linux/fs/btrfs/extent_map.c

1054 lines
28 KiB
C
Raw Normal View History

License cleanup: add SPDX GPL-2.0 license identifier to files with no license Many source files in the tree are missing licensing information, which makes it harder for compliance tools to determine the correct license. By default all files without license information are under the default license of the kernel, which is GPL version 2. Update the files which contain no license information with the 'GPL-2.0' SPDX license identifier. The SPDX identifier is a legally binding shorthand, which can be used instead of the full boiler plate text. This patch is based on work done by Thomas Gleixner and Kate Stewart and Philippe Ombredanne. How this work was done: Patches were generated and checked against linux-4.14-rc6 for a subset of the use cases: - file had no licensing information it it. - file was a */uapi/* one with no licensing information in it, - file was a */uapi/* one with existing licensing information, Further patches will be generated in subsequent months to fix up cases where non-standard license headers were used, and references to license had to be inferred by heuristics based on keywords. The analysis to determine which SPDX License Identifier to be applied to a file was done in a spreadsheet of side by side results from of the output of two independent scanners (ScanCode & Windriver) producing SPDX tag:value files created by Philippe Ombredanne. Philippe prepared the base worksheet, and did an initial spot review of a few 1000 files. The 4.13 kernel was the starting point of the analysis with 60,537 files assessed. Kate Stewart did a file by file comparison of the scanner results in the spreadsheet to determine which SPDX license identifier(s) to be applied to the file. She confirmed any determination that was not immediately clear with lawyers working with the Linux Foundation. Criteria used to select files for SPDX license identifier tagging was: - Files considered eligible had to be source code files. - Make and config files were included as candidates if they contained >5 lines of source - File already had some variant of a license header in it (even if <5 lines). All documentation files were explicitly excluded. The following heuristics were used to determine which SPDX license identifiers to apply. - when both scanners couldn't find any license traces, file was considered to have no license information in it, and the top level COPYING file license applied. For non */uapi/* files that summary was: SPDX license identifier # files ---------------------------------------------------|------- GPL-2.0 11139 and resulted in the first patch in this series. If that file was a */uapi/* path one, it was "GPL-2.0 WITH Linux-syscall-note" otherwise it was "GPL-2.0". Results of that was: SPDX license identifier # files ---------------------------------------------------|------- GPL-2.0 WITH Linux-syscall-note 930 and resulted in the second patch in this series. - if a file had some form of licensing information in it, and was one of the */uapi/* ones, it was denoted with the Linux-syscall-note if any GPL family license was found in the file or had no licensing in it (per prior point). Results summary: SPDX license identifier # files ---------------------------------------------------|------ GPL-2.0 WITH Linux-syscall-note 270 GPL-2.0+ WITH Linux-syscall-note 169 ((GPL-2.0 WITH Linux-syscall-note) OR BSD-2-Clause) 21 ((GPL-2.0 WITH Linux-syscall-note) OR BSD-3-Clause) 17 LGPL-2.1+ WITH Linux-syscall-note 15 GPL-1.0+ WITH Linux-syscall-note 14 ((GPL-2.0+ WITH Linux-syscall-note) OR BSD-3-Clause) 5 LGPL-2.0+ WITH Linux-syscall-note 4 LGPL-2.1 WITH Linux-syscall-note 3 ((GPL-2.0 WITH Linux-syscall-note) OR MIT) 3 ((GPL-2.0 WITH Linux-syscall-note) AND MIT) 1 and that resulted in the third patch in this series. - when the two scanners agreed on the detected license(s), that became the concluded license(s). - when there was disagreement between the two scanners (one detected a license but the other didn't, or they both detected different licenses) a manual inspection of the file occurred. - In most cases a manual inspection of the information in the file resulted in a clear resolution of the license that should apply (and which scanner probably needed to revisit its heuristics). - When it was not immediately clear, the license identifier was confirmed with lawyers working with the Linux Foundation. - If there was any question as to the appropriate license identifier, the file was flagged for further research and to be revisited later in time. In total, over 70 hours of logged manual review was done on the spreadsheet to determine the SPDX license identifiers to apply to the source files by Kate, Philippe, Thomas and, in some cases, confirmation by lawyers working with the Linux Foundation. Kate also obtained a third independent scan of the 4.13 code base from FOSSology, and compared selected files where the other two scanners disagreed against that SPDX file, to see if there was new insights. The Windriver scanner is based on an older version of FOSSology in part, so they are related. Thomas did random spot checks in about 500 files from the spreadsheets for the uapi headers and agreed with SPDX license identifier in the files he inspected. For the non-uapi files Thomas did random spot checks in about 15000 files. In initial set of patches against 4.14-rc6, 3 files were found to have copy/paste license identifier errors, and have been fixed to reflect the correct identifier. Additionally Philippe spent 10 hours this week doing a detailed manual inspection and review of the 12,461 patched files from the initial patch version early this week with: - a full scancode scan run, collecting the matched texts, detected license ids and scores - reviewing anything where there was a license detected (about 500+ files) to ensure that the applied SPDX license was correct - reviewing anything where there was no detection but the patch license was not GPL-2.0 WITH Linux-syscall-note to ensure that the applied SPDX license was correct This produced a worksheet with 20 files needing minor correction. This worksheet was then exported into 3 different .csv files for the different types of files to be modified. These .csv files were then reviewed by Greg. Thomas wrote a script to parse the csv files and add the proper SPDX tag to the file, in the format that the file expected. This script was further refined by Greg based on the output to detect more types of files automatically and to distinguish between header and source .c files (which need different comment types.) Finally Greg ran the script using the .csv files to generate the patches. Reviewed-by: Kate Stewart <kstewart@linuxfoundation.org> Reviewed-by: Philippe Ombredanne <pombredanne@nexb.com> Reviewed-by: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2017-11-01 14:07:57 +00:00
// SPDX-License-Identifier: GPL-2.0
#include <linux/err.h>
#include <linux/slab.h>
#include <linux/spinlock.h>
#include "messages.h"
#include "ctree.h"
#include "volumes.h"
#include "extent_map.h"
#include "compression.h"
#include "btrfs_inode.h"
static struct kmem_cache *extent_map_cache;
int __init extent_map_init(void)
{
extent_map_cache = kmem_cache_create("btrfs_extent_map",
sizeof(struct extent_map), 0,
SLAB_MEM_SPREAD, NULL);
if (!extent_map_cache)
return -ENOMEM;
return 0;
}
void __cold extent_map_exit(void)
{
kmem_cache_destroy(extent_map_cache);
}
/*
* Initialize the extent tree @tree. Should be called for each new inode or
* other user of the extent_map interface.
*/
void extent_map_tree_init(struct extent_map_tree *tree)
{
tree->map = RB_ROOT_CACHED;
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
INIT_LIST_HEAD(&tree->modified_extents);
rwlock_init(&tree->lock);
}
/*
* Allocate a new extent_map structure. The new structure is returned with a
* reference count of one and needs to be freed using free_extent_map()
*/
struct extent_map *alloc_extent_map(void)
{
struct extent_map *em;
em = kmem_cache_zalloc(extent_map_cache, GFP_NOFS);
if (!em)
return NULL;
RB_CLEAR_NODE(&em->rb_node);
em->compress_type = BTRFS_COMPRESS_NONE;
refcount_set(&em->refs, 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
INIT_LIST_HEAD(&em->list);
return em;
}
/*
* Drop the reference out on @em by one and free the structure if the reference
* count hits zero.
*/
void free_extent_map(struct extent_map *em)
{
if (!em)
return;
if (refcount_dec_and_test(&em->refs)) {
WARN_ON(extent_map_in_tree(em));
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
WARN_ON(!list_empty(&em->list));
Btrfs: fix NULL pointer crash when running balance and scrub concurrently While running balance, scrub, fsstress concurrently we hit the following kernel crash: [56561.448845] BTRFS info (device sde): relocating block group 11005853696 flags 132 [56561.524077] BUG: unable to handle kernel NULL pointer dereference at 0000000000000078 [56561.524237] IP: [<ffffffffa038956d>] scrub_chunk.isra.12+0xdd/0x130 [btrfs] [56561.524297] PGD 9be28067 PUD 7f3dd067 PMD 0 [56561.524325] Oops: 0000 [#1] SMP [....] [56561.527237] Call Trace: [56561.527309] [<ffffffffa038980e>] scrub_enumerate_chunks+0x24e/0x490 [btrfs] [56561.527392] [<ffffffff810abe00>] ? abort_exclusive_wait+0x50/0xb0 [56561.527476] [<ffffffffa038add4>] btrfs_scrub_dev+0x1a4/0x530 [btrfs] [56561.527561] [<ffffffffa0368107>] btrfs_ioctl+0x13f7/0x2a90 [btrfs] [56561.527639] [<ffffffff811c82f0>] do_vfs_ioctl+0x2e0/0x4c0 [56561.527712] [<ffffffff8109c384>] ? vtime_account_user+0x54/0x60 [56561.527788] [<ffffffff810f768c>] ? __audit_syscall_entry+0x9c/0xf0 [56561.527870] [<ffffffff811c8551>] SyS_ioctl+0x81/0xa0 [56561.527941] [<ffffffff815707f7>] tracesys+0xdd/0xe2 [...] [56561.528304] RIP [<ffffffffa038956d>] scrub_chunk.isra.12+0xdd/0x130 [btrfs] [56561.528395] RSP <ffff88004c0f5be8> [56561.528454] CR2: 0000000000000078 This is because in btrfs_relocate_chunk(), we will free @bdev directly while scrub may still hold extent mapping, and may access freed memory. Fix this problem by wrapping freeing @bdev work into free_extent_map() which is based on reference count. Reported-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-06-19 02:42:52 +00:00
if (test_bit(EXTENT_FLAG_FS_MAPPING, &em->flags))
kfree(em->map_lookup);
kmem_cache_free(extent_map_cache, em);
}
}
/* Do the math around the end of an extent, handling wrapping. */
static u64 range_end(u64 start, u64 len)
{
if (start + len < start)
return (u64)-1;
return start + len;
}
static int tree_insert(struct rb_root_cached *root, struct extent_map *em)
{
struct rb_node **p = &root->rb_root.rb_node;
struct rb_node *parent = NULL;
struct extent_map *entry = NULL;
struct rb_node *orig_parent = NULL;
u64 end = range_end(em->start, em->len);
bool leftmost = true;
while (*p) {
parent = *p;
entry = rb_entry(parent, struct extent_map, rb_node);
if (em->start < entry->start) {
p = &(*p)->rb_left;
} else if (em->start >= extent_map_end(entry)) {
p = &(*p)->rb_right;
leftmost = false;
} else {
return -EEXIST;
}
}
orig_parent = parent;
while (parent && em->start >= extent_map_end(entry)) {
parent = rb_next(parent);
entry = rb_entry(parent, struct extent_map, rb_node);
}
if (parent)
if (end > entry->start && em->start < extent_map_end(entry))
return -EEXIST;
parent = orig_parent;
entry = rb_entry(parent, struct extent_map, rb_node);
while (parent && em->start < entry->start) {
parent = rb_prev(parent);
entry = rb_entry(parent, struct extent_map, rb_node);
}
if (parent)
if (end > entry->start && em->start < extent_map_end(entry))
return -EEXIST;
rb_link_node(&em->rb_node, orig_parent, p);
rb_insert_color_cached(&em->rb_node, root, leftmost);
return 0;
}
/*
* Search through the tree for an extent_map with a given offset. If it can't
* be found, try to find some neighboring extents
*/
static struct rb_node *__tree_search(struct rb_root *root, u64 offset,
struct rb_node **prev_or_next_ret)
{
struct rb_node *n = root->rb_node;
struct rb_node *prev = NULL;
struct rb_node *orig_prev = NULL;
struct extent_map *entry;
struct extent_map *prev_entry = NULL;
ASSERT(prev_or_next_ret);
while (n) {
entry = rb_entry(n, struct extent_map, rb_node);
prev = n;
prev_entry = entry;
if (offset < entry->start)
n = n->rb_left;
else if (offset >= extent_map_end(entry))
n = n->rb_right;
else
return n;
}
orig_prev = prev;
while (prev && offset >= extent_map_end(prev_entry)) {
prev = rb_next(prev);
prev_entry = rb_entry(prev, struct extent_map, rb_node);
}
/*
* Previous extent map found, return as in this case the caller does not
* care about the next one.
*/
if (prev) {
*prev_or_next_ret = prev;
return NULL;
}
prev = orig_prev;
prev_entry = rb_entry(prev, struct extent_map, rb_node);
while (prev && offset < prev_entry->start) {
prev = rb_prev(prev);
prev_entry = rb_entry(prev, struct extent_map, rb_node);
}
*prev_or_next_ret = prev;
return NULL;
}
/* Check to see if two extent_map structs are adjacent and safe to merge. */
static int mergable_maps(struct extent_map *prev, struct extent_map *next)
{
if (test_bit(EXTENT_FLAG_PINNED, &prev->flags))
return 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
/*
* don't merge compressed extents, we need to know their
* actual size
*/
if (test_bit(EXTENT_FLAG_COMPRESSED, &prev->flags))
return 0;
if (test_bit(EXTENT_FLAG_LOGGING, &prev->flags) ||
test_bit(EXTENT_FLAG_LOGGING, &next->flags))
return 0;
2013-04-05 20:51:15 +00:00
/*
* We don't want to merge stuff that hasn't been written to the log yet
* since it may not reflect exactly what is on disk, and that would be
* bad.
