file_write will lock down page cache pages, make sure any writeback is finished, do the copy_from_user and then release them. Before the loop runs we check the first and last pages in the write to see if they are only being partially modified. If the start or end of the write isn't aligned, we make sure the corresponding pages are up to date so that we don't introduce garbage into the file. With the copy_from_user changes, we're allowing the VM to reclaim the pages after a partial update from copy_from_user, but we're not making sure the page cache page is up to date when we loop around to resume the write. We deal with this by pushing the up to date checks down into the page prep code. This fits better with how the rest of file_write works. Signed-off-by: Chris Mason <chris.mason@oracle.com> Reported-by: Mitch Harder <mitch.harder@sabayonlinux.org> cc: stable@kernel.org
2011-02-28 14:52:08 +00:00
{
int ret = 0;
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
if (((pos & (PAGE_SIZE - 1)) || force_uptodate) &&
!PageUptodate(page)) {
Btrfs: fix regressions in copy_from_user handling Commit 914ee295af418e936ec20a08c1663eaabe4cd07a fixed deadlocks in btrfs_file_write where we would catch page faults on pages we had locked. But, there were a few problems: 1) The x86-32 iov_iter_copy_from_user_atomic code always fails to copy data when the amount to copy is more than 4K and the offset to start copying from is not page aligned. The result was btrfs_file_write looping forever retrying the iov_iter_copy_from_user_atomic We deal with this by changing btrfs_file_write to drop down to single page copies when iov_iter_copy_from_user_atomic starts returning failure. 2) The btrfs_file_write code was leaking delalloc reservations when iov_iter_copy_from_user_atomic returned zero. The looping above would result in the entire filesystem running out of delalloc reservations and constantly trying to flush things to disk. 3) btrfs_file_write will lock down page cache pages, make sure any writeback is finished, do the copy_from_user and then release them. Before the loop runs we check the first and last pages in the write to see if they are only being partially modified. If the start or end of the write isn't aligned, we make sure the corresponding pages are up to date so that we don't introduce garbage into the file. With the copy_from_user changes, we're allowing the VM to reclaim the pages after a partial update from copy_from_user, but we're not making sure the page cache page is up to date when we loop around to resume the write. We deal with this by pushing the up to date checks down into the page prep code. This fits better with how the rest of file_write works. Signed-off-by: Chris Mason <chris.mason@oracle.com> Reported-by: Mitch Harder <mitch.harder@sabayonlinux.org> cc: stable@kernel.org
2011-02-28 14:52:08 +00:00
ret = btrfs_readpage(NULL, page);
if (ret)
return ret;
lock_page(page);
if (!PageUptodate(page)) {
unlock_page(page);
return -EIO;
}
if (page->mapping != inode->i_mapping) {
unlock_page(page);
return -EAGAIN;
}
Btrfs: fix regressions in copy_from_user handling Commit 914ee295af418e936ec20a08c1663eaabe4cd07a fixed deadlocks in btrfs_file_write where we would catch page faults on pages we had locked. But, there were a few problems: 1) The x86-32 iov_iter_copy_from_user_atomic code always fails to copy data when the amount to copy is more than 4K and the offset to start copying from is not page aligned. The result was btrfs_file_write looping forever retrying the iov_iter_copy_from_user_atomic We deal with this by changing btrfs_file_write to drop down to single page copies when iov_iter_copy_from_user_atomic starts returning failure. 2) The btrfs_file_write code was leaking delalloc reservations when iov_iter_copy_from_user_atomic returned zero. The looping above would result in the entire filesystem running out of delalloc reservations and constantly trying to flush things to disk. 3) btrfs_file_write will lock down page cache pages, make sure any writeback is finished, do the copy_from_user and then release them. Before the loop runs we check the first and last pages in the write to see if they are only being partially modified. If the start or end of the write isn't aligned, we make sure the corresponding pages are up to date so that we don't introduce garbage into the file. With the copy_from_user changes, we're allowing the VM to reclaim the pages after a partial update from copy_from_user, but we're not making sure the page cache page is up to date when we loop around to resume the write. We deal with this by pushing the up to date checks down into the page prep code. This fits better with how the rest of file_write works. Signed-off-by: Chris Mason <chris.mason@oracle.com> Reported-by: Mitch Harder <mitch.harder@sabayonlinux.org> cc: stable@kernel.org
2011-02-28 14:52:08 +00:00
}
return 0;
}
/*
Btrfs: fix the reserved space leak caused by the race between nonlock dio and buffered io When we ran sysbench on the fs with compression, the following WARN_ONs were triggered: fs/btrfs/inode.c:7829 WARN_ON(BTRFS_I(inode)->outstanding_extents); fs/btrfs/inode.c:7830 WARN_ON(BTRFS_I(inode)->reserved_extents); fs/btrfs/inode.c:7832 WARN_ON(BTRFS_I(inode)->csum_bytes); Steps to reproduce: # mkfs.btrfs -f <dev> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync prepare # cd - # umount <mnt> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync run # cd - # umount <mnt> The reason of this problem is: Task0 Task1 btrfs_direct_IO unlock(&inode->i_mutex) lock(&inode->i_mutex) reserve_space() prepare_pages() lock_extent() clear_extent() unlock_extent() lock_extent() test_extent(uptodate) return false copy_data() set_delalloc_extent() extent need compress go back to buffered write clear_extent(DELALLOC | DIRTY) unlock_extent() Task 0 and 1 wrote the same place, and task0 cleared the delalloc flag which was set by task1, it made the dirty pages in that extents couldn't be flushed into the disk, so the reserved space for that extent was not released at the end. This patch fixes the above bug by unlocking the extent after the delalloc. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-10 11:25:04 +00:00
* this just gets pages into the page cache and locks them down.
*/
static noinline int prepare_pages(struct inode *inode, struct page **pages,
size_t num_pages, loff_t pos,
size_t write_bytes, bool force_uptodate)
{
int i;
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
unsigned long index = pos >> PAGE_SHIFT;
gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping);
int err = 0;
Btrfs: fix the reserved space leak caused by the race between nonlock dio and buffered io When we ran sysbench on the fs with compression, the following WARN_ONs were triggered: fs/btrfs/inode.c:7829 WARN_ON(BTRFS_I(inode)->outstanding_extents); fs/btrfs/inode.c:7830 WARN_ON(BTRFS_I(inode)->reserved_extents); fs/btrfs/inode.c:7832 WARN_ON(BTRFS_I(inode)->csum_bytes); Steps to reproduce: # mkfs.btrfs -f <dev> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync prepare # cd - # umount <mnt> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync run # cd - # umount <mnt> The reason of this problem is: Task0 Task1 btrfs_direct_IO unlock(&inode->i_mutex) lock(&inode->i_mutex) reserve_space() prepare_pages() lock_extent() clear_extent() unlock_extent() lock_extent() test_extent(uptodate) return false copy_data() set_delalloc_extent() extent need compress go back to buffered write clear_extent(DELALLOC | DIRTY) unlock_extent() Task 0 and 1 wrote the same place, and task0 cleared the delalloc flag which was set by task1, it made the dirty pages in that extents couldn't be flushed into the disk, so the reserved space for that extent was not released at the end. This patch fixes the above bug by unlocking the extent after the delalloc. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-10 11:25:04 +00:00
int faili;
for (i = 0; i < num_pages; i++) {
again:
pages[i] = find_or_create_page(inode->i_mapping, index + i,
mask | __GFP_WRITE);
if (!pages[i]) {
Btrfs: fix regressions in copy_from_user handling Commit 914ee295af418e936ec20a08c1663eaabe4cd07a fixed deadlocks in btrfs_file_write where we would catch page faults on pages we had locked. But, there were a few problems: 1) The x86-32 iov_iter_copy_from_user_atomic code always fails to copy data when the amount to copy is more than 4K and the offset to start copying from is not page aligned. The result was btrfs_file_write looping forever retrying the iov_iter_copy_from_user_atomic We deal with this by changing btrfs_file_write to drop down to single page copies when iov_iter_copy_from_user_atomic starts returning failure. 2) The btrfs_file_write code was leaking delalloc reservations when iov_iter_copy_from_user_atomic returned zero. The looping above would result in the entire filesystem running out of delalloc reservations and constantly trying to flush things to disk. 3) btrfs_file_write will lock down page cache pages, make sure any writeback is finished, do the copy_from_user and then release them. Before the loop runs we check the first and last pages in the write to see if they are only being partially modified. If the start or end of the write isn't aligned, we make sure the corresponding pages are up to date so that we don't introduce garbage into the file. With the copy_from_user changes, we're allowing the VM to reclaim the pages after a partial update from copy_from_user, but we're not making sure the page cache page is up to date when we loop around to resume the write. We deal with this by pushing the up to date checks down into the page prep code. This fits better with how the rest of file_write works. Signed-off-by: Chris Mason <chris.mason@oracle.com> Reported-by: Mitch Harder <mitch.harder@sabayonlinux.org> cc: stable@kernel.org
2011-02-28 14:52:08 +00:00
faili = i - 1;
err = -ENOMEM;
goto fail;
}
if (i == 0)
err = prepare_uptodate_page(inode, pages[i], pos,
force_uptodate);
if (!err && i == num_pages - 1)
err = prepare_uptodate_page(inode, pages[i],
pos + write_bytes, false);
Btrfs: fix regressions in copy_from_user handling Commit 914ee295af418e936ec20a08c1663eaabe4cd07a fixed deadlocks in btrfs_file_write where we would catch page faults on pages we had locked. But, there were a few problems: 1) The x86-32 iov_iter_copy_from_user_atomic code always fails to copy data when the amount to copy is more than 4K and the offset to start copying from is not page aligned. The result was btrfs_file_write looping forever retrying the iov_iter_copy_from_user_atomic We deal with this by changing btrfs_file_write to drop down to single page copies when iov_iter_copy_from_user_atomic starts returning failure. 2) The btrfs_file_write code was leaking delalloc reservations when iov_iter_copy_from_user_atomic returned zero. The looping above would result in the entire filesystem running out of delalloc reservations and constantly trying to flush things to disk. 3) btrfs_file_write will lock down page cache pages, make sure any writeback is finished, do the copy_from_user and then release them. Before the loop runs we check the first and last pages in the write to see if they are only being partially modified. If the start or end of the write isn't aligned, we make sure the corresponding pages are up to date so that we don't introduce garbage into the file. With the copy_from_user changes, we're allowing the VM to reclaim the pages after a partial update from copy_from_user, but we're not making sure the page cache page is up to date when we loop around to resume the write. We deal with this by pushing the up to date checks down into the page prep code. This fits better with how the rest of file_write works. Signed-off-by: Chris Mason <chris.mason@oracle.com> Reported-by: Mitch Harder <mitch.harder@sabayonlinux.org> cc: stable@kernel.org
2011-02-28 14:52:08 +00:00
if (err) {
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
put_page(pages[i]);
if (err == -EAGAIN) {
err = 0;
goto again;
}
Btrfs: fix regressions in copy_from_user handling Commit 914ee295af418e936ec20a08c1663eaabe4cd07a fixed deadlocks in btrfs_file_write where we would catch page faults on pages we had locked. But, there were a few problems: 1) The x86-32 iov_iter_copy_from_user_atomic code always fails to copy data when the amount to copy is more than 4K and the offset to start copying from is not page aligned. The result was btrfs_file_write looping forever retrying the iov_iter_copy_from_user_atomic We deal with this by changing btrfs_file_write to drop down to single page copies when iov_iter_copy_from_user_atomic starts returning failure. 2) The btrfs_file_write code was leaking delalloc reservations when iov_iter_copy_from_user_atomic returned zero. The looping above would result in the entire filesystem running out of delalloc reservations and constantly trying to flush things to disk. 3) btrfs_file_write will lock down page cache pages, make sure any writeback is finished, do the copy_from_user and then release them. Before the loop runs we check the first and last pages in the write to see if they are only being partially modified. If the start or end of the write isn't aligned, we make sure the corresponding pages are up to date so that we don't introduce garbage into the file. With the copy_from_user changes, we're allowing the VM to reclaim the pages after a partial update from copy_from_user, but we're not making sure the page cache page is up to date when we loop around to resume the write. We deal with this by pushing the up to date checks down into the page prep code. This fits better with how the rest of file_write works. Signed-off-by: Chris Mason <chris.mason@oracle.com> Reported-by: Mitch Harder <mitch.harder@sabayonlinux.org> cc: stable@kernel.org
2011-02-28 14:52:08 +00:00
faili = i - 1;
goto fail;
}
wait_on_page_writeback(pages[i]);
}
Btrfs: fix the reserved space leak caused by the race between nonlock dio and buffered io When we ran sysbench on the fs with compression, the following WARN_ONs were triggered: fs/btrfs/inode.c:7829 WARN_ON(BTRFS_I(inode)->outstanding_extents); fs/btrfs/inode.c:7830 WARN_ON(BTRFS_I(inode)->reserved_extents); fs/btrfs/inode.c:7832 WARN_ON(BTRFS_I(inode)->csum_bytes); Steps to reproduce: # mkfs.btrfs -f <dev> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync prepare # cd - # umount <mnt> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync run # cd - # umount <mnt> The reason of this problem is: Task0 Task1 btrfs_direct_IO unlock(&inode->i_mutex) lock(&inode->i_mutex) reserve_space() prepare_pages() lock_extent() clear_extent() unlock_extent() lock_extent() test_extent(uptodate) return false copy_data() set_delalloc_extent() extent need compress go back to buffered write clear_extent(DELALLOC | DIRTY) unlock_extent() Task 0 and 1 wrote the same place, and task0 cleared the delalloc flag which was set by task1, it made the dirty pages in that extents couldn't be flushed into the disk, so the reserved space for that extent was not released at the end. This patch fixes the above bug by unlocking the extent after the delalloc. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-10 11:25:04 +00:00
return 0;
fail:
while (faili >= 0) {
unlock_page(pages[faili]);
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
put_page(pages[faili]);
Btrfs: fix the reserved space leak caused by the race between nonlock dio and buffered io When we ran sysbench on the fs with compression, the following WARN_ONs were triggered: fs/btrfs/inode.c:7829 WARN_ON(BTRFS_I(inode)->outstanding_extents); fs/btrfs/inode.c:7830 WARN_ON(BTRFS_I(inode)->reserved_extents); fs/btrfs/inode.c:7832 WARN_ON(BTRFS_I(inode)->csum_bytes); Steps to reproduce: # mkfs.btrfs -f <dev> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync prepare # cd - # umount <mnt> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync run # cd - # umount <mnt> The reason of this problem is: Task0 Task1 btrfs_direct_IO unlock(&inode->i_mutex) lock(&inode->i_mutex) reserve_space() prepare_pages() lock_extent() clear_extent() unlock_extent() lock_extent() test_extent(uptodate) return false copy_data() set_delalloc_extent() extent need compress go back to buffered write clear_extent(DELALLOC | DIRTY) unlock_extent() Task 0 and 1 wrote the same place, and task0 cleared the delalloc flag which was set by task1, it made the dirty pages in that extents couldn't be flushed into the disk, so the reserved space for that extent was not released at the end. This patch fixes the above bug by unlocking the extent after the delalloc. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-10 11:25:04 +00:00
faili--;
}
return err;
}
Btrfs: fix reported number of inode blocks Currently when there are buffered writes that were not yet flushed and they fall within allocated ranges of the file (that is, not in holes or beyond eof assuming there are no prealloc extents beyond eof), btrfs simply reports an incorrect number of used blocks through the stat(2) system call (or any of its variants), regardless of mount options or inode flags (compress, compress-force, nodatacow). This is because the number of blocks used that is reported is based on the current number of bytes in the vfs inode plus the number of dealloc bytes in the btrfs inode. The later covers bytes that both fall within allocated regions of the file and holes. Example scenarios where the number of reported blocks is wrong while the buffered writes are not flushed: $ mkfs.btrfs -f /dev/sdc $ mount /dev/sdc /mnt/sdc $ xfs_io -f -c "pwrite -S 0xaa 0 64K" /mnt/sdc/foo1 wrote 65536/65536 bytes at offset 0 64 KiB, 16 ops; 0.0000 sec (259.336 MiB/sec and 66390.0415 ops/sec) $ sync $ xfs_io -c "pwrite -S 0xbb 0 64K" /mnt/sdc/foo1 wrote 65536/65536 bytes at offset 0 64 KiB, 16 ops; 0.0000 sec (192.308 MiB/sec and 49230.7692 ops/sec) # The following should have reported 64K... $ du -h /mnt/sdc/foo1 128K /mnt/sdc/foo1 $ sync # After flushing the buffered write, it now reports the correct value. $ du -h /mnt/sdc/foo1 64K /mnt/sdc/foo1 $ xfs_io -f -c "falloc -k 0 128K" -c "pwrite -S 0xaa 0 64K" /mnt/sdc/foo2 wrote 65536/65536 bytes at offset 0 64 KiB, 16 ops; 0.0000 sec (520.833 MiB/sec and 133333.3333 ops/sec) $ sync $ xfs_io -c "pwrite -S 0xbb 64K 64K" /mnt/sdc/foo2 wrote 65536/65536 bytes at offset 65536 64 KiB, 16 ops; 0.0000 sec (260.417 MiB/sec and 66666.6667 ops/sec) # The following should have reported 128K... $ du -h /mnt/sdc/foo2 192K /mnt/sdc/foo2 $ sync # After flushing the buffered write, it now reports the correct value. $ du -h /mnt/sdc/foo2 128K /mnt/sdc/foo2 So the number of used file blocks is simply incorrect, unlike in other filesystems such as ext4 and xfs for example, but only while the buffered writes are not flushed. Fix this by tracking the number of delalloc bytes that fall within holes and beyond eof of a file, and use instead this new counter when reporting the number of used blocks for an inode. Another different problem that exists is that the delalloc bytes counter is reset when writeback starts (by clearing the EXTENT_DEALLOC flag from the respective range in the inode's iotree) and the vfs inode's bytes counter is only incremented when writeback finishes (through insert_reserved_file_extent()). Therefore while writeback is ongoing we simply report a wrong number of blocks used by an inode if the write operation covers a range previously unallocated. While this change does not fix this problem, it does minimizes it a lot by shortening that time window, as the new dealloc bytes counter (new_delalloc_bytes) is only decremented when writeback finishes right before updating the vfs inode's bytes counter. Fully fixing this second problem is not trivial and will be addressed later by a different patch. Signed-off-by: Filipe Manana <fdmanana@suse.com>
2017-04-03 09:45:46 +00:00
static int btrfs_find_new_delalloc_bytes(struct btrfs_inode *inode,
const u64 start,
const u64 len,
struct extent_state **cached_state)
{
u64 search_start = start;
const u64 end = start + len - 1;
while (search_start < end) {
const u64 search_len = end - search_start + 1;
struct extent_map *em;
u64 em_len;
int ret = 0;
em = btrfs_get_extent(inode, NULL, 0, search_start,
search_len, 0);
if (IS_ERR(em))
return PTR_ERR(em);
if (em->block_start != EXTENT_MAP_HOLE)
goto next;
em_len = em->len;
if (em->start < search_start)
em_len -= search_start - em->start;
if (em_len > search_len)
em_len = search_len;
ret = set_extent_bit(&inode->io_tree, search_start,
search_start + em_len - 1,
EXTENT_DELALLOC_NEW,
NULL, cached_state, GFP_NOFS);
next:
search_start = extent_map_end(em);
free_extent_map(em);
if (ret)
return ret;
}
return 0;
}
Btrfs: fix the reserved space leak caused by the race between nonlock dio and buffered io When we ran sysbench on the fs with compression, the following WARN_ONs were triggered: fs/btrfs/inode.c:7829 WARN_ON(BTRFS_I(inode)->outstanding_extents); fs/btrfs/inode.c:7830 WARN_ON(BTRFS_I(inode)->reserved_extents); fs/btrfs/inode.c:7832 WARN_ON(BTRFS_I(inode)->csum_bytes); Steps to reproduce: # mkfs.btrfs -f <dev> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync prepare # cd - # umount <mnt> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync run # cd - # umount <mnt> The reason of this problem is: Task0 Task1 btrfs_direct_IO unlock(&inode->i_mutex) lock(&inode->i_mutex) reserve_space() prepare_pages() lock_extent() clear_extent() unlock_extent() lock_extent() test_extent(uptodate) return false copy_data() set_delalloc_extent() extent need compress go back to buffered write clear_extent(DELALLOC | DIRTY) unlock_extent() Task 0 and 1 wrote the same place, and task0 cleared the delalloc flag which was set by task1, it made the dirty pages in that extents couldn't be flushed into the disk, so the reserved space for that extent was not released at the end. This patch fixes the above bug by unlocking the extent after the delalloc. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-10 11:25:04 +00:00
/*
* This function locks the extent and properly waits for data=ordered extents
* to finish before allowing the pages to be modified if need.
*
* The return value:
* 1 - the extent is locked
* 0 - the extent is not locked, and everything is OK
* -EAGAIN - need re-prepare the pages
* the other < 0 number - Something wrong happens
*/
static noinline int
lock_and_cleanup_extent_if_need(struct btrfs_inode *inode, struct page **pages,
Btrfs: fix the reserved space leak caused by the race between nonlock dio and buffered io When we ran sysbench on the fs with compression, the following WARN_ONs were triggered: fs/btrfs/inode.c:7829 WARN_ON(BTRFS_I(inode)->outstanding_extents); fs/btrfs/inode.c:7830 WARN_ON(BTRFS_I(inode)->reserved_extents); fs/btrfs/inode.c:7832 WARN_ON(BTRFS_I(inode)->csum_bytes); Steps to reproduce: # mkfs.btrfs -f <dev> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync prepare # cd - # umount <mnt> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync run # cd - # umount <mnt> The reason of this problem is: Task0 Task1 btrfs_direct_IO unlock(&inode->i_mutex) lock(&inode->i_mutex) reserve_space() prepare_pages() lock_extent() clear_extent() unlock_extent() lock_extent() test_extent(uptodate) return false copy_data() set_delalloc_extent() extent need compress go back to buffered write clear_extent(DELALLOC | DIRTY) unlock_extent() Task 0 and 1 wrote the same place, and task0 cleared the delalloc flag which was set by task1, it made the dirty pages in that extents couldn't be flushed into the disk, so the reserved space for that extent was not released at the end. This patch fixes the above bug by unlocking the extent after the delalloc. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-10 11:25:04 +00:00
size_t num_pages, loff_t pos,
size_t write_bytes,
Btrfs: fix the reserved space leak caused by the race between nonlock dio and buffered io When we ran sysbench on the fs with compression, the following WARN_ONs were triggered: fs/btrfs/inode.c:7829 WARN_ON(BTRFS_I(inode)->outstanding_extents); fs/btrfs/inode.c:7830 WARN_ON(BTRFS_I(inode)->reserved_extents); fs/btrfs/inode.c:7832 WARN_ON(BTRFS_I(inode)->csum_bytes); Steps to reproduce: # mkfs.btrfs -f <dev> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync prepare # cd - # umount <mnt> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync run # cd - # umount <mnt> The reason of this problem is: Task0 Task1 btrfs_direct_IO unlock(&inode->i_mutex) lock(&inode->i_mutex) reserve_space() prepare_pages() lock_extent() clear_extent() unlock_extent() lock_extent() test_extent(uptodate) return false copy_data() set_delalloc_extent() extent need compress go back to buffered write clear_extent(DELALLOC | DIRTY) unlock_extent() Task 0 and 1 wrote the same place, and task0 cleared the delalloc flag which was set by task1, it made the dirty pages in that extents couldn't be flushed into the disk, so the reserved space for that extent was not released at the end. This patch fixes the above bug by unlocking the extent after the delalloc. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-10 11:25:04 +00:00
u64 *lockstart, u64 *lockend,
struct extent_state **cached_state)
{
struct btrfs_fs_info *fs_info = btrfs_sb(inode->vfs_inode.i_sb);
Btrfs: fix the reserved space leak caused by the race between nonlock dio and buffered io When we ran sysbench on the fs with compression, the following WARN_ONs were triggered: fs/btrfs/inode.c:7829 WARN_ON(BTRFS_I(inode)->outstanding_extents); fs/btrfs/inode.c:7830 WARN_ON(BTRFS_I(inode)->reserved_extents); fs/btrfs/inode.c:7832 WARN_ON(BTRFS_I(inode)->csum_bytes); Steps to reproduce: # mkfs.btrfs -f <dev> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync prepare # cd - # umount <mnt> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync run # cd - # umount <mnt> The reason of this problem is: Task0 Task1 btrfs_direct_IO unlock(&inode->i_mutex) lock(&inode->i_mutex) reserve_space() prepare_pages() lock_extent() clear_extent() unlock_extent() lock_extent() test_extent(uptodate) return false copy_data() set_delalloc_extent() extent need compress go back to buffered write clear_extent(DELALLOC | DIRTY) unlock_extent() Task 0 and 1 wrote the same place, and task0 cleared the delalloc flag which was set by task1, it made the dirty pages in that extents couldn't be flushed into the disk, so the reserved space for that extent was not released at the end. This patch fixes the above bug by unlocking the extent after the delalloc. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-10 11:25:04 +00:00
u64 start_pos;
u64 last_pos;
int i;
int ret = 0;
start_pos = round_down(pos, fs_info->sectorsize);
last_pos = start_pos
+ round_up(pos + write_bytes - start_pos,
fs_info->sectorsize) - 1;
Btrfs: fix the reserved space leak caused by the race between nonlock dio and buffered io When we ran sysbench on the fs with compression, the following WARN_ONs were triggered: fs/btrfs/inode.c:7829 WARN_ON(BTRFS_I(inode)->outstanding_extents); fs/btrfs/inode.c:7830 WARN_ON(BTRFS_I(inode)->reserved_extents); fs/btrfs/inode.c:7832 WARN_ON(BTRFS_I(inode)->csum_bytes); Steps to reproduce: # mkfs.btrfs -f <dev> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync prepare # cd - # umount <mnt> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync run # cd - # umount <mnt> The reason of this problem is: Task0 Task1 btrfs_direct_IO unlock(&inode->i_mutex) lock(&inode->i_mutex) reserve_space() prepare_pages() lock_extent() clear_extent() unlock_extent() lock_extent() test_extent(uptodate) return false copy_data() set_delalloc_extent() extent need compress go back to buffered write clear_extent(DELALLOC | DIRTY) unlock_extent() Task 0 and 1 wrote the same place, and task0 cleared the delalloc flag which was set by task1, it made the dirty pages in that extents couldn't be flushed into the disk, so the reserved space for that extent was not released at the end. This patch fixes the above bug by unlocking the extent after the delalloc. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-10 11:25:04 +00:00
Btrfs: fix reported number of inode blocks Currently when there are buffered writes that were not yet flushed and they fall within allocated ranges of the file (that is, not in holes or beyond eof assuming there are no prealloc extents beyond eof), btrfs simply reports an incorrect number of used blocks through the stat(2) system call (or any of its variants), regardless of mount options or inode flags (compress, compress-force, nodatacow). This is because the number of blocks used that is reported is based on the current number of bytes in the vfs inode plus the number of dealloc bytes in the btrfs inode. The later covers bytes that both fall within allocated regions of the file and holes. Example scenarios where the number of reported blocks is wrong while the buffered writes are not flushed: $ mkfs.btrfs -f /dev/sdc $ mount /dev/sdc /mnt/sdc $ xfs_io -f -c "pwrite -S 0xaa 0 64K" /mnt/sdc/foo1 wrote 65536/65536 bytes at offset 0 64 KiB, 16 ops; 0.0000 sec (259.336 MiB/sec and 66390.0415 ops/sec) $ sync $ xfs_io -c "pwrite -S 0xbb 0 64K" /mnt/sdc/foo1 wrote 65536/65536 bytes at offset 0 64 KiB, 16 ops; 0.0000 sec (192.308 MiB/sec and 49230.7692 ops/sec) # The following should have reported 64K... $ du -h /mnt/sdc/foo1 128K /mnt/sdc/foo1 $ sync # After flushing the buffered write, it now reports the correct value. $ du -h /mnt/sdc/foo1 64K /mnt/sdc/foo1 $ xfs_io -f -c "falloc -k 0 128K" -c "pwrite -S 0xaa 0 64K" /mnt/sdc/foo2 wrote 65536/65536 bytes at offset 0 64 KiB, 16 ops; 0.0000 sec (520.833 MiB/sec and 133333.3333 ops/sec) $ sync $ xfs_io -c "pwrite -S 0xbb 64K 64K" /mnt/sdc/foo2 wrote 65536/65536 bytes at offset 65536 64 KiB, 16 ops; 0.0000 sec (260.417 MiB/sec and 66666.6667 ops/sec) # The following should have reported 128K... $ du -h /mnt/sdc/foo2 192K /mnt/sdc/foo2 $ sync # After flushing the buffered write, it now reports the correct value. $ du -h /mnt/sdc/foo2 128K /mnt/sdc/foo2 So the number of used file blocks is simply incorrect, unlike in other filesystems such as ext4 and xfs for example, but only while the buffered writes are not flushed. Fix this by tracking the number of delalloc bytes that fall within holes and beyond eof of a file, and use instead this new counter when reporting the number of used blocks for an inode. Another different problem that exists is that the delalloc bytes counter is reset when writeback starts (by clearing the EXTENT_DEALLOC flag from the respective range in the inode's iotree) and the vfs inode's bytes counter is only incremented when writeback finishes (through insert_reserved_file_extent()). Therefore while writeback is ongoing we simply report a wrong number of blocks used by an inode if the write operation covers a range previously unallocated. While this change does not fix this problem, it does minimizes it a lot by shortening that time window, as the new dealloc bytes counter (new_delalloc_bytes) is only decremented when writeback finishes right before updating the vfs inode's bytes counter. Fully fixing this second problem is not trivial and will be addressed later by a different patch. Signed-off-by: Filipe Manana <fdmanana@suse.com>
