The inode structure layout is largely random, and some of the vfs paths really do care. The path lookup in particular is already quite D$ intensive, and profiles show that accessing the 'inode->i_op->xyz' fields is quite costly. We already optimized the dcache to not unnecessarily load the d_op structure for members that are often NULL using the DCACHE_OP_xyz bits in dentry->d_flags, and this does something very similar for the inode ops that are used during pathname lookup. It also re-orders the fields so that the fields accessed by 'stat' are together at the beginning of the inode structure, and roughly in the order accessed. The effect of this seems to be in the 1-2% range for an empty kernel "make -j" run (which is fairly kernel-intensive, mostly in filename lookup), so it's visible. The numbers are fairly noisy, though, and likely depend a lot on exact microarchitecture. So there's more tuning to be done. Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
inode_lru_list_del() is expensive because of per superblock lru locking, while some inodes are not in lru list. Adding a check in iput_final() can speedup pipe/sockets workloads on SMP. Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Some inodes (pipes, sockets, ...) are not hashed, no need to take contended inode_hash_lock at dismantle time. nice speedup on SMP machines on socket intensive workloads. Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Some inodes (pipes, sockets, ...) are not in bdi writeback list. evict() can avoid calling inode_wb_list_del() and its expensive spinlock by checking inode i_wb_list being empty or not. At this point, no other cpu/user can concurrently manipulate this inode i_wb_list Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs-2.6: merge fchmod() and fchmodat() guts, kill ancient broken kludge xfs: fix misspelled S_IS...() xfs: get rid of open-coded S_ISREG(), etc. vfs: document locking requirements for d_move, __d_move and d_materialise_unique omfs: fix (mode & S_IFDIR) abuse btrfs: S_ISREG(mode) is not mode & S_IFREG... ima: fmode_t misspelled as mode_t... pci-label.c: size_t misspelled as mode_t jffs2: S_ISLNK(mode & S_IFMT) is pointless snd_msnd ->mode is fmode_t, not mode_t v9fs_iop_get_acl: get rid of unused variable vfs: dont chain pipe/anon/socket on superblock s_inodes list Documentation: Exporting: update description of d_splice_alias fs: add missing unlock in default_llseek()
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/wfg/writeback: (27 commits) mm: properly reflect task dirty limits in dirty_exceeded logic writeback: don't busy retry writeback on new/freeing inodes writeback: scale IO chunk size up to half device bandwidth writeback: trace global_dirty_state writeback: introduce max-pause and pass-good dirty limits writeback: introduce smoothed global dirty limit writeback: consolidate variable names in balance_dirty_pages() writeback: show bdi write bandwidth in debugfs writeback: bdi write bandwidth estimation writeback: account per-bdi accumulated written pages writeback: make writeback_control.nr_to_write straight writeback: skip tmpfs early in balance_dirty_pages_ratelimited_nr() writeback: trace event writeback_queue_io writeback: trace event writeback_single_inode writeback: remove .nonblocking and .encountered_congestion writeback: remove writeback_control.more_io writeback: skip balance_dirty_pages() for in-memory fs writeback: add bdi_dirty_limit() kernel-doc writeback: avoid extra sync work at enqueue time writeback: elevate queue_io() into wb_writeback() ... Fix up trivial conflicts in fs/fs-writeback.c and mm/filemap.c
Workloads using pipes and sockets hit inode_sb_list_lock contention. superblock s_inodes list is needed for quota, dirty, pagecache and fsnotify management. pipe/anon/socket fs are clearly not candidates for these. Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
i_alloc_sem is a rather special rw_semaphore. It's the last one that may be released by a non-owner, and it's write side is always mirrored by real exclusion. It's intended use it to wait for all pending direct I/O requests to finish before starting a truncate. Replace it with a hand-grown construct: - exclusion for truncates is already guaranteed by i_mutex, so it can simply fall way - the reader side is replaced by an i_dio_count member in struct inode that counts the number of pending direct I/O requests. Truncate can't proceed as long as it's non-zero - when i_dio_count reaches non-zero we wake up a pending truncate using wake_up_bit on a new bit in i_flags - new references to i_dio_count can't appear while we are waiting for it to read zero because the direct I/O count always needs i_mutex (or an equivalent like XFS's i_iolock) for starting a new operation. This scheme is much simpler, and saves the space of a spinlock_t and a struct list_head in struct inode (typically 160 bits on a non-debug 64-bit system). Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Now that we have per-sb shrinkers with a lifecycle that is a subset of the superblock lifecycle and can reliably detect a filesystem being unmounted, there is not longer any race condition for the iprune_sem to protect against. Hence we can remove it. Signed-off-by: Dave Chinner <dchinner@redhat.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
With context based shrinkers, we can implement a per-superblock shrinker that shrinks the caches attached to the superblock. We currently have global shrinkers for the inode and dentry caches that split up into per-superblock operations via a coarse proportioning method that does not batch very well. The global shrinkers also have a dependency - dentries pin inodes - so we have to be very careful about how we register the global shrinkers so that the implicit call order is always correct. With a per-sb shrinker callout, we can encode this dependency directly into the per-sb shrinker, hence avoiding the need for strictly ordering shrinker registrations. We also have no need for any proportioning code for the shrinker subsystem already provides this functionality across all shrinkers. Allowing the shrinker to operate on a single superblock at a time means that we do less superblock list traversals and locking and reclaim should batch more effectively. This should result in less CPU overhead for reclaim and potentially faster reclaim of items from each filesystem. Signed-off-by: Dave Chinner <dchinner@redhat.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>