*/
if (!list_empty(&prev->list) || !list_empty(&next->list))
return 0;
ASSERT(next->block_start != EXTENT_MAP_DELALLOC &&
prev->block_start != EXTENT_MAP_DELALLOC);
if (prev->map_lookup || next->map_lookup)
ASSERT(test_bit(EXTENT_FLAG_FS_MAPPING, &prev->flags) &&
test_bit(EXTENT_FLAG_FS_MAPPING, &next->flags));
if (extent_map_end(prev) == next->start &&
prev->flags == next->flags &&
prev->map_lookup == next->map_lookup &&
((next->block_start == EXTENT_MAP_HOLE &&
prev->block_start == EXTENT_MAP_HOLE) ||
(next->block_start == EXTENT_MAP_INLINE &&
prev->block_start == EXTENT_MAP_INLINE) ||
(next->block_start < EXTENT_MAP_LAST_BYTE - 1 &&
next->block_start == extent_map_block_end(prev)))) {
return 1;
}
return 0;
}
static void try_merge_map(struct extent_map_tree *tree, struct extent_map *em)
{
struct extent_map *merge = NULL;
struct rb_node *rb;
Btrfs: fix race between using extent maps and merging them We have a few cases where we allow an extent map that is in an extent map tree to be merged with other extents in the tree. Such cases include the unpinning of an extent after the respective ordered extent completed or after logging an extent during a fast fsync. This can lead to subtle and dangerous problems because when doing the merge some other task might be using the same extent map and as consequence see an inconsistent state of the extent map - for example sees the new length but has seen the old start offset. With luck this triggers a BUG_ON(), and not some silent bug, such as the following one in __do_readpage(): $ cat -n fs/btrfs/extent_io.c 3061 static int __do_readpage(struct extent_io_tree *tree, 3062 struct page *page, (...) 3127 em = __get_extent_map(inode, page, pg_offset, cur, 3128 end - cur + 1, get_extent, em_cached); 3129 if (IS_ERR_OR_NULL(em)) { 3130 SetPageError(page); 3131 unlock_extent(tree, cur, end); 3132 break; 3133 } 3134 extent_offset = cur - em->start; 3135 BUG_ON(extent_map_end(em) <= cur); (...) Consider the following example scenario, where we end up hitting the BUG_ON() in __do_readpage(). We have an inode with a size of 8KiB and 2 extent maps: extent A: file offset 0, length 4KiB, disk_bytenr = X, persisted on disk by a previous transaction extent B: file offset 4KiB, length 4KiB, disk_bytenr = X + 4KiB, not yet persisted but writeback started for it already. The extent map is pinned since there's writeback and an ordered extent in progress, so it can not be merged with extent map A yet The following sequence of steps leads to the BUG_ON(): 1) The ordered extent for extent B completes, the respective page gets its writeback bit cleared and the extent map is unpinned, at that point it is not yet merged with extent map A because it's in the list of modified extents; 2) Due to memory pressure, or some other reason, the MM subsystem releases the page corresponding to extent B - btrfs_releasepage() is called and returns 1, meaning the page can be released as it's not dirty, not under writeback anymore and the extent range is not locked in the inode's iotree. However the extent map is not released, either because we are not in a context that allows memory allocations to block or because the inode's size is smaller than 16MiB - in this case our inode has a size of 8KiB; 3) Task B needs to read extent B and ends up __do_readpage() through the btrfs_readpage() callback. At __do_readpage() it gets a reference to extent map B; 4) Task A, doing a fast fsync, calls clear_em_loggin() against extent map B while holding the write lock on the inode's extent map tree - this results in try_merge_map() being called and since it's possible to merge extent map B with extent map A now (the extent map B was removed from the list of modified extents), the merging begins - it sets extent map B's start offset to 0 (was 4KiB), but before it increments the map's length to 8KiB (4kb + 4KiB), task A is at: BUG_ON(extent_map_end(em) <= cur); The call to extent_map_end() sees the extent map has a start of 0 and a length still at 4KiB, so it returns 4KiB and 'cur' is 4KiB, so the BUG_ON() is triggered. So it's dangerous to modify an extent map that is in the tree, because some other task might have got a reference to it before and still using it, and needs to see a consistent map while using it. Generally this is very rare since most paths that lookup and use extent maps also have the file range locked in the inode's iotree. The fsync path is pretty much the only exception where we don't do it to avoid serialization with concurrent reads. Fix this by not allowing an extent map do be merged if if it's being used by tasks other then the one attempting to merge the extent map (when the reference count of the extent map is greater than 2). Reported-by: ryusuke1925 <st13s20@gm.ibaraki-ct.ac.jp> Reported-by: Koki Mitani <koki.mitani.xg@hco.ntt.co.jp> Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=206211 CC: stable@vger.kernel.org # 4.4+ Reviewed-by: Josef Bacik <josef@toxicpanda.com> Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: David Sterba <dsterba@suse.com>
2020-01-31 14:06:07 +00:00
/*
* We can't modify an extent map that is in the tree and that is being
* used by another task, as it can cause that other task to see it in
* inconsistent state during the merging. We always have 1 reference for
* the tree and 1 for this task (which is unpinning the extent map or
* clearing the logging flag), so anything > 2 means it's being used by
* other tasks too.
*/
if (refcount_read(&em->refs) > 2)
return;
if (em->start != 0) {
rb = rb_prev(&em->rb_node);
if (rb)
merge = rb_entry(rb, struct extent_map, rb_node);
if (rb && mergable_maps(merge, em)) {
em->start = merge->start;
em->orig_start = merge->orig_start;
em->len += merge->len;
em->block_len += merge->block_len;
em->block_start = merge->block_start;
em->mod_len = (em->mod_len + em->mod_start) - merge->mod_start;
em->mod_start = merge->mod_start;
em->generation = max(em->generation, merge->generation);
set_bit(EXTENT_FLAG_MERGED, &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
rb_erase_cached(&merge->rb_node, &tree->map);
RB_CLEAR_NODE(&merge->rb_node);
free_extent_map(merge);
}
}
rb = rb_next(&em->rb_node);
if (rb)
merge = rb_entry(rb, struct extent_map, rb_node);
if (rb && mergable_maps(em, merge)) {
em->len += merge->len;
em->block_len += merge->block_len;
rb_erase_cached(&merge->rb_node, &tree->map);
RB_CLEAR_NODE(&merge->rb_node);
em->mod_len = (merge->mod_start + merge->mod_len) - em->mod_start;
em->generation = max(em->generation, merge->generation);
set_bit(EXTENT_FLAG_MERGED, &em->flags);
free_extent_map(merge);
}
}
/*
* Unpin an extent from the 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
* @tree: tree to unpin the extent in
* @start: logical offset in the file
* @len: length of the extent
* @gen: generation that this extent has been modified in
*
* Called after an extent has been written to disk properly. Set the generation
* to the generation that actually added the file item to the inode so we know
* we need to sync this extent when we call fsync().
*/
int unpin_extent_cache(struct extent_map_tree *tree, u64 start, u64 len,
u64 gen)
{
int ret = 0;
struct extent_map *em;
bool prealloc = false;
write_lock(&tree->lock);
em = lookup_extent_mapping(tree, start, len);
WARN_ON(!em || em->start != start);
if (!em)
goto out;
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
em->generation = gen;
clear_bit(EXTENT_FLAG_PINNED, &em->flags);
em->mod_start = em->start;
em->mod_len = em->len;
if (test_bit(EXTENT_FLAG_FILLING, &em->flags)) {
prealloc = true;
clear_bit(EXTENT_FLAG_FILLING, &em->flags);
}
try_merge_map(tree, em);
if (prealloc) {
em->mod_start = em->start;
em->mod_len = em->len;
}
free_extent_map(em);
out:
write_unlock(&tree->lock);
return ret;
}
void clear_em_logging(struct extent_map_tree *tree, struct extent_map *em)
{
lockdep_assert_held_write(&tree->lock);
clear_bit(EXTENT_FLAG_LOGGING, &em->flags);
if (extent_map_in_tree(em))
try_merge_map(tree, em);
}
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
static inline void setup_extent_mapping(struct extent_map_tree *tree,
struct extent_map *em,
int modified)
{
refcount_inc(&em->refs);
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
em->mod_start = em->start;
em->mod_len = em->len;
if (modified)
list_move(&em->list, &tree->modified_extents);
else
try_merge_map(tree, em);
}
static void extent_map_device_set_bits(struct extent_map *em, unsigned bits)
{
struct map_lookup *map = em->map_lookup;
u64 stripe_size = em->orig_block_len;
int i;
for (i = 0; i < map->num_stripes; i++) {
struct btrfs_io_stripe *stripe = &map->stripes[i];
struct btrfs_device *device = stripe->dev;
set_extent_bit(&device->alloc_state, stripe->physical,
stripe->physical + stripe_size - 1,
bits | EXTENT_NOWAIT, NULL);
}
}
static void extent_map_device_clear_bits(struct extent_map *em, unsigned bits)
{
struct map_lookup *map = em->map_lookup;
u64 stripe_size = em->orig_block_len;
int i;
for (i = 0; i < map->num_stripes; i++) {
struct btrfs_io_stripe *stripe = &map->stripes[i];
struct btrfs_device *device = stripe->dev;
__clear_extent_bit(&device->alloc_state, stripe->physical,
stripe->physical + stripe_size - 1,
bits | EXTENT_NOWAIT,
NULL, NULL);
}
}
/*
* Add new extent map to the extent tree
*
* @tree: tree to insert new map in
* @em: map to insert
* @modified: indicate whether the given @em should be added to the
* modified list, which indicates the extent needs to be logged
*
* Insert @em into @tree or perform a simple forward/backward merge with
* existing mappings. The extent_map struct passed in will be inserted
* into the tree directly, with an additional reference taken, or a
* reference dropped if the merge attempt was successful.
*/
int add_extent_mapping(struct extent_map_tree *tree,
2013-04-05 20:51:15 +00:00
struct extent_map *em, int modified)
{
int ret = 0;
lockdep_assert_held_write(&tree->lock);
ret = tree_insert(&tree->map, em);
if (ret)
goto out;
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
setup_extent_mapping(tree, em, modified);
if (test_bit(EXTENT_FLAG_FS_MAPPING, &em->flags)) {
extent_map_device_set_bits(em, CHUNK_ALLOCATED);
extent_map_device_clear_bits(em, CHUNK_TRIMMED);
}
out:
return ret;
}
static struct extent_map *
__lookup_extent_mapping(struct extent_map_tree *tree,
u64 start, u64 len, int strict)
{
struct extent_map *em;
struct rb_node *rb_node;
struct rb_node *prev_or_next = NULL;
u64 end = range_end(start, len);
rb_node = __tree_search(&tree->map.rb_root, start, &prev_or_next);
if (!rb_node) {
if (prev_or_next)
rb_node = prev_or_next;
else
return NULL;
}
em = rb_entry(rb_node, struct extent_map, rb_node);
if (strict && !(end > em->start && start < extent_map_end(em)))
return NULL;
refcount_inc(&em->refs);
return em;
}
/*
* Lookup extent_map that intersects @start + @len range.
*
* @tree: tree to lookup in
* @start: byte offset to start the search
* @len: length of the lookup range
*
* Find and return the first extent_map struct in @tree that intersects the
* [start, len] range. There may be additional objects in the tree that
* intersect, so check the object returned carefully to make sure that no
* additional lookups are needed.
*/
struct extent_map *lookup_extent_mapping(struct extent_map_tree *tree,
u64 start, u64 len)
{
return __lookup_extent_mapping(tree, start, len, 1);
}
/*
* Find a nearby extent map intersecting @start + @len (not an exact search).
*
* @tree: tree to lookup in
* @start: byte offset to start the search
* @len: length of the lookup range
*
* Find and return the first extent_map struct in @tree that intersects the
* [start, len] range.
*
* If one can't be found, any nearby extent may be returned
*/
struct extent_map *search_extent_mapping(struct extent_map_tree *tree,
u64 start, u64 len)
{
return __lookup_extent_mapping(tree, start, len, 0);
}
/*
* Remove an extent_map from the extent tree.
*
* @tree: extent tree to remove from
* @em: extent map being removed
*
* Remove @em from @tree. No reference counts are dropped, and no checks
* are done to see if the range is in use.