2017-04-03 09:45:46 +00:00
if (start_pos < inode->vfs_inode.i_size ||
(inode->flags & BTRFS_INODE_PREALLOC)) {
struct btrfs_ordered_extent *ordered;
Btrfs: fix reported number of inode blocks Currently when there are buffered writes that were not yet flushed and they fall within allocated ranges of the file (that is, not in holes or beyond eof assuming there are no prealloc extents beyond eof), btrfs simply reports an incorrect number of used blocks through the stat(2) system call (or any of its variants), regardless of mount options or inode flags (compress, compress-force, nodatacow). This is because the number of blocks used that is reported is based on the current number of bytes in the vfs inode plus the number of dealloc bytes in the btrfs inode. The later covers bytes that both fall within allocated regions of the file and holes. Example scenarios where the number of reported blocks is wrong while the buffered writes are not flushed: $ mkfs.btrfs -f /dev/sdc $ mount /dev/sdc /mnt/sdc $ xfs_io -f -c "pwrite -S 0xaa 0 64K" /mnt/sdc/foo1 wrote 65536/65536 bytes at offset 0 64 KiB, 16 ops; 0.0000 sec (259.336 MiB/sec and 66390.0415 ops/sec) $ sync $ xfs_io -c "pwrite -S 0xbb 0 64K" /mnt/sdc/foo1 wrote 65536/65536 bytes at offset 0 64 KiB, 16 ops; 0.0000 sec (192.308 MiB/sec and 49230.7692 ops/sec) # The following should have reported 64K... $ du -h /mnt/sdc/foo1 128K /mnt/sdc/foo1 $ sync # After flushing the buffered write, it now reports the correct value. $ du -h /mnt/sdc/foo1 64K /mnt/sdc/foo1 $ xfs_io -f -c "falloc -k 0 128K" -c "pwrite -S 0xaa 0 64K" /mnt/sdc/foo2 wrote 65536/65536 bytes at offset 0 64 KiB, 16 ops; 0.0000 sec (520.833 MiB/sec and 133333.3333 ops/sec) $ sync $ xfs_io -c "pwrite -S 0xbb 64K 64K" /mnt/sdc/foo2 wrote 65536/65536 bytes at offset 65536 64 KiB, 16 ops; 0.0000 sec (260.417 MiB/sec and 66666.6667 ops/sec) # The following should have reported 128K... $ du -h /mnt/sdc/foo2 192K /mnt/sdc/foo2 $ sync # After flushing the buffered write, it now reports the correct value. $ du -h /mnt/sdc/foo2 128K /mnt/sdc/foo2 So the number of used file blocks is simply incorrect, unlike in other filesystems such as ext4 and xfs for example, but only while the buffered writes are not flushed. Fix this by tracking the number of delalloc bytes that fall within holes and beyond eof of a file, and use instead this new counter when reporting the number of used blocks for an inode. Another different problem that exists is that the delalloc bytes counter is reset when writeback starts (by clearing the EXTENT_DEALLOC flag from the respective range in the inode's iotree) and the vfs inode's bytes counter is only incremented when writeback finishes (through insert_reserved_file_extent()). Therefore while writeback is ongoing we simply report a wrong number of blocks used by an inode if the write operation covers a range previously unallocated. While this change does not fix this problem, it does minimizes it a lot by shortening that time window, as the new dealloc bytes counter (new_delalloc_bytes) is only decremented when writeback finishes right before updating the vfs inode's bytes counter. Fully fixing this second problem is not trivial and will be addressed later by a different patch. Signed-off-by: Filipe Manana <fdmanana@suse.com>
2017-04-03 09:45:46 +00:00
unsigned int clear_bits;
lock_extent_bits(&inode->io_tree, start_pos, last_pos,
cached_state);
ordered = btrfs_lookup_ordered_range(inode, start_pos,
last_pos - start_pos + 1);
if (ordered &&
ordered->file_offset + ordered->len > start_pos &&
Btrfs: fix the reserved space leak caused by the race between nonlock dio and buffered io When we ran sysbench on the fs with compression, the following WARN_ONs were triggered: fs/btrfs/inode.c:7829 WARN_ON(BTRFS_I(inode)->outstanding_extents); fs/btrfs/inode.c:7830 WARN_ON(BTRFS_I(inode)->reserved_extents); fs/btrfs/inode.c:7832 WARN_ON(BTRFS_I(inode)->csum_bytes); Steps to reproduce: # mkfs.btrfs -f <dev> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync prepare # cd - # umount <mnt> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync run # cd - # umount <mnt> The reason of this problem is: Task0 Task1 btrfs_direct_IO unlock(&inode->i_mutex) lock(&inode->i_mutex) reserve_space() prepare_pages() lock_extent() clear_extent() unlock_extent() lock_extent() test_extent(uptodate) return false copy_data() set_delalloc_extent() extent need compress go back to buffered write clear_extent(DELALLOC | DIRTY) unlock_extent() Task 0 and 1 wrote the same place, and task0 cleared the delalloc flag which was set by task1, it made the dirty pages in that extents couldn't be flushed into the disk, so the reserved space for that extent was not released at the end. This patch fixes the above bug by unlocking the extent after the delalloc. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-10 11:25:04 +00:00
ordered->file_offset <= last_pos) {
unlock_extent_cached(&inode->io_tree, start_pos,
last_pos, cached_state, GFP_NOFS);
for (i = 0; i < num_pages; i++) {
unlock_page(pages[i]);
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
put_page(pages[i]);
}
btrfs_start_ordered_extent(&inode->vfs_inode,
ordered, 1);
btrfs_put_ordered_extent(ordered);
return -EAGAIN;
}
if (ordered)
btrfs_put_ordered_extent(ordered);
Btrfs: fix reported number of inode blocks Currently when there are buffered writes that were not yet flushed and they fall within allocated ranges of the file (that is, not in holes or beyond eof assuming there are no prealloc extents beyond eof), btrfs simply reports an incorrect number of used blocks through the stat(2) system call (or any of its variants), regardless of mount options or inode flags (compress, compress-force, nodatacow). This is because the number of blocks used that is reported is based on the current number of bytes in the vfs inode plus the number of dealloc bytes in the btrfs inode. The later covers bytes that both fall within allocated regions of the file and holes. Example scenarios where the number of reported blocks is wrong while the buffered writes are not flushed: $ mkfs.btrfs -f /dev/sdc $ mount /dev/sdc /mnt/sdc $ xfs_io -f -c "pwrite -S 0xaa 0 64K" /mnt/sdc/foo1 wrote 65536/65536 bytes at offset 0 64 KiB, 16 ops; 0.0000 sec (259.336 MiB/sec and 66390.0415 ops/sec) $ sync $ xfs_io -c "pwrite -S 0xbb 0 64K" /mnt/sdc/foo1 wrote 65536/65536 bytes at offset 0 64 KiB, 16 ops; 0.0000 sec (192.308 MiB/sec and 49230.7692 ops/sec) # The following should have reported 64K... $ du -h /mnt/sdc/foo1 128K /mnt/sdc/foo1 $ sync # After flushing the buffered write, it now reports the correct value. $ du -h /mnt/sdc/foo1 64K /mnt/sdc/foo1 $ xfs_io -f -c "falloc -k 0 128K" -c "pwrite -S 0xaa 0 64K" /mnt/sdc/foo2 wrote 65536/65536 bytes at offset 0 64 KiB, 16 ops; 0.0000 sec (520.833 MiB/sec and 133333.3333 ops/sec) $ sync $ xfs_io -c "pwrite -S 0xbb 64K 64K" /mnt/sdc/foo2 wrote 65536/65536 bytes at offset 65536 64 KiB, 16 ops; 0.0000 sec (260.417 MiB/sec and 66666.6667 ops/sec) # The following should have reported 128K... $ du -h /mnt/sdc/foo2 192K /mnt/sdc/foo2 $ sync # After flushing the buffered write, it now reports the correct value. $ du -h /mnt/sdc/foo2 128K /mnt/sdc/foo2 So the number of used file blocks is simply incorrect, unlike in other filesystems such as ext4 and xfs for example, but only while the buffered writes are not flushed. Fix this by tracking the number of delalloc bytes that fall within holes and beyond eof of a file, and use instead this new counter when reporting the number of used blocks for an inode. Another different problem that exists is that the delalloc bytes counter is reset when writeback starts (by clearing the EXTENT_DEALLOC flag from the respective range in the inode's iotree) and the vfs inode's bytes counter is only incremented when writeback finishes (through insert_reserved_file_extent()). Therefore while writeback is ongoing we simply report a wrong number of blocks used by an inode if the write operation covers a range previously unallocated. While this change does not fix this problem, it does minimizes it a lot by shortening that time window, as the new dealloc bytes counter (new_delalloc_bytes) is only decremented when writeback finishes right before updating the vfs inode's bytes counter. Fully fixing this second problem is not trivial and will be addressed later by a different patch. Signed-off-by: Filipe Manana <fdmanana@suse.com>
2017-04-03 09:45:46 +00:00
ret = btrfs_find_new_delalloc_bytes(inode, start_pos,
last_pos - start_pos + 1,
cached_state);
clear_bits = EXTENT_DIRTY | EXTENT_DELALLOC |
EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG;
if (ret)
clear_bits |= EXTENT_DELALLOC_NEW | EXTENT_LOCKED;
clear_extent_bit(&inode->io_tree, start_pos,
Btrfs: fix reported number of inode blocks Currently when there are buffered writes that were not yet flushed and they fall within allocated ranges of the file (that is, not in holes or beyond eof assuming there are no prealloc extents beyond eof), btrfs simply reports an incorrect number of used blocks through the stat(2) system call (or any of its variants), regardless of mount options or inode flags (compress, compress-force, nodatacow). This is because the number of blocks used that is reported is based on the current number of bytes in the vfs inode plus the number of dealloc bytes in the btrfs inode. The later covers bytes that both fall within allocated regions of the file and holes. Example scenarios where the number of reported blocks is wrong while the buffered writes are not flushed: $ mkfs.btrfs -f /dev/sdc $ mount /dev/sdc /mnt/sdc $ xfs_io -f -c "pwrite -S 0xaa 0 64K" /mnt/sdc/foo1 wrote 65536/65536 bytes at offset 0 64 KiB, 16 ops; 0.0000 sec (259.336 MiB/sec and 66390.0415 ops/sec) $ sync $ xfs_io -c "pwrite -S 0xbb 0 64K" /mnt/sdc/foo1 wrote 65536/65536 bytes at offset 0 64 KiB, 16 ops; 0.0000 sec (192.308 MiB/sec and 49230.7692 ops/sec) # The following should have reported 64K... $ du -h /mnt/sdc/foo1 128K /mnt/sdc/foo1 $ sync # After flushing the buffered write, it now reports the correct value. $ du -h /mnt/sdc/foo1 64K /mnt/sdc/foo1 $ xfs_io -f -c "falloc -k 0 128K" -c "pwrite -S 0xaa 0 64K" /mnt/sdc/foo2 wrote 65536/65536 bytes at offset 0 64 KiB, 16 ops; 0.0000 sec (520.833 MiB/sec and 133333.3333 ops/sec) $ sync $ xfs_io -c "pwrite -S 0xbb 64K 64K" /mnt/sdc/foo2 wrote 65536/65536 bytes at offset 65536 64 KiB, 16 ops; 0.0000 sec (260.417 MiB/sec and 66666.6667 ops/sec) # The following should have reported 128K... $ du -h /mnt/sdc/foo2 192K /mnt/sdc/foo2 $ sync # After flushing the buffered write, it now reports the correct value. $ du -h /mnt/sdc/foo2 128K /mnt/sdc/foo2 So the number of used file blocks is simply incorrect, unlike in other filesystems such as ext4 and xfs for example, but only while the buffered writes are not flushed. Fix this by tracking the number of delalloc bytes that fall within holes and beyond eof of a file, and use instead this new counter when reporting the number of used blocks for an inode. Another different problem that exists is that the delalloc bytes counter is reset when writeback starts (by clearing the EXTENT_DEALLOC flag from the respective range in the inode's iotree) and the vfs inode's bytes counter is only incremented when writeback finishes (through insert_reserved_file_extent()). Therefore while writeback is ongoing we simply report a wrong number of blocks used by an inode if the write operation covers a range previously unallocated. While this change does not fix this problem, it does minimizes it a lot by shortening that time window, as the new dealloc bytes counter (new_delalloc_bytes) is only decremented when writeback finishes right before updating the vfs inode's bytes counter. Fully fixing this second problem is not trivial and will be addressed later by a different patch. Signed-off-by: Filipe Manana <fdmanana@suse.com>
2017-04-03 09:45:46 +00:00
last_pos, clear_bits,
(clear_bits & EXTENT_LOCKED) ? 1 : 0,
0, cached_state, GFP_NOFS);
if (ret)
return ret;
Btrfs: fix the reserved space leak caused by the race between nonlock dio and buffered io When we ran sysbench on the fs with compression, the following WARN_ONs were triggered: fs/btrfs/inode.c:7829 WARN_ON(BTRFS_I(inode)->outstanding_extents); fs/btrfs/inode.c:7830 WARN_ON(BTRFS_I(inode)->reserved_extents); fs/btrfs/inode.c:7832 WARN_ON(BTRFS_I(inode)->csum_bytes); Steps to reproduce: # mkfs.btrfs -f <dev> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync prepare # cd - # umount <mnt> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync run # cd - # umount <mnt> The reason of this problem is: Task0 Task1 btrfs_direct_IO unlock(&inode->i_mutex) lock(&inode->i_mutex) reserve_space() prepare_pages() lock_extent() clear_extent() unlock_extent() lock_extent() test_extent(uptodate) return false copy_data() set_delalloc_extent() extent need compress go back to buffered write clear_extent(DELALLOC | DIRTY) unlock_extent() Task 0 and 1 wrote the same place, and task0 cleared the delalloc flag which was set by task1, it made the dirty pages in that extents couldn't be flushed into the disk, so the reserved space for that extent was not released at the end. This patch fixes the above bug by unlocking the extent after the delalloc. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-10 11:25:04 +00:00
*lockstart = start_pos;
*lockend = last_pos;
ret = 1;
}
Btrfs: fix the reserved space leak caused by the race between nonlock dio and buffered io When we ran sysbench on the fs with compression, the following WARN_ONs were triggered: fs/btrfs/inode.c:7829 WARN_ON(BTRFS_I(inode)->outstanding_extents); fs/btrfs/inode.c:7830 WARN_ON(BTRFS_I(inode)->reserved_extents); fs/btrfs/inode.c:7832 WARN_ON(BTRFS_I(inode)->csum_bytes); Steps to reproduce: # mkfs.btrfs -f <dev> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync prepare # cd - # umount <mnt> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync run # cd - # umount <mnt> The reason of this problem is: Task0 Task1 btrfs_direct_IO unlock(&inode->i_mutex) lock(&inode->i_mutex) reserve_space() prepare_pages() lock_extent() clear_extent() unlock_extent() lock_extent() test_extent(uptodate) return false copy_data() set_delalloc_extent() extent need compress go back to buffered write clear_extent(DELALLOC | DIRTY) unlock_extent() Task 0 and 1 wrote the same place, and task0 cleared the delalloc flag which was set by task1, it made the dirty pages in that extents couldn't be flushed into the disk, so the reserved space for that extent was not released at the end. This patch fixes the above bug by unlocking the extent after the delalloc. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-10 11:25:04 +00:00
for (i = 0; i < num_pages; i++) {
if (clear_page_dirty_for_io(pages[i]))
account_page_redirty(pages[i]);
set_page_extent_mapped(pages[i]);
WARN_ON(!PageLocked(pages[i]));
}
Btrfs: fix regressions in copy_from_user handling Commit 914ee295af418e936ec20a08c1663eaabe4cd07a fixed deadlocks in btrfs_file_write where we would catch page faults on pages we had locked. But, there were a few problems: 1) The x86-32 iov_iter_copy_from_user_atomic code always fails to copy data when the amount to copy is more than 4K and the offset to start copying from is not page aligned. The result was btrfs_file_write looping forever retrying the iov_iter_copy_from_user_atomic We deal with this by changing btrfs_file_write to drop down to single page copies when iov_iter_copy_from_user_atomic starts returning failure. 2) The btrfs_file_write code was leaking delalloc reservations when iov_iter_copy_from_user_atomic returned zero. The looping above would result in the entire filesystem running out of delalloc reservations and constantly trying to flush things to disk. 3) btrfs_file_write will lock down page cache pages, make sure any writeback is finished, do the copy_from_user and then release them. Before the loop runs we check the first and last pages in the write to see if they are only being partially modified. If the start or end of the write isn't aligned, we make sure the corresponding pages are up to date so that we don't introduce garbage into the file. With the copy_from_user changes, we're allowing the VM to reclaim the pages after a partial update from copy_from_user, but we're not making sure the page cache page is up to date when we loop around to resume the write. We deal with this by pushing the up to date checks down into the page prep code. This fits better with how the rest of file_write works. Signed-off-by: Chris Mason <chris.mason@oracle.com> Reported-by: Mitch Harder <mitch.harder@sabayonlinux.org> cc: stable@kernel.org
2011-02-28 14:52:08 +00:00
Btrfs: fix the reserved space leak caused by the race between nonlock dio and buffered io When we ran sysbench on the fs with compression, the following WARN_ONs were triggered: fs/btrfs/inode.c:7829 WARN_ON(BTRFS_I(inode)->outstanding_extents); fs/btrfs/inode.c:7830 WARN_ON(BTRFS_I(inode)->reserved_extents); fs/btrfs/inode.c:7832 WARN_ON(BTRFS_I(inode)->csum_bytes); Steps to reproduce: # mkfs.btrfs -f <dev> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync prepare # cd - # umount <mnt> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync run # cd - # umount <mnt> The reason of this problem is: Task0 Task1 btrfs_direct_IO unlock(&inode->i_mutex) lock(&inode->i_mutex) reserve_space() prepare_pages() lock_extent() clear_extent() unlock_extent() lock_extent() test_extent(uptodate) return false copy_data() set_delalloc_extent() extent need compress go back to buffered write clear_extent(DELALLOC | DIRTY) unlock_extent() Task 0 and 1 wrote the same place, and task0 cleared the delalloc flag which was set by task1, it made the dirty pages in that extents couldn't be flushed into the disk, so the reserved space for that extent was not released at the end. This patch fixes the above bug by unlocking the extent after the delalloc. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-10 11:25:04 +00:00
return ret;
}
static noinline int check_can_nocow(struct btrfs_inode *inode, loff_t pos,
size_t *write_bytes)
{
struct btrfs_fs_info *fs_info = btrfs_sb(inode->vfs_inode.i_sb);
struct btrfs_root *root = inode->root;
struct btrfs_ordered_extent *ordered;
u64 lockstart, lockend;
u64 num_bytes;
int ret;
Btrfs: fix snapshot inconsistency after a file write followed by truncate If right after starting the snapshot creation ioctl we perform a write against a file followed by a truncate, with both operations increasing the file's size, we can get a snapshot tree that reflects a state of the source subvolume's tree where the file truncation happened but the write operation didn't. This leaves a gap between 2 file extent items of the inode, which makes btrfs' fsck complain about it. For example, if we perform the following file operations: $ mkfs.btrfs -f /dev/vdd $ mount /dev/vdd /mnt $ xfs_io -f \ -c "pwrite -S 0xaa -b 32K 0 32K" \ -c "fsync" \ -c "pwrite -S 0xbb -b 32770 16K 32770" \ -c "truncate 90123" \ /mnt/foobar and the snapshot creation ioctl was just called before the second write, we often can get the following inode items in the snapshot's btree: item 120 key (257 INODE_ITEM 0) itemoff 7987 itemsize 160 inode generation 146 transid 7 size 90123 block group 0 mode 100600 links 1 uid 0 gid 0 rdev 0 flags 0x0 item 121 key (257 INODE_REF 256) itemoff 7967 itemsize 20 inode ref index 282 namelen 10 name: foobar item 122 key (257 EXTENT_DATA 0) itemoff 7914 itemsize 53 extent data disk byte 1104855040 nr 32768 extent data offset 0 nr 32768 ram 32768 extent compression 0 item 123 key (257 EXTENT_DATA 53248) itemoff 7861 itemsize 53 extent data disk byte 0 nr 0 extent data offset 0 nr 40960 ram 40960 extent compression 0 There's a file range, corresponding to the interval [32K; ALIGN(16K + 32770, 4096)[ for which there's no file extent item covering it. This is because the file write and file truncate operations happened both right after the snapshot creation ioctl called btrfs_start_delalloc_inodes(), which means we didn't start and wait for the ordered extent that matches the write and, in btrfs_setsize(), we were able to call btrfs_cont_expand() before being able to commit the current transaction in the snapshot creation ioctl. So this made it possibe to insert the hole file extent item in the source subvolume (which represents the region added by the truncate) right before the transaction commit from the snapshot creation ioctl. Btrfs' fsck tool complains about such cases with a message like the following: "root 331 inode 257 errors 100, file extent discount" >From a user perspective, the expectation when a snapshot is created while those file operations are being performed is that the snapshot will have a file that either: 1) is empty 2) only the first write was captured 3) only the 2 writes were captured 4) both writes and the truncation were captured But never capture a state where only the first write and the truncation were captured (since the second write was performed before the truncation). A test case for xfstests follows. Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-10-29 11:57:59 +00:00
ret = btrfs_start_write_no_snapshoting(root);
if (!ret)
return -ENOSPC;
lockstart = round_down(pos, fs_info->sectorsize);
lockend = round_up(pos + *write_bytes,
fs_info->sectorsize) - 1;
while (1) {
lock_extent(&inode->io_tree, lockstart, lockend);
ordered = btrfs_lookup_ordered_range(inode, lockstart,
lockend - lockstart + 1);
if (!ordered) {
break;
}
unlock_extent(&inode->io_tree, lockstart, lockend);
btrfs_start_ordered_extent(&inode->vfs_inode, ordered, 1);
btrfs_put_ordered_extent(ordered);
}
num_bytes = lockend - lockstart + 1;
ret = can_nocow_extent(&inode->vfs_inode, lockstart, &num_bytes,
NULL, NULL, NULL);
if (ret <= 0) {
ret = 0;
Btrfs: fix snapshot inconsistency after a file write followed by truncate If right after starting the snapshot creation ioctl we perform a write against a file followed by a truncate, with both operations increasing the file's size, we can get a snapshot tree that reflects a state of the source subvolume's tree where the file truncation happened but the write operation didn't. This leaves a gap between 2 file extent items of the inode, which makes btrfs' fsck complain about it. For example, if we perform the following file operations: $ mkfs.btrfs -f /dev/vdd $ mount /dev/vdd /mnt $ xfs_io -f \ -c "pwrite -S 0xaa -b 32K 0 32K" \ -c "fsync" \ -c "pwrite -S 0xbb -b 32770 16K 32770" \ -c "truncate 90123" \ /mnt/foobar and the snapshot creation ioctl was just called before the second write, we often can get the following inode items in the snapshot's btree: item 120 key (257 INODE_ITEM 0) itemoff 7987 itemsize 160 inode generation 146 transid 7 size 90123 block group 0 mode 100600 links 1 uid 0 gid 0 rdev 0 flags 0x0 item 121 key (257 INODE_REF 256) itemoff 7967 itemsize 20 inode ref index 282 namelen 10 name: foobar item 122 key (257 EXTENT_DATA 0) itemoff 7914 itemsize 53 extent data disk byte 1104855040 nr 32768 extent data offset 0 nr 32768 ram 32768 extent compression 0 item 123 key (257 EXTENT_DATA 53248) itemoff 7861 itemsize 53 extent data disk byte 0 nr 0 extent data offset 0 nr 40960 ram 40960 extent compression 0 There's a file range, corresponding to the interval [32K; ALIGN(16K + 32770, 4096)[ for which there's no file extent item covering it. This is because the file write and file truncate operations happened both right after the snapshot creation ioctl called btrfs_start_delalloc_inodes(), which means we didn't start and wait for the ordered extent that matches the write and, in btrfs_setsize(), we were able to call btrfs_cont_expand() before being able to commit the current transaction in the snapshot creation ioctl. So this made it possibe to insert the hole file extent item in the source subvolume (which represents the region added by the truncate) right before the transaction commit from the snapshot creation ioctl. Btrfs' fsck tool complains about such cases with a message like the following: "root 331 inode 257 errors 100, file extent discount" >From a user perspective, the expectation when a snapshot is created while those file operations are being performed is that the snapshot will have a file that either: 1) is empty 2) only the first write was captured 3) only the 2 writes were captured 4) both writes and the truncation were captured But never capture a state where only the first write and the truncation were captured (since the second write was performed before the truncation). A test case for xfstests follows. Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-10-29 11:57:59 +00:00
btrfs_end_write_no_snapshoting(root);
} else {
*write_bytes = min_t(size_t, *write_bytes ,
num_bytes - pos + lockstart);
}
unlock_extent(&inode->io_tree, lockstart, lockend);
return ret;
}
static noinline ssize_t __btrfs_buffered_write(struct file *file,
struct iov_iter *i,
loff_t pos)
{
struct inode *inode = file_inode(file);
struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
struct btrfs_root *root = BTRFS_I(inode)->root;
struct page **pages = NULL;
Btrfs: fix the reserved space leak caused by the race between nonlock dio and buffered io When we ran sysbench on the fs with compression, the following WARN_ONs were triggered: fs/btrfs/inode.c:7829 WARN_ON(BTRFS_I(inode)->outstanding_extents); fs/btrfs/inode.c:7830 WARN_ON(BTRFS_I(inode)->reserved_extents); fs/btrfs/inode.c:7832 WARN_ON(BTRFS_I(inode)->csum_bytes); Steps to reproduce: # mkfs.btrfs -f <dev> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync prepare # cd - # umount <mnt> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync run # cd - # umount <mnt> The reason of this problem is: Task0 Task1 btrfs_direct_IO unlock(&inode->i_mutex) lock(&inode->i_mutex) reserve_space() prepare_pages() lock_extent() clear_extent() unlock_extent() lock_extent() test_extent(uptodate) return false copy_data() set_delalloc_extent() extent need compress go back to buffered write clear_extent(DELALLOC | DIRTY) unlock_extent() Task 0 and 1 wrote the same place, and task0 cleared the delalloc flag which was set by task1, it made the dirty pages in that extents couldn't be flushed into the disk, so the reserved space for that extent was not released at the end. This patch fixes the above bug by unlocking the extent after the delalloc. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-10 11:25:04 +00:00
struct extent_state *cached_state = NULL;
u64 release_bytes = 0;
Btrfs: fix the reserved space leak caused by the race between nonlock dio and buffered io When we ran sysbench on the fs with compression, the following WARN_ONs were triggered: fs/btrfs/inode.c:7829 WARN_ON(BTRFS_I(inode)->outstanding_extents); fs/btrfs/inode.c:7830 WARN_ON(BTRFS_I(inode)->reserved_extents); fs/btrfs/inode.c:7832 WARN_ON(BTRFS_I(inode)->csum_bytes); Steps to reproduce: # mkfs.btrfs -f <dev> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync prepare # cd - # umount <mnt> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync run # cd - # umount <mnt> The reason of this problem is: Task0 Task1 btrfs_direct_IO unlock(&inode->i_mutex) lock(&inode->i_mutex) reserve_space() prepare_pages() lock_extent() clear_extent() unlock_extent() lock_extent() test_extent(uptodate) return false copy_data() set_delalloc_extent() extent need compress go back to buffered write clear_extent(DELALLOC | DIRTY) unlock_extent() Task 0 and 1 wrote the same place, and task0 cleared the delalloc flag which was set by task1, it made the dirty pages in that extents couldn't be flushed into the disk, so the reserved space for that extent was not released at the end. This patch fixes the above bug by unlocking the extent after the delalloc. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-10 11:25:04 +00:00
u64 lockstart;
u64 lockend;
size_t num_written = 0;
int nrptrs;
int ret = 0;
bool only_release_metadata = false;
bool force_page_uptodate = false;
Btrfs: fix the reserved space leak caused by the race between nonlock dio and buffered io When we ran sysbench on the fs with compression, the following WARN_ONs were triggered: fs/btrfs/inode.c:7829 WARN_ON(BTRFS_I(inode)->outstanding_extents); fs/btrfs/inode.c:7830 WARN_ON(BTRFS_I(inode)->reserved_extents); fs/btrfs/inode.c:7832 WARN_ON(BTRFS_I(inode)->csum_bytes); Steps to reproduce: # mkfs.btrfs -f <dev> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync prepare # cd - # umount <mnt> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync run # cd - # umount <mnt> The reason of this problem is: Task0 Task1 btrfs_direct_IO unlock(&inode->i_mutex) lock(&inode->i_mutex) reserve_space() prepare_pages() lock_extent() clear_extent() unlock_extent() lock_extent() test_extent(uptodate) return false copy_data() set_delalloc_extent() extent need compress go back to buffered write clear_extent(DELALLOC | DIRTY) unlock_extent() Task 0 and 1 wrote the same place, and task0 cleared the delalloc flag which was set by task1, it made the dirty pages in that extents couldn't be flushed into the disk, so the reserved space for that extent was not released at the end. This patch fixes the above bug by unlocking the extent after the delalloc. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-10 11:25:04 +00:00
bool need_unlock;
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
nrptrs = min(DIV_ROUND_UP(iov_iter_count(i), PAGE_SIZE),
PAGE_SIZE / (sizeof(struct page *)));
nrptrs = min(nrptrs, current->nr_dirtied_pause - current->nr_dirtied);
nrptrs = max(nrptrs, 8);
pages = kmalloc_array(nrptrs, sizeof(struct page *), GFP_KERNEL);
if (!pages)
return -ENOMEM;
while (iov_iter_count(i) > 0) {
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
size_t offset = pos & (PAGE_SIZE - 1);
size_t sector_offset;
size_t write_bytes = min(iov_iter_count(i),
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
nrptrs * (size_t)PAGE_SIZE -
offset);
size_t num_pages = DIV_ROUND_UP(write_bytes + offset,
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
PAGE_SIZE);
size_t reserve_bytes;
size_t dirty_pages;
size_t copied;
size_t dirty_sectors;
size_t num_sectors;
WARN_ON(num_pages > nrptrs);
/*
* Fault pages before locking them in prepare_pages
* to avoid recursive lock
*/
if (unlikely(iov_iter_fault_in_readable(i, write_bytes))) {
ret = -EFAULT;
break;
}
sector_offset = pos & (fs_info->sectorsize - 1);
reserve_bytes = round_up(write_bytes + sector_offset,
fs_info->sectorsize);
ret = btrfs_check_data_free_space(inode, pos, write_bytes);
if (ret < 0) {
if ((BTRFS_I(inode)->flags & (BTRFS_INODE_NODATACOW |
BTRFS_INODE_PREALLOC)) &&
check_can_nocow(BTRFS_I(inode), pos,
&write_bytes) > 0) {
/*
* For nodata cow case, no need to reserve
* data space.
*/
only_release_metadata = true;
/*
* our prealloc extent may be smaller than
* write_bytes, so scale down.
