linux-bcache.git - Unnamed repository; edit this file 'description' to name the repository.

Age	Commit message (Collapse)	Author
2017-01-18	bcache: rw aux search tree improvement	Kent Overstreet

2017-01-18	bcache: bkey_unpack_pos()	Kent Overstreet
	slightly more efficient when we don't have a dynamically generated bkey_unpack_key()
2017-01-18	bcache: add end_offset to struct bset_tree	Kent Overstreet
	duplicates bset->u64s - for better data locality
2017-01-18	bcache: drop a comparison in bset_search_tree()	Kent Overstreet

2017-01-18	bcache: squish struct btree some more	Kent Overstreet
	finally under half a kilobyte
2017-01-18	bcache: kill struct btree_keys	Kent Overstreet

2017-01-18	bcache: larger bfloats for base of aux search tree	Kent Overstreet

2017-01-18	bcache: kill bset_tree->prev_offset	Kent Overstreet

2017-01-18	bcache: squish struct btree some more	Kent Overstreet

2017-01-18	bcache: use a single buffer for btree node aux trees	Kent Overstreet

2017-01-18	bcache: uninline bkey_cmp_left_packed()	Kent Overstreet
	this is generally a slowpath
2017-01-18	bcache: more aux search tree microoptimizations	Kent Overstreet

2017-01-18	Revert "bcache: Optimization for aux search tree slowpath"	Kent Overstreet
	This reverts commit 27b3474087af56511602a52050a3f2ec0fc69325. Turned out to not really help, gonna go with a different approach.
2017-01-18	bcache: prefetch btree root in bch_btree_iter_init()	Kent Overstreet

2017-01-18	bcache: inline some fastpaths	Kent Overstreet

2017-01-18	bcache: btree_node_iter_init() optimizations	Kent Overstreet

2017-01-18	bcache: round packed bkey format fields up to nearest byte when possible	Kent Overstreet
	wasn't useful with the old bkey_unpack_key(), but with the new dynamic codegen bkey_unpack_key() we can take advantage of this to avoid shifts and generate smaller code
2017-01-18	bcache: jit bkey_unpack_key()	Kent Overstreet

2017-01-18	bcache: packed bkeys refactoring	Kent Overstreet

2017-01-18	bcache: asm bkey_cmp_packed()	Kent Overstreet
	shouldn't need asm for this, but gcc is generating retarded code... later, we should precompute bkey_format_key_bits() and stash that somewhere
2017-01-18	bcache: drop trace_bkey_pack_pos_fail	Kent Overstreet
	this isn't actually a slowpath - move the tracepoint to the actual slowpath
2017-01-18	bcache: fix another copygc deadlock	Kent Overstreet
	since btree node reserves got reworked, btree updates can end up blocking on copygc - which means we have to run index updates for copygc out of a separate workqueue. also fix another deadlock related to rwsems being unfair.
2017-01-18	bcache: change copygc assertion to just print a warning	Kent Overstreet
	also, fix the assertion...
2017-01-18	bcache: fix btree node write tracepoint	Kent Overstreet
	b->nsets was changed, and it wasn't updated
2017-01-18	bcachefs: fix a regression from extent whiteouts optimizations	Kent Overstreet
	checking bkey_whiteout(&s->insert->k->k) in extent_insert_committed() was incorrect, because if we were already finished, the insert key's size would be 0 and 0 size extents are always KEY_TYPE_DELETED (i.e. whiteouts).
2017-01-18	bcache: allow __bch_write() to fragment io more than twice without slowpath loop	Kent Overstreet

2017-01-18	bcache: Whiteout optimizations: extents	Kent Overstreet

2017-01-18	bcache: extent insert path refactoring	Kent Overstreet

2017-01-18	bcache: re-enable btree_node_write() idx_to_write	Kent Overstreet

2017-01-18	bcache: btree_node_verify() improvements	Kent Overstreet

2017-01-18	bcache: Move unwritten whiteouts to a separate area	Kent Overstreet

2017-01-18	bcache: Optimize btree node writes	Kent Overstreet

2017-01-18	bcache: fix a race with journal_seq_blacklist flushing	Kent Overstreet
	the code was checking for missing journal entries in journal replay - but by that point, journal_seq_blacklist entries could already have been flushed and removed, causing us to spuriously think we were missing journal entries
2017-01-18	bcache: New sorting infrastructure	Kent Overstreet
	The upcoming whiteout optimizations have to do new kinds of sorting, add some new infrastructure to abstract out the comparison function Later, we should use this for the sorts we do when reading in a btree node (bch_key_sort_fix_overlapping and bch_extent_sort_fix_overlapping).
2017-01-18	bcache: Ensure we create all required journal seq blacklist entries	Kent Overstreet

2017-01-18	bcache: btree insert path refactoring	Kent Overstreet

2017-01-18	bcache: slim struct btree a bit more	Kent Overstreet

2017-01-18	bcache: kill b->c	Kent Overstreet

2017-01-18	bcache: btree write path improvements, kill struct bbio	Kent Overstreet

2017-01-18	bcache: More sort refactoring	Kent Overstreet
	reduce b->c usage
2017-01-18	bcache: Smarter journal flushing	Kent Overstreet

2017-01-18	bcache: add support for BSET_SEPARATE_WHITEOUTS()	Kent Overstreet

2017-01-18	bcache: Kill btree write work item, flush via journal reclaim	Kent Overstreet

2017-01-18	bcache: more aux search tree cleanups	Kent Overstreet

2017-01-18	bcache: make b->nsets normal	Kent Overstreet

2017-01-18	bcache: Clean up sorting, kill btree_keys_ops, etc.	Kent Overstreet

2017-01-18	bcache: more fighting with gcc inlining	Kent Overstreet
	gcc isn't inlining calls to constant fn pointers now...
2017-01-18	bcachefs: use BCH_HASH_SET_MUST_CREATE in bch_dirent_create()	Kent Overstreet
	Not a functional change currently - the VFS locks the directory during creates - but this means we don't rely on that and can maybe relax that locking later.
2017-01-18	bcache: Another journal assertion	Kent Overstreet

2017-01-18	bcache: Inline journal fastpaths	Kent Overstreet