*/
void remove_extent_mapping(struct extent_map_tree *tree, struct extent_map *em)
{
lockdep_assert_held_write(&tree->lock);
WARN_ON(test_bit(EXTENT_FLAG_PINNED, &em->flags));
rb_erase_cached(&em->rb_node, &tree->map);
if (!test_bit(EXTENT_FLAG_LOGGING, &em->flags))
list_del_init(&em->list);
if (test_bit(EXTENT_FLAG_FS_MAPPING, &em->flags))
extent_map_device_clear_bits(em, CHUNK_ALLOCATED);
RB_CLEAR_NODE(&em->rb_node);
}
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
static void replace_extent_mapping(struct extent_map_tree *tree,
struct extent_map *cur,
struct extent_map *new,
int modified)
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
{
lockdep_assert_held_write(&tree->lock);
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
WARN_ON(test_bit(EXTENT_FLAG_PINNED, &cur->flags));
ASSERT(extent_map_in_tree(cur));
if (!test_bit(EXTENT_FLAG_LOGGING, &cur->flags))
list_del_init(&cur->list);
rb_replace_node_cached(&cur->rb_node, &new->rb_node, &tree->map);
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
RB_CLEAR_NODE(&cur->rb_node);
setup_extent_mapping(tree, new, modified);
}
static struct extent_map *next_extent_map(const struct extent_map *em)
{
struct rb_node *next;
next = rb_next(&em->rb_node);
if (!next)
return NULL;
return container_of(next, struct extent_map, rb_node);
}
static struct extent_map *prev_extent_map(struct extent_map *em)
{
struct rb_node *prev;
prev = rb_prev(&em->rb_node);
if (!prev)
return NULL;
return container_of(prev, struct extent_map, rb_node);
}
/*
* Helper for btrfs_get_extent. Given an existing extent in the tree,
* the existing extent is the nearest extent to map_start,
* and an extent that you want to insert, deal with overlap and insert
* the best fitted new extent into the tree.
*/
static noinline int merge_extent_mapping(struct extent_map_tree *em_tree,
struct extent_map *existing,
struct extent_map *em,
u64 map_start)
{
struct extent_map *prev;
struct extent_map *next;
u64 start;
u64 end;
u64 start_diff;
BUG_ON(map_start < em->start || map_start >= extent_map_end(em));
if (existing->start > map_start) {
next = existing;
prev = prev_extent_map(next);
} else {
prev = existing;
next = next_extent_map(prev);
}
start = prev ? extent_map_end(prev) : em->start;
start = max_t(u64, start, em->start);
end = next ? next->start : extent_map_end(em);
end = min_t(u64, end, extent_map_end(em));
start_diff = start - em->start;
em->start = start;
em->len = end - start;
if (em->block_start < EXTENT_MAP_LAST_BYTE &&
!test_bit(EXTENT_FLAG_COMPRESSED, &em->flags)) {
em->block_start += start_diff;
em->block_len = em->len;
}
return add_extent_mapping(em_tree, em, 0);
}
/*
* Add extent mapping into em_tree.
*
* @fs_info: the filesystem
* @em_tree: extent tree into which we want to insert the extent mapping
* @em_in: extent we are inserting
* @start: start of the logical range btrfs_get_extent() is requesting
* @len: length of the logical range btrfs_get_extent() is requesting
*
* Note that @em_in's range may be different from [start, start+len),
* but they must be overlapped.
*
* Insert @em_in into @em_tree. In case there is an overlapping range, handle
* the -EEXIST by either:
* a) Returning the existing extent in @em_in if @start is within the
* existing em.
* b) Merge the existing extent with @em_in passed in.
*
* Return 0 on success, otherwise -EEXIST.
*
*/
int btrfs_add_extent_mapping(struct btrfs_fs_info *fs_info,
struct extent_map_tree *em_tree,
struct extent_map **em_in, u64 start, u64 len)
{
int ret;
struct extent_map *em = *em_in;
/*
* Tree-checker should have rejected any inline extent with non-zero
* file offset. Here just do a sanity check.
*/
if (em->block_start == EXTENT_MAP_INLINE)
ASSERT(em->start == 0);
ret = add_extent_mapping(em_tree, em, 0);
/* it is possible that someone inserted the extent into the tree
* while we had the lock dropped. It is also possible that
* an overlapping map exists in the tree
*/
if (ret == -EEXIST) {
struct extent_map *existing;
ret = 0;
existing = search_extent_mapping(em_tree, start, len);
trace_btrfs_handle_em_exist(fs_info, existing, em, start, len);
/*
* existing will always be non-NULL, since there must be
* extent causing the -EEXIST.
*/
if (start >= existing->start &&
start < extent_map_end(existing)) {
free_extent_map(em);
*em_in = existing;
ret = 0;
} else {
u64 orig_start = em->start;
u64 orig_len = em->len;
/*
* The existing extent map is the one nearest to
* the [start, start + len) range which overlaps
*/
ret = merge_extent_mapping(em_tree, existing,
em, start);
if (ret) {
free_extent_map(em);
*em_in = NULL;
WARN_ONCE(ret,
"unexpected error %d: merge existing(start %llu len %llu) with em(start %llu len %llu)\n",
ret, existing->start, existing->len,
orig_start, orig_len);
}
free_extent_map(existing);
}
}
ASSERT(ret == 0 || ret == -EEXIST);
return ret;
}
/*
* Drop all extent maps from a tree in the fastest possible way, rescheduling
* if needed. This avoids searching the tree, from the root down to the first
* extent map, before each deletion.
*/
static void drop_all_extent_maps_fast(struct extent_map_tree *tree)
{
write_lock(&tree->lock);
while (!RB_EMPTY_ROOT(&tree->map.rb_root)) {
struct extent_map *em;
struct rb_node *node;
node = rb_first_cached(&tree->map);
em = rb_entry(node, struct extent_map, rb_node);
clear_bit(EXTENT_FLAG_PINNED, &em->flags);
clear_bit(EXTENT_FLAG_LOGGING, &em->flags);
remove_extent_mapping(tree, em);
free_extent_map(em);
cond_resched_rwlock_write(&tree->lock);
}
write_unlock(&tree->lock);
}
/*
* Drop all extent maps in a given range.
*
* @inode: The target inode.
* @start: Start offset of the range.
* @end: End offset of the range (inclusive value).
* @skip_pinned: Indicate if pinned extent maps should be ignored or not.
*
* This drops all the extent maps that intersect the given range [@start, @end].
* Extent maps that partially overlap the range and extend behind or beyond it,
* are split.
* The caller should have locked an appropriate file range in the inode's io
* tree before calling this function.
*/
void btrfs_drop_extent_map_range(struct btrfs_inode *inode, u64 start, u64 end,
bool skip_pinned)
{
btrfs: drop extent map range more efficiently Currently when dropping extent maps for a file range, through btrfs_drop_extent_map_range(), we do the following non-optimal things: 1) We lookup for extent maps one by one, always starting the search from the root of the extent map tree. This is not efficient if we have multiple extent maps in the range; 2) We check on every iteration if we have the 'split' and 'split2' spare extent maps in case we need to split an extent map that intersects our range but also crosses its boundaries (to the left, to the right or both cases). If our target range is for example: [2M, 8M) And we have 3 extents maps in the range: [1M, 3M) [3M, 6M) [6M, 10M[ The on the first iteration we allocate two extent maps for 'split' and 'split2', and use the 'split' to split the first extent map, so after the split we set 'split' to 'split2' and then set 'split2' to NULL. On the second iteration, we don't need to split the second extent map, but because 'split2' is now NULL, we allocate a new extent map for 'split2'. On the third iteration we need to split the third extent map, so we use the extent map pointed by 'split'. So we ended up allocating 3 extent maps for splitting, but all we needed was 2 extent maps. We never need to allocate more than 2, because extent maps that need to be split are always the first one and the last one in the target range. Improve on this by: 1) Using rb_next() to move on to the next extent map. This results in iterating over less nodes of the tree and it does not require comparing the ranges of nodes to our start/end offset; 2) Allocate the 2 extent maps for splitting before entering the loop and never allocate more than 2. In practice it's very rare to have the combination of both extent map allocations fail, since we have a dedicated slab for extent maps, and also have the need to split two extent maps. This patch is part of a patchset comprised of the following patches: btrfs: fix missed extent on fsync after dropping extent maps btrfs: move btrfs_drop_extent_cache() to extent_map.c btrfs: use extent_map_end() at btrfs_drop_extent_map_range() btrfs: use cond_resched_rwlock_write() during inode eviction btrfs: move open coded extent map tree deletion out of inode eviction btrfs: add helper to replace extent map range with a new extent map btrfs: remove the refcount warning/check at free_extent_map() btrfs: remove unnecessary extent map initializations btrfs: assert tree is locked when clearing extent map from logging btrfs: remove unnecessary NULL pointer checks when searching extent maps btrfs: remove unnecessary next extent map search btrfs: avoid pointless extent map tree search when flushing delalloc btrfs: drop extent map range more efficiently And the following fio test was done before and after applying the whole patchset, on a non-debug kernel (Debian's default kernel config) on a 12 cores Intel box with 64G of ram: $ cat test.sh #!/bin/bash DEV=/dev/nvme0n1 MNT=/mnt/nvme0n1 MOUNT_OPTIONS="-o ssd" MKFS_OPTIONS="-R free-space-tree -O no-holes" cat <<EOF > /tmp/fio-job.ini [writers] rw=randwrite fsync=8 fallocate=none group_reporting=1 direct=0 bssplit=4k/20:8k/20:16k/20:32k/10:64k/10:128k/5:256k/5:512k/5:1m/5 ioengine=psync filesize=2G runtime=300 time_based directory=$MNT numjobs=8 thread EOF echo performance | \ tee /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor echo echo "Using config:" echo cat /tmp/fio-job.ini echo umount $MNT &> /dev/null mkfs.btrfs -f $MKFS_OPTIONS $DEV mount $MOUNT_OPTIONS $DEV $MNT fio /tmp/fio-job.ini umount $MNT Result before applying the patchset: WRITE: bw=197MiB/s (206MB/s), 197MiB/s-197MiB/s (206MB/s-206MB/s), io=57.7GiB (61.9GB), run=300188-300188msec Result after applying the patchset: WRITE: bw=203MiB/s (213MB/s), 203MiB/s-203MiB/s (213MB/s-213MB/s), io=59.5GiB (63.9GB), run=300019-300019msec Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: David Sterba <dsterba@suse.com>
2022-09-19 14:06:40 +00:00
struct extent_map *split;
struct extent_map *split2;
struct extent_map *em;
struct extent_map_tree *em_tree = &inode->extent_tree;
u64 len = end - start + 1;
WARN_ON(end < start);
if (end == (u64)-1) {
if (start == 0 && !skip_pinned) {
drop_all_extent_maps_fast(em_tree);
return;
}
len = (u64)-1;
btrfs: drop extent map range more efficiently Currently when dropping extent maps for a file range, through btrfs_drop_extent_map_range(), we do the following non-optimal things: 1) We lookup for extent maps one by one, always starting the search from the root of the extent map tree. This is not efficient if we have multiple extent maps in the range; 2) We check on every iteration if we have the 'split' and 'split2' spare extent maps in case we need to split an extent map that intersects our range but also crosses its boundaries (to the left, to the right or both cases). If our target range is for example: [2M, 8M) And we have 3 extents maps in the range: [1M, 3M) [3M, 6M) [6M, 10M[ The on the first iteration we allocate two extent maps for 'split' and 'split2', and use the 'split' to split the first extent map, so after the split we set 'split' to 'split2' and then set 'split2' to NULL. On the second iteration, we don't need to split the second extent map, but because 'split2' is now NULL, we allocate a new extent map for 'split2'. On the third iteration we need to split the third extent map, so we use the extent map pointed by 'split'. So we ended up allocating 3 extent maps for splitting, but all we needed was 2 extent maps. We never need to allocate more than 2, because extent maps that need to be split are always the first one and the last one in the target range. Improve on this by: 1) Using rb_next() to move on to the next extent map. This results in iterating over less nodes of the tree and it does not require comparing the ranges of nodes to our start/end offset; 2) Allocate the 2 extent maps for splitting before entering the loop and never allocate more than 2. In practice it's very rare to have the combination of both extent map allocations fail, since we have a dedicated slab for extent maps, and also have the need to split two extent maps. This patch is part of a patchset comprised of the following patches: btrfs: fix missed extent on fsync after dropping extent maps btrfs: move btrfs_drop_extent_cache() to extent_map.c btrfs: use extent_map_end() at btrfs_drop_extent_map_range() btrfs: use cond_resched_rwlock_write() during inode eviction btrfs: move open coded extent map tree deletion out of inode eviction btrfs: add helper to replace extent map range with a new extent map btrfs: remove the refcount warning/check at free_extent_map() btrfs: remove unnecessary extent map initializations btrfs: assert tree is locked when clearing extent map from logging btrfs: remove unnecessary NULL pointer checks when searching extent maps btrfs: remove unnecessary next extent map search btrfs: avoid pointless extent map tree search when flushing delalloc btrfs: drop extent map range more efficiently And the following fio test was done before and after applying the whole patchset, on a non-debug kernel (Debian's default kernel config) on a 12 cores Intel box with 64G of ram: $ cat test.sh #!/bin/bash DEV=/dev/nvme0n1 MNT=/mnt/nvme0n1 MOUNT_OPTIONS="-o ssd" MKFS_OPTIONS="-R free-space-tree -O no-holes" cat <<EOF > /tmp/fio-job.ini [writers] rw=randwrite fsync=8 fallocate=none group_reporting=1 direct=0 bssplit=4k/20:8k/20:16k/20:32k/10:64k/10:128k/5:256k/5:512k/5:1m/5 ioengine=psync filesize=2G runtime=300 time_based directory=$MNT numjobs=8 thread EOF echo performance | \ tee /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor echo echo "Using config:" echo cat /tmp/fio-job.ini echo umount $MNT &> /dev/null mkfs.btrfs -f $MKFS_OPTIONS $DEV mount $MOUNT_OPTIONS $DEV $MNT fio /tmp/fio-job.ini umount $MNT Result before applying the patchset: WRITE: bw=197MiB/s (206MB/s), 197MiB/s-197MiB/s (206MB/s-206MB/s), io=57.7GiB (61.9GB), run=300188-300188msec Result after applying the patchset: WRITE: bw=203MiB/s (213MB/s), 203MiB/s-203MiB/s (213MB/s-213MB/s), io=59.5GiB (63.9GB), run=300019-300019msec Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: David Sterba <dsterba@suse.com>
