fs: icache RCU free inodes
RCU free the struct inode. This will allow:
- Subsequent store-free path walking patch. The inode must be consulted for
permissions when walking, so an RCU inode reference is a must.
- sb_inode_list_lock to be moved inside i_lock because sb list walkers who want
to take i_lock no longer need to take sb_inode_list_lock to walk the list in
the first place. This will simplify and optimize locking.
- Could remove some nested trylock loops in dcache code
- Could potentially simplify things a bit in VM land. Do not need to take the
page lock to follow page->mapping.
The downsides of this is the performance cost of using RCU. In a simple
creat/unlink microbenchmark, performance drops by about 10% due to inability to
reuse cache-hot slab objects. As iterations increase and RCU freeing starts
kicking over, this increases to about 20%.
In cases where inode lifetimes are longer (ie. many inodes may be allocated
during the average life span of a single inode), a lot of this cache reuse is
not applicable, so the regression caused by this patch is smaller.
The cache-hot regression could largely be avoided by using SLAB_DESTROY_BY_RCU,
however this adds some complexity to list walking and store-free path walking,
so I prefer to implement this at a later date, if it is shown to be a win in
real situations. I haven't found a regression in any non-micro benchmark so I
doubt it will be a problem.
Signed-off-by: Nick Piggin <npiggin@kernel.dk>
diff --git a/Documentation/filesystems/porting b/Documentation/filesystems/porting
index 1eb76959..ccf0ce7 100644
--- a/Documentation/filesystems/porting
+++ b/Documentation/filesystems/porting
@@ -346,3 +346,17 @@
for details of what locks to replace dcache_lock with in order to protect
particular things. Most of the time, a filesystem only needs ->d_lock, which
protects *all* the dcache state of a given dentry.
+
+--
+[mandatory]
+
+ Filesystems must RCU-free their inodes, if they can have been accessed
+via rcu-walk path walk (basically, if the file can have had a path name in the
+vfs namespace).
+
+ i_dentry and i_rcu share storage in a union, and the vfs expects
+i_dentry to be reinitialized before it is freed, so an:
+
+ INIT_LIST_HEAD(&inode->i_dentry);
+
+must be done in the RCU callback.