*/
num_pages = DIV_ROUND_UP(write_bytes + offset,
PAGE_SIZE);
reserve_bytes = round_up(write_bytes +
sector_offset,
fs_info->sectorsize);
} else {
break;
}
}
ret = btrfs_delalloc_reserve_metadata(BTRFS_I(inode),
reserve_bytes);
if (ret) {
if (!only_release_metadata)
btrfs_free_reserved_data_space(inode, pos,
write_bytes);
else
Btrfs: fix snapshot inconsistency after a file write followed by truncate If right after starting the snapshot creation ioctl we perform a write against a file followed by a truncate, with both operations increasing the file's size, we can get a snapshot tree that reflects a state of the source subvolume's tree where the file truncation happened but the write operation didn't. This leaves a gap between 2 file extent items of the inode, which makes btrfs' fsck complain about it. For example, if we perform the following file operations: $ mkfs.btrfs -f /dev/vdd $ mount /dev/vdd /mnt $ xfs_io -f \ -c "pwrite -S 0xaa -b 32K 0 32K" \ -c "fsync" \ -c "pwrite -S 0xbb -b 32770 16K 32770" \ -c "truncate 90123" \ /mnt/foobar and the snapshot creation ioctl was just called before the second write, we often can get the following inode items in the snapshot's btree: item 120 key (257 INODE_ITEM 0) itemoff 7987 itemsize 160 inode generation 146 transid 7 size 90123 block group 0 mode 100600 links 1 uid 0 gid 0 rdev 0 flags 0x0 item 121 key (257 INODE_REF 256) itemoff 7967 itemsize 20 inode ref index 282 namelen 10 name: foobar item 122 key (257 EXTENT_DATA 0) itemoff 7914 itemsize 53 extent data disk byte 1104855040 nr 32768 extent data offset 0 nr 32768 ram 32768 extent compression 0 item 123 key (257 EXTENT_DATA 53248) itemoff 7861 itemsize 53 extent data disk byte 0 nr 0 extent data offset 0 nr 40960 ram 40960 extent compression 0 There's a file range, corresponding to the interval [32K; ALIGN(16K + 32770, 4096)[ for which there's no file extent item covering it. This is because the file write and file truncate operations happened both right after the snapshot creation ioctl called btrfs_start_delalloc_inodes(), which means we didn't start and wait for the ordered extent that matches the write and, in btrfs_setsize(), we were able to call btrfs_cont_expand() before being able to commit the current transaction in the snapshot creation ioctl. So this made it possibe to insert the hole file extent item in the source subvolume (which represents the region added by the truncate) right before the transaction commit from the snapshot creation ioctl. Btrfs' fsck tool complains about such cases with a message like the following: "root 331 inode 257 errors 100, file extent discount" >From a user perspective, the expectation when a snapshot is created while those file operations are being performed is that the snapshot will have a file that either: 1) is empty 2) only the first write was captured 3) only the 2 writes were captured 4) both writes and the truncation were captured But never capture a state where only the first write and the truncation were captured (since the second write was performed before the truncation). A test case for xfstests follows. Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-10-29 11:57:59 +00:00
btrfs_end_write_no_snapshoting(root);
break;
}
release_bytes = reserve_bytes;
Btrfs: fix the reserved space leak caused by the race between nonlock dio and buffered io When we ran sysbench on the fs with compression, the following WARN_ONs were triggered: fs/btrfs/inode.c:7829 WARN_ON(BTRFS_I(inode)->outstanding_extents); fs/btrfs/inode.c:7830 WARN_ON(BTRFS_I(inode)->reserved_extents); fs/btrfs/inode.c:7832 WARN_ON(BTRFS_I(inode)->csum_bytes); Steps to reproduce: # mkfs.btrfs -f <dev> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync prepare # cd - # umount <mnt> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync run # cd - # umount <mnt> The reason of this problem is: Task0 Task1 btrfs_direct_IO unlock(&inode->i_mutex) lock(&inode->i_mutex) reserve_space() prepare_pages() lock_extent() clear_extent() unlock_extent() lock_extent() test_extent(uptodate) return false copy_data() set_delalloc_extent() extent need compress go back to buffered write clear_extent(DELALLOC | DIRTY) unlock_extent() Task 0 and 1 wrote the same place, and task0 cleared the delalloc flag which was set by task1, it made the dirty pages in that extents couldn't be flushed into the disk, so the reserved space for that extent was not released at the end. This patch fixes the above bug by unlocking the extent after the delalloc. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-10 11:25:04 +00:00
need_unlock = false;
again:
/*
* This is going to setup the pages array with the number of
* pages we want, so we don't really need to worry about the
* contents of pages from loop to loop
*/
ret = prepare_pages(inode, pages, num_pages,
pos, write_bytes,
force_page_uptodate);
if (ret)
break;
ret = lock_and_cleanup_extent_if_need(BTRFS_I(inode), pages,
num_pages, pos, write_bytes, &lockstart,
&lockend, &cached_state);
Btrfs: fix the reserved space leak caused by the race between nonlock dio and buffered io When we ran sysbench on the fs with compression, the following WARN_ONs were triggered: fs/btrfs/inode.c:7829 WARN_ON(BTRFS_I(inode)->outstanding_extents); fs/btrfs/inode.c:7830 WARN_ON(BTRFS_I(inode)->reserved_extents); fs/btrfs/inode.c:7832 WARN_ON(BTRFS_I(inode)->csum_bytes); Steps to reproduce: # mkfs.btrfs -f <dev> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync prepare # cd - # umount <mnt> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync run # cd - # umount <mnt> The reason of this problem is: Task0 Task1 btrfs_direct_IO unlock(&inode->i_mutex) lock(&inode->i_mutex) reserve_space() prepare_pages() lock_extent() clear_extent() unlock_extent() lock_extent() test_extent(uptodate) return false copy_data() set_delalloc_extent() extent need compress go back to buffered write clear_extent(DELALLOC | DIRTY) unlock_extent() Task 0 and 1 wrote the same place, and task0 cleared the delalloc flag which was set by task1, it made the dirty pages in that extents couldn't be flushed into the disk, so the reserved space for that extent was not released at the end. This patch fixes the above bug by unlocking the extent after the delalloc. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-10 11:25:04 +00:00
if (ret < 0) {
if (ret == -EAGAIN)
goto again;
break;
} else if (ret > 0) {
need_unlock = true;
ret = 0;
}
copied = btrfs_copy_from_user(pos, write_bytes, pages, i);
Btrfs: fix regressions in copy_from_user handling Commit 914ee295af418e936ec20a08c1663eaabe4cd07a fixed deadlocks in btrfs_file_write where we would catch page faults on pages we had locked. But, there were a few problems: 1) The x86-32 iov_iter_copy_from_user_atomic code always fails to copy data when the amount to copy is more than 4K and the offset to start copying from is not page aligned. The result was btrfs_file_write looping forever retrying the iov_iter_copy_from_user_atomic We deal with this by changing btrfs_file_write to drop down to single page copies when iov_iter_copy_from_user_atomic starts returning failure. 2) The btrfs_file_write code was leaking delalloc reservations when iov_iter_copy_from_user_atomic returned zero. The looping above would result in the entire filesystem running out of delalloc reservations and constantly trying to flush things to disk. 3) btrfs_file_write will lock down page cache pages, make sure any writeback is finished, do the copy_from_user and then release them. Before the loop runs we check the first and last pages in the write to see if they are only being partially modified. If the start or end of the write isn't aligned, we make sure the corresponding pages are up to date so that we don't introduce garbage into the file. With the copy_from_user changes, we're allowing the VM to reclaim the pages after a partial update from copy_from_user, but we're not making sure the page cache page is up to date when we loop around to resume the write. We deal with this by pushing the up to date checks down into the page prep code. This fits better with how the rest of file_write works. Signed-off-by: Chris Mason <chris.mason@oracle.com> Reported-by: Mitch Harder <mitch.harder@sabayonlinux.org> cc: stable@kernel.org
2011-02-28 14:52:08 +00:00
num_sectors = BTRFS_BYTES_TO_BLKS(fs_info, reserve_bytes);
Btrfs: fix handling of faults from btrfs_copy_from_user When btrfs_copy_from_user isn't able to copy all of the pages, we need to adjust our accounting to reflect the work that was actually done. Commit 2e78c927d79 changed around the decisions a little and we ended up skipping the accounting adjustments some of the time. This commit makes sure that when we don't copy anything at all, we still hop into the adjustments, and switches to release_bytes instead of write_bytes, since write_bytes isn't aligned. The accounting errors led to warnings during btrfs_destroy_inode: [ 70.847532] WARNING: CPU: 10 PID: 514 at fs/btrfs/inode.c:9350 btrfs_destroy_inode+0x2b3/0x2c0 [ 70.847536] Modules linked in: i2c_piix4 virtio_net i2c_core input_leds button led_class serio_raw acpi_cpufreq sch_fq_codel autofs4 virtio_blk [ 70.847538] CPU: 10 PID: 514 Comm: umount Tainted: G W 4.6.0-rc6_00062_g2997da1-dirty #23 [ 70.847539] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.9.0-1.fc24 04/01/2014 [ 70.847542] 0000000000000000 ffff880ff5cafab8 ffffffff8149d5e9 0000000000000202 [ 70.847543] 0000000000000000 0000000000000000 0000000000000000 ffff880ff5cafb08 [ 70.847547] ffffffff8107bdfd ffff880ff5cafaf8 000024868120013d ffff880ff5cafb28 [ 70.847547] Call Trace: [ 70.847550] [<ffffffff8149d5e9>] dump_stack+0x51/0x78 [ 70.847551] [<ffffffff8107bdfd>] __warn+0xfd/0x120 [ 70.847553] [<ffffffff8107be3d>] warn_slowpath_null+0x1d/0x20 [ 70.847555] [<ffffffff8139c9e3>] btrfs_destroy_inode+0x2b3/0x2c0 [ 70.847556] [<ffffffff812003a1>] ? __destroy_inode+0x71/0x140 [ 70.847558] [<ffffffff812004b3>] destroy_inode+0x43/0x70 [ 70.847559] [<ffffffff810b7b5f>] ? wake_up_bit+0x2f/0x40 [ 70.847560] [<ffffffff81200c68>] evict+0x148/0x1d0 [ 70.847562] [<ffffffff81398ade>] ? start_transaction+0x3de/0x460 [ 70.847564] [<ffffffff81200d49>] dispose_list+0x59/0x80 [ 70.847565] [<ffffffff81201ba0>] evict_inodes+0x180/0x190 [ 70.847566] [<ffffffff812191ff>] ? __sync_filesystem+0x3f/0x50 [ 70.847568] [<ffffffff811e95f8>] generic_shutdown_super+0x48/0x100 [ 70.847569] [<ffffffff810b75c0>] ? woken_wake_function+0x20/0x20 [ 70.847571] [<ffffffff811e9796>] kill_anon_super+0x16/0x30 [ 70.847573] [<ffffffff81365cde>] btrfs_kill_super+0x1e/0x130 [ 70.847574] [<ffffffff811e99be>] deactivate_locked_super+0x4e/0x90 [ 70.847576] [<ffffffff811e9e61>] deactivate_super+0x51/0x70 [ 70.847577] [<ffffffff8120536f>] cleanup_mnt+0x3f/0x80 [ 70.847579] [<ffffffff81205402>] __cleanup_mnt+0x12/0x20 [ 70.847581] [<ffffffff81098358>] task_work_run+0x68/0xa0 [ 70.847582] [<ffffffff810022b6>] exit_to_usermode_loop+0xd6/0xe0 [ 70.847583] [<ffffffff81002e1d>] do_syscall_64+0xbd/0x170 [ 70.847586] [<ffffffff817d4dbc>] entry_SYSCALL64_slow_path+0x25/0x25 This is the test program I used to force short returns from btrfs_copy_from_user void *dontneed(void *arg) { char *p = arg; int ret; while(1) { ret = madvise(p, BUFSIZE/4, MADV_DONTNEED); if (ret) { perror("madvise"); exit(1); } } } int main(int ac, char **av) { int ret; int fd; char *filename; unsigned long offset; char *buf; int i; pthread_t tid; if (ac != 2) { fprintf(stderr, "usage: dammitdave filename\n"); exit(1); } buf = mmap(NULL, BUFSIZE, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0); if (buf == MAP_FAILED) { perror("mmap"); exit(1); } memset(buf, 'a', BUFSIZE); filename = av[1]; ret = pthread_create(&tid, NULL, dontneed, buf); if (ret) { fprintf(stderr, "error %d from pthread_create\n", ret); exit(1); } ret = pthread_detach(tid); if (ret) { fprintf(stderr, "pthread detach failed %d\n", ret); exit(1); } while (1) { fd = open(filename, O_RDWR | O_CREAT, 0600); if (fd < 0) { perror("open"); exit(1); } for (i = 0; i < ROUNDS; i++) { int this_write = BUFSIZE; offset = rand() % MAXSIZE; ret = pwrite(fd, buf, this_write, offset); if (ret < 0) { perror("pwrite"); exit(1); } else if (ret != this_write) { fprintf(stderr, "short write to %s offset %lu ret %d\n", filename, offset, ret); exit(1); } if (i == ROUNDS - 1) { ret = sync_file_range(fd, offset, 4096, SYNC_FILE_RANGE_WRITE); if (ret < 0) { perror("sync_file_range"); exit(1); } } } ret = ftruncate(fd, 0); if (ret < 0) { perror("ftruncate"); exit(1); } ret = close(fd); if (ret) { perror("close"); exit(1); } ret = unlink(filename); if (ret) { perror("unlink"); exit(1); } } return 0; } Signed-off-by: Chris Mason <clm@fb.com> Reported-by: Dave Jones <dsj@fb.com> Fixes: 2e78c927d79333f299a8ac81c2fd2952caeef335 cc: stable@vger.kernel.org # v4.6 Signed-off-by: Chris Mason <clm@fb.com>
2016-05-16 16:21:01 +00:00
dirty_sectors = round_up(copied + sector_offset,
fs_info->sectorsize);
dirty_sectors = BTRFS_BYTES_TO_BLKS(fs_info, dirty_sectors);
Btrfs: fix handling of faults from btrfs_copy_from_user When btrfs_copy_from_user isn't able to copy all of the pages, we need to adjust our accounting to reflect the work that was actually done. Commit 2e78c927d79 changed around the decisions a little and we ended up skipping the accounting adjustments some of the time. This commit makes sure that when we don't copy anything at all, we still hop into the adjustments, and switches to release_bytes instead of write_bytes, since write_bytes isn't aligned. The accounting errors led to warnings during btrfs_destroy_inode: [ 70.847532] WARNING: CPU: 10 PID: 514 at fs/btrfs/inode.c:9350 btrfs_destroy_inode+0x2b3/0x2c0 [ 70.847536] Modules linked in: i2c_piix4 virtio_net i2c_core input_leds button led_class serio_raw acpi_cpufreq sch_fq_codel autofs4 virtio_blk [ 70.847538] CPU: 10 PID: 514 Comm: umount Tainted: G W 4.6.0-rc6_00062_g2997da1-dirty #23 [ 70.847539] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.9.0-1.fc24 04/01/2014 [ 70.847542] 0000000000000000 ffff880ff5cafab8 ffffffff8149d5e9 0000000000000202 [ 70.847543] 0000000000000000 0000000000000000 0000000000000000 ffff880ff5cafb08 [ 70.847547] ffffffff8107bdfd ffff880ff5cafaf8 000024868120013d ffff880ff5cafb28 [ 70.847547] Call Trace: [ 70.847550] [<ffffffff8149d5e9>] dump_stack+0x51/0x78 [ 70.847551] [<ffffffff8107bdfd>] __warn+0xfd/0x120 [ 70.847553] [<ffffffff8107be3d>] warn_slowpath_null+0x1d/0x20 [ 70.847555] [<ffffffff8139c9e3>] btrfs_destroy_inode+0x2b3/0x2c0 [ 70.847556] [<ffffffff812003a1>] ? __destroy_inode+0x71/0x140 [ 70.847558] [<ffffffff812004b3>] destroy_inode+0x43/0x70 [ 70.847559] [<ffffffff810b7b5f>] ? wake_up_bit+0x2f/0x40 [ 70.847560] [<ffffffff81200c68>] evict+0x148/0x1d0 [ 70.847562] [<ffffffff81398ade>] ? start_transaction+0x3de/0x460 [ 70.847564] [<ffffffff81200d49>] dispose_list+0x59/0x80 [ 70.847565] [<ffffffff81201ba0>] evict_inodes+0x180/0x190 [ 70.847566] [<ffffffff812191ff>] ? __sync_filesystem+0x3f/0x50 [ 70.847568] [<ffffffff811e95f8>] generic_shutdown_super+0x48/0x100 [ 70.847569] [<ffffffff810b75c0>] ? woken_wake_function+0x20/0x20 [ 70.847571] [<ffffffff811e9796>] kill_anon_super+0x16/0x30 [ 70.847573] [<ffffffff81365cde>] btrfs_kill_super+0x1e/0x130 [ 70.847574] [<ffffffff811e99be>] deactivate_locked_super+0x4e/0x90 [ 70.847576] [<ffffffff811e9e61>] deactivate_super+0x51/0x70 [ 70.847577] [<ffffffff8120536f>] cleanup_mnt+0x3f/0x80 [ 70.847579] [<ffffffff81205402>] __cleanup_mnt+0x12/0x20 [ 70.847581] [<ffffffff81098358>] task_work_run+0x68/0xa0 [ 70.847582] [<ffffffff810022b6>] exit_to_usermode_loop+0xd6/0xe0 [ 70.847583] [<ffffffff81002e1d>] do_syscall_64+0xbd/0x170 [ 70.847586] [<ffffffff817d4dbc>] entry_SYSCALL64_slow_path+0x25/0x25 This is the test program I used to force short returns from btrfs_copy_from_user void *dontneed(void *arg) { char *p = arg; int ret; while(1) { ret = madvise(p, BUFSIZE/4, MADV_DONTNEED); if (ret) { perror("madvise"); exit(1); } } } int main(int ac, char **av) { int ret; int fd; char *filename; unsigned long offset; char *buf; int i; pthread_t tid; if (ac != 2) { fprintf(stderr, "usage: dammitdave filename\n"); exit(1); } buf = mmap(NULL, BUFSIZE, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0); if (buf == MAP_FAILED) { perror("mmap"); exit(1); } memset(buf, 'a', BUFSIZE); filename = av[1]; ret = pthread_create(&tid, NULL, dontneed, buf); if (ret) { fprintf(stderr, "error %d from pthread_create\n", ret); exit(1); } ret = pthread_detach(tid); if (ret) { fprintf(stderr, "pthread detach failed %d\n", ret); exit(1); } while (1) { fd = open(filename, O_RDWR | O_CREAT, 0600); if (fd < 0) { perror("open"); exit(1); } for (i = 0; i < ROUNDS; i++) { int this_write = BUFSIZE; offset = rand() % MAXSIZE; ret = pwrite(fd, buf, this_write, offset); if (ret < 0) { perror("pwrite"); exit(1); } else if (ret != this_write) { fprintf(stderr, "short write to %s offset %lu ret %d\n", filename, offset, ret); exit(1); } if (i == ROUNDS - 1) { ret = sync_file_range(fd, offset, 4096, SYNC_FILE_RANGE_WRITE); if (ret < 0) { perror("sync_file_range"); exit(1); } } } ret = ftruncate(fd, 0); if (ret < 0) { perror("ftruncate"); exit(1); } ret = close(fd); if (ret) { perror("close"); exit(1); } ret = unlink(filename); if (ret) { perror("unlink"); exit(1); } } return 0; } Signed-off-by: Chris Mason <clm@fb.com> Reported-by: Dave Jones <dsj@fb.com> Fixes: 2e78c927d79333f299a8ac81c2fd2952caeef335 cc: stable@vger.kernel.org # v4.6 Signed-off-by: Chris Mason <clm@fb.com>
2016-05-16 16:21:01 +00:00
Btrfs: fix regressions in copy_from_user handling Commit 914ee295af418e936ec20a08c1663eaabe4cd07a fixed deadlocks in btrfs_file_write where we would catch page faults on pages we had locked. But, there were a few problems: 1) The x86-32 iov_iter_copy_from_user_atomic code always fails to copy data when the amount to copy is more than 4K and the offset to start copying from is not page aligned. The result was btrfs_file_write looping forever retrying the iov_iter_copy_from_user_atomic We deal with this by changing btrfs_file_write to drop down to single page copies when iov_iter_copy_from_user_atomic starts returning failure. 2) The btrfs_file_write code was leaking delalloc reservations when iov_iter_copy_from_user_atomic returned zero. The looping above would result in the entire filesystem running out of delalloc reservations and constantly trying to flush things to disk. 3) btrfs_file_write will lock down page cache pages, make sure any writeback is finished, do the copy_from_user and then release them. Before the loop runs we check the first and last pages in the write to see if they are only being partially modified. If the start or end of the write isn't aligned, we make sure the corresponding pages are up to date so that we don't introduce garbage into the file. With the copy_from_user changes, we're allowing the VM to reclaim the pages after a partial update from copy_from_user, but we're not making sure the page cache page is up to date when we loop around to resume the write. We deal with this by pushing the up to date checks down into the page prep code. This fits better with how the rest of file_write works. Signed-off-by: Chris Mason <chris.mason@oracle.com> Reported-by: Mitch Harder <mitch.harder@sabayonlinux.org> cc: stable@kernel.org
2011-02-28 14:52:08 +00:00
/*
* if we have trouble faulting in the pages, fall
* back to one page at a time
*/
if (copied < write_bytes)
nrptrs = 1;
if (copied == 0) {
force_page_uptodate = true;
Btrfs: fix handling of faults from btrfs_copy_from_user When btrfs_copy_from_user isn't able to copy all of the pages, we need to adjust our accounting to reflect the work that was actually done. Commit 2e78c927d79 changed around the decisions a little and we ended up skipping the accounting adjustments some of the time. This commit makes sure that when we don't copy anything at all, we still hop into the adjustments, and switches to release_bytes instead of write_bytes, since write_bytes isn't aligned. The accounting errors led to warnings during btrfs_destroy_inode: [ 70.847532] WARNING: CPU: 10 PID: 514 at fs/btrfs/inode.c:9350 btrfs_destroy_inode+0x2b3/0x2c0 [ 70.847536] Modules linked in: i2c_piix4 virtio_net i2c_core input_leds button led_class serio_raw acpi_cpufreq sch_fq_codel autofs4 virtio_blk [ 70.847538] CPU: 10 PID: 514 Comm: umount Tainted: G W 4.6.0-rc6_00062_g2997da1-dirty #23 [ 70.847539] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.9.0-1.fc24 04/01/2014 [ 70.847542] 0000000000000000 ffff880ff5cafab8 ffffffff8149d5e9 0000000000000202 [ 70.847543] 0000000000000000 0000000000000000 0000000000000000 ffff880ff5cafb08 [ 70.847547] ffffffff8107bdfd ffff880ff5cafaf8 000024868120013d ffff880ff5cafb28 [ 70.847547] Call Trace: [ 70.847550] [<ffffffff8149d5e9>] dump_stack+0x51/0x78 [ 70.847551] [<ffffffff8107bdfd>] __warn+0xfd/0x120 [ 70.847553] [<ffffffff8107be3d>] warn_slowpath_null+0x1d/0x20 [ 70.847555] [<ffffffff8139c9e3>] btrfs_destroy_inode+0x2b3/0x2c0 [ 70.847556] [<ffffffff812003a1>] ? __destroy_inode+0x71/0x140 [ 70.847558] [<ffffffff812004b3>] destroy_inode+0x43/0x70 [ 70.847559] [<ffffffff810b7b5f>] ? wake_up_bit+0x2f/0x40 [ 70.847560] [<ffffffff81200c68>] evict+0x148/0x1d0 [ 70.847562] [<ffffffff81398ade>] ? start_transaction+0x3de/0x460 [ 70.847564] [<ffffffff81200d49>] dispose_list+0x59/0x80 [ 70.847565] [<ffffffff81201ba0>] evict_inodes+0x180/0x190 [ 70.847566] [<ffffffff812191ff>] ? __sync_filesystem+0x3f/0x50 [ 70.847568] [<ffffffff811e95f8>] generic_shutdown_super+0x48/0x100 [ 70.847569] [<ffffffff810b75c0>] ? woken_wake_function+0x20/0x20 [ 70.847571] [<ffffffff811e9796>] kill_anon_super+0x16/0x30 [ 70.847573] [<ffffffff81365cde>] btrfs_kill_super+0x1e/0x130 [ 70.847574] [<ffffffff811e99be>] deactivate_locked_super+0x4e/0x90 [ 70.847576] [<ffffffff811e9e61>] deactivate_super+0x51/0x70 [ 70.847577] [<ffffffff8120536f>] cleanup_mnt+0x3f/0x80 [ 70.847579] [<ffffffff81205402>] __cleanup_mnt+0x12/0x20 [ 70.847581] [<ffffffff81098358>] task_work_run+0x68/0xa0 [ 70.847582] [<ffffffff810022b6>] exit_to_usermode_loop+0xd6/0xe0 [ 70.847583] [<ffffffff81002e1d>] do_syscall_64+0xbd/0x170 [ 70.847586] [<ffffffff817d4dbc>] entry_SYSCALL64_slow_path+0x25/0x25 This is the test program I used to force short returns from btrfs_copy_from_user void *dontneed(void *arg) { char *p = arg; int ret; while(1) { ret = madvise(p, BUFSIZE/4, MADV_DONTNEED); if (ret) { perror("madvise"); exit(1); } } } int main(int ac, char **av) { int ret; int fd; char *filename; unsigned long offset; char *buf; int i; pthread_t tid; if (ac != 2) { fprintf(stderr, "usage: dammitdave filename\n"); exit(1); } buf = mmap(NULL, BUFSIZE, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0); if (buf == MAP_FAILED) { perror("mmap"); exit(1); } memset(buf, 'a', BUFSIZE); filename = av[1]; ret = pthread_create(&tid, NULL, dontneed, buf); if (ret) { fprintf(stderr, "error %d from pthread_create\n", ret); exit(1); } ret = pthread_detach(tid); if (ret) { fprintf(stderr, "pthread detach failed %d\n", ret); exit(1); } while (1) { fd = open(filename, O_RDWR | O_CREAT, 0600); if (fd < 0) { perror("open"); exit(1); } for (i = 0; i < ROUNDS; i++) { int this_write = BUFSIZE; offset = rand() % MAXSIZE; ret = pwrite(fd, buf, this_write, offset); if (ret < 0) { perror("pwrite"); exit(1); } else if (ret != this_write) { fprintf(stderr, "short write to %s offset %lu ret %d\n", filename, offset, ret); exit(1); } if (i == ROUNDS - 1) { ret = sync_file_range(fd, offset, 4096, SYNC_FILE_RANGE_WRITE); if (ret < 0) { perror("sync_file_range"); exit(1); } } } ret = ftruncate(fd, 0); if (ret < 0) { perror("ftruncate"); exit(1); } ret = close(fd); if (ret) { perror("close"); exit(1); } ret = unlink(filename); if (ret) { perror("unlink"); exit(1); } } return 0; } Signed-off-by: Chris Mason <clm@fb.com> Reported-by: Dave Jones <dsj@fb.com> Fixes: 2e78c927d79333f299a8ac81c2fd2952caeef335 cc: stable@vger.kernel.org # v4.6 Signed-off-by: Chris Mason <clm@fb.com>
2016-05-16 16:21:01 +00:00
dirty_sectors = 0;
Btrfs: fix regressions in copy_from_user handling Commit 914ee295af418e936ec20a08c1663eaabe4cd07a fixed deadlocks in btrfs_file_write where we would catch page faults on pages we had locked. But, there were a few problems: 1) The x86-32 iov_iter_copy_from_user_atomic code always fails to copy data when the amount to copy is more than 4K and the offset to start copying from is not page aligned. The result was btrfs_file_write looping forever retrying the iov_iter_copy_from_user_atomic We deal with this by changing btrfs_file_write to drop down to single page copies when iov_iter_copy_from_user_atomic starts returning failure. 2) The btrfs_file_write code was leaking delalloc reservations when iov_iter_copy_from_user_atomic returned zero. The looping above would result in the entire filesystem running out of delalloc reservations and constantly trying to flush things to disk. 3) btrfs_file_write will lock down page cache pages, make sure any writeback is finished, do the copy_from_user and then release them. Before the loop runs we check the first and last pages in the write to see if they are only being partially modified. If the start or end of the write isn't aligned, we make sure the corresponding pages are up to date so that we don't introduce garbage into the file. With the copy_from_user changes, we're allowing the VM to reclaim the pages after a partial update from copy_from_user, but we're not making sure the page cache page is up to date when we loop around to resume the write. We deal with this by pushing the up to date checks down into the page prep code. This fits better with how the rest of file_write works. Signed-off-by: Chris Mason <chris.mason@oracle.com> Reported-by: Mitch Harder <mitch.harder@sabayonlinux.org> cc: stable@kernel.org
2011-02-28 14:52:08 +00:00
dirty_pages = 0;
} else {
force_page_uptodate = false;
dirty_pages = DIV_ROUND_UP(copied + offset,
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
PAGE_SIZE);
}
/*
* If we had a short copy we need to release the excess delaloc
* bytes we reserved. We need to increment outstanding_extents
Btrfs: fix handling of faults from btrfs_copy_from_user When btrfs_copy_from_user isn't able to copy all of the pages, we need to adjust our accounting to reflect the work that was actually done. Commit 2e78c927d79 changed around the decisions a little and we ended up skipping the accounting adjustments some of the time. This commit makes sure that when we don't copy anything at all, we still hop into the adjustments, and switches to release_bytes instead of write_bytes, since write_bytes isn't aligned. The accounting errors led to warnings during btrfs_destroy_inode: [ 70.847532] WARNING: CPU: 10 PID: 514 at fs/btrfs/inode.c:9350 btrfs_destroy_inode+0x2b3/0x2c0 [ 70.847536] Modules linked in: i2c_piix4 virtio_net i2c_core input_leds button led_class serio_raw acpi_cpufreq sch_fq_codel autofs4 virtio_blk [ 70.847538] CPU: 10 PID: 514 Comm: umount Tainted: G W 4.6.0-rc6_00062_g2997da1-dirty #23 [ 70.847539] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.9.0-1.fc24 04/01/2014 [ 70.847542] 0000000000000000 ffff880ff5cafab8 ffffffff8149d5e9 0000000000000202 [ 70.847543] 0000000000000000 0000000000000000 0000000000000000 ffff880ff5cafb08 [ 70.847547] ffffffff8107bdfd ffff880ff5cafaf8 000024868120013d ffff880ff5cafb28 [ 70.847547] Call Trace: [ 70.847550] [<ffffffff8149d5e9>] dump_stack+0x51/0x78 [ 70.847551] [<ffffffff8107bdfd>] __warn+0xfd/0x120 [ 70.847553] [<ffffffff8107be3d>] warn_slowpath_null+0x1d/0x20 [ 70.847555] [<ffffffff8139c9e3>] btrfs_destroy_inode+0x2b3/0x2c0 [ 70.847556] [<ffffffff812003a1>] ? __destroy_inode+0x71/0x140 [ 70.847558] [<ffffffff812004b3>] destroy_inode+0x43/0x70 [ 70.847559] [<ffffffff810b7b5f>] ? wake_up_bit+0x2f/0x40 [ 70.847560] [<ffffffff81200c68>] evict+0x148/0x1d0 [ 70.847562] [<ffffffff81398ade>] ? start_transaction+0x3de/0x460 [ 70.847564] [<ffffffff81200d49>] dispose_list+0x59/0x80 [ 70.847565] [<ffffffff81201ba0>] evict_inodes+0x180/0x190 [ 70.847566] [<ffffffff812191ff>] ? __sync_filesystem+0x3f/0x50 [ 70.847568] [<ffffffff811e95f8>] generic_shutdown_super+0x48/0x100 [ 70.847569] [<ffffffff810b75c0>] ? woken_wake_function+0x20/0x20 [ 70.847571] [<ffffffff811e9796>] kill_anon_super+0x16/0x30 [ 70.847573] [<ffffffff81365cde>] btrfs_kill_super+0x1e/0x130 [ 70.847574] [<ffffffff811e99be>] deactivate_locked_super+0x4e/0x90 [ 70.847576] [<ffffffff811e9e61>] deactivate_super+0x51/0x70 [ 70.847577] [<ffffffff8120536f>] cleanup_mnt+0x3f/0x80 [ 70.847579] [<ffffffff81205402>] __cleanup_mnt+0x12/0x20 [ 70.847581] [<ffffffff81098358>] task_work_run+0x68/0xa0 [ 70.847582] [<ffffffff810022b6>] exit_to_usermode_loop+0xd6/0xe0 [ 70.847583] [<ffffffff81002e1d>] do_syscall_64+0xbd/0x170 [ 70.847586] [<ffffffff817d4dbc>] entry_SYSCALL64_slow_path+0x25/0x25 This is the test program I used to force short returns from btrfs_copy_from_user void *dontneed(void *arg) { char *p = arg; int ret; while(1) { ret = madvise(p, BUFSIZE/4, MADV_DONTNEED); if (ret) { perror("madvise"); exit(1); } } } int main(int ac, char **av) { int ret; int fd; char *filename; unsigned long offset; char *buf; int i; pthread_t tid; if (ac != 2) { fprintf(stderr, "usage: dammitdave filename\n"); exit(1); } buf = mmap(NULL, BUFSIZE, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0); if (buf == MAP_FAILED) { perror("mmap"); exit(1); } memset(buf, 'a', BUFSIZE); filename = av[1]; ret = pthread_create(&tid, NULL, dontneed, buf); if (ret) { fprintf(stderr, "error %d from pthread_create\n", ret); exit(1); } ret = pthread_detach(tid); if (ret) { fprintf(stderr, "pthread detach failed %d\n", ret); exit(1); } while (1) { fd = open(filename, O_RDWR | O_CREAT, 0600); if (fd < 0) { perror("open"); exit(1); } for (i = 0; i < ROUNDS; i++) { int this_write = BUFSIZE; offset = rand() % MAXSIZE; ret = pwrite(fd, buf, this_write, offset); if (ret < 0) { perror("pwrite"); exit(1); } else if (ret != this_write) { fprintf(stderr, "short write to %s offset %lu ret %d\n", filename, offset, ret); exit(1); } if (i == ROUNDS - 1) { ret = sync_file_range(fd, offset, 4096, SYNC_FILE_RANGE_WRITE); if (ret < 0) { perror("sync_file_range"); exit(1); } } } ret = ftruncate(fd, 0); if (ret < 0) { perror("ftruncate"); exit(1); } ret = close(fd); if (ret) { perror("close"); exit(1); } ret = unlink(filename); if (ret) { perror("unlink"); exit(1); } } return 0; } Signed-off-by: Chris Mason <clm@fb.com> Reported-by: Dave Jones <dsj@fb.com> Fixes: 2e78c927d79333f299a8ac81c2fd2952caeef335 cc: stable@vger.kernel.org # v4.6 Signed-off-by: Chris Mason <clm@fb.com>
2016-05-16 16:21:01 +00:00
* because btrfs_delalloc_release_space and
* btrfs_delalloc_release_metadata will decrement it, but
* we still have an outstanding extent for the chunk we actually
* managed to copy.