2022-09-19 14:06:40 +00:00
} else {
/* Make end offset exclusive for use in the loop below. */
end++;
}
btrfs: drop extent map range more efficiently Currently when dropping extent maps for a file range, through btrfs_drop_extent_map_range(), we do the following non-optimal things: 1) We lookup for extent maps one by one, always starting the search from the root of the extent map tree. This is not efficient if we have multiple extent maps in the range; 2) We check on every iteration if we have the 'split' and 'split2' spare extent maps in case we need to split an extent map that intersects our range but also crosses its boundaries (to the left, to the right or both cases). If our target range is for example: [2M, 8M) And we have 3 extents maps in the range: [1M, 3M) [3M, 6M) [6M, 10M[ The on the first iteration we allocate two extent maps for 'split' and 'split2', and use the 'split' to split the first extent map, so after the split we set 'split' to 'split2' and then set 'split2' to NULL. On the second iteration, we don't need to split the second extent map, but because 'split2' is now NULL, we allocate a new extent map for 'split2'. On the third iteration we need to split the third extent map, so we use the extent map pointed by 'split'. So we ended up allocating 3 extent maps for splitting, but all we needed was 2 extent maps. We never need to allocate more than 2, because extent maps that need to be split are always the first one and the last one in the target range. Improve on this by: 1) Using rb_next() to move on to the next extent map. This results in iterating over less nodes of the tree and it does not require comparing the ranges of nodes to our start/end offset; 2) Allocate the 2 extent maps for splitting before entering the loop and never allocate more than 2. In practice it's very rare to have the combination of both extent map allocations fail, since we have a dedicated slab for extent maps, and also have the need to split two extent maps. This patch is part of a patchset comprised of the following patches: btrfs: fix missed extent on fsync after dropping extent maps btrfs: move btrfs_drop_extent_cache() to extent_map.c btrfs: use extent_map_end() at btrfs_drop_extent_map_range() btrfs: use cond_resched_rwlock_write() during inode eviction btrfs: move open coded extent map tree deletion out of inode eviction btrfs: add helper to replace extent map range with a new extent map btrfs: remove the refcount warning/check at free_extent_map() btrfs: remove unnecessary extent map initializations btrfs: assert tree is locked when clearing extent map from logging btrfs: remove unnecessary NULL pointer checks when searching extent maps btrfs: remove unnecessary next extent map search btrfs: avoid pointless extent map tree search when flushing delalloc btrfs: drop extent map range more efficiently And the following fio test was done before and after applying the whole patchset, on a non-debug kernel (Debian's default kernel config) on a 12 cores Intel box with 64G of ram: $ cat test.sh #!/bin/bash DEV=/dev/nvme0n1 MNT=/mnt/nvme0n1 MOUNT_OPTIONS="-o ssd" MKFS_OPTIONS="-R free-space-tree -O no-holes" cat <<EOF > /tmp/fio-job.ini [writers] rw=randwrite fsync=8 fallocate=none group_reporting=1 direct=0 bssplit=4k/20:8k/20:16k/20:32k/10:64k/10:128k/5:256k/5:512k/5:1m/5 ioengine=psync filesize=2G runtime=300 time_based directory=$MNT numjobs=8 thread EOF echo performance | \ tee /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor echo echo "Using config:" echo cat /tmp/fio-job.ini echo umount $MNT &> /dev/null mkfs.btrfs -f $MKFS_OPTIONS $DEV mount $MOUNT_OPTIONS $DEV $MNT fio /tmp/fio-job.ini umount $MNT Result before applying the patchset: WRITE: bw=197MiB/s (206MB/s), 197MiB/s-197MiB/s (206MB/s-206MB/s), io=57.7GiB (61.9GB), run=300188-300188msec Result after applying the patchset: WRITE: bw=203MiB/s (213MB/s), 203MiB/s-203MiB/s (213MB/s-213MB/s), io=59.5GiB (63.9GB), run=300019-300019msec Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: David Sterba <dsterba@suse.com>
2022-09-19 14:06:40 +00:00
/*
* It's ok if we fail to allocate the extent maps, see the comment near
* the bottom of the loop below. We only need two spare extent maps in
* the worst case, where the first extent map that intersects our range
* starts before the range and the last extent map that intersects our
* range ends after our range (and they might be the same extent map),
* because we need to split those two extent maps at the boundaries.
*/
split = alloc_extent_map();
split2 = alloc_extent_map();
write_lock(&em_tree->lock);
em = lookup_extent_mapping(em_tree, start, len);
while (em) {
/* extent_map_end() returns exclusive value (last byte + 1). */
const u64 em_end = extent_map_end(em);
struct extent_map *next_em = NULL;
u64 gen;
unsigned long flags;
bool modified;
bool compressed;
btrfs: drop extent map range more efficiently Currently when dropping extent maps for a file range, through btrfs_drop_extent_map_range(), we do the following non-optimal things: 1) We lookup for extent maps one by one, always starting the search from the root of the extent map tree. This is not efficient if we have multiple extent maps in the range; 2) We check on every iteration if we have the 'split' and 'split2' spare extent maps in case we need to split an extent map that intersects our range but also crosses its boundaries (to the left, to the right or both cases). If our target range is for example: [2M, 8M) And we have 3 extents maps in the range: [1M, 3M) [3M, 6M) [6M, 10M[ The on the first iteration we allocate two extent maps for 'split' and 'split2', and use the 'split' to split the first extent map, so after the split we set 'split' to 'split2' and then set 'split2' to NULL. On the second iteration, we don't need to split the second extent map, but because 'split2' is now NULL, we allocate a new extent map for 'split2'. On the third iteration we need to split the third extent map, so we use the extent map pointed by 'split'. So we ended up allocating 3 extent maps for splitting, but all we needed was 2 extent maps. We never need to allocate more than 2, because extent maps that need to be split are always the first one and the last one in the target range. Improve on this by: 1) Using rb_next() to move on to the next extent map. This results in iterating over less nodes of the tree and it does not require comparing the ranges of nodes to our start/end offset; 2) Allocate the 2 extent maps for splitting before entering the loop and never allocate more than 2. In practice it's very rare to have the combination of both extent map allocations fail, since we have a dedicated slab for extent maps, and also have the need to split two extent maps. This patch is part of a patchset comprised of the following patches: btrfs: fix missed extent on fsync after dropping extent maps btrfs: move btrfs_drop_extent_cache() to extent_map.c btrfs: use extent_map_end() at btrfs_drop_extent_map_range() btrfs: use cond_resched_rwlock_write() during inode eviction btrfs: move open coded extent map tree deletion out of inode eviction btrfs: add helper to replace extent map range with a new extent map btrfs: remove the refcount warning/check at free_extent_map() btrfs: remove unnecessary extent map initializations btrfs: assert tree is locked when clearing extent map from logging btrfs: remove unnecessary NULL pointer checks when searching extent maps btrfs: remove unnecessary next extent map search btrfs: avoid pointless extent map tree search when flushing delalloc btrfs: drop extent map range more efficiently And the following fio test was done before and after applying the whole patchset, on a non-debug kernel (Debian's default kernel config) on a 12 cores Intel box with 64G of ram: $ cat test.sh #!/bin/bash DEV=/dev/nvme0n1 MNT=/mnt/nvme0n1 MOUNT_OPTIONS="-o ssd" MKFS_OPTIONS="-R free-space-tree -O no-holes" cat <<EOF > /tmp/fio-job.ini [writers] rw=randwrite fsync=8 fallocate=none group_reporting=1 direct=0 bssplit=4k/20:8k/20:16k/20:32k/10:64k/10:128k/5:256k/5:512k/5:1m/5 ioengine=psync filesize=2G runtime=300 time_based directory=$MNT numjobs=8 thread EOF echo performance | \ tee /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor echo echo "Using config:" echo cat /tmp/fio-job.ini echo umount $MNT &> /dev/null mkfs.btrfs -f $MKFS_OPTIONS $DEV mount $MOUNT_OPTIONS $DEV $MNT fio /tmp/fio-job.ini umount $MNT Result before applying the patchset: WRITE: bw=197MiB/s (206MB/s), 197MiB/s-197MiB/s (206MB/s-206MB/s), io=57.7GiB (61.9GB), run=300188-300188msec Result after applying the patchset: WRITE: bw=203MiB/s (213MB/s), 203MiB/s-203MiB/s (213MB/s-213MB/s), io=59.5GiB (63.9GB), run=300019-300019msec Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: David Sterba <dsterba@suse.com>
2022-09-19 14:06:40 +00:00
if (em_end < end) {
next_em = next_extent_map(em);
if (next_em) {
if (next_em->start < end)
refcount_inc(&next_em->refs);
else
next_em = NULL;
}
}
btrfs: drop extent map range more efficiently Currently when dropping extent maps for a file range, through btrfs_drop_extent_map_range(), we do the following non-optimal things: 1) We lookup for extent maps one by one, always starting the search from the root of the extent map tree. This is not efficient if we have multiple extent maps in the range; 2) We check on every iteration if we have the 'split' and 'split2' spare extent maps in case we need to split an extent map that intersects our range but also crosses its boundaries (to the left, to the right or both cases). If our target range is for example: [2M, 8M) And we have 3 extents maps in the range: [1M, 3M) [3M, 6M) [6M, 10M[ The on the first iteration we allocate two extent maps for 'split' and 'split2', and use the 'split' to split the first extent map, so after the split we set 'split' to 'split2' and then set 'split2' to NULL. On the second iteration, we don't need to split the second extent map, but because 'split2' is now NULL, we allocate a new extent map for 'split2'. On the third iteration we need to split the third extent map, so we use the extent map pointed by 'split'. So we ended up allocating 3 extent maps for splitting, but all we needed was 2 extent maps. We never need to allocate more than 2, because extent maps that need to be split are always the first one and the last one in the target range. Improve on this by: 1) Using rb_next() to move on to the next extent map. This results in iterating over less nodes of the tree and it does not require comparing the ranges of nodes to our start/end offset; 2) Allocate the 2 extent maps for splitting before entering the loop and never allocate more than 2. In practice it's very rare to have the combination of both extent map allocations fail, since we have a dedicated slab for extent maps, and also have the need to split two extent maps. This patch is part of a patchset comprised of the following patches: btrfs: fix missed extent on fsync after dropping extent maps btrfs: move btrfs_drop_extent_cache() to extent_map.c btrfs: use extent_map_end() at btrfs_drop_extent_map_range() btrfs: use cond_resched_rwlock_write() during inode eviction btrfs: move open coded extent map tree deletion out of inode eviction btrfs: add helper to replace extent map range with a new extent map btrfs: remove the refcount warning/check at free_extent_map() btrfs: remove unnecessary extent map initializations btrfs: assert tree is locked when clearing extent map from logging btrfs: remove unnecessary NULL pointer checks when searching extent maps btrfs: remove unnecessary next extent map search btrfs: avoid pointless extent map tree search when flushing delalloc btrfs: drop extent map range more efficiently And the following fio test was done before and after applying the whole patchset, on a non-debug kernel (Debian's default kernel config) on a 12 cores Intel box with 64G of ram: $ cat test.sh #!/bin/bash DEV=/dev/nvme0n1 MNT=/mnt/nvme0n1 MOUNT_OPTIONS="-o ssd" MKFS_OPTIONS="-R free-space-tree -O no-holes" cat <<EOF > /tmp/fio-job.ini [writers] rw=randwrite fsync=8 fallocate=none group_reporting=1 direct=0 bssplit=4k/20:8k/20:16k/20:32k/10:64k/10:128k/5:256k/5:512k/5:1m/5 ioengine=psync filesize=2G runtime=300 time_based directory=$MNT numjobs=8 thread EOF echo performance | \ tee /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor echo echo "Using config:" echo cat /tmp/fio-job.ini echo umount $MNT &> /dev/null mkfs.btrfs -f $MKFS_OPTIONS $DEV mount $MOUNT_OPTIONS $DEV $MNT fio /tmp/fio-job.ini umount $MNT Result before applying the patchset: WRITE: bw=197MiB/s (206MB/s), 197MiB/s-197MiB/s (206MB/s-206MB/s), io=57.7GiB (61.9GB), run=300188-300188msec Result after applying the patchset: WRITE: bw=203MiB/s (213MB/s), 203MiB/s-203MiB/s (213MB/s-213MB/s), io=59.5GiB (63.9GB), run=300019-300019msec Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: David Sterba <dsterba@suse.com>
