]> git.proxmox.com Git - mirror_ubuntu-bionic-kernel.git/commit - lib/radix-tree.c
radix-tree: improve multiorder iterators
authorMatthew Wilcox <mawilcox@microsoft.com>
Wed, 14 Dec 2016 23:08:49 +0000 (15:08 -0800)
committerLinus Torvalds <torvalds@linux-foundation.org>
Thu, 15 Dec 2016 00:04:10 +0000 (16:04 -0800)
commit148deab223b23734069abcacb5c7118b0e7deadc
tree56aae9c91802e9262312f7fcbb3571eb9c8ec0e9
parentb35df27a39f40e39fabf1b1e9569c7b24e1add6a
radix-tree: improve multiorder iterators

This fixes several interlinked problems with the iterators in the
presence of multiorder entries.

1. radix_tree_iter_next() would only advance by one slot, which would
   result in the iterators returning the same entry more than once if
   there were sibling entries.

2. radix_tree_next_slot() could return an internal pointer instead of
   a user pointer if a tagged multiorder entry was immediately followed by
   an entry of lower order.

3. radix_tree_next_slot() expanded to a lot more code than it used to
   when multiorder support was compiled in.  And I wasn't comfortable with
   entry_to_node() being in a header file.

Fixing radix_tree_iter_next() for the presence of sibling entries
necessarily involves examining the contents of the radix tree, so we now
need to pass 'slot' to radix_tree_iter_next(), and we need to change the
calling convention so it is called *before* dropping the lock which
protects the tree.  Also rename it to radix_tree_iter_resume(), as some
people thought it was necessary to call radix_tree_iter_next() each time
around the loop.

radix_tree_next_slot() becomes closer to how it looked before multiorder
support was introduced.  It only checks to see if the next entry in the
chunk is a sibling entry or a pointer to a node; this should be rare
enough that handling this case out of line is not a performance impact
(and such impact is amortised by the fact that the entry we just
processed was a multiorder entry).  Also, radix_tree_next_slot() used to
force a new chunk lookup for untagged entries, which is more expensive
than the out of line sibling entry skipping.

Link: http://lkml.kernel.org/r/1480369871-5271-55-git-send-email-mawilcox@linuxonhyperv.com
Signed-off-by: Matthew Wilcox <mawilcox@microsoft.com>
Tested-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Konstantin Khlebnikov <koct9i@gmail.com>
Cc: Ross Zwisler <ross.zwisler@linux.intel.com>
Cc: Matthew Wilcox <mawilcox@microsoft.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
fs/btrfs/tests/btrfs-tests.c
include/linux/radix-tree.h
lib/radix-tree.c
mm/khugepaged.c
mm/shmem.c
tools/testing/radix-tree/iteration_check.c
tools/testing/radix-tree/multiorder.c
tools/testing/radix-tree/regression3.c
tools/testing/radix-tree/test.h