*/
if (num_sectors > dirty_sectors) {
/* release everything except the sectors we dirtied */
release_bytes -= dirty_sectors <<
fs_info->sb->s_blocksize_bits;
if (copied > 0) {
spin_lock(&BTRFS_I(inode)->lock);
BTRFS_I(inode)->outstanding_extents++;
spin_unlock(&BTRFS_I(inode)->lock);
}
if (only_release_metadata) {
btrfs_delalloc_release_metadata(BTRFS_I(inode),
release_bytes);
} else {
u64 __pos;
__pos = round_down(pos,
fs_info->sectorsize) +
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
(dirty_pages << PAGE_SHIFT);
btrfs_delalloc_release_space(inode, __pos,
release_bytes);
}
}
release_bytes = round_up(copied + sector_offset,
fs_info->sectorsize);
Btrfs: fix the reserved space leak caused by the race between nonlock dio and buffered io When we ran sysbench on the fs with compression, the following WARN_ONs were triggered: fs/btrfs/inode.c:7829 WARN_ON(BTRFS_I(inode)->outstanding_extents); fs/btrfs/inode.c:7830 WARN_ON(BTRFS_I(inode)->reserved_extents); fs/btrfs/inode.c:7832 WARN_ON(BTRFS_I(inode)->csum_bytes); Steps to reproduce: # mkfs.btrfs -f <dev> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync prepare # cd - # umount <mnt> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync run # cd - # umount <mnt> The reason of this problem is: Task0 Task1 btrfs_direct_IO unlock(&inode->i_mutex) lock(&inode->i_mutex) reserve_space() prepare_pages() lock_extent() clear_extent() unlock_extent() lock_extent() test_extent(uptodate) return false copy_data() set_delalloc_extent() extent need compress go back to buffered write clear_extent(DELALLOC | DIRTY) unlock_extent() Task 0 and 1 wrote the same place, and task0 cleared the delalloc flag which was set by task1, it made the dirty pages in that extents couldn't be flushed into the disk, so the reserved space for that extent was not released at the end. This patch fixes the above bug by unlocking the extent after the delalloc. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-10 11:25:04 +00:00
if (copied > 0)
ret = btrfs_dirty_pages(inode, pages, dirty_pages,
pos, copied, NULL);
Btrfs: fix the reserved space leak caused by the race between nonlock dio and buffered io When we ran sysbench on the fs with compression, the following WARN_ONs were triggered: fs/btrfs/inode.c:7829 WARN_ON(BTRFS_I(inode)->outstanding_extents); fs/btrfs/inode.c:7830 WARN_ON(BTRFS_I(inode)->reserved_extents); fs/btrfs/inode.c:7832 WARN_ON(BTRFS_I(inode)->csum_bytes); Steps to reproduce: # mkfs.btrfs -f <dev> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync prepare # cd - # umount <mnt> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync run # cd - # umount <mnt> The reason of this problem is: Task0 Task1 btrfs_direct_IO unlock(&inode->i_mutex) lock(&inode->i_mutex) reserve_space() prepare_pages() lock_extent() clear_extent() unlock_extent() lock_extent() test_extent(uptodate) return false copy_data() set_delalloc_extent() extent need compress go back to buffered write clear_extent(DELALLOC | DIRTY) unlock_extent() Task 0 and 1 wrote the same place, and task0 cleared the delalloc flag which was set by task1, it made the dirty pages in that extents couldn't be flushed into the disk, so the reserved space for that extent was not released at the end. This patch fixes the above bug by unlocking the extent after the delalloc. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-10 11:25:04 +00:00
if (need_unlock)
unlock_extent_cached(&BTRFS_I(inode)->io_tree,
lockstart, lockend, &cached_state,
GFP_NOFS);
if (ret) {
btrfs_drop_pages(pages, num_pages);
Btrfs: fix the reserved space leak caused by the race between nonlock dio and buffered io When we ran sysbench on the fs with compression, the following WARN_ONs were triggered: fs/btrfs/inode.c:7829 WARN_ON(BTRFS_I(inode)->outstanding_extents); fs/btrfs/inode.c:7830 WARN_ON(BTRFS_I(inode)->reserved_extents); fs/btrfs/inode.c:7832 WARN_ON(BTRFS_I(inode)->csum_bytes); Steps to reproduce: # mkfs.btrfs -f <dev> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync prepare # cd - # umount <mnt> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync run # cd - # umount <mnt> The reason of this problem is: Task0 Task1 btrfs_direct_IO unlock(&inode->i_mutex) lock(&inode->i_mutex) reserve_space() prepare_pages() lock_extent() clear_extent() unlock_extent() lock_extent() test_extent(uptodate) return false copy_data() set_delalloc_extent() extent need compress go back to buffered write clear_extent(DELALLOC | DIRTY) unlock_extent() Task 0 and 1 wrote the same place, and task0 cleared the delalloc flag which was set by task1, it made the dirty pages in that extents couldn't be flushed into the disk, so the reserved space for that extent was not released at the end. This patch fixes the above bug by unlocking the extent after the delalloc. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-10 11:25:04 +00:00
break;
}
Btrfs: fix the reserved space leak caused by the race between nonlock dio and buffered io When we ran sysbench on the fs with compression, the following WARN_ONs were triggered: fs/btrfs/inode.c:7829 WARN_ON(BTRFS_I(inode)->outstanding_extents); fs/btrfs/inode.c:7830 WARN_ON(BTRFS_I(inode)->reserved_extents); fs/btrfs/inode.c:7832 WARN_ON(BTRFS_I(inode)->csum_bytes); Steps to reproduce: # mkfs.btrfs -f <dev> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync prepare # cd - # umount <mnt> # mount -o compress <dev> <mnt> # cd <mnt> # sysbench --test=fileio --num-threads=8 --file-total-size=8G \ > --file-block-size=32K --file-io-mode=rndwr --file-fsync-freq=0 \ > --file-fsync-end=no --max-requests=300000 --file-extra-flags=direct \ > --file-test-mode=sync run # cd - # umount <mnt> The reason of this problem is: Task0 Task1 btrfs_direct_IO unlock(&inode->i_mutex) lock(&inode->i_mutex) reserve_space() prepare_pages() lock_extent() clear_extent() unlock_extent() lock_extent() test_extent(uptodate) return false copy_data() set_delalloc_extent() extent need compress go back to buffered write clear_extent(DELALLOC | DIRTY) unlock_extent() Task 0 and 1 wrote the same place, and task0 cleared the delalloc flag which was set by task1, it made the dirty pages in that extents couldn't be flushed into the disk, so the reserved space for that extent was not released at the end. This patch fixes the above bug by unlocking the extent after the delalloc. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-10 11:25:04 +00:00
release_bytes = 0;
if (only_release_metadata)
Btrfs: fix snapshot inconsistency after a file write followed by truncate If right after starting the snapshot creation ioctl we perform a write against a file followed by a truncate, with both operations increasing the file's size, we can get a snapshot tree that reflects a state of the source subvolume's tree where the file truncation happened but the write operation didn't. This leaves a gap between 2 file extent items of the inode, which makes btrfs' fsck complain about it. For example, if we perform the following file operations: $ mkfs.btrfs -f /dev/vdd $ mount /dev/vdd /mnt $ xfs_io -f \ -c "pwrite -S 0xaa -b 32K 0 32K" \ -c "fsync" \ -c "pwrite -S 0xbb -b 32770 16K 32770" \ -c "truncate 90123" \ /mnt/foobar and the snapshot creation ioctl was just called before the second write, we often can get the following inode items in the snapshot's btree: item 120 key (257 INODE_ITEM 0) itemoff 7987 itemsize 160 inode generation 146 transid 7 size 90123 block group 0 mode 100600 links 1 uid 0 gid 0 rdev 0 flags 0x0 item 121 key (257 INODE_REF 256) itemoff 7967 itemsize 20 inode ref index 282 namelen 10 name: foobar item 122 key (257 EXTENT_DATA 0) itemoff 7914 itemsize 53 extent data disk byte 1104855040 nr 32768 extent data offset 0 nr 32768 ram 32768 extent compression 0 item 123 key (257 EXTENT_DATA 53248) itemoff 7861 itemsize 53 extent data disk byte 0 nr 0 extent data offset 0 nr 40960 ram 40960 extent compression 0 There's a file range, corresponding to the interval [32K; ALIGN(16K + 32770, 4096)[ for which there's no file extent item covering it. This is because the file write and file truncate operations happened both right after the snapshot creation ioctl called btrfs_start_delalloc_inodes(), which means we didn't start and wait for the ordered extent that matches the write and, in btrfs_setsize(), we were able to call btrfs_cont_expand() before being able to commit the current transaction in the snapshot creation ioctl. So this made it possibe to insert the hole file extent item in the source subvolume (which represents the region added by the truncate) right before the transaction commit from the snapshot creation ioctl. Btrfs' fsck tool complains about such cases with a message like the following: "root 331 inode 257 errors 100, file extent discount" >From a user perspective, the expectation when a snapshot is created while those file operations are being performed is that the snapshot will have a file that either: 1) is empty 2) only the first write was captured 3) only the 2 writes were captured 4) both writes and the truncation were captured But never capture a state where only the first write and the truncation were captured (since the second write was performed before the truncation). A test case for xfstests follows. Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-10-29 11:57:59 +00:00
btrfs_end_write_no_snapshoting(root);
if (only_release_metadata && copied > 0) {
lockstart = round_down(pos,
fs_info->sectorsize);
lockend = round_up(pos + copied,
fs_info->sectorsize) - 1;
set_extent_bit(&BTRFS_I(inode)->io_tree, lockstart,
lockend, EXTENT_NORESERVE, NULL,
NULL, GFP_NOFS);
only_release_metadata = false;
}
btrfs_drop_pages(pages, num_pages);
cond_resched();
balance_dirty_pages_ratelimited(inode->i_mapping);
if (dirty_pages < (fs_info->nodesize >> PAGE_SHIFT) + 1)
btrfs_btree_balance_dirty(fs_info);
pos += copied;
num_written += copied;
}
kfree(pages);
if (release_bytes) {
if (only_release_metadata) {
Btrfs: fix snapshot inconsistency after a file write followed by truncate If right after starting the snapshot creation ioctl we perform a write against a file followed by a truncate, with both operations increasing the file's size, we can get a snapshot tree that reflects a state of the source subvolume's tree where the file truncation happened but the write operation didn't. This leaves a gap between 2 file extent items of the inode, which makes btrfs' fsck complain about it. For example, if we perform the following file operations: $ mkfs.btrfs -f /dev/vdd $ mount /dev/vdd /mnt $ xfs_io -f \ -c "pwrite -S 0xaa -b 32K 0 32K" \ -c "fsync" \ -c "pwrite -S 0xbb -b 32770 16K 32770" \ -c "truncate 90123" \ /mnt/foobar and the snapshot creation ioctl was just called before the second write, we often can get the following inode items in the snapshot's btree: item 120 key (257 INODE_ITEM 0) itemoff 7987 itemsize 160 inode generation 146 transid 7 size 90123 block group 0 mode 100600 links 1 uid 0 gid 0 rdev 0 flags 0x0 item 121 key (257 INODE_REF 256) itemoff 7967 itemsize 20 inode ref index 282 namelen 10 name: foobar item 122 key (257 EXTENT_DATA 0) itemoff 7914 itemsize 53 extent data disk byte 1104855040 nr 32768 extent data offset 0 nr 32768 ram 32768 extent compression 0 item 123 key (257 EXTENT_DATA 53248) itemoff 7861 itemsize 53 extent data disk byte 0 nr 0 extent data offset 0 nr 40960 ram 40960 extent compression 0 There's a file range, corresponding to the interval [32K; ALIGN(16K + 32770, 4096)[ for which there's no file extent item covering it. This is because the file write and file truncate operations happened both right after the snapshot creation ioctl called btrfs_start_delalloc_inodes(), which means we didn't start and wait for the ordered extent that matches the write and, in btrfs_setsize(), we were able to call btrfs_cont_expand() before being able to commit the current transaction in the snapshot creation ioctl. So this made it possibe to insert the hole file extent item in the source subvolume (which represents the region added by the truncate) right before the transaction commit from the snapshot creation ioctl. Btrfs' fsck tool complains about such cases with a message like the following: "root 331 inode 257 errors 100, file extent discount" >From a user perspective, the expectation when a snapshot is created while those file operations are being performed is that the snapshot will have a file that either: 1) is empty 2) only the first write was captured 3) only the 2 writes were captured 4) both writes and the truncation were captured But never capture a state where only the first write and the truncation were captured (since the second write was performed before the truncation). A test case for xfstests follows. Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-10-29 11:57:59 +00:00
btrfs_end_write_no_snapshoting(root);
btrfs_delalloc_release_metadata(BTRFS_I(inode),
release_bytes);
} else {
btrfs_delalloc_release_space(inode,
round_down(pos, fs_info->sectorsize),
release_bytes);
}
}
return num_written ? num_written : ret;
}
static ssize_t __btrfs_direct_write(struct kiocb *iocb, struct iov_iter *from)
{
struct file *file = iocb->ki_filp;
struct inode *inode = file_inode(file);
loff_t pos = iocb->ki_pos;
ssize_t written;
ssize_t written_buffered;
loff_t endbyte;
int err;
written = generic_file_direct_write(iocb, from);
if (written < 0 || !iov_iter_count(from))
return written;
pos += written;
written_buffered = __btrfs_buffered_write(file, from, pos);
if (written_buffered < 0) {
err = written_buffered;
goto out;
}
/*
* Ensure all data is persisted. We want the next direct IO read to be
* able to read what was just written.
*/
endbyte = pos + written_buffered - 1;
err = btrfs_fdatawrite_range(inode, pos, endbyte);
if (err)
goto out;
err = filemap_fdatawait_range(inode->i_mapping, pos, endbyte);
if (err)
goto out;
written += written_buffered;
iocb->ki_pos = pos + written_buffered;
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
invalidate_mapping_pages(file->f_mapping, pos >> PAGE_SHIFT,
endbyte >> PAGE_SHIFT);
out:
return written ? written : err;
}
static void update_time_for_write(struct inode *inode)
{
struct timespec now;
if (IS_NOCMTIME(inode))
return;
now = current_time(inode);
if (!timespec_equal(&inode->i_mtime, &now))
inode->i_mtime = now;
if (!timespec_equal(&inode->i_ctime, &now))
inode->i_ctime = now;
if (IS_I_VERSION(inode))
inode_inc_iversion(inode);
}
static ssize_t btrfs_file_write_iter(struct kiocb *iocb,
struct iov_iter *from)
{
struct file *file = iocb->ki_filp;
struct inode *inode = file_inode(file);
struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
struct btrfs_root *root = BTRFS_I(inode)->root;
u64 start_pos;
u64 end_pos;
ssize_t num_written = 0;
bool sync = (file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host);
ssize_t err;
loff_t pos;
size_t count;
loff_t oldsize;
int clean_page = 0;
inode_lock(inode);
err = generic_write_checks(iocb, from);
if (err <= 0) {
inode_unlock(inode);
return err;
}
current->backing_dev_info = inode_to_bdi(inode);
err = file_remove_privs(file);
if (err) {
inode_unlock(inode);
goto out;
}
/*
* If BTRFS flips readonly due to some impossible error
* (fs_info->fs_state now has BTRFS_SUPER_FLAG_ERROR),
* although we have opened a file as writable, we have
* to stop this write operation to ensure FS consistency.
*/
if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) {
inode_unlock(inode);
err = -EROFS;
goto out;
}
/*
* We reserve space for updating the inode when we reserve space for the
* extent we are going to write, so we will enospc out there. We don't
* need to start yet another transaction to update the inode as we will
* update the inode when we finish writing whatever data we write.
*/
update_time_for_write(inode);
pos = iocb->ki_pos;
count = iov_iter_count(from);
start_pos = round_down(pos, fs_info->sectorsize);
oldsize = i_size_read(inode);
if (start_pos > oldsize) {
/* Expand hole size to cover write data, preventing empty gap */
end_pos = round_up(pos + count,
fs_info->sectorsize);
err = btrfs_cont_expand(inode, oldsize, end_pos);
if (err) {
inode_unlock(inode);
goto out;
}
if (start_pos > round_up(oldsize, fs_info->sectorsize))
clean_page = 1;
}
if (sync)
atomic_inc(&BTRFS_I(inode)->sync_writers);
if (iocb->ki_flags & IOCB_DIRECT) {
num_written = __btrfs_direct_write(iocb, from);
} else {
num_written = __btrfs_buffered_write(file, from, pos);
if (num_written > 0)
iocb->ki_pos = pos + num_written;
if (clean_page)
pagecache_isize_extended(inode, oldsize,
i_size_read(inode));
}
inode_unlock(inode);
Btrfs: add extra flushing for renames and truncates Renames and truncates are both common ways to replace old data with new data. The filesystem can make an effort to make sure the new data is on disk before actually replacing the old data. This is especially important for rename, which many application use as though it were atomic for both the data and the metadata involved. The current btrfs code will happily replace a file that is fully on disk with one that was just created and still has pending IO. If we crash after transaction commit but before the IO is done, we'll end up replacing a good file with a zero length file. The solution used here is to create a list of inodes that need special ordering and force them to disk before the commit is done. This is similar to the ext3 style data=ordering, except it is only done on selected files. Btrfs is able to get away with this because it does not wait on commits very often, even for fsync (which use a sub-commit). For renames, we order the file when it wasn't already on disk and when it is replacing an existing file. Larger files are sent to filemap_flush right away (before the transaction handle is opened). For truncates, we order if the file goes from non-zero size down to zero size. This is a little different, because at the time of the truncate the file has no dirty bytes to order. But, we flag the inode so that it is added to the ordered list on close (via release method). We also immediately add it to the ordered list of the current transaction so that we can try to flush down any writes the application sneaks in before commit. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2009-03-31 17:27:11 +00:00
/*
* We also have to set last_sub_trans to the current log transid,
* otherwise subsequent syncs to a file that's been synced in this
* transaction will appear to have already occurred.
Btrfs: add extra flushing for renames and truncates Renames and truncates are both common ways to replace old data with new data. The filesystem can make an effort to make sure the new data is on disk before actually replacing the old data. This is especially important for rename, which many application use as though it were atomic for both the data and the metadata involved. The current btrfs code will happily replace a file that is fully on disk with one that was just created and still has pending IO. If we crash after transaction commit but before the IO is done, we'll end up replacing a good file with a zero length file. The solution used here is to create a list of inodes that need special ordering and force them to disk before the commit is done. This is similar to the ext3 style data=ordering, except it is only done on selected files. Btrfs is able to get away with this because it does not wait on commits very often, even for fsync (which use a sub-commit). For renames, we order the file when it wasn't already on disk and when it is replacing an existing file. Larger files are sent to filemap_flush right away (before the transaction handle is opened). For truncates, we order if the file goes from non-zero size down to zero size. This is a little different, because at the time of the truncate the file has no dirty bytes to order. But, we flag the inode so that it is added to the ordered list on close (via release method). We also immediately add it to the ordered list of the current transaction so that we can try to flush down any writes the application sneaks in before commit. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2009-03-31 17:27:11 +00:00
*/
Btrfs: fix metadata inconsistencies after directory fsync We can get into inconsistency between inodes and directory entries after fsyncing a directory. The issue is that while a directory gets the new dentries persisted in the fsync log and replayed at mount time, the link count of the inode that directory entries point to doesn't get updated, staying with an incorrect link count (smaller then the correct value). This later leads to stale file handle errors when accessing (including attempt to delete) some of the links if all the other ones are removed, which also implies impossibility to delete the parent directories, since the dentries can not be removed. Another issue is that (unlike ext3/4, xfs, f2fs, reiserfs, nilfs2), when fsyncing a directory, new files aren't logged (their metadata and dentries) nor any child directories. So this patch fixes this issue too, since it has the same resolution as the incorrect inode link count issue mentioned before. This is very easy to reproduce, and the following excerpt from my test case for xfstests shows how: _scratch_mkfs >> $seqres.full 2>&1 _init_flakey _mount_flakey # Create our main test file and directory. $XFS_IO_PROG -f -c "pwrite -S 0xaa 0 8K" $SCRATCH_MNT/foo | _filter_xfs_io mkdir $SCRATCH_MNT/mydir # Make sure all metadata and data are durably persisted. sync # Add a hard link to 'foo' inside our test directory and fsync only the # directory. The btrfs fsync implementation had a bug that caused the new # directory entry to be visible after the fsync log replay but, the inode # of our file remained with a link count of 1. ln $SCRATCH_MNT/foo $SCRATCH_MNT/mydir/foo_2 # Add a few more links and new files. # This is just to verify nothing breaks or gives incorrect results after the # fsync log is replayed. ln $SCRATCH_MNT/foo $SCRATCH_MNT/mydir/foo_3 $XFS_IO_PROG -f -c "pwrite -S 0xff 0 64K" $SCRATCH_MNT/hello | _filter_xfs_io ln $SCRATCH_MNT/hello $SCRATCH_MNT/mydir/hello_2 # Add some subdirectories and new files and links to them. This is to verify # that after fsyncing our top level directory 'mydir', all the subdirectories # and their files/links are registered in the fsync log and exist after the # fsync log is replayed. mkdir -p $SCRATCH_MNT/mydir/x/y/z ln $SCRATCH_MNT/foo $SCRATCH_MNT/mydir/x/y/foo_y_link ln $SCRATCH_MNT/foo $SCRATCH_MNT/mydir/x/y/z/foo_z_link touch $SCRATCH_MNT/mydir/x/y/z/qwerty # Now fsync only our top directory. $XFS_IO_PROG -c "fsync" $SCRATCH_MNT/mydir # And fsync now our new file named 'hello', just to verify later that it has # the expected content and that the previous fsync on the directory 'mydir' had # no bad influence on this fsync. $XFS_IO_PROG -c "fsync" $SCRATCH_MNT/hello # Simulate a crash/power loss. _load_flakey_table $FLAKEY_DROP_WRITES _unmount_flakey _load_flakey_table $FLAKEY_ALLOW_WRITES _mount_flakey # Verify the content of our file 'foo' remains the same as before, 8192 bytes, # all with the value 0xaa. echo "File 'foo' content after log replay:" od -t x1 $SCRATCH_MNT/foo # Remove the first name of our inode. Because of the directory fsync bug, the # inode's link count was 1 instead of 5, so removing the 'foo' name ended up # deleting the inode and the other names became stale directory entries (still # visible to applications). Attempting to remove or access the remaining # dentries pointing to that inode resulted in stale file handle errors and # made it impossible to remove the parent directories since it was impossible # for them to become empty. echo "file 'foo' link count after log replay: $(stat -c %h $SCRATCH_MNT/foo)" rm -f $SCRATCH_MNT/foo # Now verify that all files, links and directories created before fsyncing our # directory exist after the fsync log was replayed. [ -f $SCRATCH_MNT/mydir/foo_2 ] || echo "Link mydir/foo_2 is missing" [ -f $SCRATCH_MNT/mydir/foo_3 ] || echo "Link mydir/foo_3 is missing" [ -f $SCRATCH_MNT/hello ] || echo "File hello is missing" [ -f $SCRATCH_MNT/mydir/hello_2 ] || echo "Link mydir/hello_2 is missing" [ -f $SCRATCH_MNT/mydir/x/y/foo_y_link ] || \ echo "Link mydir/x/y/foo_y_link is missing" [ -f $SCRATCH_MNT/mydir/x/y/z/foo_z_link ] || \ echo "Link mydir/x/y/z/foo_z_link is missing" [ -f $SCRATCH_MNT/mydir/x/y/z/qwerty ] || \ echo "File mydir/x/y/z/qwerty is missing" # We expect our file here to have a size of 64Kb and all the bytes having the # value 0xff. echo "file 'hello' content after log replay:" od -t x1 $SCRATCH_MNT/hello # Now remove all files/links, under our test directory 'mydir', and verify we # can remove all the directories. rm -f $SCRATCH_MNT/mydir/x/y/z/* rmdir $SCRATCH_MNT/mydir/x/y/z rm -f $SCRATCH_MNT/mydir/x/y/* rmdir $SCRATCH_MNT/mydir/x/y rmdir $SCRATCH_MNT/mydir/x rm -f $SCRATCH_MNT/mydir/* rmdir $SCRATCH_MNT/mydir # An fsck, run by the fstests framework everytime a test finishes, also detected # the inconsistency and printed the following error message: # # root 5 inode 257 errors 2001, no inode item, link count wrong # unresolved ref dir 258 index 2 namelen 5 name foo_2 filetype 1 errors 4, no inode ref # unresolved ref dir 258 index 3 namelen 5 name foo_3 filetype 1 errors 4, no inode ref status=0 exit The expected golden output for the test is: wrote 8192/8192 bytes at offset 0 XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) wrote 65536/65536 bytes at offset 0 XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) File 'foo' content after log replay: 0000000 aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa * 0020000 file 'foo' link count after log replay: 5 file 'hello' content after log replay: 0000000 ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff * 0200000 Which is the output after this patch and when running the test against ext3/4, xfs, f2fs, reiserfs or nilfs2. Without this patch, the test's output is: wrote 8192/8192 bytes at offset 0 XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) wrote 65536/65536 bytes at offset 0 XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) File 'foo' content after log replay: 0000000 aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa * 0020000 file 'foo' link count after log replay: 1 Link mydir/foo_2 is missing Link mydir/foo_3 is missing Link mydir/x/y/foo_y_link is missing Link mydir/x/y/z/foo_z_link is missing File mydir/x/y/z/qwerty is missing file 'hello' content after log replay: 0000000 ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff * 0200000 rmdir: failed to remove '/home/fdmanana/btrfs-tests/scratch_1/mydir/x/y/z': No such file or directory rmdir: failed to remove '/home/fdmanana/btrfs-tests/scratch_1/mydir/x/y': No such file or directory rmdir: failed to remove '/home/fdmanana/btrfs-tests/scratch_1/mydir/x': No such file or directory rm: cannot remove '/home/fdmanana/btrfs-tests/scratch_1/mydir/foo_2': Stale file handle rm: cannot remove '/home/fdmanana/btrfs-tests/scratch_1/mydir/foo_3': Stale file handle rmdir: failed to remove '/home/fdmanana/btrfs-tests/scratch_1/mydir': Directory not empty Fsck, without this fix, also complains about the wrong link count: root 5 inode 257 errors 2001, no inode item, link count wrong unresolved ref dir 258 index 2 namelen 5 name foo_2 filetype 1 errors 4, no inode ref unresolved ref dir 258 index 3 namelen 5 name foo_3 filetype 1 errors 4, no inode ref So fix this by logging the inodes that the dentries point to when fsyncing a directory. A test case for xfstests follows. Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-03-20 17:19:46 +00:00
spin_lock(&BTRFS_I(inode)->lock);
BTRFS_I(inode)->last_sub_trans = root->log_transid;
Btrfs: fix metadata inconsistencies after directory fsync We can get into inconsistency between inodes and directory entries after fsyncing a directory. The issue is that while a directory gets the new dentries persisted in the fsync log and replayed at mount time, the link count of the inode that directory entries point to doesn't get updated, staying with an incorrect link count (smaller then the correct value). This later leads to stale file handle errors when accessing (including attempt to delete) some of the links if all the other ones are removed, which also implies impossibility to delete the parent directories, since the dentries can not be removed. Another issue is that (unlike ext3/4, xfs, f2fs, reiserfs, nilfs2), when fsyncing a directory, new files aren't logged (their metadata and dentries) nor any child directories. So this patch fixes this issue too, since it has the same resolution as the incorrect inode link count issue mentioned before. This is very easy to reproduce, and the following excerpt from my test case for xfstests shows how: _scratch_mkfs >> $seqres.full 2>&1 _init_flakey _mount_flakey # Create our main test file and directory. $XFS_IO_PROG -f -c "pwrite -S 0xaa 0 8K" $SCRATCH_MNT/foo | _filter_xfs_io mkdir $SCRATCH_MNT/mydir # Make sure all metadata and data are durably persisted. sync # Add a hard link to 'foo' inside our test directory and fsync only the # directory. The btrfs fsync implementation had a bug that caused the new # directory entry to be visible after the fsync log replay but, the inode # of our file remained with a link count of 1. ln $SCRATCH_MNT/foo $SCRATCH_MNT/mydir/foo_2 # Add a few more links and new files. # This is just to verify nothing breaks or gives incorrect results after the # fsync log is replayed. ln $SCRATCH_MNT/foo $SCRATCH_MNT/mydir/foo_3 $XFS_IO_PROG -f -c "pwrite -S 0xff 0 64K" $SCRATCH_MNT/hello | _filter_xfs_io ln $SCRATCH_MNT/hello $SCRATCH_MNT/mydir/hello_2 # Add some subdirectories and new files and links to them. This is to verify # that after fsyncing our top level directory 'mydir', all the subdirectories # and their files/links are registered in the fsync log and exist after the # fsync log is replayed. mkdir -p $SCRATCH_MNT/mydir/x/y/z ln $SCRATCH_MNT/foo $SCRATCH_MNT/mydir/x/y/foo_y_link ln $SCRATCH_MNT/foo $SCRATCH_MNT/mydir/x/y/z/foo_z_link touch $SCRATCH_MNT/mydir/x/y/z/qwerty # Now fsync only our top directory. $XFS_IO_PROG -c "fsync" $SCRATCH_MNT/mydir # And fsync now our new file named 'hello', just to verify later that it has # the expected content and that the previous fsync on the directory 'mydir' had # no bad influence on this fsync. $XFS_IO_PROG -c "fsync" $SCRATCH_MNT/hello # Simulate a crash/power loss. _load_flakey_table $FLAKEY_DROP_WRITES _unmount_flakey _load_flakey_table $FLAKEY_ALLOW_WRITES _mount_flakey # Verify the content of our file 'foo' remains the same as before, 8192 bytes, # all with the value 0xaa. echo "File 'foo' content after log replay:" od -t x1 $SCRATCH_MNT/foo # Remove the first name of our inode. Because of the directory fsync bug, the # inode's link count was 1 instead of 5, so removing the 'foo' name ended up # deleting the inode and the other names became stale directory entries (still # visible to applications). Attempting to remove or access the remaining # dentries pointing to that inode resulted in stale file handle errors and # made it impossible to remove the parent directories since it was impossible # for them to become empty. echo "file 'foo' link count after log replay: $(stat -c %h $SCRATCH_MNT/foo)" rm -f $SCRATCH_MNT/foo # Now verify that all files, links and directories created before fsyncing our # directory exist after the fsync log was replayed. [ -f $SCRATCH_MNT/mydir/foo_2 ] || echo "Link mydir/foo_2 is missing" [ -f $SCRATCH_MNT/mydir/foo_3 ] || echo "Link mydir/foo_3 is missing" [ -f $SCRATCH_MNT/hello ] || echo "File hello is missing" [ -f $SCRATCH_MNT/mydir/hello_2 ] || echo "Link mydir/hello_2 is missing" [ -f $SCRATCH_MNT/mydir/x/y/foo_y_link ] || \ echo "Link mydir/x/y/foo_y_link is missing" [ -f $SCRATCH_MNT/mydir/x/y/z/foo_z_link ] || \ echo "Link mydir/x/y/z/foo_z_link is missing" [ -f $SCRATCH_MNT/mydir/x/y/z/qwerty ] || \ echo "File mydir/x/y/z/qwerty is missing" # We expect our file here to have a size of 64Kb and all the bytes having the # value 0xff. echo "file 'hello' content after log replay:" od -t x1 $SCRATCH_MNT/hello # Now remove all files/links, under our test directory 'mydir', and verify we # can remove all the directories. rm -f $SCRATCH_MNT/mydir/x/y/z/* rmdir $SCRATCH_MNT/mydir/x/y/z rm -f $SCRATCH_MNT/mydir/x/y/* rmdir $SCRATCH_MNT/mydir/x/y rmdir $SCRATCH_MNT/mydir/x rm -f $SCRATCH_MNT/mydir/* rmdir $SCRATCH_MNT/mydir # An fsck, run by the fstests framework everytime a test finishes, also detected # the inconsistency and printed the following error message: # # root 5 inode 257 errors 2001, no inode item, link count wrong # unresolved ref dir 258 index 2 namelen 5 name foo_2 filetype 1 errors 4, no inode ref # unresolved ref dir 258 index 3 namelen 5 name foo_3 filetype 1 errors 4, no inode ref status=0 exit The expected golden output for the test is: wrote 8192/8192 bytes at offset 0 XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) wrote 65536/65536 bytes at offset 0 XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) File 'foo' content after log replay: 0000000 aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa * 0020000 file 'foo' link count after log replay: 5 file 'hello' content after log replay: 0000000 ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff * 0200000 Which is the output after this patch and when running the test against ext3/4, xfs, f2fs, reiserfs or nilfs2. Without this patch, the test's output is: wrote 8192/8192 bytes at offset 0 XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) wrote 65536/65536 bytes at offset 0 XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) File 'foo' content after log replay: 0000000 aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa * 0020000 file 'foo' link count after log replay: 1 Link mydir/foo_2 is missing Link mydir/foo_3 is missing Link mydir/x/y/foo_y_link is missing Link mydir/x/y/z/foo_z_link is missing File mydir/x/y/z/qwerty is missing file 'hello' content after log replay: 0000000 ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff * 0200000 rmdir: failed to remove '/home/fdmanana/btrfs-tests/scratch_1/mydir/x/y/z': No such file or directory rmdir: failed to remove '/home/fdmanana/btrfs-tests/scratch_1/mydir/x/y': No such file or directory rmdir: failed to remove '/home/fdmanana/btrfs-tests/scratch_1/mydir/x': No such file or directory rm: cannot remove '/home/fdmanana/btrfs-tests/scratch_1/mydir/foo_2': Stale file handle rm: cannot remove '/home/fdmanana/btrfs-tests/scratch_1/mydir/foo_3': Stale file handle rmdir: failed to remove '/home/fdmanana/btrfs-tests/scratch_1/mydir': Directory not empty Fsck, without this fix, also complains about the wrong link count: root 5 inode 257 errors 2001, no inode item, link count wrong unresolved ref dir 258 index 2 namelen 5 name foo_2 filetype 1 errors 4, no inode ref unresolved ref dir 258 index 3 namelen 5 name foo_3 filetype 1 errors 4, no inode ref So fix this by logging the inodes that the dentries point to when fsyncing a directory. A test case for xfstests follows. Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-03-20 17:19:46 +00:00
spin_unlock(&BTRFS_I(inode)->lock);
if (num_written > 0)
num_written = generic_write_sync(iocb, num_written);
if (sync)
atomic_dec(&BTRFS_I(inode)->sync_writers);
out:
current->backing_dev_info = NULL;
return num_written ? num_written : err;
}
int btrfs_release_file(struct inode *inode, struct file *filp)
{
if (filp->private_data)
btrfs_ioctl_trans_end(filp);
/*
* ordered_data_close is set by settattr when we are about to truncate
* a file from a non-zero size to a zero size. This tries to
* flush down new bytes that may have been written if the
* application were using truncate to replace a file in place.
