- 11 Jul, 2024 40 commits
-
-
Qu Wenruo authored
The extent_map::block_len is either extent_map::len (non-compressed extent) or extent_map::disk_num_bytes (compressed extent). Since we already have sanity checks to do the cross-checks between the new and old members, we can drop the old extent_map::block_len now. For most call sites, they can manually select extent_map::len or extent_map::disk_num_bytes, since most if not all of them have checked if the extent is compressed. Reviewed-by:
Johannes Thumshirn <johannes.thumshirn@wdc.com> Reviewed-by:
Filipe Manana <fdmanana@suse.com> Signed-off-by:
Qu Wenruo <wqu@suse.com> Reviewed-by:
David Sterba <dsterba@suse.com> Signed-off-by:
-
Qu Wenruo authored
Since we have extent_map::offset, the old extent_map::orig_start is just extent_map::start - extent_map::offset for non-hole/inline extents. And since the new extent_map::offset is already verified by validate_extent_map() while the old orig_start is not, let's just remove the old member from all call sites. Reviewed-by:
-
Qu Wenruo authored
Since extent_map structure has the all the needed members to represent a file extent directly, we can apply all the file extent sanity checks to an extent map. The new sanity checks will cross check both the old members (block_start/block_len/orig_start) and the new members (disk_bytenr/disk_num_bytes/offset). There is a special case for offset/orig_start/start cross check, we only do such sanity check for compressed extent, as only compressed read/encoded write really utilize orig_start. This can be proved by the cleanup patch of orig_start. The checks happens at the following times: - add_extent_mapping() This is for newly added extent map - replace_extent_mapping() This is for btrfs_drop_extent_map_range() and split_extent_map() - try_merge_map() For a lot of call sites we have to properly populate all the members to pass the sanity check, meanwhile the following code needs extra modification: - setup_file_extents() from inode-tests The file extents layout of setup_file_extents() is already too invalid that tree-checker would reject most of them in real world. However there is just a special unaligned regular extent which has mismatched disk_num_bytes (4096) and ram_bytes (4096 - 1). So instead of dropping the whole test case, here we just unify disk_num_bytes and ram_bytes to 4096 - 1. - test_case_7() from extent-map-tests An extent is inserted with 16K length, but on-disk extent size is only 4K. This means it must be a compressed extent, so set the compressed flag for it. Reviewed-by:
-
Qu Wenruo authored
Introduce two new members for extent_map: - disk_bytenr - offset Both are matching the members with the same name inside btrfs_file_extent_items. For now this patch only touches those members when: - Reading btrfs_file_extent_items from disk - Inserting new holes - Merging two extent maps With the new disk_bytenr and disk_num_bytes, doing merging would be a little more complex, as we have 3 different cases: * Both extent maps are referring to the same data extents |<----- data extent A ----->| |<- em 1 ->|<- em 2 ->| * Both extent maps are referring to different data extents |<-- data extent A -->|<-- data extent B -->| |<- em 1 ->|<- em 2 ->| * One of the extent maps is referring to a merged and larger data extent that covers both extent maps This is not really valid case other than some selftests. So this test case would be removed. A new helper merge_ondisk_extents() is introduced to handle the above valid cases. To properly assign values for those new members, a new btrfs_file_extent parameter is introduced to all the involved call sites. - For NOCOW writes the btrfs_file_extent would be exposed from can_nocow_file_extent(). - For other writes, the members can be easily calculated As most of them have 0 offset and utilizing the whole on-disk data extent. The exception is encoded write, but thankfully that interface provided offset directly and all other needed info. For now, both the old members (block_start/block_len/orig_start) are co-existing with the new members (disk_bytenr/offset), meanwhile all the critical code is still using the old members only. The cleanup will happen later after all the old and new members are properly validated. There would be some re-ordering for the assignment of the extent_map members, now we follow the new ordering: - start and len Or file_pos and num_bytes for other structures. - disk_bytenr and disk_num_bytes - offset and ram_bytes - compression So expect some seemingly unrelated line movement. Reviewed-by: -
