summaryrefslogtreecommitdiff
path: root/drivers
AgeCommit message (Collapse)Author
2016-11-02bcache: get rid of a memcmp() in bch_btree_node_get()Kent Overstreet
2016-11-02bcache: don't touch journal flags in fast pathsKent Overstreet
2016-11-02bcache: optimize bch_btree_node_iter_init()Kent Overstreet
Avoid the memmove in bch_btree_node_iter_push() by just doing a sort at the end.
2016-10-27bcache: delete some duplicated codeKent Overstreet
bch_btree_node_fill() does the unlock
2016-10-27bcache: rename some bset aux tree stuffKent Overstreet
2016-10-27bcachefs: be consistent with generic_write_sync()Kent Overstreet
2016-10-25bcache: fix handling of ptr_offset in allocation codeKent Overstreet
only affects replication
2016-10-24bcache: fix for btree code getting wrong journal sequence numberKent Overstreet
2016-10-21bcache: shrink struct btree a bitKent Overstreet
2016-10-20bcache: Make btree coalescing triggerable from sysfsKent Overstreet
2016-10-20bcache: improve btree node coalesce tracepointsKent Overstreet
2016-10-18bcache: Make debugfs interface a bit more informativeKent Overstreet
2016-10-14bcache: workaround for buggy FUAKent Overstreet
2016-10-14bcache: fix a shutdown deadlockKent Overstreet
2016-10-10bcache: fix a race in the journalling codeKent Overstreet
2016-10-10bcache: hack around a race with journal reclaimKent Overstreet
2016-10-10bcache: rip out a bunch more data move codeKent Overstreet
switch to a much much simpler mechanism for perserving io ordering/locality
2016-10-10bcache: rip out bch_queue_start()/stop()Kent Overstreet
this code wasn't even doing anything, since when shutting down tiering/copygc we were stopping the kthreads before stopping the queues...
2016-10-10bcache: rip out generational gcKent Overstreet
generational copygc has the side effect of potentially fragmenting data, we probably don't do it by default, if at all
2016-10-10bcache: move ratelimit accounting to more sensible locationKent Overstreet
2016-10-10bcache: create a sensible crc entry for ptrs lacking crcsKent Overstreet
2016-10-07bcache: More fsck improvementsKent Overstreet
2016-10-07bcache: copygc fixesKent Overstreet
Copygc now passes really nasty torture tests without deadlocking
2016-10-07bcache: another allocator fixupKent Overstreet
2016-10-07bcache: tweak a fault injection thingKent Overstreet
copygc doesn't retry stale reads, because if the read was stale that means the data was overwritten and doesn't have to be moved anymore - so injecting a fault there causes it to not move data it should have, and breaks tests.
2016-10-07bcache: Simpler keylistsKent Overstreet
With scan keylists gone, keylists don't have to hold more than a few keys anymore, and we can go back to a much simpler implementation. Yay for deleting code. This also means we can finally fix the definition of BKEY_EXTENT_VAL_U64s_MAX...
2016-10-07bcache: scan keylists get to die in a fireKent Overstreet
2016-10-07bcache: add a journal error check to __bch_btree_reserve_get()Kent Overstreet
2016-10-07bcache: better error messagesKent Overstreet
2016-10-07bcache: fix an awesome deadlockKent Overstreet
2016-10-07bcache: fix a divide by zeroKent Overstreet
2016-10-07bcache: fix tiering + compressionKent Overstreet
2016-10-07bcache: fix an assertion in bch_extent_invalid()Kent Overstreet
2016-10-07bcache: fix corner case in allocation code with replicationKent Overstreet
2016-10-07bcache: add mount option to fix errorsKent Overstreet
2016-10-07bcache: don't mark as clean when there were errorsKent Overstreet
2016-10-07bcache: Make allocation code more intelligentKent Overstreet
- rework bch_bucket_alloc_set so that it doesn't fail unless it can't allocate from any devices - split out the policy crap - "foreground writes try to allocate from tier 0 first" - into a separate wrapper function - make btree allocations also just prefer tier 0, instead of _only_ allocating from tier 0
2016-10-07bcache: Track inodes in a journal write with a bloom filterKent Overstreet
This lets us make fsync work when the inode's been evicted - fixes xfstests 106.
2016-10-07bcache: ensure that data writes aren't cached if journal write won't do flushKent Overstreet
2016-10-07bcache: improve alloc_debugKent Overstreet
2016-10-07bcache: more useful checksum errorsKent Overstreet
2016-10-07bcache: git rid of a branch in bkey_pack_key()Kent Overstreet
2016-10-07bcache: fix a lost wakeupKent Overstreet
the way bch_next_cache() was using buckets_free_cache() was racy
2016-10-07bcachefs: fs-gc, fsck improvementsKent Overstreet
Make some checks tighter
2016-10-07bcache: fix bch_btree_iter_advance_pos()Kent Overstreet
iter->k doesn't get updated when bch_btree_iter_peek() returns null
2016-10-07bcachefs: update i_nlink synchronouslyKent Overstreet
Fixes a bug uncovered by the previous patch
2016-10-07bcache: add flag to superblock indicating if we shutdown cleanlyKent Overstreet
2016-10-07bcache: fix ordering of index updates after unclean shutdownKent Overstreet
the journal sequence number blacklist machinery doesn't work if we don't actually hit all the btree nodes before starting journal replay...
2016-10-07bcache: fix an ordering issue in btree_interior_update codeKent Overstreet
Also a bunch of refactoring
2016-10-07bcache: add bch_journal_pin_add_if_older()Kent Overstreet
also uninline the journal pin stuff, there's enough callers (or will be soon) that it shouldn't be inlined.