2022-09-19 14:06:40 +00:00
if (skip_pinned && test_bit(EXTENT_FLAG_PINNED, &em->flags)) {
start = em_end;
btrfs: drop extent map range more efficiently Currently when dropping extent maps for a file range, through btrfs_drop_extent_map_range(), we do the following non-optimal things: 1) We lookup for extent maps one by one, always starting the search from the root of the extent map tree. This is not efficient if we have multiple extent maps in the range; 2) We check on every iteration if we have the 'split' and 'split2' spare extent maps in case we need to split an extent map that intersects our range but also crosses its boundaries (to the left, to the right or both cases). If our target range is for example: [2M, 8M) And we have 3 extents maps in the range: [1M, 3M) [3M, 6M) [6M, 10M[ The on the first iteration we allocate two extent maps for 'split' and 'split2', and use the 'split' to split the first extent map, so after the split we set 'split' to 'split2' and then set 'split2' to NULL. On the second iteration, we don't need to split the second extent map, but because 'split2' is now NULL, we allocate a new extent map for 'split2'. On the third iteration we need to split the third extent map, so we use the extent map pointed by 'split'. So we ended up allocating 3 extent maps for splitting, but all we needed was 2 extent maps. We never need to allocate more than 2, because extent maps that need to be split are always the first one and the last one in the target range. Improve on this by: 1) Using rb_next() to move on to the next extent map. This results in iterating over less nodes of the tree and it does not require comparing the ranges of nodes to our start/end offset; 2) Allocate the 2 extent maps for splitting before entering the loop and never allocate more than 2. In practice it's very rare to have the combination of both extent map allocations fail, since we have a dedicated slab for extent maps, and also have the need to split two extent maps. This patch is part of a patchset comprised of the following patches: btrfs: fix missed extent on fsync after dropping extent maps btrfs: move btrfs_drop_extent_cache() to extent_map.c btrfs: use extent_map_end() at btrfs_drop_extent_map_range() btrfs: use cond_resched_rwlock_write() during inode eviction btrfs: move open coded extent map tree deletion out of inode eviction btrfs: add helper to replace extent map range with a new extent map btrfs: remove the refcount warning/check at free_extent_map() btrfs: remove unnecessary extent map initializations btrfs: assert tree is locked when clearing extent map from logging btrfs: remove unnecessary NULL pointer checks when searching extent maps btrfs: remove unnecessary next extent map search btrfs: avoid pointless extent map tree search when flushing delalloc btrfs: drop extent map range more efficiently And the following fio test was done before and after applying the whole patchset, on a non-debug kernel (Debian's default kernel config) on a 12 cores Intel box with 64G of ram: $ cat test.sh #!/bin/bash DEV=/dev/nvme0n1 MNT=/mnt/nvme0n1 MOUNT_OPTIONS="-o ssd" MKFS_OPTIONS="-R free-space-tree -O no-holes" cat <<EOF > /tmp/fio-job.ini [writers] rw=randwrite fsync=8 fallocate=none group_reporting=1 direct=0 bssplit=4k/20:8k/20:16k/20:32k/10:64k/10:128k/5:256k/5:512k/5:1m/5 ioengine=psync filesize=2G runtime=300 time_based directory=$MNT numjobs=8 thread EOF echo performance | \ tee /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor echo echo "Using config:" echo cat /tmp/fio-job.ini echo umount $MNT &> /dev/null mkfs.btrfs -f $MKFS_OPTIONS $DEV mount $MOUNT_OPTIONS $DEV $MNT fio /tmp/fio-job.ini umount $MNT Result before applying the patchset: WRITE: bw=197MiB/s (206MB/s), 197MiB/s-197MiB/s (206MB/s-206MB/s), io=57.7GiB (61.9GB), run=300188-300188msec Result after applying the patchset: WRITE: bw=203MiB/s (213MB/s), 203MiB/s-203MiB/s (213MB/s-213MB/s), io=59.5GiB (63.9GB), run=300019-300019msec Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: David Sterba <dsterba@suse.com>
2022-09-19 14:06:40 +00:00
goto next;
}
btrfs: drop extent map range more efficiently Currently when dropping extent maps for a file range, through btrfs_drop_extent_map_range(), we do the following non-optimal things: 1) We lookup for extent maps one by one, always starting the search from the root of the extent map tree. This is not efficient if we have multiple extent maps in the range; 2) We check on every iteration if we have the 'split' and 'split2' spare extent maps in case we need to split an extent map that intersects our range but also crosses its boundaries (to the left, to the right or both cases). If our target range is for example: [2M, 8M) And we have 3 extents maps in the range: [1M, 3M) [3M, 6M) [6M, 10M[ The on the first iteration we allocate two extent maps for 'split' and 'split2', and use the 'split' to split the first extent map, so after the split we set 'split' to 'split2' and then set 'split2' to NULL. On the second iteration, we don't need to split the second extent map, but because 'split2' is now NULL, we allocate a new extent map for 'split2'. On the third iteration we need to split the third extent map, so we use the extent map pointed by 'split'. So we ended up allocating 3 extent maps for splitting, but all we needed was 2 extent maps. We never need to allocate more than 2, because extent maps that need to be split are always the first one and the last one in the target range. Improve on this by: 1) Using rb_next() to move on to the next extent map. This results in iterating over less nodes of the tree and it does not require comparing the ranges of nodes to our start/end offset; 2) Allocate the 2 extent maps for splitting before entering the loop and never allocate more than 2. In practice it's very rare to have the combination of both extent map allocations fail, since we have a dedicated slab for extent maps, and also have the need to split two extent maps. This patch is part of a patchset comprised of the following patches: btrfs: fix missed extent on fsync after dropping extent maps btrfs: move btrfs_drop_extent_cache() to extent_map.c btrfs: use extent_map_end() at btrfs_drop_extent_map_range() btrfs: use cond_resched_rwlock_write() during inode eviction btrfs: move open coded extent map tree deletion out of inode eviction btrfs: add helper to replace extent map range with a new extent map btrfs: remove the refcount warning/check at free_extent_map() btrfs: remove unnecessary extent map initializations btrfs: assert tree is locked when clearing extent map from logging btrfs: remove unnecessary NULL pointer checks when searching extent maps btrfs: remove unnecessary next extent map search btrfs: avoid pointless extent map tree search when flushing delalloc btrfs: drop extent map range more efficiently And the following fio test was done before and after applying the whole patchset, on a non-debug kernel (Debian's default kernel config) on a 12 cores Intel box with 64G of ram: $ cat test.sh #!/bin/bash DEV=/dev/nvme0n1 MNT=/mnt/nvme0n1 MOUNT_OPTIONS="-o ssd" MKFS_OPTIONS="-R free-space-tree -O no-holes" cat <<EOF > /tmp/fio-job.ini [writers] rw=randwrite fsync=8 fallocate=none group_reporting=1 direct=0 bssplit=4k/20:8k/20:16k/20:32k/10:64k/10:128k/5:256k/5:512k/5:1m/5 ioengine=psync filesize=2G runtime=300 time_based directory=$MNT numjobs=8 thread EOF echo performance | \ tee /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor echo echo "Using config:" echo cat /tmp/fio-job.ini echo umount $MNT &> /dev/null mkfs.btrfs -f $MKFS_OPTIONS $DEV mount $MOUNT_OPTIONS $DEV $MNT fio /tmp/fio-job.ini umount $MNT Result before applying the patchset: WRITE: bw=197MiB/s (206MB/s), 197MiB/s-197MiB/s (206MB/s-206MB/s), io=57.7GiB (61.9GB), run=300188-300188msec Result after applying the patchset: WRITE: bw=203MiB/s (213MB/s), 203MiB/s-203MiB/s (213MB/s-213MB/s), io=59.5GiB (63.9GB), run=300019-300019msec Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: David Sterba <dsterba@suse.com>
2022-09-19 14:06:40 +00:00
flags = em->flags;
clear_bit(EXTENT_FLAG_PINNED, &em->flags);
/*
* In case we split the extent map, we want to preserve the
* EXTENT_FLAG_LOGGING flag on our extent map, but we don't want
* it on the new extent maps.
*/
clear_bit(EXTENT_FLAG_LOGGING, &flags);
modified = !list_empty(&em->list);
btrfs: drop extent map range more efficiently Currently when dropping extent maps for a file range, through btrfs_drop_extent_map_range(), we do the following non-optimal things: 1) We lookup for extent maps one by one, always starting the search from the root of the extent map tree. This is not efficient if we have multiple extent maps in the range; 2) We check on every iteration if we have the 'split' and 'split2' spare extent maps in case we need to split an extent map that intersects our range but also crosses its boundaries (to the left, to the right or both cases). If our target range is for example: [2M, 8M) And we have 3 extents maps in the range: [1M, 3M) [3M, 6M) [6M, 10M[ The on the first iteration we allocate two extent maps for 'split' and 'split2', and use the 'split' to split the first extent map, so after the split we set 'split' to 'split2' and then set 'split2' to NULL. On the second iteration, we don't need to split the second extent map, but because 'split2' is now NULL, we allocate a new extent map for 'split2'. On the third iteration we need to split the third extent map, so we use the extent map pointed by 'split'. So we ended up allocating 3 extent maps for splitting, but all we needed was 2 extent maps. We never need to allocate more than 2, because extent maps that need to be split are always the first one and the last one in the target range. Improve on this by: 1) Using rb_next() to move on to the next extent map. This results in iterating over less nodes of the tree and it does not require comparing the ranges of nodes to our start/end offset; 2) Allocate the 2 extent maps for splitting before entering the loop and never allocate more than 2. In practice it's very rare to have the combination of both extent map allocations fail, since we have a dedicated slab for extent maps, and also have the need to split two extent maps. This patch is part of a patchset comprised of the following patches: btrfs: fix missed extent on fsync after dropping extent maps btrfs: move btrfs_drop_extent_cache() to extent_map.c btrfs: use extent_map_end() at btrfs_drop_extent_map_range() btrfs: use cond_resched_rwlock_write() during inode eviction btrfs: move open coded extent map tree deletion out of inode eviction btrfs: add helper to replace extent map range with a new extent map btrfs: remove the refcount warning/check at free_extent_map() btrfs: remove unnecessary extent map initializations btrfs: assert tree is locked when clearing extent map from logging btrfs: remove unnecessary NULL pointer checks when searching extent maps btrfs: remove unnecessary next extent map search btrfs: avoid pointless extent map tree search when flushing delalloc btrfs: drop extent map range more efficiently And the following fio test was done before and after applying the whole patchset, on a non-debug kernel (Debian's default kernel config) on a 12 cores Intel box with 64G of ram: $ cat test.sh #!/bin/bash DEV=/dev/nvme0n1 MNT=/mnt/nvme0n1 MOUNT_OPTIONS="-o ssd" MKFS_OPTIONS="-R free-space-tree -O no-holes" cat <<EOF > /tmp/fio-job.ini [writers] rw=randwrite fsync=8 fallocate=none group_reporting=1 direct=0 bssplit=4k/20:8k/20:16k/20:32k/10:64k/10:128k/5:256k/5:512k/5:1m/5 ioengine=psync filesize=2G runtime=300 time_based directory=$MNT numjobs=8 thread EOF echo performance | \ tee /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor echo echo "Using config:" echo cat /tmp/fio-job.ini echo umount $MNT &> /dev/null mkfs.btrfs -f $MKFS_OPTIONS $DEV mount $MOUNT_OPTIONS $DEV $MNT fio /tmp/fio-job.ini umount $MNT Result before applying the patchset: WRITE: bw=197MiB/s (206MB/s), 197MiB/s-197MiB/s (206MB/s-206MB/s), io=57.7GiB (61.9GB), run=300188-300188msec Result after applying the patchset: WRITE: bw=203MiB/s (213MB/s), 203MiB/s-203MiB/s (213MB/s-213MB/s), io=59.5GiB (63.9GB), run=300019-300019msec Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: David Sterba <dsterba@suse.com>
2022-09-19 14:06:40 +00:00
/*
* The extent map does not cross our target range, so no need to
* split it, we can remove it directly.