Qu Wenruo authored
Currently function can_nocow_extent() only returns members needed for extent_map. However since we will soon change the extent_map structure to be more like btrfs_file_extent_item, we want to expose the expected file extent caused by the NOCOW write for future usage. This introduces a new structure, btrfs_file_extent, to be a more memory access friendly representation of btrfs_file_extent_item. And use that structure to expose the expected file extent caused by the NOCOW write. For now there is no user of the new structure yet. Reviewed-by:
-
Qu Wenruo authored
This would make it very obvious that the member just matches btrfs_file_extent_item::disk_num_bytes. Reviewed-by:
-
Filipe Manana authored
Currently the core of the fiemap code lives in extent_io.c, which does not make any sense because it's not related to extent IO at all (and it was not as well before the big rewrite of fiemap I did some time ago). The entry point for fiemap, btrfs_fiemap(), lives in inode.c since it's an inode operation. Since there's a significant amount of fiemap code, move all of it into a dedicated file, including its entry point inode.c:btrfs_fiemap(). Reviewed-by:
-
Filipe Manana authored
Now that there is a helper to commit the current transaction and we are using it, there's no need for the label and goto statements at ensure_commit_roots_uptodate(). So replace them with direct return statements that call btrfs_commit_current_transaction(). Reviewed-by:
Josef Bacik <josef@toxicpanda.com> Reviewed-by:
-
Filipe Manana authored
We have several places that attach to the current transaction with btrfs_attach_transaction_barrier() and then commit the transaction if there is one. Add a helper and use it to deduplicate this pattern. Reviewed-by:
-
Filipe Manana authored
At finish_extent_writes_for_zoned() we use btrfs_join_transaction() to catch any running transaction and then commit it. This will however create a new and empty transaction in case there's no running transaction anymore (got committed by the transaction kthread or other task for example) or there's a running transaction finishing its commit and with a state >= TRANS_STATE_UNBLOCKED. In the former case we don't need to do anything while in the second case we just need to wait for the transaction to complete its commit. So improve this by using btrfs_attach_transaction_barrier() instead, which does not create a new transaction if there's none running, and if there's a current transaction that is committing, it will wait for it to fully commit and not create a new transaction. This helps avoiding creating and committing empty transactions, saving IO, time and unnecessary rotation of the backup roots in the super block. Reviewed-by:
-
Filipe Manana authored
At ensure_commit_roots_uptodate() we use btrfs_join_transaction() to catch any running transaction and then commit it. This will however create a new and empty transaction in case there's no running transaction anymore (got committed by the transaction kthread or other task for example) or there's a running transaction finishing its commit and with a state >= TRANS_STATE_UNBLOCKED. In the former case we don't need to do anything while in the second case we just need to wait for the transaction to complete its commit. So improve this by using btrfs_attach_transaction_barrier() instead, which does not create a new transaction if there's none running, and if there's a current transaction that is committing, it will wait for it to fully commit and not create a new transaction. This helps avoiding creating and committing empty transactions, saving IO, time and unnecessary rotation of the backup roots in the super block. Reviewed-by:
-
Filipe Manana authored
