[PATCH v3 00/25] Lockless update of reference count protected by spinlock

From: Waiman Long
Date: Wed Jul 03 2013 - 16:27:22 EST

v2->v3:
- Completely revamp the packaging by adding a new lockref data
structure that combines the spinlock with the reference
count. Helper functions are also added to manipulate the new data
structure. That results in modifying over 50 files, but the changes
were trivial in most of them.
- Change initial spinlock wait to use a timeout.
- Force 64-bit alignment of the spinlock & reference count structure.
- Add a new way to use the combo by using a new union and helper
functions.

v1->v2:
- Add one more layer of indirection to LOCK_WITH_REFCOUNT macro.
- Add __LINUX_SPINLOCK_REFCOUNT_H protection to spinlock_refcount.h.
- Add some generic get/put macros into spinlock_refcount.h.

This patchset supports a generic mechanism to atomically update
a reference count that is protected by a spinlock without actually
acquiring the lock itself. If the update doesn't succeeed, the caller
will have to acquire the lock and update the reference count in the
the old way. This will help in situation where there is a lot of
spinlock contention because of frequent reference count update.

The d_lock and d_count fields of the struct dentry in dcache.h was
modified to use the new lockref data structure and the associated
helper functions to access the spinlock and the reference count. This
change cause significant performance improvement in the short workload
of the AIM7 benchmark on a 8-socket x86-64 machine with 80 cores.

patch 1: Introduce the new lockref data structure
patch 2: Enable x86 architecture to use the feature
patch 3: Change the dentry structure to use lockref to improve
performance for high contention situation
patches 4-25: Use new helper functions to access the d_lock and
d_count fields of the original dentry structure.

Thank to Thomas Gleixner, Andi Kleen and Linus for their valuable
input in shaping this patchset.

Signed-off-by: Waiman Long <Waiman.Long@xxxxxx>

Waiman Long (25):
spinlock: A new lockref structure for lockless update of refcount
spinlock: Enable x86 architecture to do lockless refcount update
dcache: Enable lockless update of d_count in dentry structure
powerpc: Change how dentry's d_lock field is accessed
infiniband: Change how dentry's d_lock field is accessed
9p-fs: Change how dentry's d_lock field is accessed
afs-fs: Change how dentry's d_lock field is accessed
auto-fs: Change how dentry's d_lock and d_count fields are accessed
ceph-fs: Change how dentry's d_lock and d_count fields are accessed
cifs: Change how dentry's d_lock field is accessed
coda-fs: Change how dentry's d_lock and d_count fields are accessed
config-fs: Change how dentry's d_lock and d_count fields are accessed
vfs: Change how dentry's d_lock and d_count fields are accessed
ecrypt-fs: Change how dentry's d_count field is accessed
export-fs: Change how dentry's d_lock field is accessed
vfat: Change how dentry's d_lock field is accessed
file locking: Change how dentry's d_lock and d_count fields are
accessed
ncp-fs: Change how dentry's d_lock field is accessed
nfs: Change how dentry's d_lock and d_count fields are accessed
nilfs2: Change how dentry's d_count field is accessed
fsnotify: Change how dentry's d_lock field is accessed
ocfs2: Change how dentry's d_lock field is accessed
cgroup: Change how dentry's d_lock field is accessed
sunrpc: Change how dentry's d_lock field is accessed
selinux: Change how dentry's d_lock field is accessed

arch/powerpc/platforms/cell/spufs/inode.c | 6 +-
arch/x86/Kconfig | 3 +
arch/x86/include/asm/spinlock_refcount.h | 1 +
drivers/infiniband/hw/ipath/ipath_fs.c | 6 +-
drivers/infiniband/hw/qib/qib_fs.c | 6 +-
fs/9p/fid.c | 14 +-
fs/afs/dir.c | 4 +-
fs/autofs4/autofs_i.h | 24 +-
fs/autofs4/expire.c | 48 ++--
fs/autofs4/root.c | 14 +-
fs/ceph/caps.c | 8 +-
fs/ceph/debugfs.c | 8 +-
fs/ceph/dir.c | 34 ++--
fs/ceph/export.c | 4 +-
fs/ceph/inode.c | 24 +-
fs/ceph/mds_client.c | 22 +-
fs/cifs/dir.c | 10 +-
fs/coda/cache.c | 4 +-
fs/coda/dir.c | 4 +-
fs/configfs/configfs_internal.h | 4 +-
fs/configfs/dir.c | 2 +-
fs/configfs/inode.c | 6 +-
fs/dcache.c | 324 +++++++++++++++--------------
fs/dcookies.c | 8 +-
fs/ecryptfs/inode.c | 2 +-
fs/exportfs/expfs.c | 8 +-
fs/fat/namei_vfat.c | 4 +-
fs/fs-writeback.c | 4 +-
fs/libfs.c | 36 ++--
fs/locks.c | 2 +-
fs/namei.c | 28 ++--
fs/namespace.c | 10 +-
fs/ncpfs/dir.c | 6 +-
fs/ncpfs/ncplib_kernel.h | 8 +-
fs/nfs/dir.c | 14 +-
fs/nfs/getroot.c | 8 +-
fs/nfs/namespace.c | 16 +-
fs/nfs/unlink.c | 22 +-
fs/nilfs2/super.c | 2 +-
fs/notify/fsnotify.c | 8 +-
fs/notify/vfsmount_mark.c | 24 +-
fs/ocfs2/dcache.c | 6 +-
include/asm-generic/spinlock_refcount.h | 98 +++++++++
include/linux/dcache.h | 52 ++++--
include/linux/fs.h | 4 +-
include/linux/fsnotify_backend.h | 6 +-
include/linux/spinlock_refcount.h | 159 ++++++++++++++
kernel/Kconfig.locks | 5 +
kernel/cgroup.c | 10 +-
lib/Makefile | 2 +
lib/spinlock_refcount.c | 229 ++++++++++++++++++++
net/sunrpc/rpc_pipe.c | 6 +-
security/selinux/selinuxfs.c | 14 +-
53 files changed, 958 insertions(+), 423 deletions(-)
create mode 100644 arch/x86/include/asm/spinlock_refcount.h
create mode 100644 include/asm-generic/spinlock_refcount.h
create mode 100644 include/linux/spinlock_refcount.h
create mode 100644 lib/spinlock_refcount.c

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/