]> git.ipfire.org Git - thirdparty/kernel/stable.git/commit
libfs: Convert simple directory offsets to use a Maple Tree
authorChuck Lever <chuck.lever@oracle.com>
Sat, 17 Feb 2024 20:24:16 +0000 (15:24 -0500)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Thu, 30 May 2024 07:48:59 +0000 (09:48 +0200)
commit3d8babce24a7bad53dacae3182b071700cc28940
tree9933171b0613c01a14591cc36d4552d13e1d32c7
parentc1ab3bf4969f041009a078a1d15e9c2cf96318a2
libfs: Convert simple directory offsets to use a Maple Tree

[ Upstream commit 0e4a862174f2a8d1653a8a9cf0815020e1d3af24 ]

Test robot reports:
> kernel test robot noticed a -19.0% regression of aim9.disk_src.ops_per_sec on:
>
> commit: a2e459555c5f9da3e619b7e47a63f98574dc75f1 ("shmem: stable directory offsets")
> https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git master

Feng Tang further clarifies that:
> ... the new simple_offset_add()
> called by shmem_mknod() brings extra cost related with slab,
> specifically the 'radix_tree_node', which cause the regression.

Willy's analysis is that, over time, the test workload causes
xa_alloc_cyclic() to fragment the underlying SLAB cache.

This patch replaces the offset_ctx's xarray with a Maple Tree in the
hope that Maple Tree's dense node mode will handle this scenario
more scalably.

In addition, we can widen the simple directory offset maximum to
signed long (as loff_t is also signed).

Suggested-by: Matthew Wilcox <willy@infradead.org>
Reported-by: kernel test robot <oliver.sang@intel.com>
Closes: https://lore.kernel.org/oe-lkp/202309081306.3ecb3734-oliver.sang@intel.com
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Link: https://lore.kernel.org/r/170820145616.6328.12620992971699079156.stgit@91.116.238.104.host.secureserver.net
Reviewed-by: Jan Kara <jack@suse.cz>
Signed-off-by: Christian Brauner <brauner@kernel.org>
Stable-dep-of: 23cdd0eed3f1 ("libfs: Fix simple_offset_rename_exchange()")
Signed-off-by: Sasha Levin <sashal@kernel.org>
fs/libfs.c
include/linux/fs.h