Before starting a send operation we have to make sure that every root has its commit root matching the regular root, to that send doesn't find stale inodes in the commit root (inodes that were deleted in the regular root) and fails the inode lookups with -ESTALE. Currently we keep looking for roots used by the send operation and as soon as we find one we commit the current transaction (or a new one since btrfs_join_transaction() creates one if there isn't any running or the running one is in a state >= TRANS_STATE_UNBLOCKED). It's pointless to keep looking until we don't find any, because after the first transaction commit all the other roots are updated too, as they were already tagged in the fs_info->fs_roots_radix radix tree when they were modified in order to have a commit root different from the regular root. Currently we are also always passing the main send root into btrfs_join_transaction(), which despite not having any functional issue, it is not optimal because in case the root wasn't modified we end up adding it to fs_info->fs_roots_radix and then update its root item in the root tree when committing the transaction, causing unnecessary work. So simplify and make this more efficient by removing the looping and by passing the first root we found that is modified as the argument to btrfs_join_transaction(). Reviewed-by:
-
Filipe Manana authored
At btrfs_commit_super(), called in a few contexts such as when unmounting a filesystem, we use btrfs_join_transaction() to catch any running transaction and then commit it. This will however create a new and empty transaction in case there's no running transaction or there's a running transaction with a state >= TRANS_STATE_UNBLOCKED. As we just want to be sure that any existing transaction is fully committed, we can use btrfs_attach_transaction_barrier() instead of btrfs_join_transaction(), therefore avoiding the creation and commit of empty transactions, which only waste IO and causes rotation of the precious backup roots. Example where we create and commit a pointless empty transaction: $ mkfs.btrfs -f /dev/sdj $ btrfs inspect-internal dump-super /dev/sdj | grep -e '^generation' generation 6 $ mount /dev/sdj /mnt/sdj $ touch /mnt/sdj/foo # Commit the currently open transaction. Just 'sync' or wait ~30 # seconds for the transaction kthread to commit it. $ sync $ btrfs inspect-internal dump-super /dev/sdj | grep -e '^generation' generation 7 $ umount /mnt/sdj $ btrfs inspect-internal dump-super /dev/sdj | grep -e '^generation' generation 8 The transaction with id 8 was pointless, an empty transaction that did not achieve anything. Reviewed-by:
-
Filipe Manana authored
When flushing reservations we are using btrfs_join_transaction() to get a handle for the current transaction and then commit it to try to release space. However btrfs_join_transaction() has some undesirable consequences: 1) If there's no running transaction, it will create one, and we will commit it right after. This is unnecessary because it will not release any space, and it will result in unnecessary IO and rotation of backup roots in the superblock; 2) If there's a current transaction and that transaction is committing (its state is >= TRANS_STATE_COMMIT_DOING), it will wait for that transaction to almost finish its commit (for its state to be >= TRANS_STATE_UNBLOCKED) and then start and return a new transaction. We will then commit that new transaction, which is pointless because all we wanted was to wait for the current (previous) transaction to fully finish its commit (state == TRANS_STATE_COMPLETED), and by starting and committing a new transaction we are wasting IO too and causing unnecessary rotation of backup roots in the superblock. So improve this by using btrfs_attach_transaction_barrier() instead, which does not create a new transaction if there's none running, and if there's a current transaction that is committing, it will wait for it to fully commit and not create a new transaction. Reviewed-by:
-
David Sterba authored
The range is specified only in two ways, we can simplify the case for the whole filesystem range as a NULL block group parameter. Signed-off-by: -
Qu Wenruo authored
Currently if we fully clean a subvolume (not only delete its directory, but fully clean all it's related data and root item), the associated qgroup would not be removed. We have "btrfs qgroup clear-stale" to handle such 0 level qgroups. Change the behavior to automatically removie the qgroup of a fully cleaned subvolume when possible: - Full qgroup but still consistent We can and should remove the qgroup. The qgroup numbers should be 0, without any rsv. - Full qgroup but inconsistent Can happen with drop_subtree_threshold feature (skip accounting and mark qgroup inconsistent). We can and should remove the qgroup. Higher level qgroup numbers will be incorrect, but since qgroup is already inconsistent, it should not be a problem. - Squota mode This is the special case, we can only drop the qgroup if its numbers are all 0. This would be handled by can_delete_qgroup(), so we only need to check the return value and ignore the -EBUSY error. Link: https://bugzilla.suse.com/show_bug.cgi?id=1222847Reviewed-by:
Boris Burkov <boris@bur.io> Signed-off-by:
-
Qu Wenruo authored
[BUG] Currently if one is utilizing "qgroups/drop_subtree_threshold" sysfs, and a snapshot with level higher than that value is dropped, we will not be able to delete the qgroup until next qgroup rescan: uuid=ffffffff-eeee-dddd-cccc-000000000000 wipefs -fa $dev mkfs.btrfs -f $dev -O quota -s 4k -n 4k -U $uuid mount $dev $mnt btrfs subvolume create $mnt/subv1/ for (( i = 0; i < 1024; i++ )); do xfs_io -f -c "pwrite 0 2k" $mnt/subv1/file_$i > /dev/null done sync btrfs subvolume snapshot $mnt/subv1 $mnt/snapshot btrfs quota enable $mnt btrfs quota rescan -w $mnt sync echo 1 > /sys/fs/btrfs/$uuid/qgroups/drop_subtree_threshold btrfs subvolume delete $mnt/snapshot btrfs subvolume sync $mnt btrfs qgroup show -prce --sync $mnt btrfs qgroup destroy 0/257 $mnt umount $mnt The final qgroup removal would fail with the following error: ERROR: unable to destroy quota group: Device or resource busy [CAUSE] The above script would generate a subvolume of level 2, then snapshot it, enable qgroup, set the drop_subtree_threshold, then drop the snapshot. Since the subvolume drop would meet the threshold, qgroup would be marked inconsistent and skip accounting to avoid hanging the system at transaction commit. But currently we do not allow a qgroup with any rfer/excl numbers to be dropped, and this is not really compatible with the new drop_subtree_threshold behavior. [FIX] Only require the strict zero rfer/excl/rfer_cmpr/excl_cmpr for squota mode. This is due to the fact that squota can never go inconsistent, and it can have dropped subvolume but with non-zero qgroup numbers for future accounting. For full qgroup mode, we only check if there is a subvolume for it. Reviewed-by:
-
David Sterba authored
Reported by 'gcc -Wcast-qual', the argument from which write_extent_buffer() reads data to write to the eb should be const. In addition the const needs to be also added to __write_extent_buffer() local buffers. All callers of write_eb_member() can now be updated to use const for the input buffer structure or type. Reviewed-by:
Anand Jain <anand.jain@oracle.com> Reviewed-by:
Naohiro Aota <naohiro.aota@wdc.com> Signed-off-by:
-
David Sterba authored
This was reported by 'gcc -Wcast-qual', the get_unaligned_le8() simply returns the argument and there's no reason to drop the cast. Reviewed-by:
-
David Sterba authored
This was added in c61a16a7 ("Btrfs: fix the confusion between delalloc bytes and metadata bytes") and removed in 03fe78cc ("btrfs: use delalloc_bytes to determine flush amount for shrink_delalloc") where the calculation was reworked to use a non-constant numbers. This was found by 'make W=2'. Reviewed-by:
-
David Sterba authored
We've started to use for-loop local variables and in a few places this shadows a function variable. Convert a few cases reported by 'make W=2'. If applicable also change the style to post-increment, that's the preferred one. Reviewed-by:
-
David Sterba authored
Fix variable names in two macros where there's a local function variable of the same name. In subpage_calc_start_bit() it's in several callers, in btrfs_abort_transaction() it's only in replace_file_extents(). Found by 'make W=2'. Reviewed-by:
-
David Sterba authored
When running 'make W=2' there are a few reports where a variable of the same name is declared in a nested block. In all the cases we can use the one declared in the parent block, no problematic cases were found. Reviewed-by:
-
Filipe Manana authored
Instead of using a VFS inode local pointer and then doing many BTRFS_I() calls inside btrfs_sync_file(), use a btrfs_inode pointer instead. This makes everything a bit easier to read and less confusing, allowing to make some statements shorter. Reviewed-by:
-
Filipe Manana authored
Instead of passing a (VFS) inode pointer argument, pass a btrfs_inode instead, as this is generally what we do for internal APIs, making it more consistent with most of the code base. This will later allow to help to remove a lot of BTRFS_I() calls in btrfs_sync_file(). Reviewed-by:
-
Filipe Manana authored
Instead of passing a (VFS) inode pointer argument, pass a btrfs_inode instead, as this is generally what we do for internal APIs, making it more consistent with most of the code base. This will later allow to help to remove a lot of BTRFS_I() calls in btrfs_sync_file(). Reviewed-by:
-
Filipe Manana authored
Instead of using a inode pointer, use a btrfs_inode pointer in the log context structure, as this is generally what we need and allows for some internal APIs to take a btrfs_inode instead, making them more consistent with most of the code base. This will later allow to help to remove a lot of BTRFS_I() calls in btrfs_sync_file(). Reviewed-by:
-
Filipe Manana authored
Currently btrfs_finish_ordered_extent() returns a boolean indicating if the ordered extent was added to the work queue for completion, but none of its callers cares about it, so make it return void. Reviewed-by:
-
Anand Jain authored
The function btrfs_block_group_root() is declared in disk-io.c; however, all its callers are in block-group.c. Move it to the latter file and declare it static. Reviewed-by:
-
Anand Jain authored
Drop the single-use variable bytenr_orig and instead use btrfs_sb_offset() in the function argument passing. Fix a stale comment about not automatically fixing a bad primary superblock from the backup mirror copies. Also, move the comment closer to where the primary superblock read occurs. Signed-off-by:
-
Filipe Manana authored
We are currently using a cached rb_root (struct rb_root_cached) for the rb root of struct extent_map_tree. This doesn't offer much of an advantage here because: 1) It's only advantage over the regular rb_root is that it caches a pointer to the left most node (first node), so a call to rb_first_cached() doesn't have to chase pointers until it reaches the left most node; 2) We only have two scenarios that access left most node with rb_first_cached(): When dropping all extent maps from an inode, during inode eviction; When iterating over extent maps during the extent map shrinker; 3) In both cases we keep removing extent maps, which causes deletion of the left most node so rb_erase_cached() has to call rb_next() to find out what's the next left most node and assign it to struct rb_root_cached::rb_leftmost; 4) We can do that ourselves in those two uses cases and stop using a rb_root_cached rb tree and use instead a regular rb_root rb tree. This reduces the size of struct extent_map_tree by 8 bytes and, since this structure is embedded in struct btrfs_inode, it also reduces the size of that structure by 8 bytes. So on a 64 bits platform the size of btrfs_inode is reduced from 1032 bytes down to 1024 bytes. This means we will be able to have 4 inodes per 4K page instead of 3. Signed-off-by: -
Filipe Manana authored
Currently we name the rb_root member of struct extent_map_tree as 'map', which is odd and confusing. Since it's a root node, rename it to 'root'. Signed-off-by:
-
Filipe Manana authored
On 64 bits platforms we don't really need to have a dedicated member (the objectid field) for the inode's number since we store in the VFS inode's i_ino member, which is an unsigned long and this type is 64 bits wide on 64 bits platforms. We only need that field in case we are on a 32 bits platform because the unsigned long type is 32 bits wide on such platforms See commit 33345d01 ("Btrfs: Always use 64bit inode number") regarding this 64/32 bits detail. The objectid field of struct btrfs_inode is also used to store the ID of a root for directories that are stubs for unreferenced roots. In such cases the inode is a directory and has the BTRFS_INODE_ROOT_STUB runtime flag set. So in order to reduce the size of btrfs_inode structure on 64 bits platforms we can remove the objectid member and use the VFS inode's i_ino member instead whenever we need to get the inode number. In case the inode is a root stub (BTRFS_INODE_ROOT_STUB set) we can use the member last_reflink_trans to store the ID of the unreferenced root, since such inode is a directory and reflinks can't be done against directories. So remove the objectid fields for 64 bits platforms and alias the last_reflink_trans field with a name of ref_root_id in a union. On a release kernel config, this reduces the size of struct btrfs_inode from 1040 bytes down to 1032 bytes. Signed-off-by:
-
Filipe Manana authored
Currently struct btrfs_inode has a key member, named "location", that is either: 1) The key of the inode's item. In this case the objectid is the number of the inode; 2) A key stored in a dir entry with a type of BTRFS_ROOT_ITEM_KEY, for the case where we have a root that is a snapshot of a subvolume that points to other subvolumes. In this case the objectid is the ID of a subvolume inside the snapshotted parent subvolume. The key is only used to lookup the inode item for the first case, while for the second it's never used since it corresponds to directory stubs created with new_simple_dir() and which are marked as dummy, so there's no actual inode item to ever update. In the second case we only check the key type at btrfs_ino() for 32 bits platforms and its objectid is only needed for unlink. Instead of using a key we can do fine with just the objectid, since we can generate the key whenever we need it having only the objectid, as in all use cases the type is always BTRFS_INODE_ITEM_KEY and the offset is always 0. So use only an objectid instead of a full key. This reduces the size of struct btrfs_inode from 1048 bytes down to 1040 bytes on a release kernel. Signed-off-by:
-
Filipe Manana authored
When not using the NO_HOLES feature we always allocate an io tree for an inode's file_extent_tree. This is wasteful because that io tree is only used for regular files, so we allocate more memory than needed for inodes that represent directories or symlinks for example, or for inodes that correspond to free space inodes. So improve on this by allocating the io tree only for inodes of regular files that are not free space inodes. Reviewed-by:
-
Filipe Manana authored
The index_cnt field of struct btrfs_inode is used only for two purposes: 1) To store the index for the next entry added to a directory; 2) For the data relocation inode to track the logical start address of the block group currently being relocated. For the relocation case we use index_cnt because it's not used for anything else in the relocation use case - we could have used other fields that are not used by relocation such as defrag_bytes, last_unlink_trans or last_reflink_trans for example (among others). Since the csum_bytes field is not used for directories, do the following changes: 1) Put index_cnt and csum_bytes in a union, and index_cnt is only initialized when the inode is a directory. The csum_bytes is only accessed in IO paths for regular files, so we're fine here; 2) Use the defrag_bytes field for relocation, since the data relocation inode is never used for defrag purposes. And to make the naming better, alias it to reloc_block_group_start by using a union. This reduces the size of struct btrfs_inode by 8 bytes in a release kernel, from 1056 bytes down to 1048 bytes. Reviewed-by:
-
Filipe Manana authored
Currently we use the spinlock inode_lock from struct btrfs_root to serialize access to two different data structures: 1) The delayed inodes xarray (struct btrfs_root::delayed_nodes); 2) The inodes xarray (struct btrfs_root::inodes). Instead of using our own lock, we can use the spinlock that is part of the xarray implementation, by using the xa_lock() and xa_unlock() APIs and using the xarray APIs with the double underscore prefix that don't take the xarray locks and assume the caller is using xa_lock() and xa_unlock(). So remove the spinlock inode_lock from struct btrfs_root and use the corresponding xarray locks. This brings 2 benefits: 1) We reduce the size of struct btrfs_root, from 1336 bytes down to 1328 bytes on a 64 bits release kernel config; 2) We reduce lock contention by not using anymore the same lock for changing two different and unrelated xarrays. Reviewed-by:
-
Filipe Manana authored
Make btrfs_iget_path() simpler and easier to read by avoiding nesting of if-then-else statements and having an error label to do all the error handling instead of repeating it a couple times. Reviewed-by:
-
Filipe Manana authored