*/
if (test_and_clear_bit(BTRFS_INODE_ORDERED_DATA_CLOSE,
&BTRFS_I(inode)->runtime_flags))
filemap_flush(inode->i_mapping);
return 0;
}
Btrfs: fix fsync race leading to invalid data after log replay When the fsync callback (btrfs_sync_file) starts, it first waits for the writeback of any dirty pages to start and finish without holding the inode's mutex (to reduce contention). After this it acquires the inode's mutex and repeats that process via btrfs_wait_ordered_range only if we're doing a full sync (BTRFS_INODE_NEEDS_FULL_SYNC flag is set on the inode). This is not safe for a non full sync - we need to start and wait for writeback to finish for any pages that might have been made dirty before acquiring the inode's mutex and after that first step mentioned before. Why this is needed is explained by the following comment added to btrfs_sync_file: "Right before acquiring the inode's mutex, we might have new writes dirtying pages, which won't immediately start the respective ordered operations - that is done through the fill_delalloc callbacks invoked from the writepage and writepages address space operations. So make sure we start all ordered operations before starting to log our inode. Not doing this means that while logging the inode, writeback could start and invoke writepage/writepages, which would call the fill_delalloc callbacks (cow_file_range, submit_compressed_extents). These callbacks add first an extent map to the modified list of extents and then create the respective ordered operation, which means in tree-log.c:btrfs_log_inode() we might capture all existing ordered operations (with btrfs_get_logged_extents()) before the fill_delalloc callback adds its ordered operation, and by the time we visit the modified list of extent maps (with btrfs_log_changed_extents()), we see and process the extent map they created. We then use the extent map to construct a file extent item for logging without waiting for the respective ordered operation to finish - this file extent item points to a disk location that might not have yet been written to, containing random data - so after a crash a log replay will make our inode have file extent items that point to disk locations containing invalid data, as we returned success to userspace without waiting for the respective ordered operation to finish, because it wasn't captured by btrfs_get_logged_extents()." Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-09-02 10:09:58 +00:00
static int start_ordered_ops(struct inode *inode, loff_t start, loff_t end)
{
int ret;
atomic_inc(&BTRFS_I(inode)->sync_writers);
ret = btrfs_fdatawrite_range(inode, start, end);
Btrfs: fix fsync race leading to invalid data after log replay When the fsync callback (btrfs_sync_file) starts, it first waits for the writeback of any dirty pages to start and finish without holding the inode's mutex (to reduce contention). After this it acquires the inode's mutex and repeats that process via btrfs_wait_ordered_range only if we're doing a full sync (BTRFS_INODE_NEEDS_FULL_SYNC flag is set on the inode). This is not safe for a non full sync - we need to start and wait for writeback to finish for any pages that might have been made dirty before acquiring the inode's mutex and after that first step mentioned before. Why this is needed is explained by the following comment added to btrfs_sync_file: "Right before acquiring the inode's mutex, we might have new writes dirtying pages, which won't immediately start the respective ordered operations - that is done through the fill_delalloc callbacks invoked from the writepage and writepages address space operations. So make sure we start all ordered operations before starting to log our inode. Not doing this means that while logging the inode, writeback could start and invoke writepage/writepages, which would call the fill_delalloc callbacks (cow_file_range, submit_compressed_extents). These callbacks add first an extent map to the modified list of extents and then create the respective ordered operation, which means in tree-log.c:btrfs_log_inode() we might capture all existing ordered operations (with btrfs_get_logged_extents()) before the fill_delalloc callback adds its ordered operation, and by the time we visit the modified list of extent maps (with btrfs_log_changed_extents()), we see and process the extent map they created. We then use the extent map to construct a file extent item for logging without waiting for the respective ordered operation to finish - this file extent item points to a disk location that might not have yet been written to, containing random data - so after a crash a log replay will make our inode have file extent items that point to disk locations containing invalid data, as we returned success to userspace without waiting for the respective ordered operation to finish, because it wasn't captured by btrfs_get_logged_extents()." Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-09-02 10:09:58 +00:00
atomic_dec(&BTRFS_I(inode)->sync_writers);
return ret;
}
/*
* fsync call for both files and directories. This logs the inode into
* the tree log instead of forcing full commits whenever possible.
*
* It needs to call filemap_fdatawait so that all ordered extent updates are
* in the metadata btree are up to date for copying to the log.
*
* It drops the inode mutex before doing the tree log commit. This is an
* important optimization for directories because holding the mutex prevents
* new operations on the dir while we write to disk.
*/
int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
{
btrfs: fix crash/invalid memory access on fsync when using overlayfs If the lower or upper directory of an overlayfs mount belong to a btrfs file system and we fsync the file through the overlayfs' merged directory we ended up accessing an inode that didn't belong to btrfs as if it were a btrfs inode at btrfs_sync_file() resulting in a crash like the following: [ 7782.588845] BUG: unable to handle kernel NULL pointer dereference at 0000000000000544 [ 7782.590624] IP: [<ffffffffa030b7ab>] btrfs_sync_file+0x11b/0x3e9 [btrfs] [ 7782.591931] PGD 4d954067 PUD 1e878067 PMD 0 [ 7782.592016] Oops: 0002 [#6] PREEMPT SMP DEBUG_PAGEALLOC [ 7782.592016] Modules linked in: btrfs overlay ppdev crc32c_generic evdev xor raid6_pq psmouse pcspkr sg serio_raw acpi_cpufreq parport_pc parport tpm_tis i2c_piix4 tpm i2c_core processor button loop autofs4 ext4 crc16 mbcache jbd2 sr_mod cdrom sd_mod ata_generic virtio_scsi ata_piix virtio_pci libata virtio_ring virtio scsi_mod e1000 floppy [last unloaded: btrfs] [ 7782.592016] CPU: 10 PID: 16437 Comm: xfs_io Tainted: G D 4.5.0-rc6-btrfs-next-26+ #1 [ 7782.592016] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS by qemu-project.org 04/01/2014 [ 7782.592016] task: ffff88001b8d40c0 ti: ffff880137488000 task.ti: ffff880137488000 [ 7782.592016] RIP: 0010:[<ffffffffa030b7ab>] [<ffffffffa030b7ab>] btrfs_sync_file+0x11b/0x3e9 [btrfs] [ 7782.592016] RSP: 0018:ffff88013748be40 EFLAGS: 00010286 [ 7782.592016] RAX: 0000000080000000 RBX: ffff880133b30c88 RCX: 0000000000000001 [ 7782.592016] RDX: 0000000000000001 RSI: ffffffff8148fec0 RDI: 00000000ffffffff [ 7782.592016] RBP: ffff88013748bec0 R08: 0000000000000001 R09: 0000000000000000 [ 7782.624248] R10: ffff88013748be40 R11: 0000000000000246 R12: 0000000000000000 [ 7782.624248] R13: 0000000000000000 R14: 00000000009305a0 R15: ffff880015e3be40 [ 7782.624248] FS: 00007fa83b9cb700(0000) GS:ffff88023ed40000(0000) knlGS:0000000000000000 [ 7782.624248] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 7782.624248] CR2: 0000000000000544 CR3: 00000001fa652000 CR4: 00000000000006e0 [ 7782.624248] Stack: [ 7782.624248] ffffffff8108b5cc ffff88013748bec0 0000000000000246 ffff8800b005ded0 [ 7782.624248] ffff880133b30d60 8000000000000000 7fffffffffffffff 0000000000000246 [ 7782.624248] 0000000000000246 ffffffff81074f9b ffffffff8104357c ffff880015e3be40 [ 7782.624248] Call Trace: [ 7782.624248] [<ffffffff8108b5cc>] ? arch_local_irq_save+0x9/0xc [ 7782.624248] [<ffffffff81074f9b>] ? ___might_sleep+0xce/0x217 [ 7782.624248] [<ffffffff8104357c>] ? __do_page_fault+0x3c0/0x43a [ 7782.624248] [<ffffffff811a2351>] vfs_fsync_range+0x8c/0x9e [ 7782.624248] [<ffffffff811a237f>] vfs_fsync+0x1c/0x1e [ 7782.624248] [<ffffffff811a24d6>] do_fsync+0x31/0x4a [ 7782.624248] [<ffffffff811a2700>] SyS_fsync+0x10/0x14 [ 7782.624248] [<ffffffff81493617>] entry_SYSCALL_64_fastpath+0x12/0x6b [ 7782.624248] Code: 85 c0 0f 85 e2 02 00 00 48 8b 45 b0 31 f6 4c 29 e8 48 ff c0 48 89 45 a8 48 8d 83 d8 00 00 00 48 89 c7 48 89 45 a0 e8 fc 43 18 e1 <f0> 41 ff 84 24 44 05 00 00 48 8b 83 58 ff ff ff 48 c1 e8 07 83 [ 7782.624248] RIP [<ffffffffa030b7ab>] btrfs_sync_file+0x11b/0x3e9 [btrfs] [ 7782.624248] RSP <ffff88013748be40> [ 7782.624248] CR2: 0000000000000544 [ 7782.661994] ---[ end trace 721e14960eb939bc ]--- This started happening since commit 4bacc9c9234 (overlayfs: Make f_path always point to the overlay and f_inode to the underlay) and even though after this change we could still access the btrfs inode through struct file->f_mapping->host or struct file->f_inode, we would end up resulting in more similar issues later on at check_parent_dirs_for_sync() because the dentry we got (from struct file->f_path.dentry) was from overlayfs and not from btrfs, that is, we had no way of getting the dentry that belonged to btrfs (we always got the dentry that belonged to overlayfs). The new patch from Miklos Szeredi, titled "vfs: add file_dentry()" and recently submitted to linux-fsdevel, adds a file_dentry() API that allows us to get the btrfs dentry from the input file and therefore being able to fsync when the upper and lower directories belong to btrfs filesystems. This issue has been reported several times by users in the mailing list and bugzilla. A test case for xfstests is being submitted as well. Fixes: 4bacc9c9234c ("overlayfs: Make f_path always point to the overlay and f_inode to the underlay") Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=101951 Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=109791 Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com> Cc: stable@vger.kernel.org
2016-03-30 23:03:13 +00:00
struct dentry *dentry = file_dentry(file);
struct inode *inode = d_inode(dentry);
struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
struct btrfs_root *root = BTRFS_I(inode)->root;
struct btrfs_trans_handle *trans;
struct btrfs_log_ctx ctx;
int ret = 0;
bool full_sync = 0;
u64 len;
/*
* The range length can be represented by u64, we have to do the typecasts
* to avoid signed overflow if it's [0, LLONG_MAX] eg. from fsync()
*/
len = (u64)end - (u64)start + 1;
Btrfs: add initial tracepoint support for btrfs Tracepoints can provide insight into why btrfs hits bugs and be greatly helpful for debugging, e.g dd-7822 [000] 2121.641088: btrfs_inode_request: root = 5(FS_TREE), gen = 4, ino = 256, blocks = 8, disk_i_size = 0, last_trans = 8, logged_trans = 0 dd-7822 [000] 2121.641100: btrfs_inode_new: root = 5(FS_TREE), gen = 8, ino = 257, blocks = 0, disk_i_size = 0, last_trans = 0, logged_trans = 0 btrfs-transacti-7804 [001] 2146.935420: btrfs_cow_block: root = 2(EXTENT_TREE), refs = 2, orig_buf = 29368320 (orig_level = 0), cow_buf = 29388800 (cow_level = 0) btrfs-transacti-7804 [001] 2146.935473: btrfs_cow_block: root = 1(ROOT_TREE), refs = 2, orig_buf = 29364224 (orig_level = 0), cow_buf = 29392896 (cow_level = 0) btrfs-transacti-7804 [001] 2146.972221: btrfs_transaction_commit: root = 1(ROOT_TREE), gen = 8 flush-btrfs-2-7821 [001] 2155.824210: btrfs_chunk_alloc: root = 3(CHUNK_TREE), offset = 1103101952, size = 1073741824, num_stripes = 1, sub_stripes = 0, type = DATA flush-btrfs-2-7821 [001] 2155.824241: btrfs_cow_block: root = 2(EXTENT_TREE), refs = 2, orig_buf = 29388800 (orig_level = 0), cow_buf = 29396992 (cow_level = 0) flush-btrfs-2-7821 [001] 2155.824255: btrfs_cow_block: root = 4(DEV_TREE), refs = 2, orig_buf = 29372416 (orig_level = 0), cow_buf = 29401088 (cow_level = 0) flush-btrfs-2-7821 [000] 2155.824329: btrfs_cow_block: root = 3(CHUNK_TREE), refs = 2, orig_buf = 20971520 (orig_level = 0), cow_buf = 20975616 (cow_level = 0) btrfs-endio-wri-7800 [001] 2155.898019: btrfs_cow_block: root = 5(FS_TREE), refs = 2, orig_buf = 29384704 (orig_level = 0), cow_buf = 29405184 (cow_level = 0) btrfs-endio-wri-7800 [001] 2155.898043: btrfs_cow_block: root = 7(CSUM_TREE), refs = 2, orig_buf = 29376512 (orig_level = 0), cow_buf = 29409280 (cow_level = 0) Here is what I have added: 1) ordere_extent: btrfs_ordered_extent_add btrfs_ordered_extent_remove btrfs_ordered_extent_start btrfs_ordered_extent_put These provide critical information to understand how ordered_extents are updated. 2) extent_map: btrfs_get_extent extent_map is used in both read and write cases, and it is useful for tracking how btrfs specific IO is running. 3) writepage: __extent_writepage btrfs_writepage_end_io_hook Pages are cirtical resourses and produce a lot of corner cases during writeback, so it is valuable to know how page is written to disk. 4) inode: btrfs_inode_new btrfs_inode_request btrfs_inode_evict These can show where and when a inode is created, when a inode is evicted. 5) sync: btrfs_sync_file btrfs_sync_fs These show sync arguments. 6) transaction: btrfs_transaction_commit In transaction based filesystem, it will be useful to know the generation and who does commit. 7) back reference and cow: btrfs_delayed_tree_ref btrfs_delayed_data_ref btrfs_delayed_ref_head btrfs_cow_block Btrfs natively supports back references, these tracepoints are helpful on understanding btrfs's COW mechanism. 8) chunk: btrfs_chunk_alloc btrfs_chunk_free Chunk is a link between physical offset and logical offset, and stands for space infomation in btrfs, and these are helpful on tracing space things. 9) reserved_extent: btrfs_reserved_extent_alloc btrfs_reserved_extent_free These can show how btrfs uses its space. Signed-off-by: Liu Bo <liubo2009@cn.fujitsu.com> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2011-03-24 11:18:59 +00:00
trace_btrfs_sync_file(file, datasync);
/*
* We write the dirty pages in the range and wait until they complete
* out of the ->i_mutex. If so, we can flush the dirty pages by
* multi-task, and make the performance up. See
* btrfs_wait_ordered_range for an explanation of the ASYNC check.
*/
Btrfs: fix fsync race leading to invalid data after log replay When the fsync callback (btrfs_sync_file) starts, it first waits for the writeback of any dirty pages to start and finish without holding the inode's mutex (to reduce contention). After this it acquires the inode's mutex and repeats that process via btrfs_wait_ordered_range only if we're doing a full sync (BTRFS_INODE_NEEDS_FULL_SYNC flag is set on the inode). This is not safe for a non full sync - we need to start and wait for writeback to finish for any pages that might have been made dirty before acquiring the inode's mutex and after that first step mentioned before. Why this is needed is explained by the following comment added to btrfs_sync_file: "Right before acquiring the inode's mutex, we might have new writes dirtying pages, which won't immediately start the respective ordered operations - that is done through the fill_delalloc callbacks invoked from the writepage and writepages address space operations. So make sure we start all ordered operations before starting to log our inode. Not doing this means that while logging the inode, writeback could start and invoke writepage/writepages, which would call the fill_delalloc callbacks (cow_file_range, submit_compressed_extents). These callbacks add first an extent map to the modified list of extents and then create the respective ordered operation, which means in tree-log.c:btrfs_log_inode() we might capture all existing ordered operations (with btrfs_get_logged_extents()) before the fill_delalloc callback adds its ordered operation, and by the time we visit the modified list of extent maps (with btrfs_log_changed_extents()), we see and process the extent map they created. We then use the extent map to construct a file extent item for logging without waiting for the respective ordered operation to finish - this file extent item points to a disk location that might not have yet been written to, containing random data - so after a crash a log replay will make our inode have file extent items that point to disk locations containing invalid data, as we returned success to userspace without waiting for the respective ordered operation to finish, because it wasn't captured by btrfs_get_logged_extents()." Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-09-02 10:09:58 +00:00
ret = start_ordered_ops(inode, start, end);
if (ret)
return ret;
inode_lock(inode);
atomic_inc(&root->log_batch);
full_sync = test_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
&BTRFS_I(inode)->runtime_flags);
Btrfs: fix fsync race leading to invalid data after log replay When the fsync callback (btrfs_sync_file) starts, it first waits for the writeback of any dirty pages to start and finish without holding the inode's mutex (to reduce contention). After this it acquires the inode's mutex and repeats that process via btrfs_wait_ordered_range only if we're doing a full sync (BTRFS_INODE_NEEDS_FULL_SYNC flag is set on the inode). This is not safe for a non full sync - we need to start and wait for writeback to finish for any pages that might have been made dirty before acquiring the inode's mutex and after that first step mentioned before. Why this is needed is explained by the following comment added to btrfs_sync_file: "Right before acquiring the inode's mutex, we might have new writes dirtying pages, which won't immediately start the respective ordered operations - that is done through the fill_delalloc callbacks invoked from the writepage and writepages address space operations. So make sure we start all ordered operations before starting to log our inode. Not doing this means that while logging the inode, writeback could start and invoke writepage/writepages, which would call the fill_delalloc callbacks (cow_file_range, submit_compressed_extents). These callbacks add first an extent map to the modified list of extents and then create the respective ordered operation, which means in tree-log.c:btrfs_log_inode() we might capture all existing ordered operations (with btrfs_get_logged_extents()) before the fill_delalloc callback adds its ordered operation, and by the time we visit the modified list of extent maps (with btrfs_log_changed_extents()), we see and process the extent map they created. We then use the extent map to construct a file extent item for logging without waiting for the respective ordered operation to finish - this file extent item points to a disk location that might not have yet been written to, containing random data - so after a crash a log replay will make our inode have file extent items that point to disk locations containing invalid data, as we returned success to userspace without waiting for the respective ordered operation to finish, because it wasn't captured by btrfs_get_logged_extents()." Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-09-02 10:09:58 +00:00
/*
* We might have have had more pages made dirty after calling
* start_ordered_ops and before acquiring the inode's i_mutex.
*/
if (full_sync) {
Btrfs: fix fsync race leading to invalid data after log replay When the fsync callback (btrfs_sync_file) starts, it first waits for the writeback of any dirty pages to start and finish without holding the inode's mutex (to reduce contention). After this it acquires the inode's mutex and repeats that process via btrfs_wait_ordered_range only if we're doing a full sync (BTRFS_INODE_NEEDS_FULL_SYNC flag is set on the inode). This is not safe for a non full sync - we need to start and wait for writeback to finish for any pages that might have been made dirty before acquiring the inode's mutex and after that first step mentioned before. Why this is needed is explained by the following comment added to btrfs_sync_file: "Right before acquiring the inode's mutex, we might have new writes dirtying pages, which won't immediately start the respective ordered operations - that is done through the fill_delalloc callbacks invoked from the writepage and writepages address space operations. So make sure we start all ordered operations before starting to log our inode. Not doing this means that while logging the inode, writeback could start and invoke writepage/writepages, which would call the fill_delalloc callbacks (cow_file_range, submit_compressed_extents). These callbacks add first an extent map to the modified list of extents and then create the respective ordered operation, which means in tree-log.c:btrfs_log_inode() we might capture all existing ordered operations (with btrfs_get_logged_extents()) before the fill_delalloc callback adds its ordered operation, and by the time we visit the modified list of extent maps (with btrfs_log_changed_extents()), we see and process the extent map they created. We then use the extent map to construct a file extent item for logging without waiting for the respective ordered operation to finish - this file extent item points to a disk location that might not have yet been written to, containing random data - so after a crash a log replay will make our inode have file extent items that point to disk locations containing invalid data, as we returned success to userspace without waiting for the respective ordered operation to finish, because it wasn't captured by btrfs_get_logged_extents()." Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-09-02 10:09:58 +00:00
/*
* For a full sync, we need to make sure any ordered operations
* start and finish before we start logging the inode, so that
* all extents are persisted and the respective file extent
* items are in the fs/subvol btree.
*/
Btrfs: avoid syncing log in the fast fsync path when not necessary Commit 3a8b36f37806 ("Btrfs: fix data loss in the fast fsync path") added a performance regression for that causes an unnecessary sync of the log trees (fs/subvol and root log trees) when 2 consecutive fsyncs are done against a file, without no writes or any metadata updates to the inode in between them and if a transaction is committed before the second fsync is called. Huang Ying reported this to lkml (https://lkml.org/lkml/2015/3/18/99) after a test sysbench test that measured a -62% decrease of file io requests per second for that tests' workload. The test is: echo performance > /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor echo performance > /sys/devices/system/cpu/cpu1/cpufreq/scaling_governor echo performance > /sys/devices/system/cpu/cpu2/cpufreq/scaling_governor echo performance > /sys/devices/system/cpu/cpu3/cpufreq/scaling_governor mkfs -t btrfs /dev/sda2 mount -t btrfs /dev/sda2 /fs/sda2 cd /fs/sda2 for ((i = 0; i < 1024; i++)); do fallocate -l 67108864 testfile.$i; done sysbench --test=fileio --max-requests=0 --num-threads=4 --max-time=600 \ --file-test-mode=rndwr --file-total-size=68719476736 --file-io-mode=sync \ --file-num=1024 run A test on kvm guest, running a debug kernel gave me the following results: Without 3a8b36f378060d: 16.01 reqs/sec With 3a8b36f378060d: 3.39 reqs/sec With 3a8b36f378060d and this patch: 16.04 reqs/sec Reported-by: Huang Ying <ying.huang@intel.com> Tested-by: Huang, Ying <ying.huang@intel.com> Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-03-31 13:16:52 +00:00
ret = btrfs_wait_ordered_range(inode, start, len);
Btrfs: fix fsync race leading to invalid data after log replay When the fsync callback (btrfs_sync_file) starts, it first waits for the writeback of any dirty pages to start and finish without holding the inode's mutex (to reduce contention). After this it acquires the inode's mutex and repeats that process via btrfs_wait_ordered_range only if we're doing a full sync (BTRFS_INODE_NEEDS_FULL_SYNC flag is set on the inode). This is not safe for a non full sync - we need to start and wait for writeback to finish for any pages that might have been made dirty before acquiring the inode's mutex and after that first step mentioned before. Why this is needed is explained by the following comment added to btrfs_sync_file: "Right before acquiring the inode's mutex, we might have new writes dirtying pages, which won't immediately start the respective ordered operations - that is done through the fill_delalloc callbacks invoked from the writepage and writepages address space operations. So make sure we start all ordered operations before starting to log our inode. Not doing this means that while logging the inode, writeback could start and invoke writepage/writepages, which would call the fill_delalloc callbacks (cow_file_range, submit_compressed_extents). These callbacks add first an extent map to the modified list of extents and then create the respective ordered operation, which means in tree-log.c:btrfs_log_inode() we might capture all existing ordered operations (with btrfs_get_logged_extents()) before the fill_delalloc callback adds its ordered operation, and by the time we visit the modified list of extent maps (with btrfs_log_changed_extents()), we see and process the extent map they created. We then use the extent map to construct a file extent item for logging without waiting for the respective ordered operation to finish - this file extent item points to a disk location that might not have yet been written to, containing random data - so after a crash a log replay will make our inode have file extent items that point to disk locations containing invalid data, as we returned success to userspace without waiting for the respective ordered operation to finish, because it wasn't captured by btrfs_get_logged_extents()." Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-09-02 10:09:58 +00:00
} else {
/*
* Start any new ordered operations before starting to log the
* inode. We will wait for them to finish in btrfs_sync_log().
*
* Right before acquiring the inode's mutex, we might have new
* writes dirtying pages, which won't immediately start the
* respective ordered operations - that is done through the
* fill_delalloc callbacks invoked from the writepage and
* writepages address space operations. So make sure we start
* all ordered operations before starting to log our inode. Not
* doing this means that while logging the inode, writeback
* could start and invoke writepage/writepages, which would call
* the fill_delalloc callbacks (cow_file_range,
* submit_compressed_extents). These callbacks add first an
* extent map to the modified list of extents and then create
* the respective ordered operation, which means in
* tree-log.c:btrfs_log_inode() we might capture all existing
* ordered operations (with btrfs_get_logged_extents()) before
* the fill_delalloc callback adds its ordered operation, and by
* the time we visit the modified list of extent maps (with
* btrfs_log_changed_extents()), we see and process the extent
* map they created. We then use the extent map to construct a
* file extent item for logging without waiting for the
* respective ordered operation to finish - this file extent
* item points to a disk location that might not have yet been
* written to, containing random data - so after a crash a log
* replay will make our inode have file extent items that point
* to disk locations containing invalid data, as we returned
* success to userspace without waiting for the respective
* ordered operation to finish, because it wasn't captured by
* btrfs_get_logged_extents().
*/
ret = start_ordered_ops(inode, start, end);
}
if (ret) {
inode_unlock(inode);
Btrfs: fix fsync race leading to invalid data after log replay When the fsync callback (btrfs_sync_file) starts, it first waits for the writeback of any dirty pages to start and finish without holding the inode's mutex (to reduce contention). After this it acquires the inode's mutex and repeats that process via btrfs_wait_ordered_range only if we're doing a full sync (BTRFS_INODE_NEEDS_FULL_SYNC flag is set on the inode). This is not safe for a non full sync - we need to start and wait for writeback to finish for any pages that might have been made dirty before acquiring the inode's mutex and after that first step mentioned before. Why this is needed is explained by the following comment added to btrfs_sync_file: "Right before acquiring the inode's mutex, we might have new writes dirtying pages, which won't immediately start the respective ordered operations - that is done through the fill_delalloc callbacks invoked from the writepage and writepages address space operations. So make sure we start all ordered operations before starting to log our inode. Not doing this means that while logging the inode, writeback could start and invoke writepage/writepages, which would call the fill_delalloc callbacks (cow_file_range, submit_compressed_extents). These callbacks add first an extent map to the modified list of extents and then create the respective ordered operation, which means in tree-log.c:btrfs_log_inode() we might capture all existing ordered operations (with btrfs_get_logged_extents()) before the fill_delalloc callback adds its ordered operation, and by the time we visit the modified list of extent maps (with btrfs_log_changed_extents()), we see and process the extent map they created. We then use the extent map to construct a file extent item for logging without waiting for the respective ordered operation to finish - this file extent item points to a disk location that might not have yet been written to, containing random data - so after a crash a log replay will make our inode have file extent items that point to disk locations containing invalid data, as we returned success to userspace without waiting for the respective ordered operation to finish, because it wasn't captured by btrfs_get_logged_extents()." Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-09-02 10:09:58 +00:00
goto out;
}
atomic_inc(&root->log_batch);
/*
Btrfs: fix data loss in the fast fsync path When using the fast file fsync code path we can miss the fact that new writes happened since the last file fsync and therefore return without waiting for the IO to finish and write the new extents to the fsync log. Here's an example scenario where the fsync will miss the fact that new file data exists that wasn't yet durably persisted: 1. fs_info->last_trans_committed == N - 1 and current transaction is transaction N (fs_info->generation == N); 2. do a buffered write; 3. fsync our inode, this clears our inode's full sync flag, starts an ordered extent and waits for it to complete - when it completes at btrfs_finish_ordered_io(), the inode's last_trans is set to the value N (via btrfs_update_inode_fallback -> btrfs_update_inode -> btrfs_set_inode_last_trans); 4. transaction N is committed, so fs_info->last_trans_committed is now set to the value N and fs_info->generation remains with the value N; 5. do another buffered write, when this happens btrfs_file_write_iter sets our inode's last_trans to the value N + 1 (that is fs_info->generation + 1 == N + 1); 6. transaction N + 1 is started and fs_info->generation now has the value N + 1; 7. transaction N + 1 is committed, so fs_info->last_trans_committed is set to the value N + 1; 8. fsync our inode - because it doesn't have the full sync flag set, we only start the ordered extent, we don't wait for it to complete (only in a later phase) therefore its last_trans field has the value N + 1 set previously by btrfs_file_write_iter(), and so we have: inode->last_trans <= fs_info->last_trans_committed (N + 1) (N + 1) Which made us not log the last buffered write and exit the fsync handler immediately, returning success (0) to user space and resulting in data loss after a crash. This can actually be triggered deterministically and the following excerpt from a testcase I made for xfstests triggers the issue. It moves a dummy file across directories and then fsyncs the old parent directory - this is just to trigger a transaction commit, so moving files around isn't directly related to the issue but it was chosen because running 'sync' for example does more than just committing the current transaction, as it flushes/waits for all file data to be persisted. The issue can also happen at random periods, since the transaction kthread periodicaly commits the current transaction (about every 30 seconds by default). The body of the test is: _scratch_mkfs >> $seqres.full 2>&1 _init_flakey _mount_flakey # Create our main test file 'foo', the one we check for data loss. # By doing an fsync against our file, it makes btrfs clear the 'needs_full_sync' # bit from its flags (btrfs inode specific flags). $XFS_IO_PROG -f -c "pwrite -S 0xaa 0 8K" \ -c "fsync" $SCRATCH_MNT/foo | _filter_xfs_io # Now create one other file and 2 directories. We will move this second file # from one directory to the other later because it forces btrfs to commit its # currently open transaction if we fsync the old parent directory. This is # necessary to trigger the data loss bug that affected btrfs. mkdir $SCRATCH_MNT/testdir_1 touch $SCRATCH_MNT/testdir_1/bar mkdir $SCRATCH_MNT/testdir_2 # Make sure everything is durably persisted. sync # Write more 8Kb of data to our file. $XFS_IO_PROG -c "pwrite -S 0xbb 8K 8K" $SCRATCH_MNT/foo | _filter_xfs_io # Move our 'bar' file into a new directory. mv $SCRATCH_MNT/testdir_1/bar $SCRATCH_MNT/testdir_2/bar # Fsync our first directory. Because it had a file moved into some other # directory, this made btrfs commit the currently open transaction. This is # a condition necessary to trigger the data loss bug. $XFS_IO_PROG -c "fsync" $SCRATCH_MNT/testdir_1 # Now fsync our main test file. If the fsync succeeds, we expect the 8Kb of # data we wrote previously to be persisted and available if a crash happens. # This did not happen with btrfs, because of the transaction commit that # happened when we fsynced the parent directory. $XFS_IO_PROG -c "fsync" $SCRATCH_MNT/foo # Simulate a crash/power loss. _load_flakey_table $FLAKEY_DROP_WRITES _unmount_flakey _load_flakey_table $FLAKEY_ALLOW_WRITES _mount_flakey # Now check that all data we wrote before are available. echo "File content after log replay:" od -t x1 $SCRATCH_MNT/foo status=0 exit The expected golden output for the test, which is what we get with this fix applied (or when running against ext3/4 and xfs), is: wrote 8192/8192 bytes at offset 0 XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) wrote 8192/8192 bytes at offset 8192 XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) File content after log replay: 0000000 aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa * 0020000 bb bb bb bb bb bb bb bb bb bb bb bb bb bb bb bb * 0040000 Without this fix applied, the output shows the test file does not have the second 8Kb extent that we successfully fsynced: wrote 8192/8192 bytes at offset 0 XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) wrote 8192/8192 bytes at offset 8192 XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) File content after log replay: 0000000 aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa * 0020000 So fix this by skipping the fsync only if we're doing a full sync and if the inode's last_trans is <= fs_info->last_trans_committed, or if the inode is already in the log. Also remove setting the inode's last_trans in btrfs_file_write_iter since it's useless/unreliable. Also because btrfs_file_write_iter no longer sets inode->last_trans to fs_info->generation + 1, don't set last_trans to 0 if we bail out and don't bail out if last_trans is 0, otherwise something as simple as the following example wouldn't log the second write on the last fsync: 1. write to file 2. fsync file 3. fsync file |--> btrfs_inode_in_log() returns true and it set last_trans to 0 4. write to file |--> btrfs_file_write_iter() no longers sets last_trans, so it remained with a value of 0 5. fsync |--> inode->last_trans == 0, so it bails out without logging the second write A test case for xfstests will be sent soon. CC: <stable@vger.kernel.org> Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-03-01 20:36:00 +00:00
* If the last transaction that changed this file was before the current
* transaction and we have the full sync flag set in our inode, we can
* bail out now without any syncing.