*/
if (em->start >= start && em_end <= end)
goto remove_em;
gen = em->generation;
compressed = test_bit(EXTENT_FLAG_COMPRESSED, &em->flags);
if (em->start < start) {
btrfs: drop extent map range more efficiently Currently when dropping extent maps for a file range, through btrfs_drop_extent_map_range(), we do the following non-optimal things: 1) We lookup for extent maps one by one, always starting the search from the root of the extent map tree. This is not efficient if we have multiple extent maps in the range; 2) We check on every iteration if we have the 'split' and 'split2' spare extent maps in case we need to split an extent map that intersects our range but also crosses its boundaries (to the left, to the right or both cases). If our target range is for example: [2M, 8M) And we have 3 extents maps in the range: [1M, 3M) [3M, 6M) [6M, 10M[ The on the first iteration we allocate two extent maps for 'split' and 'split2', and use the 'split' to split the first extent map, so after the split we set 'split' to 'split2' and then set 'split2' to NULL. On the second iteration, we don't need to split the second extent map, but because 'split2' is now NULL, we allocate a new extent map for 'split2'. On the third iteration we need to split the third extent map, so we use the extent map pointed by 'split'. So we ended up allocating 3 extent maps for splitting, but all we needed was 2 extent maps. We never need to allocate more than 2, because extent maps that need to be split are always the first one and the last one in the target range. Improve on this by: 1) Using rb_next() to move on to the next extent map. This results in iterating over less nodes of the tree and it does not require comparing the ranges of nodes to our start/end offset; 2) Allocate the 2 extent maps for splitting before entering the loop and never allocate more than 2. In practice it's very rare to have the combination of both extent map allocations fail, since we have a dedicated slab for extent maps, and also have the need to split two extent maps. This patch is part of a patchset comprised of the following patches: btrfs: fix missed extent on fsync after dropping extent maps btrfs: move btrfs_drop_extent_cache() to extent_map.c btrfs: use extent_map_end() at btrfs_drop_extent_map_range() btrfs: use cond_resched_rwlock_write() during inode eviction btrfs: move open coded extent map tree deletion out of inode eviction btrfs: add helper to replace extent map range with a new extent map btrfs: remove the refcount warning/check at free_extent_map() btrfs: remove unnecessary extent map initializations btrfs: assert tree is locked when clearing extent map from logging btrfs: remove unnecessary NULL pointer checks when searching extent maps btrfs: remove unnecessary next extent map search btrfs: avoid pointless extent map tree search when flushing delalloc btrfs: drop extent map range more efficiently And the following fio test was done before and after applying the whole patchset, on a non-debug kernel (Debian's default kernel config) on a 12 cores Intel box with 64G of ram: $ cat test.sh #!/bin/bash DEV=/dev/nvme0n1 MNT=/mnt/nvme0n1 MOUNT_OPTIONS="-o ssd" MKFS_OPTIONS="-R free-space-tree -O no-holes" cat <<EOF > /tmp/fio-job.ini [writers] rw=randwrite fsync=8 fallocate=none group_reporting=1 direct=0 bssplit=4k/20:8k/20:16k/20:32k/10:64k/10:128k/5:256k/5:512k/5:1m/5 ioengine=psync filesize=2G runtime=300 time_based directory=$MNT numjobs=8 thread EOF echo performance | \ tee /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor echo echo "Using config:" echo cat /tmp/fio-job.ini echo umount $MNT &> /dev/null mkfs.btrfs -f $MKFS_OPTIONS $DEV mount $MOUNT_OPTIONS $DEV $MNT fio /tmp/fio-job.ini umount $MNT Result before applying the patchset: WRITE: bw=197MiB/s (206MB/s), 197MiB/s-197MiB/s (206MB/s-206MB/s), io=57.7GiB (61.9GB), run=300188-300188msec Result after applying the patchset: WRITE: bw=203MiB/s (213MB/s), 203MiB/s-203MiB/s (213MB/s-213MB/s), io=59.5GiB (63.9GB), run=300019-300019msec Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: David Sterba <dsterba@suse.com>
2022-09-19 14:06:40 +00:00
if (!split) {
split = split2;
split2 = NULL;
if (!split)
goto remove_em;
}
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;
}
split->generation = gen;
split->flags = flags;
split->compress_type = em->compress_type;
replace_extent_mapping(em_tree, em, split, modified);
free_extent_map(split);
split = split2;
split2 = NULL;
}
btrfs: drop extent map range more efficiently Currently when dropping extent maps for a file range, through btrfs_drop_extent_map_range(), we do the following non-optimal things: 1) We lookup for extent maps one by one, always starting the search from the root of the extent map tree. This is not efficient if we have multiple extent maps in the range; 2) We check on every iteration if we have the 'split' and 'split2' spare extent maps in case we need to split an extent map that intersects our range but also crosses its boundaries (to the left, to the right or both cases). If our target range is for example: [2M, 8M) And we have 3 extents maps in the range: [1M, 3M) [3M, 6M) [6M, 10M[ The on the first iteration we allocate two extent maps for 'split' and 'split2', and use the 'split' to split the first extent map, so after the split we set 'split' to 'split2' and then set 'split2' to NULL. On the second iteration, we don't need to split the second extent map, but because 'split2' is now NULL, we allocate a new extent map for 'split2'. On the third iteration we need to split the third extent map, so we use the extent map pointed by 'split'. So we ended up allocating 3 extent maps for splitting, but all we needed was 2 extent maps. We never need to allocate more than 2, because extent maps that need to be split are always the first one and the last one in the target range. Improve on this by: 1) Using rb_next() to move on to the next extent map. This results in iterating over less nodes of the tree and it does not require comparing the ranges of nodes to our start/end offset; 2) Allocate the 2 extent maps for splitting before entering the loop and never allocate more than 2. In practice it's very rare to have the combination of both extent map allocations fail, since we have a dedicated slab for extent maps, and also have the need to split two extent maps. This patch is part of a patchset comprised of the following patches: btrfs: fix missed extent on fsync after dropping extent maps btrfs: move btrfs_drop_extent_cache() to extent_map.c btrfs: use extent_map_end() at btrfs_drop_extent_map_range() btrfs: use cond_resched_rwlock_write() during inode eviction btrfs: move open coded extent map tree deletion out of inode eviction btrfs: add helper to replace extent map range with a new extent map btrfs: remove the refcount warning/check at free_extent_map() btrfs: remove unnecessary extent map initializations btrfs: assert tree is locked when clearing extent map from logging btrfs: remove unnecessary NULL pointer checks when searching extent maps btrfs: remove unnecessary next extent map search btrfs: avoid pointless extent map tree search when flushing delalloc btrfs: drop extent map range more efficiently And the following fio test was done before and after applying the whole patchset, on a non-debug kernel (Debian's default kernel config) on a 12 cores Intel box with 64G of ram: $ cat test.sh #!/bin/bash DEV=/dev/nvme0n1 MNT=/mnt/nvme0n1 MOUNT_OPTIONS="-o ssd" MKFS_OPTIONS="-R free-space-tree -O no-holes" cat <<EOF > /tmp/fio-job.ini [writers] rw=randwrite fsync=8 fallocate=none group_reporting=1 direct=0 bssplit=4k/20:8k/20:16k/20:32k/10:64k/10:128k/5:256k/5:512k/5:1m/5 ioengine=psync filesize=2G runtime=300 time_based directory=$MNT numjobs=8 thread EOF echo performance | \ tee /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor echo echo "Using config:" echo cat /tmp/fio-job.ini echo umount $MNT &> /dev/null mkfs.btrfs -f $MKFS_OPTIONS $DEV mount $MOUNT_OPTIONS $DEV $MNT fio /tmp/fio-job.ini umount $MNT Result before applying the patchset: WRITE: bw=197MiB/s (206MB/s), 197MiB/s-197MiB/s (206MB/s-206MB/s), io=57.7GiB (61.9GB), run=300188-300188msec Result after applying the patchset: WRITE: bw=203MiB/s (213MB/s), 203MiB/s-203MiB/s (213MB/s-213MB/s), io=59.5GiB (63.9GB), run=300019-300019msec Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: David Sterba <dsterba@suse.com>
2022-09-19 14:06:40 +00:00
if (em_end > end) {
if (!split) {
split = split2;
split2 = NULL;
if (!split)
goto remove_em;
}
btrfs: fix incorrect splitting in btrfs_drop_extent_map_range In production we were seeing a variety of WARN_ON()'s in the extent_map code, specifically in btrfs_drop_extent_map_range() when we have to call add_extent_mapping() for our second split. Consider the following extent map layout PINNED [0 16K) [32K, 48K) and then we call btrfs_drop_extent_map_range for [0, 36K), with skip_pinned == true. The initial loop will have start = 0 end = 36K len = 36K we will find the [0, 16k) extent, but since we are pinned we will skip it, which has this code start = em_end; if (end != (u64)-1) len = start + len - em_end; em_end here is 16K, so now the values are start = 16K len = 16K + 36K - 16K = 36K len should instead be 20K. This is a problem when we find the next extent at [32K, 48K), we need to split this extent to leave [36K, 48k), however the code for the split looks like this split->start = start + len; split->len = em_end - (start + len); In this case we have em_end = 48K split->start = 16K + 36K // this should be 16K + 20K split->len = 48K - (16K + 36K) // this overflows as 16K + 36K is 52K and now we have an invalid extent_map in the tree that potentially overlaps other entries in the extent map. Even in the non-overlapping case we will have split->start set improperly, which will cause problems with any block related calculations. We don't actually need len in this loop, we can simply use end as our end point, and only adjust start up when we find a pinned extent we need to skip. Adjust the logic to do this, which keeps us from inserting an invalid extent map. We only skip_pinned in the relocation case, so this is relatively rare, except in the case where you are running relocation a lot, which can happen with auto relocation on. Fixes: 55ef68990029 ("Btrfs: Fix btrfs_drop_extent_cache for skip pinned case") CC: stable@vger.kernel.org # 4.14+ Reviewed-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Josef Bacik <josef@toxicpanda.com> Signed-off-by: David Sterba <dsterba@suse.com>
2023-08-17 20:57:30 +00:00
split->start = end;
split->len = em_end - end;
split->block_start = em->block_start;
split->flags = flags;
split->compress_type = em->compress_type;
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->orig_start = em->orig_start;
} else {
const u64 diff = start + len - em->start;
split->block_len = split->len;
split->block_start += diff;
split->orig_start = em->orig_start;
}
} else {
split->ram_bytes = split->len;
split->orig_start = split->start;
split->block_len = 0;
split->orig_block_len = 0;
}
if (extent_map_in_tree(em)) {
replace_extent_mapping(em_tree, em, split,
modified);
} else {
int ret;
ret = add_extent_mapping(em_tree, split,
modified);
/* Logic error, shouldn't happen. */
ASSERT(ret == 0);
if (WARN_ON(ret != 0) && modified)
btrfs_set_inode_full_sync(inode);
}
free_extent_map(split);
split = NULL;
}
btrfs: drop extent map range more efficiently Currently when dropping extent maps for a file range, through btrfs_drop_extent_map_range(), we do the following non-optimal things: 1) We lookup for extent maps one by one, always starting the search from the root of the extent map tree. This is not efficient if we have multiple extent maps in the range; 2) We check on every iteration if we have the 'split' and 'split2' spare extent maps in case we need to split an extent map that intersects our range but also crosses its boundaries (to the left, to the right or both cases). If our target range is for example: [2M, 8M) And we have 3 extents maps in the range: [1M, 3M) [3M, 6M) [6M, 10M[ The on the first iteration we allocate two extent maps for 'split' and 'split2', and use the 'split' to split the first extent map, so after the split we set 'split' to 'split2' and then set 'split2' to NULL. On the second iteration, we don't need to split the second extent map, but because 'split2' is now NULL, we allocate a new extent map for 'split2'. On the third iteration we need to split the third extent map, so we use the extent map pointed by 'split'. So we ended up allocating 3 extent maps for splitting, but all we needed was 2 extent maps. We never need to allocate more than 2, because extent maps that need to be split are always the first one and the last one in the target range. Improve on this by: 1) Using rb_next() to move on to the next extent map. This results in iterating over less nodes of the tree and it does not require comparing the ranges of nodes to our start/end offset; 2) Allocate the 2 extent maps for splitting before entering the loop and never allocate more than 2. In practice it's very rare to have the combination of both extent map allocations fail, since we have a dedicated slab for extent maps, and also have the need to split two extent maps. This patch is part of a patchset comprised of the following patches: btrfs: fix missed extent on fsync after dropping extent maps btrfs: move btrfs_drop_extent_cache() to extent_map.c btrfs: use extent_map_end() at btrfs_drop_extent_map_range() btrfs: use cond_resched_rwlock_write() during inode eviction btrfs: move open coded extent map tree deletion out of inode eviction btrfs: add helper to replace extent map range with a new extent map btrfs: remove the refcount warning/check at free_extent_map() btrfs: remove unnecessary extent map initializations btrfs: assert tree is locked when clearing extent map from logging btrfs: remove unnecessary NULL pointer checks when searching extent maps btrfs: remove unnecessary next extent map search btrfs: avoid pointless extent map tree search when flushing delalloc btrfs: drop extent map range more efficiently And the following fio test was done before and after applying the whole patchset, on a non-debug kernel (Debian's default kernel config) on a 12 cores Intel box with 64G of ram: $ cat test.sh #!/bin/bash DEV=/dev/nvme0n1 MNT=/mnt/nvme0n1 MOUNT_OPTIONS="-o ssd" MKFS_OPTIONS="-R free-space-tree -O no-holes" cat <<EOF > /tmp/fio-job.ini [writers] rw=randwrite fsync=8 fallocate=none group_reporting=1 direct=0 bssplit=4k/20:8k/20:16k/20:32k/10:64k/10:128k/5:256k/5:512k/5:1m/5 ioengine=psync filesize=2G runtime=300 time_based directory=$MNT numjobs=8 thread EOF echo performance | \ tee /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor echo echo "Using config:" echo cat /tmp/fio-job.ini echo umount $MNT &> /dev/null mkfs.btrfs -f $MKFS_OPTIONS $DEV mount $MOUNT_OPTIONS $DEV $MNT fio /tmp/fio-job.ini umount $MNT Result before applying the patchset: WRITE: bw=197MiB/s (206MB/s), 197MiB/s-197MiB/s (206MB/s-206MB/s), io=57.7GiB (61.9GB), run=300188-300188msec Result after applying the patchset: WRITE: bw=203MiB/s (213MB/s), 203MiB/s-203MiB/s (213MB/s-213MB/s), io=59.5GiB (63.9GB), run=300019-300019msec Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: David Sterba <dsterba@suse.com>
2022-09-19 14:06:40 +00:00
remove_em:
if (extent_map_in_tree(em)) {
/*
* If the extent map is still in the tree it means that
* either of the following is true:
*
* 1) It fits entirely in our range (doesn't end beyond
* it or starts before it);
*
* 2) It starts before our range and/or ends after our
* range, and we were not able to allocate the extent
* maps for split operations, @split and @split2.