When creating a new inode, at btrfs_create_new_inode(), one of the very last steps is to add the inode to the root's inodes xarray. This often requires allocating memory which may fail (even though xarrays have a dedicated kmem_cache which make it less likely to fail), and at that point we are forced to abort the current transaction (as some, but not all, of the inode metadata was added to its subvolume btree). To avoid a transaction abort, preallocate memory for the xarray early at btrfs_create_new_inode(), so that if we fail we don't need to abort the transaction and the insertion into the xarray is guaranteed to succeed. Reviewed-by:
-
Filipe Manana authored
Currently we use a red black tree (rb-tree) to track the currently open inodes of a root (in struct btrfs_root::inode_tree). This however is not very efficient when the number of inodes is large since rb-trees are binary trees. For example for 100K open inodes, the tree has a depth of 17. Besides that, inserting into the tree requires navigating through it and pulling useless cache lines in the process since the red black tree nodes are embedded within the btrfs inode - on the other hand, by being embedded, it requires no extra memory allocations. We can improve this by using an xarray instead, which is efficient when indices are densely clustered (such as inode numbers), is more cache friendly and behaves like a resizable array, with a much better search and insertion complexity than a red black tree. This only has one small disadvantage which is that insertion will sometimes require allocating memory for the xarray - which may fail (not that often since it uses a kmem_cache) - but on the other hand we can reduce the btrfs inode structure size by 24 bytes (from 1080 down to 1056 bytes) after removing the embedded red black tree node, which after the next patches will allow to reduce the size of the structure to 1024 bytes, meaning we will be able to store 4 inodes per 4K page instead of 3 inodes. This change does a straightforward change to use an xarray, and results in a transaction abort if we can't allocate memory for the xarray when creating an inode - but the next patch changes things so that we don't need to abort. Running the following fs_mark test showed some improvements: $ cat test.sh #!/bin/bash DEV=/dev/nullb0 MNT=/mnt/nullb0 MOUNT_OPTIONS="-o ssd" FILES=100000 THREADS=$(nproc --all) echo "performance" | \ tee /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor mkfs.btrfs -f $DEV mount $MOUNT_OPTIONS $DEV $MNT OPTS="-S 0 -L 5 -n $FILES -s 0 -t $THREADS -k" for ((i = 1; i <= $THREADS; i++)); do OPTS="$OPTS -d $MNT/d$i" done fs_mark $OPTS umount $MNT Before this patch: FSUse% Count Size Files/sec App Overhead 10 1200000 0 92081.6 12505547 16 2400000 0 138222.6 13067072 23 3600000 0 148833.1 13290336 43 4800000 0 97864.7 13931248 53 6000000 0 85597.3 14384313 After this patch: FSUse% Count Size Files/sec App Overhead 10 1200000 0 93225.1 12571078 16 2400000 0 146720.3 12805007 23 3600000 0 160626.4 13073835 46 4800000 0 116286.2 13802927 53 6000000 0 90087.9 14754892 The test was run with a release kernel config (Debian's default config). Also capturing the insertion times into the rb tree and into the xarray, that is measuring the duration of the old function inode_tree_add() and the duration of the new btrfs_add_inode_to_root() function, gave the following results (in nanoseconds): Before this patch, inode_tree_add() execution times: Count: 5000000 Range: 0.000 - 5536887.000; Mean: 775.674; Median: 729.000; Stddev: 4820.961 Percentiles: 90th: 1015.000; 95th: 1139.000; 99th: 1397.000 0.000 - 7.816: 40 | 7.816 - 37.858: 209 | 37.858 - 170.278: 6059 | 170.278 - 753.961: 2754890 ##################################################### 753.961 - 3326.728: 2232312 ########################################### 3326.728 - 14667.018: 4366 | 14667.018 - 64652.943: 852 | 64652.943 - 284981.761: 550 | 284981.761 - 1256150.914: 221 | 1256150.914 - 5536887.000: 7 | After this patch, btrfs_add_inode_to_root() execution times: Count: 5000000 Range: 0.000 - 2900652.000; Mean: 272.148; Median: 241.000; Stddev: 2873.369 Percentiles: 90th: 342.000; 95th: 432.000; 99th: 572.000 0.000 - 7.264: 104 | 7.264 - 33.145: 352 | 33.145 - 140.081: 109606 # 140.081 - 581.930: 4840090 ##################################################### 581.930 - 2407.590: 43532 | 2407.590 - 9950.979: 2245 | 9950.979 - 41119.278: 514 | 41119.278 - 169902.616: 155 | 169902.616 - 702018.539: 47 | 702018.539 - 2900652.000: 9 | Average, percentiles, standard deviation, etc, are all much better. Reviewed-by:
-