*
* Note that we can't bail out if the full sync flag isn't set. This is
* because when the full sync flag is set we start all ordered extents
* and wait for them to fully complete - when they complete they update
* the inode's last_trans field through:
*
* btrfs_finish_ordered_io() ->
* btrfs_update_inode_fallback() ->
* btrfs_update_inode() ->
* btrfs_set_inode_last_trans()
*
* So we are sure that last_trans is up to date and can do this check to
* bail out safely. For the fast path, when the full sync flag is not
* set in our inode, we can not do it because we start only our ordered
* extents and don't wait for them to complete (that is when
* btrfs_finish_ordered_io runs), so here at this point their last_trans
* value might be less than or equals to fs_info->last_trans_committed,
* and setting a speculative last_trans for an inode when a buffered
* write is made (such as fs_info->generation + 1 for example) would not
* be reliable since after setting the value and before fsync is called
* any number of transactions can start and commit (transaction kthread
* commits the current transaction periodically), and a transaction
* commit does not start nor waits for ordered extents to complete.
*/
Btrfs: kill trans_mutex We use trans_mutex for lots of things, here's a basic list 1) To serialize trans_handles joining the currently running transaction 2) To make sure that no new trans handles are started while we are committing 3) To protect the dead_roots list and the transaction lists Really the serializing trans_handles joining is not too hard, and can really get bogged down in acquiring a reference to the transaction. So replace the trans_mutex with a trans_lock spinlock and use it to do the following 1) Protect fs_info->running_transaction. All trans handles have to do is check this, and then take a reference of the transaction and keep on going. 2) Protect the fs_info->trans_list. This doesn't get used too much, basically it just holds the current transactions, which will usually just be the currently committing transaction and the currently running transaction at most. 3) Protect the dead roots list. This is only ever processed by splicing the list so this is relatively simple. 4) Protect the fs_info->reloc_ctl stuff. This is very lightweight and was using the trans_mutex before, so this is a pretty straightforward change. 5) Protect fs_info->no_trans_join. Because we don't hold the trans_lock over the entirety of the commit we need to have a way to block new people from creating a new transaction while we're doing our work. So we set no_trans_join and in join_transaction we test to see if that is set, and if it is we do a wait_on_commit. 6) Make the transaction use count atomic so we don't need to take locks to modify it when we're dropping references. 7) Add a commit_lock to the transaction to make sure multiple people trying to commit the same transaction don't race and commit at the same time. 8) Make open_ioctl_trans an atomic so we don't have to take any locks for ioctl trans. I have tested this with xfstests, but obviously it is a pretty hairy change so lots of testing is greatly appreciated. Thanks, Signed-off-by: Josef Bacik <josef@redhat.com>
2011-04-11 21:25:13 +00:00
smp_mb();
if (btrfs_inode_in_log(BTRFS_I(inode), fs_info->generation) ||
(full_sync && BTRFS_I(inode)->last_trans <=
fs_info->last_trans_committed) ||
(!btrfs_have_ordered_extents_in_range(inode, start, len) &&
BTRFS_I(inode)->last_trans
<= fs_info->last_trans_committed)) {
Btrfs: turbo charge fsync At least for the vm workload. Currently on fsync we will 1) Truncate all items in the log tree for the given inode if they exist and 2) Copy all items for a given inode into the log The problem with this is that for things like VMs you can have lots of extents from the fragmented writing behavior, and worst yet you may have only modified a few extents, not the entire thing. This patch fixes this problem by tracking which transid modified our extent, and then when we do the tree logging we find all of the extents we've modified in our current transaction, sort them and commit them. We also only truncate up to the xattrs of the inode and copy that stuff in normally, and then just drop any extents in the range we have that exist in the log already. Here are some numbers of a 50 meg fio job that does random writes and fsync()s after every write Original Patched SATA drive 82KB/s 140KB/s Fusion drive 431KB/s 2532KB/s So around 2-6 times faster depending on your hardware. There are a few corner cases, for example if you truncate at all we have to do it the old way since there is no way to be sure what is in the log is ok. This probably could be done smarter, but if you write-fsync-truncate-write-fsync you deserve what you get. All this work is in RAM of course so if your inode gets evicted from cache and you read it in and fsync it we'll do it the slow way if we are still in the same transaction that we last modified the inode in. The biggest cool part of this is that it requires no changes to the recovery code, so if you fsync with this patch and crash and load an old kernel, it will run the recovery and be a-ok. I have tested this pretty thoroughly with an fsync tester and everything comes back fine, as well as xfstests. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2012-08-17 17:14:17 +00:00
/*
* We've had everything committed since the last time we were
Btrfs: turbo charge fsync At least for the vm workload. Currently on fsync we will 1) Truncate all items in the log tree for the given inode if they exist and 2) Copy all items for a given inode into the log The problem with this is that for things like VMs you can have lots of extents from the fragmented writing behavior, and worst yet you may have only modified a few extents, not the entire thing. This patch fixes this problem by tracking which transid modified our extent, and then when we do the tree logging we find all of the extents we've modified in our current transaction, sort them and commit them. We also only truncate up to the xattrs of the inode and copy that stuff in normally, and then just drop any extents in the range we have that exist in the log already. Here are some numbers of a 50 meg fio job that does random writes and fsync()s after every write Original Patched SATA drive 82KB/s 140KB/s Fusion drive 431KB/s 2532KB/s So around 2-6 times faster depending on your hardware. There are a few corner cases, for example if you truncate at all we have to do it the old way since there is no way to be sure what is in the log is ok. This probably could be done smarter, but if you write-fsync-truncate-write-fsync you deserve what you get. All this work is in RAM of course so if your inode gets evicted from cache and you read it in and fsync it we'll do it the slow way if we are still in the same transaction that we last modified the inode in. The biggest cool part of this is that it requires no changes to the recovery code, so if you fsync with this patch and crash and load an old kernel, it will run the recovery and be a-ok. I have tested this pretty thoroughly with an fsync tester and everything comes back fine, as well as xfstests. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2012-08-17 17:14:17 +00:00
* modified so clear this flag in case it was set for whatever
* reason, it's no longer relevant.
*/
clear_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
&BTRFS_I(inode)->runtime_flags);
/*
* An ordered extent might have started before and completed
* already with io errors, in which case the inode was not
* updated and we end up here. So check the inode's mapping
* flags for any errors that might have happened while doing
* writeback of file data.
*/
ret = filemap_check_errors(inode->i_mapping);
inode_unlock(inode);
goto out;
}
/*
* ok we haven't committed the transaction yet, lets do a commit
*/
if (file->private_data)
btrfs_ioctl_trans_end(file);
/*
* We use start here because we will need to wait on the IO to complete
* in btrfs_sync_log, which could require joining a transaction (for
* example checking cross references in the nocow path). If we use join
* here we could get into a situation where we're waiting on IO to
* happen that is blocked on a transaction trying to commit. With start
* we inc the extwriter counter, so we wait for all extwriters to exit
* before we start blocking join'ers. This comment is to keep somebody
* from thinking they are super smart and changing this to
* btrfs_join_transaction *cough*Josef*cough*.
*/
trans = btrfs_start_transaction(root, 0);
if (IS_ERR(trans)) {
ret = PTR_ERR(trans);
inode_unlock(inode);
goto out;
}
trans->sync = true;
Btrfs: fix lockdep warning on deadlock against an inode's log mutex Commit 44f714dae50a ("Btrfs: improve performance on fsync against new inode after rename/unlink"), which landed in 4.8-rc2, introduced a possibility for a deadlock due to double locking of an inode's log mutex by the same task, which lockdep reports with: [23045.433975] ============================================= [23045.434748] [ INFO: possible recursive locking detected ] [23045.435426] 4.7.0-rc6-btrfs-next-34+ #1 Not tainted [23045.436044] --------------------------------------------- [23045.436044] xfs_io/3688 is trying to acquire lock: [23045.436044] (&ei->log_mutex){+.+...}, at: [<ffffffffa038552d>] btrfs_log_inode+0x13a/0xc95 [btrfs] [23045.436044] but task is already holding lock: [23045.436044] (&ei->log_mutex){+.+...}, at: [<ffffffffa038552d>] btrfs_log_inode+0x13a/0xc95 [btrfs] [23045.436044] other info that might help us debug this: [23045.436044] Possible unsafe locking scenario: [23045.436044] CPU0 [23045.436044] ---- [23045.436044] lock(&ei->log_mutex); [23045.436044] lock(&ei->log_mutex); [23045.436044] *** DEADLOCK *** [23045.436044] May be due to missing lock nesting notation [23045.436044] 3 locks held by xfs_io/3688: [23045.436044] #0: (&sb->s_type->i_mutex_key#15){+.+...}, at: [<ffffffffa035f2ae>] btrfs_sync_file+0x14e/0x425 [btrfs] [23045.436044] #1: (sb_internal#2){.+.+.+}, at: [<ffffffff8118446b>] __sb_start_write+0x5f/0xb0 [23045.436044] #2: (&ei->log_mutex){+.+...}, at: [<ffffffffa038552d>] btrfs_log_inode+0x13a/0xc95 [btrfs] [23045.436044] stack backtrace: [23045.436044] CPU: 4 PID: 3688 Comm: xfs_io Not tainted 4.7.0-rc6-btrfs-next-34+ #1 [23045.436044] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.9.1-0-gb3ef39f-prebuilt.qemu-project.org 04/01/2014 [23045.436044] 0000000000000000 ffff88022f5f7860 ffffffff8127074d ffffffff82a54b70 [23045.436044] ffffffff82a54b70 ffff88022f5f7920 ffffffff81092897 ffff880228015d68 [23045.436044] 0000000000000000 ffffffff82a54b70 ffffffff829c3f00 ffff880228015d68 [23045.436044] Call Trace: [23045.436044] [<ffffffff8127074d>] dump_stack+0x67/0x90 [23045.436044] [<ffffffff81092897>] __lock_acquire+0xcbb/0xe4e [23045.436044] [<ffffffff8109155f>] ? mark_lock+0x24/0x201 [23045.436044] [<ffffffff8109179a>] ? mark_held_locks+0x5e/0x74 [23045.436044] [<ffffffff81092de0>] lock_acquire+0x12f/0x1c3 [23045.436044] [<ffffffff81092de0>] ? lock_acquire+0x12f/0x1c3 [23045.436044] [<ffffffffa038552d>] ? btrfs_log_inode+0x13a/0xc95 [btrfs] [23045.436044] [<ffffffffa038552d>] ? btrfs_log_inode+0x13a/0xc95 [btrfs] [23045.436044] [<ffffffff814a51a4>] mutex_lock_nested+0x77/0x3a7 [23045.436044] [<ffffffffa038552d>] ? btrfs_log_inode+0x13a/0xc95 [btrfs] [23045.436044] [<ffffffffa039705e>] ? btrfs_release_delayed_node+0xb/0xd [btrfs] [23045.436044] [<ffffffffa038552d>] btrfs_log_inode+0x13a/0xc95 [btrfs] [23045.436044] [<ffffffffa038552d>] ? btrfs_log_inode+0x13a/0xc95 [btrfs] [23045.436044] [<ffffffff810a0ed1>] ? vprintk_emit+0x453/0x465 [23045.436044] [<ffffffffa0385a61>] btrfs_log_inode+0x66e/0xc95 [btrfs] [23045.436044] [<ffffffffa03c084d>] log_new_dir_dentries+0x26c/0x359 [btrfs] [23045.436044] [<ffffffffa03865aa>] btrfs_log_inode_parent+0x4a6/0x628 [btrfs] [23045.436044] [<ffffffffa0387552>] btrfs_log_dentry_safe+0x5a/0x75 [btrfs] [23045.436044] [<ffffffffa035f464>] btrfs_sync_file+0x304/0x425 [btrfs] [23045.436044] [<ffffffff811acaf4>] vfs_fsync_range+0x8c/0x9e [23045.436044] [<ffffffff811acb22>] vfs_fsync+0x1c/0x1e [23045.436044] [<ffffffff811acc79>] do_fsync+0x31/0x4a [23045.436044] [<ffffffff811ace99>] SyS_fsync+0x10/0x14 [23045.436044] [<ffffffff814a88e5>] entry_SYSCALL_64_fastpath+0x18/0xa8 [23045.436044] [<ffffffff8108f039>] ? trace_hardirqs_off_caller+0x3f/0xaa An example reproducer for this is: $ mkfs.btrfs -f /dev/sdb $ mount /dev/sdb /mnt $ mkdir /mnt/dir $ touch /mnt/dir/foo $ sync $ mv /mnt/dir/foo /mnt/dir/bar $ touch /mnt/dir/foo $ xfs_io -c "fsync" /mnt/dir/bar This is because while logging the inode of file bar we end up logging its parent directory (since its inode has an unlink_trans field matching the current transaction id due to the rename operation), which in turn logs the inodes for all its new dentries, so that the new inode for the new file named foo gets logged which in turn triggered another logging attempt for the inode we are fsync'ing, since that inode had an old name that corresponds to the name of the new inode. So fix this by ensuring that when logging the inode for a new dentry that has a name matching an old name of some other inode, we don't log again the original inode that we are fsync'ing. Fixes: 44f714dae50a ("Btrfs: improve performance on fsync against new inode after rename/unlink") Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: David Sterba <dsterba@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2016-08-23 20:13:51 +00:00
btrfs_init_log_ctx(&ctx, inode);
ret = btrfs_log_dentry_safe(trans, root, dentry, start, end, &ctx);
if (ret < 0) {
/* Fallthrough and commit/free transaction. */
ret = 1;
}
/* we've logged all the items and now have a consistent
* version of the file in the log. It is possible that
* someone will come in and modify the file, but that's
* fine because the log is consistent on disk, and we
* have references to all of the file's extents
*
* It is possible that someone will come in and log the
* file again, but that will end up using the synchronization
* inside btrfs_sync_log to keep things safe.
*/
inode_unlock(inode);
Btrfs: fix data corruption after fast fsync and writeback error When we do a fast fsync, we start all ordered operations and then while they're running in parallel we visit the list of modified extent maps and construct their matching file extent items and write them to the log btree. After that, in btrfs_sync_log() we wait for all the ordered operations to finish (via btrfs_wait_logged_extents). The problem with this is that we were completely ignoring errors that can happen in the extent write path, such as -ENOSPC, a temporary -ENOMEM or -EIO errors for example. When such error happens, it means we have parts of the on disk extent that weren't written to, and so we end up logging file extent items that point to these extents that contain garbage/random data - so after a crash/reboot plus log replay, we get our inode's metadata pointing to those extents. This worked in contrast with the full (non-fast) fsync path, where we start all ordered operations, wait for them to finish and then write to the log btree. In this path, after each ordered operation completes we check if it's flagged with an error (BTRFS_ORDERED_IOERR) and return -EIO if so (via btrfs_wait_ordered_range). So if an error happens with any ordered operation, just return a -EIO error to userspace, so that it knows that not all of its previous writes were durably persisted and the application can take proper action (like redo the writes for e.g.) - and definitely not leave any file extent items in the log refer to non fully written extents. Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-09-05 14:14:39 +00:00
/*
* If any of the ordered extents had an error, just return it to user
* space, so that the application knows some writes didn't succeed and
* can take proper action (retry for e.g.). Blindly committing the
* transaction in this case, would fool userspace that everything was
* successful. And we also want to make sure our log doesn't contain
* file extent items pointing to extents that weren't fully written to -
* just like in the non fast fsync path, where we check for the ordered
* operation's error flag before writing to the log tree and return -EIO
* if any of them had this flag set (btrfs_wait_ordered_range) -
* therefore we need to check for errors in the ordered operations,
* which are indicated by ctx.io_err.
*/
if (ctx.io_err) {
btrfs_end_transaction(trans);
Btrfs: fix data corruption after fast fsync and writeback error When we do a fast fsync, we start all ordered operations and then while they're running in parallel we visit the list of modified extent maps and construct their matching file extent items and write them to the log btree. After that, in btrfs_sync_log() we wait for all the ordered operations to finish (via btrfs_wait_logged_extents). The problem with this is that we were completely ignoring errors that can happen in the extent write path, such as -ENOSPC, a temporary -ENOMEM or -EIO errors for example. When such error happens, it means we have parts of the on disk extent that weren't written to, and so we end up logging file extent items that point to these extents that contain garbage/random data - so after a crash/reboot plus log replay, we get our inode's metadata pointing to those extents. This worked in contrast with the full (non-fast) fsync path, where we start all ordered operations, wait for them to finish and then write to the log btree. In this path, after each ordered operation completes we check if it's flagged with an error (BTRFS_ORDERED_IOERR) and return -EIO if so (via btrfs_wait_ordered_range). So if an error happens with any ordered operation, just return a -EIO error to userspace, so that it knows that not all of its previous writes were durably persisted and the application can take proper action (like redo the writes for e.g.) - and definitely not leave any file extent items in the log refer to non fully written extents. Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-09-05 14:14:39 +00:00
ret = ctx.io_err;
goto out;
}
if (ret != BTRFS_NO_LOG_SYNC) {
if (!ret) {
ret = btrfs_sync_log(trans, root, &ctx);
if (!ret) {
ret = btrfs_end_transaction(trans);
goto out;
}
}
if (!full_sync) {
ret = btrfs_wait_ordered_range(inode, start, len);
if (ret) {
btrfs_end_transaction(trans);
goto out;
}
}
ret = btrfs_commit_transaction(trans);
} else {
ret = btrfs_end_transaction(trans);
}
out:
return ret > 0 ? -EIO : ret;
}
static const struct vm_operations_struct btrfs_file_vm_ops = {
.fault = filemap_fault,
.map_pages = filemap_map_pages,
.page_mkwrite = btrfs_page_mkwrite,
};
static int btrfs_file_mmap(struct file *filp, struct vm_area_struct *vma)
{
struct address_space *mapping = filp->f_mapping;
if (!mapping->a_ops->readpage)
return -ENOEXEC;
file_accessed(filp);
vma->vm_ops = &btrfs_file_vm_ops;
return 0;
}
static int hole_mergeable(struct btrfs_inode *inode, struct extent_buffer *leaf,
int slot, u64 start, u64 end)
{
struct btrfs_file_extent_item *fi;
struct btrfs_key key;
if (slot < 0 || slot >= btrfs_header_nritems(leaf))
return 0;
btrfs_item_key_to_cpu(leaf, &key, slot);
if (key.objectid != btrfs_ino(inode) ||
key.type != BTRFS_EXTENT_DATA_KEY)
return 0;
fi = btrfs_item_ptr(leaf, slot, struct btrfs_file_extent_item);
if (btrfs_file_extent_type(leaf, fi) != BTRFS_FILE_EXTENT_REG)
return 0;
if (btrfs_file_extent_disk_bytenr(leaf, fi))
return 0;
if (key.offset == end)
return 1;
if (key.offset + btrfs_file_extent_num_bytes(leaf, fi) == start)
return 1;
return 0;
}
static int fill_holes(struct btrfs_trans_handle *trans,
struct btrfs_inode *inode,
struct btrfs_path *path, u64 offset, u64 end)
{
struct btrfs_fs_info *fs_info = btrfs_sb(inode->vfs_inode.i_sb);
struct btrfs_root *root = inode->root;
struct extent_buffer *leaf;
struct btrfs_file_extent_item *fi;
struct extent_map *hole_em;
struct extent_map_tree *em_tree = &inode->extent_tree;
struct btrfs_key key;
int ret;
if (btrfs_fs_incompat(fs_info, NO_HOLES))
goto out;
key.objectid = btrfs_ino(inode);
key.type = BTRFS_EXTENT_DATA_KEY;
key.offset = offset;
ret = btrfs_search_slot(trans, root, &key, path, 0, 1);
if (ret <= 0) {
/*
* We should have dropped this offset, so if we find it then
* something has gone horribly wrong.
*/
if (ret == 0)
ret = -EINVAL;
return ret;
}
leaf = path->nodes[0];
if (hole_mergeable(inode, leaf, path->slots[0] - 1, offset, end)) {
u64 num_bytes;
path->slots[0]--;
fi = btrfs_item_ptr(leaf, path->slots[0],
struct btrfs_file_extent_item);
num_bytes = btrfs_file_extent_num_bytes(leaf, fi) +
end - offset;
btrfs_set_file_extent_num_bytes(leaf, fi, num_bytes);
btrfs_set_file_extent_ram_bytes(leaf, fi, num_bytes);
btrfs_set_file_extent_offset(leaf, fi, 0);
btrfs_mark_buffer_dirty(leaf);
goto out;
}
if (hole_mergeable(inode, leaf, path->slots[0], offset, end)) {
u64 num_bytes;
key.offset = offset;
btrfs_set_item_key_safe(fs_info, path, &key);
fi = btrfs_item_ptr(leaf, path->slots[0],
struct btrfs_file_extent_item);
num_bytes = btrfs_file_extent_num_bytes(leaf, fi) + end -
offset;
btrfs_set_file_extent_num_bytes(leaf, fi, num_bytes);
btrfs_set_file_extent_ram_bytes(leaf, fi, num_bytes);
btrfs_set_file_extent_offset(leaf, fi, 0);
btrfs_mark_buffer_dirty(leaf);
goto out;
}
btrfs_release_path(path);
ret = btrfs_insert_file_extent(trans, root, btrfs_ino(inode),
offset, 0, 0, end - offset, 0, end - offset, 0, 0, 0);
if (ret)
return ret;
out:
btrfs_release_path(path);
hole_em = alloc_extent_map();
if (!hole_em) {
btrfs_drop_extent_cache(inode, offset, end - 1, 0);
set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, &inode->runtime_flags);
} else {
hole_em->start = offset;
hole_em->len = end - offset;
hole_em->ram_bytes = hole_em->len;
hole_em->orig_start = offset;
hole_em->block_start = EXTENT_MAP_HOLE;
hole_em->block_len = 0;
hole_em->orig_block_len = 0;
hole_em->bdev = fs_info->fs_devices->latest_bdev;
hole_em->compress_type = BTRFS_COMPRESS_NONE;
hole_em->generation = trans->transid;
do {
btrfs_drop_extent_cache(inode, offset, end - 1, 0);
write_lock(&em_tree->lock);
2013-04-05 20:51:15 +00:00
ret = add_extent_mapping(em_tree, hole_em, 1);
write_unlock(&em_tree->lock);
} while (ret == -EEXIST);
free_extent_map(hole_em);
if (ret)
set_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
&inode->runtime_flags);
}
return 0;
}
/*
* Find a hole extent on given inode and change start/len to the end of hole
* extent.(hole/vacuum extent whose em->start <= start &&
* em->start + em->len > start)
* When a hole extent is found, return 1 and modify start/len.
*/
static int find_first_non_hole(struct inode *inode, u64 *start, u64 *len)
{
struct extent_map *em;
int ret = 0;
em = btrfs_get_extent(BTRFS_I(inode), NULL, 0, *start, *len, 0);
if (IS_ERR(em))
return PTR_ERR(em);
/* Hole or vacuum extent(only exists in no-hole mode) */
if (em->block_start == EXTENT_MAP_HOLE) {
ret = 1;
*len = em->start + em->len > *start + *len ?
0 : *start + *len - em->start - em->len;
*start = em->start + em->len;
}
free_extent_map(em);
return ret;
}
static int btrfs_punch_hole(struct inode *inode, loff_t offset, loff_t len)
{
struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
struct btrfs_root *root = BTRFS_I(inode)->root;
struct extent_state *cached_state = NULL;
struct btrfs_path *path;
struct btrfs_block_rsv *rsv;
struct btrfs_trans_handle *trans;
u64 lockstart;
u64 lockend;
u64 tail_start;
u64 tail_len;
u64 orig_start = offset;
u64 cur_offset;
u64 min_size = btrfs_calc_trans_metadata_size(fs_info, 1);
u64 drop_end;
int ret = 0;
int err = 0;
unsigned int rsv_count;
bool same_block;
bool no_holes = btrfs_fs_incompat(fs_info, NO_HOLES);
u64 ino_size;
bool truncated_block = false;
Btrfs: add missing inode update when punching hole When punching a file hole if we endup only zeroing parts of a page, because the start offset isn't a multiple of the sector size or the start offset and length fall within the same page, we were not updating the inode item. This prevented an fsync from doing anything, if no other file changes happened in the current transaction, because the fields in btrfs_inode used to check if the inode needs to be fsync'ed weren't updated. This issue is easy to reproduce and the following excerpt from the xfstest case I made shows how to trigger it: _scratch_mkfs >> $seqres.full 2>&1 _init_flakey _mount_flakey # Create our test file. $XFS_IO_PROG -f -c "pwrite -S 0x22 -b 16K 0 16K" \ $SCRATCH_MNT/foo | _filter_xfs_io # Fsync the file, this makes btrfs update some btrfs inode specific fields # that are used to track if the inode needs to be written/updated to the fsync # log or not. After this fsync, the new values for those fields indicate that # a subsequent fsync does not need to touch the fsync log. $XFS_IO_PROG -c "fsync" $SCRATCH_MNT/foo # Force a commit of the current transaction. After this point, any operation # that modifies the data or metadata of our file, should update those fields in # the btrfs inode with values that make the next fsync operation write to the # fsync log. sync # Punch a hole in our file. This small range affects only 1 page. # This made the btrfs hole punching implementation write only some zeroes in # one page, but it did not update the btrfs inode fields used to determine if # the next fsync needs to write to the fsync log. $XFS_IO_PROG -c "fpunch 8000 4K" $SCRATCH_MNT/foo # Another variation of the previously mentioned case. $XFS_IO_PROG -c "fpunch 15000 100" $SCRATCH_MNT/foo # Now fsync the file. This was a no-operation because the previous hole punch # operation didn't update the inode's fields mentioned before, so they remained # with the values they had after the first fsync - that is, they indicate that # it is not needed to write to fsync log. $XFS_IO_PROG -c "fsync" $SCRATCH_MNT/foo echo "File content before:" od -t x1 $SCRATCH_MNT/foo # Simulate a crash/power loss. _load_flakey_table $FLAKEY_DROP_WRITES _unmount_flakey # Enable writes and mount the fs. This makes the fsync log replay code run. _load_flakey_table $FLAKEY_ALLOW_WRITES _mount_flakey # Because the last fsync didn't do anything, here the file content matched what # it was after the first fsync, before the holes were punched, and not what it # was after the holes were punched. echo "File content after:" od -t x1 $SCRATCH_MNT/foo This issue has been around since 2012, when the punch hole implementation was added, commit 2aaa66558172 ("Btrfs: add hole punching"). A test case for xfstests follows soon. Signed-off-by: Filipe Manana <fdmanana@suse.com> Reviewed-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-02-15 22:38:54 +00:00
bool updated_inode = false;
ret = btrfs_wait_ordered_range(inode, offset, len);
if (ret)
return ret;
inode_lock(inode);
ino_size = round_up(inode->i_size, fs_info->sectorsize);
ret = find_first_non_hole(inode, &offset, &len);
if (ret < 0)
goto out_only_mutex;
if (ret && !len) {
/* Already in a large hole */
ret = 0;
goto out_only_mutex;
}
lockstart = round_up(offset, btrfs_inode_sectorsize(inode));
lockend = round_down(offset + len,
btrfs_inode_sectorsize(inode)) - 1;
same_block = (BTRFS_BYTES_TO_BLKS(fs_info, offset))
== (BTRFS_BYTES_TO_BLKS(fs_info, offset + len - 1));
/*
* We needn't truncate any block which is beyond the end of the file
* because we are sure there is no data there.
*/
/*
* Only do this if we are in the same block and we aren't doing the
* entire block.