*
* If we are at case 2) then we just remove the entire
* extent map - this is fine since if anyone needs it to
* access the subranges outside our range, will just
* load it again from the subvolume tree's file extent
* item. However if the extent map was in the list of
* modified extents, then we must mark the inode for a
* full fsync, otherwise a fast fsync will miss this
* extent if it's new and needs to be logged.
*/
btrfs: drop extent map range more efficiently Currently when dropping extent maps for a file range, through btrfs_drop_extent_map_range(), we do the following non-optimal things: 1) We lookup for extent maps one by one, always starting the search from the root of the extent map tree. This is not efficient if we have multiple extent maps in the range; 2) We check on every iteration if we have the 'split' and 'split2' spare extent maps in case we need to split an extent map that intersects our range but also crosses its boundaries (to the left, to the right or both cases). If our target range is for example: [2M, 8M) And we have 3 extents maps in the range: [1M, 3M) [3M, 6M) [6M, 10M[ The on the first iteration we allocate two extent maps for 'split' and 'split2', and use the 'split' to split the first extent map, so after the split we set 'split' to 'split2' and then set 'split2' to NULL. On the second iteration, we don't need to split the second extent map, but because 'split2' is now NULL, we allocate a new extent map for 'split2'. On the third iteration we need to split the third extent map, so we use the extent map pointed by 'split'. So we ended up allocating 3 extent maps for splitting, but all we needed was 2 extent maps. We never need to allocate more than 2, because extent maps that need to be split are always the first one and the last one in the target range. Improve on this by: 1) Using rb_next() to move on to the next extent map. This results in iterating over less nodes of the tree and it does not require comparing the ranges of nodes to our start/end offset; 2) Allocate the 2 extent maps for splitting before entering the loop and never allocate more than 2. In practice it's very rare to have the combination of both extent map allocations fail, since we have a dedicated slab for extent maps, and also have the need to split two extent maps. This patch is part of a patchset comprised of the following patches: btrfs: fix missed extent on fsync after dropping extent maps btrfs: move btrfs_drop_extent_cache() to extent_map.c btrfs: use extent_map_end() at btrfs_drop_extent_map_range() btrfs: use cond_resched_rwlock_write() during inode eviction btrfs: move open coded extent map tree deletion out of inode eviction btrfs: add helper to replace extent map range with a new extent map btrfs: remove the refcount warning/check at free_extent_map() btrfs: remove unnecessary extent map initializations btrfs: assert tree is locked when clearing extent map from logging btrfs: remove unnecessary NULL pointer checks when searching extent maps btrfs: remove unnecessary next extent map search btrfs: avoid pointless extent map tree search when flushing delalloc btrfs: drop extent map range more efficiently And the following fio test was done before and after applying the whole patchset, on a non-debug kernel (Debian's default kernel config) on a 12 cores Intel box with 64G of ram: $ cat test.sh #!/bin/bash DEV=/dev/nvme0n1 MNT=/mnt/nvme0n1 MOUNT_OPTIONS="-o ssd" MKFS_OPTIONS="-R free-space-tree -O no-holes" cat <<EOF > /tmp/fio-job.ini [writers] rw=randwrite fsync=8 fallocate=none group_reporting=1 direct=0 bssplit=4k/20:8k/20:16k/20:32k/10:64k/10:128k/5:256k/5:512k/5:1m/5 ioengine=psync filesize=2G runtime=300 time_based directory=$MNT numjobs=8 thread EOF echo performance | \ tee /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor echo echo "Using config:" echo cat /tmp/fio-job.ini echo umount $MNT &> /dev/null mkfs.btrfs -f $MKFS_OPTIONS $DEV mount $MOUNT_OPTIONS $DEV $MNT fio /tmp/fio-job.ini umount $MNT Result before applying the patchset: WRITE: bw=197MiB/s (206MB/s), 197MiB/s-197MiB/s (206MB/s-206MB/s), io=57.7GiB (61.9GB), run=300188-300188msec Result after applying the patchset: WRITE: bw=203MiB/s (213MB/s), 203MiB/s-203MiB/s (213MB/s-213MB/s), io=59.5GiB (63.9GB), run=300019-300019msec Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: David Sterba <dsterba@suse.com>
2022-09-19 14:06:40 +00:00
if ((em->start < start || em_end > end) && modified) {
ASSERT(!split);
btrfs_set_inode_full_sync(inode);
}
remove_extent_mapping(em_tree, em);
}
btrfs: drop extent map range more efficiently Currently when dropping extent maps for a file range, through btrfs_drop_extent_map_range(), we do the following non-optimal things: 1) We lookup for extent maps one by one, always starting the search from the root of the extent map tree. This is not efficient if we have multiple extent maps in the range; 2) We check on every iteration if we have the 'split' and 'split2' spare extent maps in case we need to split an extent map that intersects our range but also crosses its boundaries (to the left, to the right or both cases). If our target range is for example: [2M, 8M) And we have 3 extents maps in the range: [1M, 3M) [3M, 6M) [6M, 10M[ The on the first iteration we allocate two extent maps for 'split' and 'split2', and use the 'split' to split the first extent map, so after the split we set 'split' to 'split2' and then set 'split2' to NULL. On the second iteration, we don't need to split the second extent map, but because 'split2' is now NULL, we allocate a new extent map for 'split2'. On the third iteration we need to split the third extent map, so we use the extent map pointed by 'split'. So we ended up allocating 3 extent maps for splitting, but all we needed was 2 extent maps. We never need to allocate more than 2, because extent maps that need to be split are always the first one and the last one in the target range. Improve on this by: 1) Using rb_next() to move on to the next extent map. This results in iterating over less nodes of the tree and it does not require comparing the ranges of nodes to our start/end offset; 2) Allocate the 2 extent maps for splitting before entering the loop and never allocate more than 2. In practice it's very rare to have the combination of both extent map allocations fail, since we have a dedicated slab for extent maps, and also have the need to split two extent maps. This patch is part of a patchset comprised of the following patches: btrfs: fix missed extent on fsync after dropping extent maps btrfs: move btrfs_drop_extent_cache() to extent_map.c btrfs: use extent_map_end() at btrfs_drop_extent_map_range() btrfs: use cond_resched_rwlock_write() during inode eviction btrfs: move open coded extent map tree deletion out of inode eviction btrfs: add helper to replace extent map range with a new extent map btrfs: remove the refcount warning/check at free_extent_map() btrfs: remove unnecessary extent map initializations btrfs: assert tree is locked when clearing extent map from logging btrfs: remove unnecessary NULL pointer checks when searching extent maps btrfs: remove unnecessary next extent map search btrfs: avoid pointless extent map tree search when flushing delalloc btrfs: drop extent map range more efficiently And the following fio test was done before and after applying the whole patchset, on a non-debug kernel (Debian's default kernel config) on a 12 cores Intel box with 64G of ram: $ cat test.sh #!/bin/bash DEV=/dev/nvme0n1 MNT=/mnt/nvme0n1 MOUNT_OPTIONS="-o ssd" MKFS_OPTIONS="-R free-space-tree -O no-holes" cat <<EOF > /tmp/fio-job.ini [writers] rw=randwrite fsync=8 fallocate=none group_reporting=1 direct=0 bssplit=4k/20:8k/20:16k/20:32k/10:64k/10:128k/5:256k/5:512k/5:1m/5 ioengine=psync filesize=2G runtime=300 time_based directory=$MNT numjobs=8 thread EOF echo performance | \ tee /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor echo echo "Using config:" echo cat /tmp/fio-job.ini echo umount $MNT &> /dev/null mkfs.btrfs -f $MKFS_OPTIONS $DEV mount $MOUNT_OPTIONS $DEV $MNT fio /tmp/fio-job.ini umount $MNT Result before applying the patchset: WRITE: bw=197MiB/s (206MB/s), 197MiB/s-197MiB/s (206MB/s-206MB/s), io=57.7GiB (61.9GB), run=300188-300188msec Result after applying the patchset: WRITE: bw=203MiB/s (213MB/s), 203MiB/s-203MiB/s (213MB/s-213MB/s), io=59.5GiB (63.9GB), run=300019-300019msec Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: David Sterba <dsterba@suse.com>
2022-09-19 14:06:40 +00:00
/*
* Once for the tree reference (we replaced or removed the
* extent map from the tree).
*/
free_extent_map(em);
btrfs: drop extent map range more efficiently Currently when dropping extent maps for a file range, through btrfs_drop_extent_map_range(), we do the following non-optimal things: 1) We lookup for extent maps one by one, always starting the search from the root of the extent map tree. This is not efficient if we have multiple extent maps in the range; 2) We check on every iteration if we have the 'split' and 'split2' spare extent maps in case we need to split an extent map that intersects our range but also crosses its boundaries (to the left, to the right or both cases). If our target range is for example: [2M, 8M) And we have 3 extents maps in the range: [1M, 3M) [3M, 6M) [6M, 10M[ The on the first iteration we allocate two extent maps for 'split' and 'split2', and use the 'split' to split the first extent map, so after the split we set 'split' to 'split2' and then set 'split2' to NULL. On the second iteration, we don't need to split the second extent map, but because 'split2' is now NULL, we allocate a new extent map for 'split2'. On the third iteration we need to split the third extent map, so we use the extent map pointed by 'split'. So we ended up allocating 3 extent maps for splitting, but all we needed was 2 extent maps. We never need to allocate more than 2, because extent maps that need to be split are always the first one and the last one in the target range. Improve on this by: 1) Using rb_next() to move on to the next extent map. This results in iterating over less nodes of the tree and it does not require comparing the ranges of nodes to our start/end offset; 2) Allocate the 2 extent maps for splitting before entering the loop and never allocate more than 2. In practice it's very rare to have the combination of both extent map allocations fail, since we have a dedicated slab for extent maps, and also have the need to split two extent maps. This patch is part of a patchset comprised of the following patches: btrfs: fix missed extent on fsync after dropping extent maps btrfs: move btrfs_drop_extent_cache() to extent_map.c btrfs: use extent_map_end() at btrfs_drop_extent_map_range() btrfs: use cond_resched_rwlock_write() during inode eviction btrfs: move open coded extent map tree deletion out of inode eviction btrfs: add helper to replace extent map range with a new extent map btrfs: remove the refcount warning/check at free_extent_map() btrfs: remove unnecessary extent map initializations btrfs: assert tree is locked when clearing extent map from logging btrfs: remove unnecessary NULL pointer checks when searching extent maps btrfs: remove unnecessary next extent map search btrfs: avoid pointless extent map tree search when flushing delalloc btrfs: drop extent map range more efficiently And the following fio test was done before and after applying the whole patchset, on a non-debug kernel (Debian's default kernel config) on a 12 cores Intel box with 64G of ram: $ cat test.sh #!/bin/bash DEV=/dev/nvme0n1 MNT=/mnt/nvme0n1 MOUNT_OPTIONS="-o ssd" MKFS_OPTIONS="-R free-space-tree -O no-holes" cat <<EOF > /tmp/fio-job.ini [writers] rw=randwrite fsync=8 fallocate=none group_reporting=1 direct=0 bssplit=4k/20:8k/20:16k/20:32k/10:64k/10:128k/5:256k/5:512k/5:1m/5 ioengine=psync filesize=2G runtime=300 time_based directory=$MNT numjobs=8 thread EOF echo performance | \ tee /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor echo echo "Using config:" echo cat /tmp/fio-job.ini echo umount $MNT &> /dev/null mkfs.btrfs -f $MKFS_OPTIONS $DEV mount $MOUNT_OPTIONS $DEV $MNT fio /tmp/fio-job.ini umount $MNT Result before applying the patchset: WRITE: bw=197MiB/s (206MB/s), 197MiB/s-197MiB/s (206MB/s-206MB/s), io=57.7GiB (61.9GB), run=300188-300188msec Result after applying the patchset: WRITE: bw=203MiB/s (213MB/s), 203MiB/s-203MiB/s (213MB/s-213MB/s), io=59.5GiB (63.9GB), run=300019-300019msec Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: David Sterba <dsterba@suse.com>
2022-09-19 14:06:40 +00:00
next:
/* Once for us (for our lookup reference). */
free_extent_map(em);
btrfs: drop extent map range more efficiently Currently when dropping extent maps for a file range, through btrfs_drop_extent_map_range(), we do the following non-optimal things: 1) We lookup for extent maps one by one, always starting the search from the root of the extent map tree. This is not efficient if we have multiple extent maps in the range; 2) We check on every iteration if we have the 'split' and 'split2' spare extent maps in case we need to split an extent map that intersects our range but also crosses its boundaries (to the left, to the right or both cases). If our target range is for example: [2M, 8M) And we have 3 extents maps in the range: [1M, 3M) [3M, 6M) [6M, 10M[ The on the first iteration we allocate two extent maps for 'split' and 'split2', and use the 'split' to split the first extent map, so after the split we set 'split' to 'split2' and then set 'split2' to NULL. On the second iteration, we don't need to split the second extent map, but because 'split2' is now NULL, we allocate a new extent map for 'split2'. On the third iteration we need to split the third extent map, so we use the extent map pointed by 'split'. So we ended up allocating 3 extent maps for splitting, but all we needed was 2 extent maps. We never need to allocate more than 2, because extent maps that need to be split are always the first one and the last one in the target range. Improve on this by: 1) Using rb_next() to move on to the next extent map. This results in iterating over less nodes of the tree and it does not require comparing the ranges of nodes to our start/end offset; 2) Allocate the 2 extent maps for splitting before entering the loop and never allocate more than 2. In practice it's very rare to have the combination of both extent map allocations fail, since we have a dedicated slab for extent maps, and also have the need to split two extent maps. This patch is part of a patchset comprised of the following patches: btrfs: fix missed extent on fsync after dropping extent maps btrfs: move btrfs_drop_extent_cache() to extent_map.c btrfs: use extent_map_end() at btrfs_drop_extent_map_range() btrfs: use cond_resched_rwlock_write() during inode eviction btrfs: move open coded extent map tree deletion out of inode eviction btrfs: add helper to replace extent map range with a new extent map btrfs: remove the refcount warning/check at free_extent_map() btrfs: remove unnecessary extent map initializations btrfs: assert tree is locked when clearing extent map from logging btrfs: remove unnecessary NULL pointer checks when searching extent maps btrfs: remove unnecessary next extent map search btrfs: avoid pointless extent map tree search when flushing delalloc btrfs: drop extent map range more efficiently And the following fio test was done before and after applying the whole patchset, on a non-debug kernel (Debian's default kernel config) on a 12 cores Intel box with 64G of ram: $ cat test.sh #!/bin/bash DEV=/dev/nvme0n1 MNT=/mnt/nvme0n1 MOUNT_OPTIONS="-o ssd" MKFS_OPTIONS="-R free-space-tree -O no-holes" cat <<EOF > /tmp/fio-job.ini [writers] rw=randwrite fsync=8 fallocate=none group_reporting=1 direct=0 bssplit=4k/20:8k/20:16k/20:32k/10:64k/10:128k/5:256k/5:512k/5:1m/5 ioengine=psync filesize=2G runtime=300 time_based directory=$MNT numjobs=8 thread EOF echo performance | \ tee /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor echo echo "Using config:" echo cat /tmp/fio-job.ini echo umount $MNT &> /dev/null mkfs.btrfs -f $MKFS_OPTIONS $DEV mount $MOUNT_OPTIONS $DEV $MNT fio /tmp/fio-job.ini umount $MNT Result before applying the patchset: WRITE: bw=197MiB/s (206MB/s), 197MiB/s-197MiB/s (206MB/s-206MB/s), io=57.7GiB (61.9GB), run=300188-300188msec Result after applying the patchset: WRITE: bw=203MiB/s (213MB/s), 203MiB/s-203MiB/s (213MB/s-213MB/s), io=59.5GiB (63.9GB), run=300019-300019msec Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: David Sterba <dsterba@suse.com>
2022-09-19 14:06:40 +00:00
em = next_em;
}
btrfs: drop extent map range more efficiently Currently when dropping extent maps for a file range, through btrfs_drop_extent_map_range(), we do the following non-optimal things: 1) We lookup for extent maps one by one, always starting the search from the root of the extent map tree. This is not efficient if we have multiple extent maps in the range; 2) We check on every iteration if we have the 'split' and 'split2' spare extent maps in case we need to split an extent map that intersects our range but also crosses its boundaries (to the left, to the right or both cases). If our target range is for example: [2M, 8M) And we have 3 extents maps in the range: [1M, 3M) [3M, 6M) [6M, 10M[ The on the first iteration we allocate two extent maps for 'split' and 'split2', and use the 'split' to split the first extent map, so after the split we set 'split' to 'split2' and then set 'split2' to NULL. On the second iteration, we don't need to split the second extent map, but because 'split2' is now NULL, we allocate a new extent map for 'split2'. On the third iteration we need to split the third extent map, so we use the extent map pointed by 'split'. So we ended up allocating 3 extent maps for splitting, but all we needed was 2 extent maps. We never need to allocate more than 2, because extent maps that need to be split are always the first one and the last one in the target range. Improve on this by: 1) Using rb_next() to move on to the next extent map. This results in iterating over less nodes of the tree and it does not require comparing the ranges of nodes to our start/end offset; 2) Allocate the 2 extent maps for splitting before entering the loop and never allocate more than 2. In practice it's very rare to have the combination of both extent map allocations fail, since we have a dedicated slab for extent maps, and also have the need to split two extent maps. This patch is part of a patchset comprised of the following patches: btrfs: fix missed extent on fsync after dropping extent maps btrfs: move btrfs_drop_extent_cache() to extent_map.c btrfs: use extent_map_end() at btrfs_drop_extent_map_range() btrfs: use cond_resched_rwlock_write() during inode eviction btrfs: move open coded extent map tree deletion out of inode eviction btrfs: add helper to replace extent map range with a new extent map btrfs: remove the refcount warning/check at free_extent_map() btrfs: remove unnecessary extent map initializations btrfs: assert tree is locked when clearing extent map from logging btrfs: remove unnecessary NULL pointer checks when searching extent maps btrfs: remove unnecessary next extent map search btrfs: avoid pointless extent map tree search when flushing delalloc btrfs: drop extent map range more efficiently And the following fio test was done before and after applying the whole patchset, on a non-debug kernel (Debian's default kernel config) on a 12 cores Intel box with 64G of ram: $ cat test.sh #!/bin/bash DEV=/dev/nvme0n1 MNT=/mnt/nvme0n1 MOUNT_OPTIONS="-o ssd" MKFS_OPTIONS="-R free-space-tree -O no-holes" cat <<EOF > /tmp/fio-job.ini [writers] rw=randwrite fsync=8 fallocate=none group_reporting=1 direct=0 bssplit=4k/20:8k/20:16k/20:32k/10:64k/10:128k/5:256k/5:512k/5:1m/5 ioengine=psync filesize=2G runtime=300 time_based directory=$MNT numjobs=8 thread EOF echo performance | \ tee /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor echo echo "Using config:" echo cat /tmp/fio-job.ini echo umount $MNT &> /dev/null mkfs.btrfs -f $MKFS_OPTIONS $DEV mount $MOUNT_OPTIONS $DEV $MNT fio /tmp/fio-job.ini umount $MNT Result before applying the patchset: WRITE: bw=197MiB/s (206MB/s), 197MiB/s-197MiB/s (206MB/s-206MB/s), io=57.7GiB (61.9GB), run=300188-300188msec Result after applying the patchset: WRITE: bw=203MiB/s (213MB/s), 203MiB/s-203MiB/s (213MB/s-213MB/s), io=59.5GiB (63.9GB), run=300019-300019msec Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: David Sterba <dsterba@suse.com>
2022-09-19 14:06:40 +00:00
write_unlock(&em_tree->lock);
free_extent_map(split);
free_extent_map(split2);
}
/*
* Replace a range in the inode's extent map tree with a new extent map.
*
* @inode: The target inode.
* @new_em: The new extent map to add to the inode's extent map tree.
* @modified: Indicate if the new extent map should be added to the list of
* modified extents (for fast fsync tracking).
*
* Drops all the extent maps in the inode's extent map tree that intersect the
* range of the new extent map and adds the new extent map to the tree.
* The caller should have locked an appropriate file range in the inode's io
* tree before calling this function.
*/
int btrfs_replace_extent_map_range(struct btrfs_inode *inode,
struct extent_map *new_em,
bool modified)
{
const u64 end = new_em->start + new_em->len - 1;
struct extent_map_tree *tree = &inode->extent_tree;
int ret;
ASSERT(!extent_map_in_tree(new_em));
/*
* The caller has locked an appropriate file range in the inode's io
* tree, but getting -EEXIST when adding the new extent map can still
* happen in case there are extents that partially cover the range, and
* this is due to two tasks operating on different parts of the extent.
* See commit 18e83ac75bfe67 ("Btrfs: fix unexpected EEXIST from
* btrfs_get_extent") for an example and details.
*/
do {
btrfs_drop_extent_map_range(inode, new_em->start, end, false);
write_lock(&tree->lock);
ret = add_extent_mapping(tree, new_em, modified);
write_unlock(&tree->lock);
} while (ret == -EEXIST);
return ret;
}
/*
* Split off the first pre bytes from the extent_map at [start, start + len],
* and set the block_start for it to new_logical.
*
* This function is used when an ordered_extent needs to be split.
*/
int split_extent_map(struct btrfs_inode *inode, u64 start, u64 len, u64 pre,
u64 new_logical)
{
struct extent_map_tree *em_tree = &inode->extent_tree;
struct extent_map *em;
struct extent_map *split_pre = NULL;
struct extent_map *split_mid = NULL;
int ret = 0;
unsigned long flags;
ASSERT(pre != 0);
ASSERT(pre < len);
split_pre = alloc_extent_map();
if (!split_pre)
return -ENOMEM;
split_mid = alloc_extent_map();
if (!split_mid) {
ret = -ENOMEM;
goto out_free_pre;
}
lock_extent(&inode->io_tree, start, start + len - 1, NULL);
write_lock(&em_tree->lock);
em = lookup_extent_mapping(em_tree, start, len);
if (!em) {
ret = -EIO;
goto out_unlock;
}
ASSERT(em->len == len);
ASSERT(!test_bit(EXTENT_FLAG_COMPRESSED, &em->flags));
ASSERT(em->block_start < EXTENT_MAP_LAST_BYTE);
ASSERT(test_bit(EXTENT_FLAG_PINNED, &em->flags));
ASSERT(!test_bit(EXTENT_FLAG_LOGGING, &em->flags));
ASSERT(!list_empty(&em->list));
flags = em->flags;
clear_bit(EXTENT_FLAG_PINNED, &em->flags);
/* First, replace the em with a new extent_map starting from * em->start */
split_pre->start = em->start;
split_pre->len = pre;
split_pre->orig_start = split_pre->start;
split_pre->block_start = new_logical;
split_pre->block_len = split_pre->len;
split_pre->orig_block_len = split_pre->block_len;
split_pre->ram_bytes = split_pre->len;
split_pre->flags = flags;
split_pre->compress_type = em->compress_type;
split_pre->generation = em->generation;
replace_extent_mapping(em_tree, em, split_pre, 1);
/*
* Now we only have an extent_map at:
* [em->start, em->start + pre]
*/
/* Insert the middle extent_map. */
split_mid->start = em->start + pre;
split_mid->len = em->len - pre;
split_mid->orig_start = split_mid->start;
split_mid->block_start = em->block_start + pre;
split_mid->block_len = split_mid->len;
split_mid->orig_block_len = split_mid->block_len;
split_mid->ram_bytes = split_mid->len;
split_mid->flags = flags;
split_mid->compress_type = em->compress_type;
split_mid->generation = em->generation;
add_extent_mapping(em_tree, split_mid, 1);
/* Once for us */
free_extent_map(em);
/* Once for the tree */
free_extent_map(em);
out_unlock:
write_unlock(&em_tree->lock);
unlock_extent(&inode->io_tree, start, start + len - 1, NULL);
free_extent_map(split_mid);
out_free_pre:
free_extent_map(split_pre);
return ret;
}