*/
if (same_block && len < fs_info->sectorsize) {
Btrfs: add missing inode update when punching hole When punching a file hole if we endup only zeroing parts of a page, because the start offset isn't a multiple of the sector size or the start offset and length fall within the same page, we were not updating the inode item. This prevented an fsync from doing anything, if no other file changes happened in the current transaction, because the fields in btrfs_inode used to check if the inode needs to be fsync'ed weren't updated. This issue is easy to reproduce and the following excerpt from the xfstest case I made shows how to trigger it: _scratch_mkfs >> $seqres.full 2>&1 _init_flakey _mount_flakey # Create our test file. $XFS_IO_PROG -f -c "pwrite -S 0x22 -b 16K 0 16K" \ $SCRATCH_MNT/foo | _filter_xfs_io # Fsync the file, this makes btrfs update some btrfs inode specific fields # that are used to track if the inode needs to be written/updated to the fsync # log or not. After this fsync, the new values for those fields indicate that # a subsequent fsync does not need to touch the fsync log. $XFS_IO_PROG -c "fsync" $SCRATCH_MNT/foo # Force a commit of the current transaction. After this point, any operation # that modifies the data or metadata of our file, should update those fields in # the btrfs inode with values that make the next fsync operation write to the # fsync log. sync # Punch a hole in our file. This small range affects only 1 page. # This made the btrfs hole punching implementation write only some zeroes in # one page, but it did not update the btrfs inode fields used to determine if # the next fsync needs to write to the fsync log. $XFS_IO_PROG -c "fpunch 8000 4K" $SCRATCH_MNT/foo # Another variation of the previously mentioned case. $XFS_IO_PROG -c "fpunch 15000 100" $SCRATCH_MNT/foo # Now fsync the file. This was a no-operation because the previous hole punch # operation didn't update the inode's fields mentioned before, so they remained # with the values they had after the first fsync - that is, they indicate that # it is not needed to write to fsync log. $XFS_IO_PROG -c "fsync" $SCRATCH_MNT/foo echo "File content before:" od -t x1 $SCRATCH_MNT/foo # Simulate a crash/power loss. _load_flakey_table $FLAKEY_DROP_WRITES _unmount_flakey # Enable writes and mount the fs. This makes the fsync log replay code run. _load_flakey_table $FLAKEY_ALLOW_WRITES _mount_flakey # Because the last fsync didn't do anything, here the file content matched what # it was after the first fsync, before the holes were punched, and not what it # was after the holes were punched. echo "File content after:" od -t x1 $SCRATCH_MNT/foo This issue has been around since 2012, when the punch hole implementation was added, commit 2aaa66558172 ("Btrfs: add hole punching"). A test case for xfstests follows soon. Signed-off-by: Filipe Manana <fdmanana@suse.com> Reviewed-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-02-15 22:38:54 +00:00
if (offset < ino_size) {
truncated_block = true;
ret = btrfs_truncate_block(inode, offset, len, 0);
Btrfs: add missing inode update when punching hole When punching a file hole if we endup only zeroing parts of a page, because the start offset isn't a multiple of the sector size or the start offset and length fall within the same page, we were not updating the inode item. This prevented an fsync from doing anything, if no other file changes happened in the current transaction, because the fields in btrfs_inode used to check if the inode needs to be fsync'ed weren't updated. This issue is easy to reproduce and the following excerpt from the xfstest case I made shows how to trigger it: _scratch_mkfs >> $seqres.full 2>&1 _init_flakey _mount_flakey # Create our test file. $XFS_IO_PROG -f -c "pwrite -S 0x22 -b 16K 0 16K" \ $SCRATCH_MNT/foo | _filter_xfs_io # Fsync the file, this makes btrfs update some btrfs inode specific fields # that are used to track if the inode needs to be written/updated to the fsync # log or not. After this fsync, the new values for those fields indicate that # a subsequent fsync does not need to touch the fsync log. $XFS_IO_PROG -c "fsync" $SCRATCH_MNT/foo # Force a commit of the current transaction. After this point, any operation # that modifies the data or metadata of our file, should update those fields in # the btrfs inode with values that make the next fsync operation write to the # fsync log. sync # Punch a hole in our file. This small range affects only 1 page. # This made the btrfs hole punching implementation write only some zeroes in # one page, but it did not update the btrfs inode fields used to determine if # the next fsync needs to write to the fsync log. $XFS_IO_PROG -c "fpunch 8000 4K" $SCRATCH_MNT/foo # Another variation of the previously mentioned case. $XFS_IO_PROG -c "fpunch 15000 100" $SCRATCH_MNT/foo # Now fsync the file. This was a no-operation because the previous hole punch # operation didn't update the inode's fields mentioned before, so they remained # with the values they had after the first fsync - that is, they indicate that # it is not needed to write to fsync log. $XFS_IO_PROG -c "fsync" $SCRATCH_MNT/foo echo "File content before:" od -t x1 $SCRATCH_MNT/foo # Simulate a crash/power loss. _load_flakey_table $FLAKEY_DROP_WRITES _unmount_flakey # Enable writes and mount the fs. This makes the fsync log replay code run. _load_flakey_table $FLAKEY_ALLOW_WRITES _mount_flakey # Because the last fsync didn't do anything, here the file content matched what # it was after the first fsync, before the holes were punched, and not what it # was after the holes were punched. echo "File content after:" od -t x1 $SCRATCH_MNT/foo This issue has been around since 2012, when the punch hole implementation was added, commit 2aaa66558172 ("Btrfs: add hole punching"). A test case for xfstests follows soon. Signed-off-by: Filipe Manana <fdmanana@suse.com> Reviewed-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-02-15 22:38:54 +00:00
} else {
ret = 0;
}
goto out_only_mutex;
}
/* zero back part of the first block */
2014-02-15 15:55:58 +00:00
if (offset < ino_size) {
truncated_block = true;
ret = btrfs_truncate_block(inode, offset, 0, 0);
if (ret) {
inode_unlock(inode);
return ret;
}
}
/* Check the aligned pages after the first unaligned page,
* if offset != orig_start, which means the first unaligned page
* including several following pages are already in holes,
* the extra check can be skipped */
if (offset == orig_start) {
/* after truncate page, check hole again */
len = offset + len - lockstart;
offset = lockstart;
ret = find_first_non_hole(inode, &offset, &len);
if (ret < 0)
goto out_only_mutex;
if (ret && !len) {
ret = 0;
goto out_only_mutex;
}
lockstart = offset;
}
/* Check the tail unaligned part is in a hole */
tail_start = lockend + 1;
tail_len = offset + len - tail_start;
if (tail_len) {
ret = find_first_non_hole(inode, &tail_start, &tail_len);
if (unlikely(ret < 0))
goto out_only_mutex;
if (!ret) {
/* zero the front end of the last page */
if (tail_start + tail_len < ino_size) {
truncated_block = true;
ret = btrfs_truncate_block(inode,
tail_start + tail_len,
0, 1);
if (ret)
goto out_only_mutex;
btrfs: Use right extent length when inserting overlap extent map. When current btrfs finds that a new extent map is going to be insereted but failed with -EEXIST, it will try again to insert the extent map but with the length of sectorsize. This is OK if we don't enable 'no-holes' feature since all extent space is continuous, we will not go into the not found->insert routine. But if we enable 'no-holes' feature, it will make things out of control. e.g. in 4K sectorsize, we pass the following args to btrfs_get_extent(): btrfs_get_extent() args: start: 27874 len 4100 28672 27874 28672 27874+4100 32768 |-----------------------| |---------hole--------------------|---------data----------| 1) not found and insert Since no extent map containing the range, btrfs_get_extent() will go into the not_found and insert routine, which will try to insert the extent map (27874, 27847 + 4100). 2) first overlap But it overlaps with (28672, 32768) extent, so -EEXIST will be returned by add_extent_mapping(). 3) retry but still overlap After catching the -EEXIST, then btrfs_get_extent() will try insert it again but with 4K length, which still overlaps, so -EEXIST will be returned. This makes the following patch fail to punch hole. d77815461f047e561f77a07754ae923ade597d4e btrfs: Avoid trucating page or punching hole in a already existed hole. This patch will use the right length, which is the (exsisting->start - em->start) to insert, making the above patch works in 'no-holes' mode. Also, some small code style problems in above patch is fixed too. Reported-by: Filipe David Manana <fdmanana@gmail.com> Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Reviewed-by: Filipe David Manana <fdmanana@suse.com> Tested-by: Filipe David Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-08-08 05:06:20 +00:00
}
}
}
if (lockend < lockstart) {
Btrfs: add missing inode update when punching hole When punching a file hole if we endup only zeroing parts of a page, because the start offset isn't a multiple of the sector size or the start offset and length fall within the same page, we were not updating the inode item. This prevented an fsync from doing anything, if no other file changes happened in the current transaction, because the fields in btrfs_inode used to check if the inode needs to be fsync'ed weren't updated. This issue is easy to reproduce and the following excerpt from the xfstest case I made shows how to trigger it: _scratch_mkfs >> $seqres.full 2>&1 _init_flakey _mount_flakey # Create our test file. $XFS_IO_PROG -f -c "pwrite -S 0x22 -b 16K 0 16K" \ $SCRATCH_MNT/foo | _filter_xfs_io # Fsync the file, this makes btrfs update some btrfs inode specific fields # that are used to track if the inode needs to be written/updated to the fsync # log or not. After this fsync, the new values for those fields indicate that # a subsequent fsync does not need to touch the fsync log. $XFS_IO_PROG -c "fsync" $SCRATCH_MNT/foo # Force a commit of the current transaction. After this point, any operation # that modifies the data or metadata of our file, should update those fields in # the btrfs inode with values that make the next fsync operation write to the # fsync log. sync # Punch a hole in our file. This small range affects only 1 page. # This made the btrfs hole punching implementation write only some zeroes in # one page, but it did not update the btrfs inode fields used to determine if # the next fsync needs to write to the fsync log. $XFS_IO_PROG -c "fpunch 8000 4K" $SCRATCH_MNT/foo # Another variation of the previously mentioned case. $XFS_IO_PROG -c "fpunch 15000 100" $SCRATCH_MNT/foo # Now fsync the file. This was a no-operation because the previous hole punch # operation didn't update the inode's fields mentioned before, so they remained # with the values they had after the first fsync - that is, they indicate that # it is not needed to write to fsync log. $XFS_IO_PROG -c "fsync" $SCRATCH_MNT/foo echo "File content before:" od -t x1 $SCRATCH_MNT/foo # Simulate a crash/power loss. _load_flakey_table $FLAKEY_DROP_WRITES _unmount_flakey # Enable writes and mount the fs. This makes the fsync log replay code run. _load_flakey_table $FLAKEY_ALLOW_WRITES _mount_flakey # Because the last fsync didn't do anything, here the file content matched what # it was after the first fsync, before the holes were punched, and not what it # was after the holes were punched. echo "File content after:" od -t x1 $SCRATCH_MNT/foo This issue has been around since 2012, when the punch hole implementation was added, commit 2aaa66558172 ("Btrfs: add hole punching"). A test case for xfstests follows soon. Signed-off-by: Filipe Manana <fdmanana@suse.com> Reviewed-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-02-15 22:38:54 +00:00
ret = 0;
goto out_only_mutex;
}
while (1) {
struct btrfs_ordered_extent *ordered;
truncate_pagecache_range(inode, lockstart, lockend);
lock_extent_bits(&BTRFS_I(inode)->io_tree, lockstart, lockend,
&cached_state);
ordered = btrfs_lookup_first_ordered_extent(inode, lockend);
/*
* We need to make sure we have no ordered extents in this range
* and nobody raced in and read a page in this range, if we did
* we need to try again.
*/
if ((!ordered ||
(ordered->file_offset + ordered->len <= lockstart ||
ordered->file_offset > lockend)) &&
!btrfs_page_exists_in_range(inode, lockstart, lockend)) {
if (ordered)
btrfs_put_ordered_extent(ordered);
break;
}
if (ordered)
btrfs_put_ordered_extent(ordered);
unlock_extent_cached(&BTRFS_I(inode)->io_tree, lockstart,
lockend, &cached_state, GFP_NOFS);
ret = btrfs_wait_ordered_range(inode, lockstart,
lockend - lockstart + 1);
if (ret) {
inode_unlock(inode);
return ret;
}
}
path = btrfs_alloc_path();
if (!path) {
ret = -ENOMEM;
goto out;
}
rsv = btrfs_alloc_block_rsv(fs_info, BTRFS_BLOCK_RSV_TEMP);
if (!rsv) {
ret = -ENOMEM;
goto out_free;
}
rsv->size = btrfs_calc_trans_metadata_size(fs_info, 1);
rsv->failfast = 1;
/*
* 1 - update the inode
* 1 - removing the extents in the range
* 1 - adding the hole extent if no_holes isn't set
*/
rsv_count = no_holes ? 2 : 3;
trans = btrfs_start_transaction(root, rsv_count);
if (IS_ERR(trans)) {
err = PTR_ERR(trans);
goto out_free;
}
ret = btrfs_block_rsv_migrate(&fs_info->trans_block_rsv, rsv,
min_size, 0);
BUG_ON(ret);
trans->block_rsv = rsv;
cur_offset = lockstart;
len = lockend - cur_offset;
while (cur_offset < lockend) {
ret = __btrfs_drop_extents(trans, root, inode, path,
cur_offset, lockend + 1,
Btrfs: faster file extent item replace operations When writing to a file we drop existing file extent items that cover the write range and then add a new file extent item that represents that write range. Before this change we were doing a tree lookup to remove the file extent items, and then after we did another tree lookup to insert the new file extent item. Most of the time all the file extent items we need to drop are located within a single leaf - this is the leaf where our new file extent item ends up at. Therefore, in this common case just combine these 2 operations into a single one. By avoiding the second btree navigation for insertion of the new file extent item, we reduce btree node/leaf lock acquisitions/releases, btree block/leaf COW operations, CPU time on btree node/leaf key binary searches, etc. Besides for file writes, this is an operation that happens for file fsync's as well. However log btrees are much less likely to big as big as regular fs btrees, therefore the impact of this change is smaller. The following benchmark was performed against an SSD drive and a HDD drive, both for random and sequential writes: sysbench --test=fileio --file-num=4096 --file-total-size=8G \ --file-test-mode=[rndwr|seqwr] --num-threads=512 \ --file-block-size=8192 \ --max-requests=1000000 \ --file-fsync-freq=0 --file-io-mode=sync [prepare|run] All results below are averages of 10 runs of the respective test. ** SSD sequential writes Before this change: 225.88 Mb/sec After this change: 277.26 Mb/sec ** SSD random writes Before this change: 49.91 Mb/sec After this change: 56.39 Mb/sec ** HDD sequential writes Before this change: 68.53 Mb/sec After this change: 69.87 Mb/sec ** HDD random writes Before this change: 13.04 Mb/sec After this change: 14.39 Mb/sec Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-01-07 11:42:27 +00:00
&drop_end, 1, 0, 0, NULL);
if (ret != -ENOSPC)
break;
trans->block_rsv = &fs_info->trans_block_rsv;
if (cur_offset < drop_end && cur_offset < ino_size) {
ret = fill_holes(trans, BTRFS_I(inode), path,
cur_offset, drop_end);
2014-02-15 15:55:58 +00:00
if (ret) {
/*
* If we failed then we didn't insert our hole
* entries for the area we dropped, so now the
* fs is corrupted, so we must abort the
* transaction.
*/
btrfs_abort_transaction(trans, ret);
2014-02-15 15:55:58 +00:00
err = ret;
break;
}
}
cur_offset = drop_end;
ret = btrfs_update_inode(trans, root, inode);
if (ret) {
err = ret;
break;
}
btrfs_end_transaction(trans);
btrfs_btree_balance_dirty(fs_info);
trans = btrfs_start_transaction(root, rsv_count);
if (IS_ERR(trans)) {
ret = PTR_ERR(trans);
trans = NULL;
break;
}
ret = btrfs_block_rsv_migrate(&fs_info->trans_block_rsv,
rsv, min_size, 0);
BUG_ON(ret); /* shouldn't happen */
trans->block_rsv = rsv;
ret = find_first_non_hole(inode, &cur_offset, &len);
if (unlikely(ret < 0))
break;
if (ret && !len) {
ret = 0;
break;
}
}
if (ret) {
err = ret;
goto out_trans;
}
trans->block_rsv = &fs_info->trans_block_rsv;
Btrfs: fix hole punching when using the no-holes feature When we are using the no-holes feature, if we punch a hole into a file range that already contains a hole which overlaps the range we are passing to fallocate(), we end up removing the extent map that represents the existing hole without adding a new one. This happens because with the no-holes feature we do not have explicit extent items to represent holes and therefore the call to __btrfs_drop_extents(), made from btrfs_punch_hole(), returns an end offset to the variable drop_end that is smaller than the end of the range passed to fallocate(), while it drops all existing extent maps in that range. Normally having a missing extent map is not a problem, for example for a readpages() operation we just end up building the extent map by looking at the fs/subvol tree for a matching extent item (or a lack of one for implicit holes). However for an fsync that uses the fast path, which needs to look at the list of modified extent maps, this means the fsync will not record information about the complete hole we had before the fallocate() call into the log tree, resulting in a file with content/layout that does not match what we had neither before nor after the hole punch operation. The following test case for fstests reproduces the issue. It fails without this change because we get a file with a different digest after the fsync log replay and also with a different extent/hole layout. seq=`basename $0` seqres=$RESULT_DIR/$seq echo "QA output created by $seq" tmp=/tmp/$$ status=1 # failure is the default! trap "_cleanup; exit \$status" 0 1 2 3 15 _cleanup() { _cleanup_flakey rm -f $tmp.* } # get standard environment, filters and checks . ./common/rc . ./common/filter . ./common/punch . ./common/dmflakey # real QA test starts here _need_to_be_root _supported_fs generic _supported_os Linux _require_scratch _require_xfs_io_command "fpunch" _require_xfs_io_command "fiemap" _require_dm_target flakey _require_metadata_journaling $SCRATCH_DEV # This test was motivated by an issue found in btrfs when the btrfs # no-holes feature is enabled (introduced in kernel 3.14). So enable # the feature if the fs being tested is btrfs. if [ $FSTYP == "btrfs" ]; then _require_btrfs_fs_feature "no_holes" _require_btrfs_mkfs_feature "no-holes" MKFS_OPTIONS="$MKFS_OPTIONS -O no-holes" fi rm -f $seqres.full _scratch_mkfs >>$seqres.full 2>&1 _init_flakey _mount_flakey # Create out test file with some data and then fsync it. # We do the fsync only to make sure the last fsync we do in this test # triggers the fast code path of btrfs' fsync implementation, a # condition necessary to trigger the bug btrfs had. $XFS_IO_PROG -f -c "pwrite -S 0xaa 0K 128K" \ -c "fsync" \ $SCRATCH_MNT/foobar | _filter_xfs_io # Now punch a hole against the range [96K, 128K[. $XFS_IO_PROG -c "fpunch 96K 32K" $SCRATCH_MNT/foobar # Punch another hole against a range that overlaps the previous range # and ends beyond eof. $XFS_IO_PROG -c "fpunch 64K 128K" $SCRATCH_MNT/foobar # Punch another hole against a range that overlaps the first range # ([96K, 128K[) and ends at eof. $XFS_IO_PROG -c "fpunch 32K 96K" $SCRATCH_MNT/foobar # Fsync our file. We want to verify that, after a power failure and # mounting the filesystem again, the file content reflects all the hole # punch operations. $XFS_IO_PROG -c "fsync" $SCRATCH_MNT/foobar echo "File digest before power failure:" md5sum $SCRATCH_MNT/foobar | _filter_scratch echo "Fiemap before power failure:" $XFS_IO_PROG -c "fiemap -v" $SCRATCH_MNT/foobar | _filter_fiemap # Silently drop all writes and umount to simulate a crash/power failure. _load_flakey_table $FLAKEY_DROP_WRITES _unmount_flakey # Allow writes again, mount to trigger log replay and validate file # contents. _load_flakey_table $FLAKEY_ALLOW_WRITES _mount_flakey echo "File digest after log replay:" # Must match the same digest we got before the power failure. md5sum $SCRATCH_MNT/foobar | _filter_scratch echo "Fiemap after log replay:" # Must match the same extent listing we got before the power failure. $XFS_IO_PROG -c "fiemap -v" $SCRATCH_MNT/foobar | _filter_fiemap _unmount_flakey status=0 exit Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-11-02 12:32:44 +00:00
/*
* If we are using the NO_HOLES feature we might have had already an
* hole that overlaps a part of the region [lockstart, lockend] and
* ends at (or beyond) lockend. Since we have no file extent items to
* represent holes, drop_end can be less than lockend and so we must
* make sure we have an extent map representing the existing hole (the
* call to __btrfs_drop_extents() might have dropped the existing extent
* map representing the existing hole), otherwise the fast fsync path
* will not record the existence of the hole region
* [existing_hole_start, lockend].
*/
if (drop_end <= lockend)
drop_end = lockend + 1;
Btrfs: fix leaf corruption caused by ENOSPC while hole punching While running a stress test with multiple threads writing to the same btrfs file system, I ended up with a situation where a leaf was corrupted in that it had 2 file extent item keys that had the same exact key. I was able to detect this quickly thanks to the following patch which triggers an assertion as soon as a leaf is marked dirty if there are duplicated keys or out of order keys: Btrfs: check if items are ordered when a leaf is marked dirty (https://patchwork.kernel.org/patch/3955431/) Basically while running the test, I got the following in dmesg: [28877.415877] WARNING: CPU: 2 PID: 10706 at fs/btrfs/file.c:553 btrfs_drop_extent_cache+0x435/0x440 [btrfs]() (...) [28877.415917] Call Trace: [28877.415922] [<ffffffff816f1189>] dump_stack+0x4e/0x68 [28877.415926] [<ffffffff8104a32c>] warn_slowpath_common+0x8c/0xc0 [28877.415929] [<ffffffff8104a37a>] warn_slowpath_null+0x1a/0x20 [28877.415944] [<ffffffffa03775a5>] btrfs_drop_extent_cache+0x435/0x440 [btrfs] [28877.415949] [<ffffffff8118e7be>] ? kmem_cache_alloc+0xfe/0x1c0 [28877.415962] [<ffffffffa03777d9>] fill_holes+0x229/0x3e0 [btrfs] [28877.415972] [<ffffffffa0345865>] ? block_rsv_add_bytes+0x55/0x80 [btrfs] [28877.415984] [<ffffffffa03792cb>] btrfs_fallocate+0xb6b/0xc20 [btrfs] (...) [29854.132560] BTRFS critical (device sdc): corrupt leaf, bad key order: block=955232256,root=1, slot=24 [29854.132565] BTRFS info (device sdc): leaf 955232256 total ptrs 40 free space 778 (...) [29854.132637] item 23 key (3486 108 667648) itemoff 2694 itemsize 53 [29854.132638] extent data disk bytenr 14574411776 nr 286720 [29854.132639] extent data offset 0 nr 286720 ram 286720 [29854.132640] item 24 key (3486 108 954368) itemoff 2641 itemsize 53 [29854.132641] extent data disk bytenr 0 nr 0 [29854.132643] extent data offset 0 nr 0 ram 0 [29854.132644] item 25 key (3486 108 954368) itemoff 2588 itemsize 53 [29854.132645] extent data disk bytenr 8699670528 nr 77824 [29854.132646] extent data offset 0 nr 77824 ram 77824 [29854.132647] item 26 key (3486 108 1146880) itemoff 2535 itemsize 53 [29854.132648] extent data disk bytenr 8699670528 nr 77824 [29854.132649] extent data offset 0 nr 77824 ram 77824 (...) [29854.132707] kernel BUG at fs/btrfs/ctree.h:3901! (...) [29854.132771] Call Trace: [29854.132779] [<ffffffffa0342b5c>] setup_items_for_insert+0x2dc/0x400 [btrfs] [29854.132791] [<ffffffffa0378537>] __btrfs_drop_extents+0xba7/0xdd0 [btrfs] [29854.132794] [<ffffffff8109c0d6>] ? trace_hardirqs_on_caller+0x16/0x1d0 [29854.132797] [<ffffffff8109c29d>] ? trace_hardirqs_on+0xd/0x10 [29854.132800] [<ffffffff8118e7be>] ? kmem_cache_alloc+0xfe/0x1c0 [29854.132810] [<ffffffffa036783b>] insert_reserved_file_extent.constprop.66+0xab/0x310 [btrfs] [29854.132820] [<ffffffffa036a6c6>] __btrfs_prealloc_file_range+0x116/0x340 [btrfs] [29854.132830] [<ffffffffa0374d53>] btrfs_prealloc_file_range+0x23/0x30 [btrfs] (...) So this is caused by getting an -ENOSPC error while punching a file hole, more specifically, we get -ENOSPC error from __btrfs_drop_extents in the while loop of file.c:btrfs_punch_hole() when it's unable to modify the btree to delete one or more file extent items due to lack of enough free space. When this happens, in btrfs_punch_hole(), we attempt to reclaim free space by switching our transaction block reservation object to root->fs_info->trans_block_rsv, end our transaction and start a new transaction basically - and, we keep increasing our current offset (cur_offset) as long as it's smaller than the end of the target range (lockend) - this makes use leave the loop with cur_offset == drop_end which in turn makes us call fill_holes() for inserting a file extent item that represents a 0 bytes range hole (and this insertion succeeds, as in the meanwhile more space became available). This 0 bytes file hole extent item is a problem because any subsequent caller of __btrfs_drop_extents (regular file writes, or fallocate calls for e.g.), with a start file offset that is equal to the offset of the hole, will not remove this extent item due to the following conditional in the while loop of __btrfs_drop_extents: if (extent_end <= search_start) { path->slots[0]++; goto next_slot; } This later makes the call to setup_items_for_insert() (at the very end of __btrfs_drop_extents), insert a new file extent item with the same offset as the 0 bytes file hole extent item that follows it. Needless is to say that this causes chaos, either when reading the leaf from disk (btree_readpage_end_io_hook), where we perform leaf sanity checks or in subsequent operations that manipulate file extent items, as in the fallocate call as shown by the dmesg trace above. Without my other patch to perform the leaf sanity checks once a leaf is marked as dirty (if the integrity checker is enabled), it would have been much harder to debug this issue. This change might fix a few similar issues reported by users in the mailing list regarding assertion failures in btrfs_set_item_key_safe calls performed by __btrfs_drop_extents, such as the following report: http://comments.gmane.org/gmane.comp.file-systems.btrfs/32938 Asking fill_holes() to create a 0 bytes wide file hole item also produced the first warning in the trace above, as we passed a range to btrfs_drop_extent_cache that has an end smaller (by -1) than its start. On 3.14 kernels this issue manifests itself through leaf corruption, as we get duplicated file extent item keys in a leaf when calling setup_items_for_insert(), but on older kernels, setup_items_for_insert() isn't called by __btrfs_drop_extents(), instead we have callers of __btrfs_drop_extents(), namely the functions inode.c:insert_inline_extent() and inode.c:insert_reserved_file_extent(), calling btrfs_insert_empty_item() to insert the new file extent item, which would fail with error -EEXIST, instead of inserting a duplicated key - which is still a serious issue as it would make all similar file extent item replace operations keep failing if they target the same file range. Cc: stable@vger.kernel.org Signed-off-by: Filipe David Borba Manana <fdmanana@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-04-29 12:18:40 +00:00
/*
* Don't insert file hole extent item if it's for a range beyond eof
* (because it's useless) or if it represents a 0 bytes range (when
* cur_offset == drop_end).
*/
if (cur_offset < ino_size && cur_offset < drop_end) {
ret = fill_holes(trans, BTRFS_I(inode), path,
cur_offset, drop_end);
2014-02-15 15:55:58 +00:00
if (ret) {
/* Same comment as above. */
btrfs_abort_transaction(trans, ret);
2014-02-15 15:55:58 +00:00
err = ret;
goto out_trans;
}
}
out_trans:
if (!trans)
goto out_free;
inode_inc_iversion(inode);
inode->i_mtime = inode->i_ctime = current_time(inode);
trans->block_rsv = &fs_info->trans_block_rsv;
ret = btrfs_update_inode(trans, root, inode);
Btrfs: add missing inode update when punching hole When punching a file hole if we endup only zeroing parts of a page, because the start offset isn't a multiple of the sector size or the start offset and length fall within the same page, we were not updating the inode item. This prevented an fsync from doing anything, if no other file changes happened in the current transaction, because the fields in btrfs_inode used to check if the inode needs to be fsync'ed weren't updated. This issue is easy to reproduce and the following excerpt from the xfstest case I made shows how to trigger it: _scratch_mkfs >> $seqres.full 2>&1 _init_flakey _mount_flakey # Create our test file. $XFS_IO_PROG -f -c "pwrite -S 0x22 -b 16K 0 16K" \ $SCRATCH_MNT/foo | _filter_xfs_io # Fsync the file, this makes btrfs update some btrfs inode specific fields # that are used to track if the inode needs to be written/updated to the fsync # log or not. After this fsync, the new values for those fields indicate that # a subsequent fsync does not need to touch the fsync log. $XFS_IO_PROG -c "fsync" $SCRATCH_MNT/foo # Force a commit of the current transaction. After this point, any operation # that modifies the data or metadata of our file, should update those fields in # the btrfs inode with values that make the next fsync operation write to the # fsync log. sync # Punch a hole in our file. This small range affects only 1 page. # This made the btrfs hole punching implementation write only some zeroes in # one page, but it did not update the btrfs inode fields used to determine if # the next fsync needs to write to the fsync log. $XFS_IO_PROG -c "fpunch 8000 4K" $SCRATCH_MNT/foo # Another variation of the previously mentioned case. $XFS_IO_PROG -c "fpunch 15000 100" $SCRATCH_MNT/foo # Now fsync the file. This was a no-operation because the previous hole punch # operation didn't update the inode's fields mentioned before, so they remained # with the values they had after the first fsync - that is, they indicate that # it is not needed to write to fsync log. $XFS_IO_PROG -c "fsync" $SCRATCH_MNT/foo echo "File content before:" od -t x1 $SCRATCH_MNT/foo # Simulate a crash/power loss. _load_flakey_table $FLAKEY_DROP_WRITES _unmount_flakey # Enable writes and mount the fs. This makes the fsync log replay code run. _load_flakey_table $FLAKEY_ALLOW_WRITES _mount_flakey # Because the last fsync didn't do anything, here the file content matched what # it was after the first fsync, before the holes were punched, and not what it # was after the holes were punched. echo "File content after:" od -t x1 $SCRATCH_MNT/foo This issue has been around since 2012, when the punch hole implementation was added, commit 2aaa66558172 ("Btrfs: add hole punching"). A test case for xfstests follows soon. Signed-off-by: Filipe Manana <fdmanana@suse.com> Reviewed-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-02-15 22:38:54 +00:00
updated_inode = true;
btrfs_end_transaction(trans);
btrfs_btree_balance_dirty(fs_info);
out_free:
btrfs_free_path(path);
btrfs_free_block_rsv(fs_info, rsv);
out:
unlock_extent_cached(&BTRFS_I(inode)->io_tree, lockstart, lockend,
&cached_state, GFP_NOFS);
out_only_mutex:
if (!updated_inode && truncated_block && !ret && !err) {
Btrfs: add missing inode update when punching hole When punching a file hole if we endup only zeroing parts of a page, because the start offset isn't a multiple of the sector size or the start offset and length fall within the same page, we were not updating the inode item. This prevented an fsync from doing anything, if no other file changes happened in the current transaction, because the fields in btrfs_inode used to check if the inode needs to be fsync'ed weren't updated. This issue is easy to reproduce and the following excerpt from the xfstest case I made shows how to trigger it: _scratch_mkfs >> $seqres.full 2>&1 _init_flakey _mount_flakey # Create our test file. $XFS_IO_PROG -f -c "pwrite -S 0x22 -b 16K 0 16K" \ $SCRATCH_MNT/foo | _filter_xfs_io # Fsync the file, this makes btrfs update some btrfs inode specific fields # that are used to track if the inode needs to be written/updated to the fsync # log or not. After this fsync, the new values for those fields indicate that # a subsequent fsync does not need to touch the fsync log. $XFS_IO_PROG -c "fsync" $SCRATCH_MNT/foo # Force a commit of the current transaction. After this point, any operation # that modifies the data or metadata of our file, should update those fields in # the btrfs inode with values that make the next fsync operation write to the # fsync log. sync # Punch a hole in our file. This small range affects only 1 page. # This made the btrfs hole punching implementation write only some zeroes in # one page, but it did not update the btrfs inode fields used to determine if # the next fsync needs to write to the fsync log. $XFS_IO_PROG -c "fpunch 8000 4K" $SCRATCH_MNT/foo # Another variation of the previously mentioned case. $XFS_IO_PROG -c "fpunch 15000 100" $SCRATCH_MNT/foo # Now fsync the file. This was a no-operation because the previous hole punch # operation didn't update the inode's fields mentioned before, so they remained # with the values they had after the first fsync - that is, they indicate that # it is not needed to write to fsync log. $XFS_IO_PROG -c "fsync" $SCRATCH_MNT/foo echo "File content before:" od -t x1 $SCRATCH_MNT/foo # Simulate a crash/power loss. _load_flakey_table $FLAKEY_DROP_WRITES _unmount_flakey # Enable writes and mount the fs. This makes the fsync log replay code run. _load_flakey_table $FLAKEY_ALLOW_WRITES _mount_flakey # Because the last fsync didn't do anything, here the file content matched what # it was after the first fsync, before the holes were punched, and not what it # was after the holes were punched. echo "File content after:" od -t x1 $SCRATCH_MNT/foo This issue has been around since 2012, when the punch hole implementation was added, commit 2aaa66558172 ("Btrfs: add hole punching"). A test case for xfstests follows soon. Signed-off-by: Filipe Manana <fdmanana@suse.com> Reviewed-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-02-15 22:38:54 +00:00
/*
* If we only end up zeroing part of a page, we still need to
* update the inode item, so that all the time fields are
* updated as well as the necessary btrfs inode in memory fields
* for detecting, at fsync time, if the inode isn't yet in the
* log tree or it's there but not up to date.
*/
trans = btrfs_start_transaction(root, 1);
if (IS_ERR(trans)) {
err = PTR_ERR(trans);
} else {
err = btrfs_update_inode(trans, root, inode);
ret = btrfs_end_transaction(trans);
Btrfs: add missing inode update when punching hole When punching a file hole if we endup only zeroing parts of a page, because the start offset isn't a multiple of the sector size or the start offset and length fall within the same page, we were not updating the inode item. This prevented an fsync from doing anything, if no other file changes happened in the current transaction, because the fields in btrfs_inode used to check if the inode needs to be fsync'ed weren't updated. This issue is easy to reproduce and the following excerpt from the xfstest case I made shows how to trigger it: _scratch_mkfs >> $seqres.full 2>&1 _init_flakey _mount_flakey # Create our test file. $XFS_IO_PROG -f -c "pwrite -S 0x22 -b 16K 0 16K" \ $SCRATCH_MNT/foo | _filter_xfs_io # Fsync the file, this makes btrfs update some btrfs inode specific fields # that are used to track if the inode needs to be written/updated to the fsync # log or not. After this fsync, the new values for those fields indicate that # a subsequent fsync does not need to touch the fsync log. $XFS_IO_PROG -c "fsync" $SCRATCH_MNT/foo # Force a commit of the current transaction. After this point, any operation # that modifies the data or metadata of our file, should update those fields in # the btrfs inode with values that make the next fsync operation write to the # fsync log. sync # Punch a hole in our file. This small range affects only 1 page. # This made the btrfs hole punching implementation write only some zeroes in # one page, but it did not update the btrfs inode fields used to determine if # the next fsync needs to write to the fsync log. $XFS_IO_PROG -c "fpunch 8000 4K" $SCRATCH_MNT/foo # Another variation of the previously mentioned case. $XFS_IO_PROG -c "fpunch 15000 100" $SCRATCH_MNT/foo # Now fsync the file. This was a no-operation because the previous hole punch # operation didn't update the inode's fields mentioned before, so they remained # with the values they had after the first fsync - that is, they indicate that # it is not needed to write to fsync log. $XFS_IO_PROG -c "fsync" $SCRATCH_MNT/foo echo "File content before:" od -t x1 $SCRATCH_MNT/foo # Simulate a crash/power loss. _load_flakey_table $FLAKEY_DROP_WRITES _unmount_flakey # Enable writes and mount the fs. This makes the fsync log replay code run. _load_flakey_table $FLAKEY_ALLOW_WRITES _mount_flakey # Because the last fsync didn't do anything, here the file content matched what # it was after the first fsync, before the holes were punched, and not what it # was after the holes were punched. echo "File content after:" od -t x1 $SCRATCH_MNT/foo This issue has been around since 2012, when the punch hole implementation was added, commit 2aaa66558172 ("Btrfs: add hole punching"). A test case for xfstests follows soon. Signed-off-by: Filipe Manana <fdmanana@suse.com> Reviewed-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-02-15 22:38:54 +00:00
}
}
inode_unlock(inode);
if (ret && !err)
err = ret;
return err;
}
/* Helper structure to record which range is already reserved */
struct falloc_range {
struct list_head list;
u64 start;
u64 len;
};
/*
* Helper function to add falloc range
*
* Caller should have locked the larger range of extent containing
* [start, len)
*/
static int add_falloc_range(struct list_head *head, u64 start, u64 len)
{
struct falloc_range *prev = NULL;
struct falloc_range *range = NULL;
if (list_empty(head))
goto insert;
/*
* As fallocate iterate by bytenr order, we only need to check
* the last range.
*/
prev = list_entry(head->prev, struct falloc_range, list);
if (prev->start + prev->len == start) {
prev->len += len;
return 0;
}
insert:
range = kmalloc(sizeof(*range), GFP_KERNEL);
if (!range)
return -ENOMEM;
range->start = start;
range->len = len;
list_add_tail(&range->list, head);
return 0;
}
static long btrfs_fallocate(struct file *file, int mode,
loff_t offset, loff_t len)
{
struct inode *inode = file_inode(file);
struct extent_state *cached_state = NULL;
struct falloc_range *range;
struct falloc_range *tmp;
struct list_head reserve_list;
u64 cur_offset;
u64 last_byte;
u64 alloc_start;
u64 alloc_end;
u64 alloc_hint = 0;
u64 locked_end;
u64 actual_end = 0;
struct extent_map *em;
int blocksize = btrfs_inode_sectorsize(inode);
int ret;
alloc_start = round_down(offset, blocksize);
alloc_end = round_up(offset + len, blocksize);
btrfs: update btrfs_space_info's bytes_may_use timely This patch can fix some false ENOSPC errors, below test script can reproduce one false ENOSPC error: #!/bin/bash dd if=/dev/zero of=fs.img bs=$((1024*1024)) count=128 dev=$(losetup --show -f fs.img) mkfs.btrfs -f -M $dev mkdir /tmp/mntpoint mount $dev /tmp/mntpoint cd /tmp/mntpoint xfs_io -f -c "falloc 0 $((64*1024*1024))" testfile Above script will fail for ENOSPC reason, but indeed fs still has free space to satisfy this request. Please see call graph: btrfs_fallocate() |-> btrfs_alloc_data_chunk_ondemand() | bytes_may_use += 64M |-> btrfs_prealloc_file_range() |-> btrfs_reserve_extent() |-> btrfs_add_reserved_bytes() | alloc_type is RESERVE_ALLOC_NO_ACCOUNT, so it does not | change bytes_may_use, and bytes_reserved += 64M. Now | bytes_may_use + bytes_reserved == 128M, which is greater | than btrfs_space_info's total_bytes, false enospc occurs. | Note, the bytes_may_use decrease operation will be done in | end of btrfs_fallocate(), which is too late. Here is another simple case for buffered write: CPU 1 | CPU 2 | |-> cow_file_range() |-> __btrfs_buffered_write() |-> btrfs_reserve_extent() | | | | | | | | | ..... | |-> btrfs_check_data_free_space() | | | | |-> extent_clear_unlock_delalloc() | In CPU 1, btrfs_reserve_extent()->find_free_extent()-> btrfs_add_reserved_bytes() do not decrease bytes_may_use, the decrease operation will be delayed to be done in extent_clear_unlock_delalloc(). Assume in this case, btrfs_reserve_extent() reserved 128MB data, CPU2's btrfs_check_data_free_space() tries to reserve 100MB data space. If 100MB > data_sinfo->total_bytes - data_sinfo->bytes_used - data_sinfo->bytes_reserved - data_sinfo->bytes_pinned - data_sinfo->bytes_readonly - data_sinfo->bytes_may_use btrfs_check_data_free_space() will try to allcate new data chunk or call btrfs_start_delalloc_roots(), or commit current transaction in order to reserve some free space, obviously a lot of work. But indeed it's not necessary as long as decreasing bytes_may_use timely, we still have free space, decreasing 128M from bytes_may_use. To fix this issue, this patch chooses to update bytes_may_use for both data and metadata in btrfs_add_reserved_bytes(). For compress path, real extent length may not be equal to file content length, so introduce a ram_bytes argument for btrfs_reserve_extent(), find_free_extent() and btrfs_add_reserved_bytes(), it's becasue bytes_may_use is increased by file content length. Then compress path can update bytes_may_use correctly. Also now we can discard RESERVE_ALLOC_NO_ACCOUNT, RESERVE_ALLOC and RESERVE_FREE. As we know, usually EXTENT_DO_ACCOUNTING is used for error path. In run_delalloc_nocow(), for inode marked as NODATACOW or extent marked as PREALLOC, we also need to update bytes_may_use, but can not pass EXTENT_DO_ACCOUNTING, because it also clears metadata reservation, so here we introduce EXTENT_CLEAR_DATA_RESV flag to indicate btrfs_clear_bit_hook() to update btrfs_space_info's bytes_may_use. Meanwhile __btrfs_prealloc_file_range() will call btrfs_free_reserved_data_space() internally for both sucessful and failed path, btrfs_prealloc_file_range()'s callers does not need to call btrfs_free_reserved_data_space() any more. Signed-off-by: Wang Xiaoguang <wangxg.fnst@cn.fujitsu.com> Reviewed-by: Josef Bacik <jbacik@fb.com> Signed-off-by: David Sterba <dsterba@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2016-07-25 07:51:40 +00:00
cur_offset = alloc_start;
/* Make sure we aren't being give some crap mode */
if (mode & ~(FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE))
return -EOPNOTSUPP;
if (mode & FALLOC_FL_PUNCH_HOLE)
return btrfs_punch_hole(inode, offset, len);
/*
* Only trigger disk allocation, don't trigger qgroup reserve
*
* For qgroup space, it will be checked later.
*/
ret = btrfs_alloc_data_chunk_ondemand(BTRFS_I(inode),
alloc_end - alloc_start);
if (ret < 0)
return ret;
inode_lock(inode);
if (!(mode & FALLOC_FL_KEEP_SIZE) && offset + len > inode->i_size) {
ret = inode_newsize_ok(inode, offset + len);
if (ret)
goto out;
}
/*
* TODO: Move these two operations after we have checked
* accurate reserved space, or fallocate can still fail but
* with page truncated or size expanded.
*
* But that's a minor problem and won't do much harm BTW.
*/
if (alloc_start > inode->i_size) {
ret = btrfs_cont_expand(inode, i_size_read(inode),
alloc_start);
if (ret)
goto out;
} else if (offset + len > inode->i_size) {
/*
* If we are fallocating from the end of the file onward we
* need to zero out the end of the block if i_size lands in the
* middle of a block.
*/
ret = btrfs_truncate_block(inode, inode->i_size, 0, 0);
if (ret)
goto out;
}
/*
* wait for ordered IO before we have any locks. We'll loop again
* below with the locks held.
*/
ret = btrfs_wait_ordered_range(inode, alloc_start,
alloc_end - alloc_start);
if (ret)
goto out;
locked_end = alloc_end - 1;
while (1) {
struct btrfs_ordered_extent *ordered;
/* the extent lock is ordered inside the running
* transaction
*/
lock_extent_bits(&BTRFS_I(inode)->io_tree, alloc_start,
locked_end, &cached_state);
ordered = btrfs_lookup_first_ordered_extent(inode,
alloc_end - 1);
if (ordered &&
ordered->file_offset + ordered->len > alloc_start &&
ordered->file_offset < alloc_end) {
btrfs_put_ordered_extent(ordered);
unlock_extent_cached(&BTRFS_I(inode)->io_tree,
alloc_start, locked_end,
&cached_state, GFP_KERNEL);
/*
* we can't wait on the range with the transaction
* running or with the extent lock held
*/
ret = btrfs_wait_ordered_range(inode, alloc_start,
alloc_end - alloc_start);
if (ret)
goto out;
} else {
if (ordered)
btrfs_put_ordered_extent(ordered);
break;
}
}
/* First, check if we exceed the qgroup limit */
INIT_LIST_HEAD(&reserve_list);
while (1) {
em = btrfs_get_extent(BTRFS_I(inode), NULL, 0, cur_offset,
alloc_end - cur_offset, 0);
if (IS_ERR(em)) {
ret = PTR_ERR(em);
break;
}
last_byte = min(extent_map_end(em), alloc_end);
actual_end = min_t(u64, extent_map_end(em), offset + len);
last_byte = ALIGN(last_byte, blocksize);
if (em->block_start == EXTENT_MAP_HOLE ||
(cur_offset >= inode->i_size &&
!test_bit(EXTENT_FLAG_PREALLOC, &em->flags))) {
ret = add_falloc_range(&reserve_list, cur_offset,
last_byte - cur_offset);
if (ret < 0) {
free_extent_map(em);
break;
}
ret = btrfs_qgroup_reserve_data(inode, cur_offset,
last_byte - cur_offset);
if (ret < 0) {
free_extent_map(em);
break;
}
btrfs: update btrfs_space_info's bytes_may_use timely This patch can fix some false ENOSPC errors, below test script can reproduce one false ENOSPC error: #!/bin/bash dd if=/dev/zero of=fs.img bs=$((1024*1024)) count=128 dev=$(losetup --show -f fs.img) mkfs.btrfs -f -M $dev mkdir /tmp/mntpoint mount $dev /tmp/mntpoint cd /tmp/mntpoint xfs_io -f -c "falloc 0 $((64*1024*1024))" testfile Above script will fail for ENOSPC reason, but indeed fs still has free space to satisfy this request. Please see call graph: btrfs_fallocate() |-> btrfs_alloc_data_chunk_ondemand() | bytes_may_use += 64M |-> btrfs_prealloc_file_range() |-> btrfs_reserve_extent() |-> btrfs_add_reserved_bytes() | alloc_type is RESERVE_ALLOC_NO_ACCOUNT, so it does not | change bytes_may_use, and bytes_reserved += 64M. Now | bytes_may_use + bytes_reserved == 128M, which is greater | than btrfs_space_info's total_bytes, false enospc occurs. | Note, the bytes_may_use decrease operation will be done in | end of btrfs_fallocate(), which is too late. Here is another simple case for buffered write: CPU 1 | CPU 2 | |-> cow_file_range() |-> __btrfs_buffered_write() |-> btrfs_reserve_extent() | | | | | | | | | ..... | |-> btrfs_check_data_free_space() | | | | |-> extent_clear_unlock_delalloc() | In CPU 1, btrfs_reserve_extent()->find_free_extent()-> btrfs_add_reserved_bytes() do not decrease bytes_may_use, the decrease operation will be delayed to be done in extent_clear_unlock_delalloc(). Assume in this case, btrfs_reserve_extent() reserved 128MB data, CPU2's btrfs_check_data_free_space() tries to reserve 100MB data space. If 100MB > data_sinfo->total_bytes - data_sinfo->bytes_used - data_sinfo->bytes_reserved - data_sinfo->bytes_pinned - data_sinfo->bytes_readonly - data_sinfo->bytes_may_use btrfs_check_data_free_space() will try to allcate new data chunk or call btrfs_start_delalloc_roots(), or commit current transaction in order to reserve some free space, obviously a lot of work. But indeed it's not necessary as long as decreasing bytes_may_use timely, we still have free space, decreasing 128M from bytes_may_use. To fix this issue, this patch chooses to update bytes_may_use for both data and metadata in btrfs_add_reserved_bytes(). For compress path, real extent length may not be equal to file content length, so introduce a ram_bytes argument for btrfs_reserve_extent(), find_free_extent() and btrfs_add_reserved_bytes(), it's becasue bytes_may_use is increased by file content length. Then compress path can update bytes_may_use correctly. Also now we can discard RESERVE_ALLOC_NO_ACCOUNT, RESERVE_ALLOC and RESERVE_FREE. As we know, usually EXTENT_DO_ACCOUNTING is used for error path. In run_delalloc_nocow(), for inode marked as NODATACOW or extent marked as PREALLOC, we also need to update bytes_may_use, but can not pass EXTENT_DO_ACCOUNTING, because it also clears metadata reservation, so here we introduce EXTENT_CLEAR_DATA_RESV flag to indicate btrfs_clear_bit_hook() to update btrfs_space_info's bytes_may_use. Meanwhile __btrfs_prealloc_file_range() will call btrfs_free_reserved_data_space() internally for both sucessful and failed path, btrfs_prealloc_file_range()'s callers does not need to call btrfs_free_reserved_data_space() any more. Signed-off-by: Wang Xiaoguang <wangxg.fnst@cn.fujitsu.com> Reviewed-by: Josef Bacik <jbacik@fb.com> Signed-off-by: David Sterba <dsterba@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2016-07-25 07:51:40 +00:00
} else {
/*
* Do not need to reserve unwritten extent for this
* range, free reserved data space first, otherwise
* it'll result in false ENOSPC error.
*/
btrfs_free_reserved_data_space(inode, cur_offset,
last_byte - cur_offset);
}
free_extent_map(em);
cur_offset = last_byte;
if (cur_offset >= alloc_end)
break;
}
/*
* If ret is still 0, means we're OK to fallocate.
* Or just cleanup the list and exit.
*/
list_for_each_entry_safe(range, tmp, &reserve_list, list) {
if (!ret)
ret = btrfs_prealloc_file_range(inode, mode,
range->start,
range->len, i_blocksize(inode),
offset + len, &alloc_hint);
btrfs: update btrfs_space_info's bytes_may_use timely This patch can fix some false ENOSPC errors, below test script can reproduce one false ENOSPC error: #!/bin/bash dd if=/dev/zero of=fs.img bs=$((1024*1024)) count=128 dev=$(losetup --show -f fs.img) mkfs.btrfs -f -M $dev mkdir /tmp/mntpoint mount $dev /tmp/mntpoint cd /tmp/mntpoint xfs_io -f -c "falloc 0 $((64*1024*1024))" testfile Above script will fail for ENOSPC reason, but indeed fs still has free space to satisfy this request. Please see call graph: btrfs_fallocate() |-> btrfs_alloc_data_chunk_ondemand() | bytes_may_use += 64M |-> btrfs_prealloc_file_range() |-> btrfs_reserve_extent() |-> btrfs_add_reserved_bytes() | alloc_type is RESERVE_ALLOC_NO_ACCOUNT, so it does not | change bytes_may_use, and bytes_reserved += 64M. Now | bytes_may_use + bytes_reserved == 128M, which is greater | than btrfs_space_info's total_bytes, false enospc occurs. | Note, the bytes_may_use decrease operation will be done in | end of btrfs_fallocate(), which is too late. Here is another simple case for buffered write: CPU 1 | CPU 2 | |-> cow_file_range() |-> __btrfs_buffered_write() |-> btrfs_reserve_extent() | | | | | | | | | ..... | |-> btrfs_check_data_free_space() | | | | |-> extent_clear_unlock_delalloc() | In CPU 1, btrfs_reserve_extent()->find_free_extent()-> btrfs_add_reserved_bytes() do not decrease bytes_may_use, the decrease operation will be delayed to be done in extent_clear_unlock_delalloc(). Assume in this case, btrfs_reserve_extent() reserved 128MB data, CPU2's btrfs_check_data_free_space() tries to reserve 100MB data space. If 100MB > data_sinfo->total_bytes - data_sinfo->bytes_used - data_sinfo->bytes_reserved - data_sinfo->bytes_pinned - data_sinfo->bytes_readonly - data_sinfo->bytes_may_use btrfs_check_data_free_space() will try to allcate new data chunk or call btrfs_start_delalloc_roots(), or commit current transaction in order to reserve some free space, obviously a lot of work. But indeed it's not necessary as long as decreasing bytes_may_use timely, we still have free space, decreasing 128M from bytes_may_use. To fix this issue, this patch chooses to update bytes_may_use for both data and metadata in btrfs_add_reserved_bytes(). For compress path, real extent length may not be equal to file content length, so introduce a ram_bytes argument for btrfs_reserve_extent(), find_free_extent() and btrfs_add_reserved_bytes(), it's becasue bytes_may_use is increased by file content length. Then compress path can update bytes_may_use correctly. Also now we can discard RESERVE_ALLOC_NO_ACCOUNT, RESERVE_ALLOC and RESERVE_FREE. As we know, usually EXTENT_DO_ACCOUNTING is used for error path. In run_delalloc_nocow(), for inode marked as NODATACOW or extent marked as PREALLOC, we also need to update bytes_may_use, but can not pass EXTENT_DO_ACCOUNTING, because it also clears metadata reservation, so here we introduce EXTENT_CLEAR_DATA_RESV flag to indicate btrfs_clear_bit_hook() to update btrfs_space_info's bytes_may_use. Meanwhile __btrfs_prealloc_file_range() will call btrfs_free_reserved_data_space() internally for both sucessful and failed path, btrfs_prealloc_file_range()'s callers does not need to call btrfs_free_reserved_data_space() any more. Signed-off-by: Wang Xiaoguang <wangxg.fnst@cn.fujitsu.com> Reviewed-by: Josef Bacik <jbacik@fb.com> Signed-off-by: David Sterba <dsterba@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2016-07-25 07:51:40 +00:00
else
btrfs_free_reserved_data_space(inode, range->start,
range->len);
list_del(&range->list);
kfree(range);
}
if (ret < 0)
goto out_unlock;
if (actual_end > inode->i_size &&
!(mode & FALLOC_FL_KEEP_SIZE)) {
struct btrfs_trans_handle *trans;
struct btrfs_root *root = BTRFS_I(inode)->root;
/*
* We didn't need to allocate any more space, but we
* still extended the size of the file so we need to
* update i_size and the inode item.
*/
trans = btrfs_start_transaction(root, 1);
if (IS_ERR(trans)) {
ret = PTR_ERR(trans);
} else {
inode->i_ctime = current_time(inode);
i_size_write(inode, actual_end);
btrfs_ordered_update_i_size(inode, actual_end, NULL);
ret = btrfs_update_inode(trans, root, inode);
if (ret)
btrfs_end_transaction(trans);
else
ret = btrfs_end_transaction(trans);
}
}
out_unlock:
unlock_extent_cached(&BTRFS_I(inode)->io_tree, alloc_start, locked_end,
&cached_state, GFP_KERNEL);
out:
inode_unlock(inode);
/* Let go of our reservation. */
btrfs: update btrfs_space_info's bytes_may_use timely This patch can fix some false ENOSPC errors, below test script can reproduce one false ENOSPC error: #!/bin/bash dd if=/dev/zero of=fs.img bs=$((1024*1024)) count=128 dev=$(losetup --show -f fs.img) mkfs.btrfs -f -M $dev mkdir /tmp/mntpoint mount $dev /tmp/mntpoint cd /tmp/mntpoint xfs_io -f -c "falloc 0 $((64*1024*1024))" testfile Above script will fail for ENOSPC reason, but indeed fs still has free space to satisfy this request. Please see call graph: btrfs_fallocate() |-> btrfs_alloc_data_chunk_ondemand() | bytes_may_use += 64M |-> btrfs_prealloc_file_range() |-> btrfs_reserve_extent() |-> btrfs_add_reserved_bytes() | alloc_type is RESERVE_ALLOC_NO_ACCOUNT, so it does not | change bytes_may_use, and bytes_reserved += 64M. Now | bytes_may_use + bytes_reserved == 128M, which is greater | than btrfs_space_info's total_bytes, false enospc occurs. | Note, the bytes_may_use decrease operation will be done in | end of btrfs_fallocate(), which is too late. Here is another simple case for buffered write: CPU 1 | CPU 2 | |-> cow_file_range() |-> __btrfs_buffered_write() |-> btrfs_reserve_extent() | | | | | | | | | ..... | |-> btrfs_check_data_free_space() | | | | |-> extent_clear_unlock_delalloc() | In CPU 1, btrfs_reserve_extent()->find_free_extent()-> btrfs_add_reserved_bytes() do not decrease bytes_may_use, the decrease operation will be delayed to be done in extent_clear_unlock_delalloc(). Assume in this case, btrfs_reserve_extent() reserved 128MB data, CPU2's btrfs_check_data_free_space() tries to reserve 100MB data space. If 100MB > data_sinfo->total_bytes - data_sinfo->bytes_used - data_sinfo->bytes_reserved - data_sinfo->bytes_pinned - data_sinfo->bytes_readonly - data_sinfo->bytes_may_use btrfs_check_data_free_space() will try to allcate new data chunk or call btrfs_start_delalloc_roots(), or commit current transaction in order to reserve some free space, obviously a lot of work. But indeed it's not necessary as long as decreasing bytes_may_use timely, we still have free space, decreasing 128M from bytes_may_use. To fix this issue, this patch chooses to update bytes_may_use for both data and metadata in btrfs_add_reserved_bytes(). For compress path, real extent length may not be equal to file content length, so introduce a ram_bytes argument for btrfs_reserve_extent(), find_free_extent() and btrfs_add_reserved_bytes(), it's becasue bytes_may_use is increased by file content length. Then compress path can update bytes_may_use correctly. Also now we can discard RESERVE_ALLOC_NO_ACCOUNT, RESERVE_ALLOC and RESERVE_FREE. As we know, usually EXTENT_DO_ACCOUNTING is used for error path. In run_delalloc_nocow(), for inode marked as NODATACOW or extent marked as PREALLOC, we also need to update bytes_may_use, but can not pass EXTENT_DO_ACCOUNTING, because it also clears metadata reservation, so here we introduce EXTENT_CLEAR_DATA_RESV flag to indicate btrfs_clear_bit_hook() to update btrfs_space_info's bytes_may_use. Meanwhile __btrfs_prealloc_file_range() will call btrfs_free_reserved_data_space() internally for both sucessful and failed path, btrfs_prealloc_file_range()'s callers does not need to call btrfs_free_reserved_data_space() any more. Signed-off-by: Wang Xiaoguang <wangxg.fnst@cn.fujitsu.com> Reviewed-by: Josef Bacik <jbacik@fb.com> Signed-off-by: David Sterba <dsterba@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2016-07-25 07:51:40 +00:00
if (ret != 0)
btrfs_free_reserved_data_space(inode, alloc_start,
alloc_end - cur_offset);
return ret;
}
static int find_desired_extent(struct inode *inode, loff_t *offset, int whence)
{
struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
struct extent_map *em = NULL;
struct extent_state *cached_state = NULL;
Btrfs: fix up bounds checking in lseek An user reported this, it is because that lseek's SEEK_SET/SEEK_CUR/SEEK_END allow a negative value for @offset, but btrfs's SEEK_DATA/SEEK_HOLE don't prepare for that and convert the negative @offset into unsigned type, so we get (end < start) warning. [ 1269.835374] ------------[ cut here ]------------ [ 1269.836809] WARNING: CPU: 0 PID: 1241 at fs/btrfs/extent_io.c:430 insert_state+0x11d/0x140() [ 1269.838816] BTRFS: end < start 4094 18446744073709551615 [ 1269.840334] CPU: 0 PID: 1241 Comm: a.out Tainted: G W 3.16.0+ #306 [ 1269.858229] Call Trace: [ 1269.858612] [<ffffffff81801a69>] dump_stack+0x4e/0x68 [ 1269.858952] [<ffffffff8107894c>] warn_slowpath_common+0x8c/0xc0 [ 1269.859416] [<ffffffff81078a36>] warn_slowpath_fmt+0x46/0x50 [ 1269.859929] [<ffffffff813b0fbd>] insert_state+0x11d/0x140 [ 1269.860409] [<ffffffff813b1396>] __set_extent_bit+0x3b6/0x4e0 [ 1269.860805] [<ffffffff813b21c7>] lock_extent_bits+0x87/0x200 [ 1269.861697] [<ffffffff813a5b28>] btrfs_file_llseek+0x148/0x2a0 [ 1269.862168] [<ffffffff811f201e>] SyS_lseek+0xae/0xc0 [ 1269.862620] [<ffffffff8180b212>] system_call_fastpath+0x16/0x1b [ 1269.862970] ---[ end trace 4d33ea885832054b ]--- This assumes that btrfs starts finding DATA/HOLE from the beginning of file if the assigned @offset is negative. Also we add alignment for lock_extent_bits 's range. Reported-by: Toralf Förster <toralf.foerster@gmx.de> Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-09-16 09:49:30 +00:00
u64 lockstart;
u64 lockend;
u64 start;
u64 len;
int ret = 0;
Btrfs: fix up bounds checking in lseek An user reported this, it is because that lseek's SEEK_SET/SEEK_CUR/SEEK_END allow a negative value for @offset, but btrfs's SEEK_DATA/SEEK_HOLE don't prepare for that and convert the negative @offset into unsigned type, so we get (end < start) warning. [ 1269.835374] ------------[ cut here ]------------ [ 1269.836809] WARNING: CPU: 0 PID: 1241 at fs/btrfs/extent_io.c:430 insert_state+0x11d/0x140() [ 1269.838816] BTRFS: end < start 4094 18446744073709551615 [ 1269.840334] CPU: 0 PID: 1241 Comm: a.out Tainted: G W 3.16.0+ #306 [ 1269.858229] Call Trace: [ 1269.858612] [<ffffffff81801a69>] dump_stack+0x4e/0x68 [ 1269.858952] [<ffffffff8107894c>] warn_slowpath_common+0x8c/0xc0 [ 1269.859416] [<ffffffff81078a36>] warn_slowpath_fmt+0x46/0x50 [ 1269.859929] [<ffffffff813b0fbd>] insert_state+0x11d/0x140 [ 1269.860409] [<ffffffff813b1396>] __set_extent_bit+0x3b6/0x4e0 [ 1269.860805] [<ffffffff813b21c7>] lock_extent_bits+0x87/0x200 [ 1269.861697] [<ffffffff813a5b28>] btrfs_file_llseek+0x148/0x2a0 [ 1269.862168] [<ffffffff811f201e>] SyS_lseek+0xae/0xc0 [ 1269.862620] [<ffffffff8180b212>] system_call_fastpath+0x16/0x1b [ 1269.862970] ---[ end trace 4d33ea885832054b ]--- This assumes that btrfs starts finding DATA/HOLE from the beginning of file if the assigned @offset is negative. Also we add alignment for lock_extent_bits 's range. Reported-by: Toralf Förster <toralf.foerster@gmx.de> Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-09-16 09:49:30 +00:00
if (inode->i_size == 0)
return -ENXIO;
/*
* *offset can be negative, in this case we start finding DATA/HOLE from
* the very start of the file.
*/
start = max_t(loff_t, 0, *offset);
lockstart = round_down(start, fs_info->sectorsize);
lockend = round_up(i_size_read(inode),
fs_info->sectorsize);
if (lockend <= lockstart)
lockend = lockstart + fs_info->sectorsize;
lockend--;
len = lockend - lockstart + 1;
lock_extent_bits(&BTRFS_I(inode)->io_tree, lockstart, lockend,
&cached_state);
while (start < inode->i_size) {
em = btrfs_get_extent_fiemap(BTRFS_I(inode), NULL, 0,
start, len, 0);
if (IS_ERR(em)) {
ret = PTR_ERR(em);
em = NULL;
break;
}
if (whence == SEEK_HOLE &&
(em->block_start == EXTENT_MAP_HOLE ||
test_bit(EXTENT_FLAG_PREALLOC, &em->flags)))
break;
else if (whence == SEEK_DATA &&
(em->block_start != EXTENT_MAP_HOLE &&
!test_bit(EXTENT_FLAG_PREALLOC, &em->flags)))
break;
start = em->start + em->len;
free_extent_map(em);
em = NULL;
cond_resched();
}
free_extent_map(em);
if (!ret) {
if (whence == SEEK_DATA && start >= inode->i_size)
ret = -ENXIO;
else
*offset = min_t(loff_t, start, inode->i_size);
}
unlock_extent_cached(&BTRFS_I(inode)->io_tree, lockstart, lockend,
&cached_state, GFP_NOFS);
return ret;
}
static loff_t btrfs_file_llseek(struct file *file, loff_t offset, int whence)
{
struct inode *inode = file->f_mapping->host;
int ret;
inode_lock(inode);
switch (whence) {
case SEEK_END:
case SEEK_CUR:
offset = generic_file_llseek(file, offset, whence);
goto out;
case SEEK_DATA:
case SEEK_HOLE:
if (offset >= i_size_read(inode)) {
inode_unlock(inode);
return -ENXIO;
}
ret = find_desired_extent(inode, &offset, whence);
if (ret) {
inode_unlock(inode);
return ret;
}
}
offset = vfs_setpos(file, offset, inode->i_sb->s_maxbytes);
out:
inode_unlock(inode);
return offset;
}
const struct file_operations btrfs_file_operations = {
.llseek = btrfs_file_llseek,
.read_iter = generic_file_read_iter,
.splice_read = generic_file_splice_read,
.write_iter = btrfs_file_write_iter,
.mmap = btrfs_file_mmap,
.open = generic_file_open,
.release = btrfs_release_file,
.fsync = btrfs_sync_file,
.fallocate = btrfs_fallocate,
.unlocked_ioctl = btrfs_ioctl,
#ifdef CONFIG_COMPAT
.compat_ioctl = btrfs_compat_ioctl,
#endif
.clone_file_range = btrfs_clone_file_range,
.dedupe_file_range = btrfs_dedupe_file_range,
};
void btrfs_auto_defrag_exit(void)
{
kmem_cache_destroy(btrfs_inode_defrag_cachep);
}
int btrfs_auto_defrag_init(void)
{
btrfs_inode_defrag_cachep = kmem_cache_create("btrfs_inode_defrag",
sizeof(struct inode_defrag), 0,
SLAB_MEM_SPREAD,
NULL);
if (!btrfs_inode_defrag_cachep)
return -ENOMEM;
return 0;
}
int btrfs_fdatawrite_range(struct inode *inode, loff_t start, loff_t end)
{
int ret;
/*
* So with compression we will find and lock a dirty page and clear the
* first one as dirty, setup an async extent, and immediately return
* with the entire range locked but with nobody actually marked with
* writeback. So we can't just filemap_write_and_wait_range() and
* expect it to work since it will just kick off a thread to do the
* actual work. So we need to call filemap_fdatawrite_range _again_
* since it will wait on the page lock, which won't be unlocked until
* after the pages have been marked as writeback and so we're good to go
* from there. We have to do this otherwise we'll miss the ordered
* extents and that results in badness. Please Josef, do not think you
* know better and pull this out at some point in the future, it is
* right and you are wrong.
*/
ret = filemap_fdatawrite_range(inode->i_mapping, start, end);
if (!ret && test_bit(BTRFS_INODE_HAS_ASYNC_EXTENT,
&BTRFS_I(inode)->runtime_flags))
ret = filemap_fdatawrite_range(inode->i_mapping, start, end);
return ret;
}